Metrics details it remains unclear how social valence is constructed on the basis of social history and what mechanisms underlie the heightened valence versatility in dynamic relationships Here we demonstrate that the ventral (v)CA1 integrates serotonin (5-HT) inputs from the dorsal raphe and neurotensin inputs from the paraventricular nucleus of the thalamus (PVT) to determine positive or negative valence of conspecific representations during an appetitive social interaction 5-HT is released into the vCA1 and disinhibits pyramidal neurons through 5-HT1B receptors whereas neurotensin is released during an aversive social interaction and potentiates vCA1 neurons directly through NTR1s Optogenetic silencing of dorsal raphe 5-HT and PVT neurotensin inputs into the vCA1 impairs positive and negative social valence and excitation flexibly switches valence assignment These results show how aversive and rewarding social experiences are linked to conspecific identity through converging dorsal raphe 5-HT and PVT neurotensin signals in the vCA1 that instruct opposing valence and represent a synaptic switch for flexible social valence computation it is poorly understood how social history influences social valence and how valence information can be flexibly updated how does the brain integrate different and even conflicting information to calculate new behavioural decisions We reasoned that distinct neuromodulatory inputs from converging circuit motifs may represent a potential neural plasticity mechanism which would allow flexible behavioural switch based on social history we established new behavioural models to determine the causal relationship between neuromodulator release and social valence Source data thereby demonstrating positive social valence association with the conspecific based on the previously appetitive interaction Source data these results indicate that positive and negative social valence information converge in the vCA1 through DR and PVT inputs Source data Source data The data suggest that separate populations of vCA1 pyramidal neurons may be preferentially potentiated for distinct valence associations Source data By manipulating 5-HT and NT release in combination with receptor antagonists we were able to invert the polarity of social valence with the same social partner during repeated exposure these results also indicate that 5-HT and NT induce opposite valence signals that are integrated for social valence calculation as presence of both ablated positive and negative social valence assignment Our experiments show that 5-HT conveys positive and NT negative valence association with a conspecific which can be flexibly switched within minutes activation of 5-HT1BRs specifically restored positive social valence assignment as well as vCA1 excitability in an ASD mouse model haplo-sufficient for Shank3 Exon 4-22 This suggests that enhanced 5-HT1BR activity can compensate for the reduction in synaptic plasticity caused by mutation of Shank3 which might be a common mechanism applicable to many ASD-risk gene mutations affecting synaptic transmission we postulate that 5-HT and NT increase excitability of a subset of vCA1 neurons so that incoming identity information is preferentially mapped onto the potentiated population and thereby overlayed with valence Since manipulation of 5-HT and NT release into the vCA1 lead to distinct behavioural outcomes it is possible that information stored in the 5-HT and NT potentiated vCA1 neurons is differentially routed to promote appropriate action-selection This represents a basic mechanism defining how the brain processes and updates many characteristics of a given entity Further studies investigating the downstream target regions receiving inputs from those cells will thus be informative are multiplexed to promote more specific types of social representation This converging neuromodulatory motif has the potential to integrate complex information into a behaviourally relevant outcome and may represent a shared principle used by behaviours including social dominance we have identified physiological correlates of a neural competition mechanism for flexible valence association which is capable of integrating social history Our findings indicate that further simultaneous exploration of several neuromodulators in converging circuit motifs will not only provide critical insights into complex social behaviours but also reveal potential therapeutic targets such as 5-HT1BR and NTR1 which can be leveraged to improve social cognitive deficits in neuropsychiatric disorders including ASD and schizophrenia dCas9-KRAB (Rosa26-LSL-dCas9-KRAB (Jackson Laboratory 033066) mice were used as experimental subjects TRAP2;Ai14 mice were generated by crossing TRAP2 (STOCK Fostm2.1(icre/ERT2)Luo/J; Jackson Laboratory 30323) and Ai14 (B6.Cg-Gt(ROSA)26Sortm14(CAG-tdTomato)Hze/J; Jackson Laboratory CFW) as well as male and female C57BL/6J mice were used as partner mice mice were housed with 2–5 mice per cage and weaned at 21 days old All behavioural experiments were conducted with 7–14-week-old mice Animals were maintained on a 12-h light–dark cycle at roughly 21 °C with 50% humidity and food and water ad libitum Behavioural experiments were performed during the same circadian period (07:00–19:00) Experiments were conducted in accordance with the National Institutes of Health Guide for Care and approved by the Use of Laboratory Animals and the Icahn School of Medicine at Mount Sinai Institutional Animal Care and Use Committee Sample sizes were not predetermined by statistical analysis but based on previous experience with the variance of the assays The sequence of control and manipulated conditions were shuffled Mice (4–8 weeks old) were anaesthetized with intramuscular injections of a drug mixture containing ketamine (100 mg kg−1) and xylazine (5 mg kg−1) and heads were then positioned on a stereotaxic apparatus (David Kopf Instruments) for virus injections (0.2–0.5 μl) at a flow rate of 0.1–0.25 μl min−1 with a microinjection pump (Harvard Apparatus) The following bregma coordinates from dura were used to target the virus solution filled glass cannula: dCA2 (bilateral anteroposterior −1.6 mm; mediolateral ±1.6 mm; dorsoventral 1.7 mm) anteroposterior −3.16; mediolateral ±3.1; dorsoventral −4.55) anteroposterior −4.36; mediolateral 0; dorsoventral −3.1) anteroposterior −0.4; mediolateral 0; dorsoventral −3.8) anteroposterior −4.4; mediolateral 0; dorsoventral −4) and reuniens (bilateral anteroposterior −0.2; mediolateral ±0.85; dorsoventral −4.3) AAV-CAG-FLEx-G and rabies virus (Janelia Research Campus) were used for monosynaptic tracing Behavioural experiments with cell body manipulations were performed 3–4 weeks after virus injections whereas the viral particles were allowed to incubate 4–8 weeks for manipulation of axon terminals numerical aperture (NA) 0.5; RWD Life Science) for optogenetic light delivery were implanted in the vCA1 bilaterally at anteroposterior −3.16; mediolateral ±3.1; dorsoventral −4.4 A 26-gauge guide cannula (Plastics One) for drug micro-infusion was implanted at anteroposterior −3.16; mediolateral ±3.1 so that the 33-gauge infuser insert reached a depth of dorsoventral −4.5 Miniature screws (thread size 00–90 × 1/16; Antrin Miniature Specialties and McMaster-Carr) light-cured dental adhesive cement (C&B Metabond Parkell) and resin (Geristore A&B paste Lang Dental) were applied to secure implants to the skull A small percentage (roughly 5%) of mice were excluded from behavioural analysis based on either off-target transgene expression or inaccurate implant placement Sigma H6278-50MG) was administered by means of intraperitoneal (i.p.) injection at 50 mg kg−1 for 15 min after pairing with partner mice CNO (Tocris Biosciences 4936) was also delivered through i.p injections at 10 mg kg−1 for 30 min before behavioural experiments Drugs used for cannula microinjection experiments were infused in a total volume of 200–400 nl at a speed of 200 nl min−1 through an injector cannula connected to a micro-infusion pump (World Precision Instruments) The following drugs and concentrations were used: methiothepin mesylate salt (Millipore Sigma M149 22 pg) and CP93129 dihydrochloride (Tocris Biosciences 1032 injection infusers stayed for 2 min until removal and sociability and social memory assays were carried out 20 min later NAS-181 (10 mg kg−1) and SR48692 (3 mg kg−1) were delivered through i.p injections 30 min before fibre photometry experiments Three groups of TRAP2;Ai14 mice (6–8 weeks old) were habituated to i.p saline injections and empty cages for two consecutive days group 1 subjects were introduced to an age-matched same-sex conspecific in a new cage group 2 to a sex-matched aggressive male CD-1 or female CFW mouse in the aggressor’s home cage and group 3 to a potential mate (opposite sex) in either the female partner’s home cage or the female subject’s home cage subjects were administered 4-OHT (50 mg kg−1) and interaction was continued for 1 h before subjects returned to their home cages mice were either perfused with 10% neutral buffered formalin (NBF) for histological analysis or brains were removed to be fresh frozen for in situ hybridization experiments test mice were habituated for 2 days to i.p infuser inserts or the fibre optic patch cord as well as the three-chamber apparatus containing two empty inverted metal grid pencil cups (10 cm diameter) in the outer two chamber for 5 min The three-chamber apparatus (60 cm long by 23 cm wide by 26 cm high) was constructed of 0.3-cm-thick white opaque acrylic sheets with two outer chambers (23 cm long by 23 cm wide) and a centre chamber (15 cm long by 23 cm wide) divided by 15-cm-long barriers extending from opposite ends of the walls White opaque acrylic dividers were used to block entrance to the two outer chambers from the centre chamber Partner mice were also habituated under the cup for 5 min for 2 days subjects were placed into the centre chamber for 2 min before the barriers were lifted whereafter the subjects were able to freely explore all three chambers Four variations of this assay were performed to test social memory The location of partner mice and objects in either chamber was counterbalanced between trials in all tests in which subjects were separated from the contents of the two outer chambers the barriers were lifted and the subjects were allowed to freely explore the three chambers for 5 min in a second session the one of the outer two chambers contained a novel mouse and the other one a familiar mouse (same mouse from previous session) under a cup Retired breeder CD-1s and female CFW group-housed with same-sex conspecifics were used as non-aggressors Only mice that did not attack in a 5-min interaction qualified as non-aggressors a barrier was placed into their home cage so the timing and pacing of the interaction with the potential mate male partner could be controlled The cardboard barrier had an opening at the bottom to easily escape to the other side of the cage The interactions with the neutral partner were without lordosis posture and barrier in a new cage It is important to note that the term neutral is used to describe a more neutral or less positive stimulus compared to the potential mate but not an absolute neutral stimulus We used two variations of the RTPP test. The basic RTPP was performed as previously described31 in which light stimulation was paired with one of the two outer chambers while the subject was allowed to freely explore the three chambers for 15 min light stimulation was paired with the opposite outer chamber while the subject freely explored all three chambers again for 15 min the two outer chambers each contained a new mouse and light stimulation was paired with one of the two chambers while the subjects freely explored all three chambers for 10 min the barriers were lifted and the subjects were allowed to move freely between the three chambers where the outer chambers contained the two previously acquainted mice (now familiar mice) under a cup Recorded videos in the three-chamber apparatus was analysed using a video tracking system (BIOBSERVE which automatically tracked the location of the subject mouse The time spent in either of the two outer chambers were scored for the duration of the tests male and female subject mice were used in roughly equal numbers Test mice were excluded from the analysis (less than 2% of the total) if they were not attacked by the aggressor if the non-aggressor attacked more than three times in 5 min or if they spent the entirety of the assay in one chamber only Mice were habituated for 2 days to behavioural chambers with optic fibres connected to light-emitting diodes (LEDs) in off states A 545-nm Dual-LED (Prizmatix) was connected to the optical implants via a rotary joint and a fibre optic patch cord (Prizmatix) to photostimulate NpHR3.0 A 450-nm Dual-LED (Prizmatix) was used to photostimulate ChR2 The LEDs were adjusted to around 15 mW for axon terminal stimulation using a digital power meter console (ThorLabs) and the frequency was controlled by Pulser Plus (Prizmatix) A cycle of 8 s on and 2 s off was used to stimulate NpHR3.0 to avoid tissue overheating ChR2 was stimulated at 20 Hz with a 5-ms pulse duration Similar to previously described52 mice were tested for baseline preference for two differently textured clear acrylic floors by spending 15 min in a white opaque polycarbonate chamber (23 cm long by 23 cm wide by 26 cm high 0.3 cm thick) on both floors (23 cm long by 13 cm wide by 2.5 cm high The ‘grid’ floor was constructed of seven evenly spaced apart rails The hole floor contained 4 × 9 round holes 8 mm in diameter mice were either administered saline or CNO (10 mg kg−1) through i.p injections 30 min before conditioning to one of the floors for 15 min mice that received saline on the previous day animals were conditioned to the alternate flooring for 15 min test mice were placed into the chamber containing both types of flooring and time spent on either type was automatically scored with video tracking software (BIOBSERVE) Mice were placed into the centre of the elevated-plus maze (74 cm long by 74 cm wide by 94 cm high; Med Associates Inc.) with white floors (each arm 35 cm long by 6 cm wide with 19 cm high walls on closed arms) for 10 min and location of the mice were scored with video tracking software (BIOBSERVE) Male subjects were placed for 5 min into a female conspecific’s home cage with one female present The female was held by the experimenter for 3 s in a mating posture so that the male could sniff the female’s rear Male mounting attempts during the 5 min of free interaction were manually quantified a different test was used to assess female receptiveness to males Male partners were placed into the female subject’s home cage which was divided into two compartments by a barrier small enough for the female to easily escape Escape times and duration of the female in each compartment were manually scored fibre photometry raw signals were fitted with a mono- or bi-exponential decay function and the resulting fluorescence trace was z scored Corresponding videos were manually analysed by frame in MATLAB to identify the time of physical contact between test mouse and partner mouse and object Peristimulus time histograms were constructed by averaging 7 s of non-overlapping epochs from the z scored trace where a time of 0 represents the time of contact The maximal z score between 0 and 4 s was used as peak z scored fluorescence One T175 flask of HEK293T cells (American Type Culture Collection) per construct was transfected with 87.6 µg PEI (1 µg ml−1; 3:1 PEI: DNA; PolySciences 24765) along with 15.4 µg of pAdΔF6 7.3 µg of pAAV2/9n (pAdΔF6 and pAAV2/9n were gifts from J respectively) and 6.5 µg of pAAV mU6-sgRNA-CR EF1α-EGFP-W3-SV40 plasmids Viral supernatant was harvested 72 h posttransfection along with cells and mixed thoroughly with 0.1 volumes of chloroform NaCl was added to a final concentration of 1 M and the sample centrifuged for 5 mins at 3,000g incubated on ice for 1 h and centrifuged for 30 min at 3,000g samples were mixed thoroughly with an equal volume of chloroform and centrifuged for 5 min at 3,000g Samples were buffer exchanged into dPBS and concentrated to roughly 50 µl using 100-kDa cut-off cellulose centrifugal filters (Millipore Sigma For the cFos immunohistochemistry experiment mice were subjected to an interaction with an aggressor (CD-1/CFW) or a potential mate for 1.5 h and immediately perfused with chilled 10% NBF 40-μm slices were prepared on a Leica VT 1000S vibratome and collected in 24-well plates filled with phosphate buffered saline (PBS) After a 1.5-h blocking in 5% normal goat serum slices were incubated with 1:1,000 diluted primary antibody and 5% normal goat serum at 4 °C for 12 h slices were incubated for 1.5 h in secondary antibody subsequently washed three times in PBS for 15 min and Fluoromount-G mounting medium was used for 4,6-diamidino-2-phenylindole (DAPI) staining The following primary antibodies were used: 1:1,000 diluted cFos antibody (Synaptic Systems 226 008) and 1:500 diluted TPH2 antibody (Abcam The following secondary antibodies were used: Alexa 488 (1:500 dilution) Alexa 568 (1:1,000 dilution) and Alexa 633 (1:1,000 dilution) AAVDJ-EF1α mCh-IRES-Cre-WPRE and AAV9-EF1α-sgTph2-EGFP or AAV9-EF1α-sgNTC-EGFP were injected into the DR (unilateral anteroposterior −4.36; mediolateral 0; dorsoventral −3.1) of dCas9-KRAB mice immunohistochemistry was performed as described above Imaging was conducted on a Zeiss LSM 780 confocal (×20 objective; NA 0.8; diode laser 405 nm diode-pumped solid-state laser 561 nm and HeNe laser 594/633 nm; detectors 32 channel GaAsP 2-PMT and Airyscan detector; bit depth of images was 8 bit) and the same cut-off threshold was applied for all replicates Polygons were manually drawn around the DR area and particles within the area of interest were analysed All infected DR cells received sgTph2 or non-targeting guides To evaluate cell body TPH2 expression in the DR An average percentage fluorescence intensity was calculated from the sgNTC replicates and the remaining ratio was used for calculation of the percentage TPH2 fluorescence intensity and knockdown efficiency Behavioural cohorts were injected with AAV9-EF1α-sgTph2-EGFP or AAV9-EF1α-sgNTC-EGFP into the DR and CAV2-Cre into the vCA1 behavioural experiments were performed followed by perfusion and immunohistochemistry Imaging and background subtraction was performed as described above Polygons were manually drawn around the vCA1 and particles in within the area were analysed In the vCA1 only the DR terminals received sgTph2 or non-targeting sgRNA we analysed the area of the TPH2+ terminals in the vCA1 An average percentage of TPH2+ area within the vCA1 area was calculated from the sgNTC replicates and the remaining ratio was used for calculation of percentage TPH2+ area Primary neuronal culture was performed as previously described56 DR nuclei were dissected from P0 pups in ice-cold Hank’s buffered saline solution (HBSS) buffer (Thermo Fisher Scientific 88284) supplemented with 2 mM Ca2+ and 0.5 mM EGTA under a dissection hood and incubated in Neuronal Isolation Enzyme (Thermo Fisher Scientific washed twice with HBSS and dissociated in prewarmed Neurobasal-A Medium (Thermo Fisher Scientific Cell yield and viability were determined by a hemocytometer and trypan blue (Thermo Fisher Scientific Neurons were plated on a poly-d-lysine (Thermo Fisher Scientific half of the serum medium was replaced with an equivalent volume of Serum-free Neurobasal-A Medium supplemented with 0.5 mM l-glutamine (Thermo Fisher Scientific A2916801) and 0.5% B-27 (Thermo Fisher Scientific half of the old medium was replaced with fresh medium supplemented with 4 mM Ara-C (Sigma and neurons were infected with AAV constructs expressing control sgRNAs or Tph2 sgRNAs the primary neuronal cultures were maintained for 1 week with medium changes every 3 days RNA extraction was performed using a Quick-RNA MicroPrep Kit (Zymo Research R1050) according to the manufacturer’s protocol Total RNA was reverse transcribed using the High-Capacity RNA-to-cDNA Kit (Thermo Fisher Scientific The resulting complementary DNA (cDNA) was then used for qPCR using PowerUp SYBR Green Master Mix (Thermo Fisher Scientific Data were recorded using a QuantStudio 7 Flex Real-Time PCR system (Thermo Fisher Scientific) The relative changes in expression were calculated using the 2-ΔΔCt method Primers used in the studies include: Tph2 (mouse): forward 5′-CAGTCCACGAAGATTTCGACTT-3′; Actb (mouse): forward Shank3Δ4-22+/− and wild-type littermates were used for whole-cell recordings Mice were euthanized and brains were sliced in a sucrose cutting solution containing (in mM): 228 sucrose 7 MgCl2 and 0.5 CaCl2 on a vibratome (Leica VT1200 S) Slicing and recording solutions were continuously equilibrated with 95% O2 and 5% CO2 Coronal vCA1 slices (300 μm) were transferred to a slice holding chamber with artificial cerebrospinal fluid containing (in mM): 119 NaCl 1.3 MgCl2 and 2.5 CaCl2 (osmolarity 289–295) for 30 min at 32 °C and then further equilibrated for 30 min at room temperature brain slices were placed in a recording chamber perfused with 28–30 °C artificial cerebrospinal fluid and visualized with a ×40 water-immersion objective on an upright fluorescent microscope (BX51WI; Olympus) equipped with infrared-differential interference contrast video microscopy and epifluorescence (CoolLED) Whole-cell current-clamp recordings (pipette opening 4–6 MΩ) were performed with pipettes filled with (in mM): 130 C6H11KO7 10 phosphocreatine (pH 7.25; osmolarity 290) Whole-cell voltage-clamp recordings (pipette opening 3–4 MΩ) of IPSCs were performed with pipettes filled with (in mM): 80 CsCl 7 phosphocreatine (pH 7.34; osmolarity 300) No CsCl but 140 CsMeSO4 was used for whole-cell voltage-clamp recordings of EPSCs Series and input resistance were monitored with −4 mV 70-ms pulse delivered through the recording pipette and experiments were excluded from the analysis if series resistance varied by more than 15% Current-clamp recordings were carried out to assess intrinsic cell excitability with a series of incremental rectangular depolarizing current pulses (20 pA 500 ms) injected into vCA1 pyramidal cells the same procedure was performed in the same cell after either CP93129 dihydrochloride (Tocris Biosciences 1032 recordings were carried out in the same cell in presence of the second drug whereby the sequence of drug application was counterbalanced between cells Current-clamp recordings in the PVT was carried out with 20-pA depolarizing current pulses and in the DR with 25-pA depolarizing current pulses Pipettes were filled with in (mM): 130 C6H11KO7 Evoked EPSCs and IPSCs were recorded from vCA1 cells with a bipolar stimulating electrode (fabricated from platinum and iridium wire) placed near the recording pipette Baseline evoked and spontaneous EPSCs were recorded in voltage-clamp settings at −70 mV the presence of picrotoxin (50 μM) Baseline evoked and spontaneous IPSCs were recorded at −70 mV in the presence of NBQX (10 μM) and D-AP5 (50 μM) CP93129 dihydrochloride (5 µM) or PD149163 (0.5 µM) were bath applied after an initial baseline recording Summary graphs of the effects of CP93129 and PD149163 over time in evoked recordings were generated by averaging 1-min bins as a percentage of the averaged 4-min baseline Cumulative probability graphs were generated from all spontaneous events before and after drug application Synaptic responses were recorded in the vCA1 8 weeks after stereotaxic injections of AAVDJ-hSyn-ChR2-eYFP into the DR or PVT To photostimulate ChR2 in vCA1 axons and/or terminals 470-nm 0.1–5.0-ms light pulses from the pE-300ultra (CoolLED) were delivered by means of a ×40 water-immersion objective to the whole slice NAS-181 (20 µM) or SR48692 (0.5 µM) were bath applied following an initial baseline recording Recordings were made using a MultiClamp 700B amplifier (Molecular Devices) digitized at 10 kHz with the Digidata 1320A or 1440A data acquisition system (Molecular Devices) and analysed with Clampfit v.10.7 software (Molecular Devices) brains were fresh frozen on dry ice and stored at −80 °C until 15-μm slices were prepared and collected directly on Superfrost plus microscopy slides using a cryostat Following fixation in prechilled 4% paraformaldehyde slices were washed in PBS and dehydrated gradually in 50% 70% and two times 100% ethanol (5 min each) After a barrier was created with a hydrophobic pen slides were incubated with roughly five drops of RNAscope hydrogen peroxide for 10 min subsequently washed with distilled water and incubated with Protease IV for 30 min Mm-Gad2-C2 (439371-C2) and Mm-Nts-C3 (420441-C3) were warmed up to 40 °C in a water bath and 1 volume of C2 and C3 were diluted in 50 volumes of C1 to create the probe mixture slices were incubated with the probe mixture for 2 h at 40 °C in a HyEZ Oven and rinsed twice at room temperature in Wash buffer (50× RNAscope Wash buffer diluted in distilled water) slides were incubated with roughly six drops of RNAscope Multiplex FL v2 Amp1 for 30 min at 40 °C in the HyEZ Oven and washed twice with wash buffer at room temperature This procedure was repeated with RNAscope Multiplex FL v2 Amp2 (30 min) and RNAscope Multiplex FL v2 Amp3 (15 min) Slides were then incubated with RNAscope Multiplex FL v2 HRP-C1 in the HyEZ Oven at 40 °C for 15 min rinsed twice in wash buffer at room temperature and incubated in a 1:1,000 diluted Opal 520 dye for 30 min at 40 °C and subsequently rinsed twice in wash buffer HRP-C3 and matching Opal dyes 570 (1:3,000 dilution) Fluoromount-G mounting medium was used for DAPI staining Imaging was performed on a Zeiss LSM 780 confocal (×20 objective; NA 0.8; diode laser 405 nm 2-PMT and Airyscan detector; bit depth of images A Fiji (ImageJ v.1.52p) macro was used for automated image analysis Polygons were manually drawn around the vCA1 area and particles in the remaining channels were automatically analysed within the masked areas inside the polygon The cut-off threshold was visually determined by comparing ascending intensity values with marked location in the images to eliminate background particles the experimenters were blinded to the virus injection the animals received Prism10 (GraphPad) was used for statistical analysis Individual data points are identified by sex in the source data Parametric statistical tests were only chosen for normally distributed samples non-parametric statistical tests were performed One-way analysis of variance (ANOVA) with Tukey’s or Dunnett’s multiple comparison post hoc test was used to determine significance for several treatment comparisons and two-way ANOVA with Tukey’s or Sidak’s multiple comparison post hoc test for several group comparisons and across many time points P values found in figure legends refer to ANOVA results of the whole group whereas the asterisks between individual samples represent results of the multiple comparison post hoc tests A two-tailed paired Student’s t-test was applied for within-group comparison of two treatments and unpaired test for comparison between two groups the Wilcoxon signed rank test was used for within-group comparison of two treatments Mann–Whitney test for between group comparison and Kruskal–Wallis with post hoc Dunn’s test for multiple comparisons 15 cells (Shank3) and 14 cells (wild-type) Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article Dynamic influences on the neural encoding of social valence The hippocampal CA2 region is essential for social memory A hippocampal circuit linking dorsal CA2 to ventral CA1 critical for social memory dynamics Ventral hippocampal projections to the medial prefrontal cortex regulate social memory A hypothalamic novelty signal modulates hippocampal memory 5-HT modulation of a medial septal circuit tunes social memory stability A circuit mechanism for differentiating positive and negative associations Neurotensin orchestrates valence assignment in the amygdala Anterior cingulate cortex and its input to the basolateral amygdala control innate fear response Neural circuit motifs in valence processing Bidirectional switch of the valence associated with a hippocampal contextual memory engram Contextual fear memory retrieval by correlated ensembles of ventral CA1 neurons Creating a false memory in the hippocampus Social odor discrimination and its enhancement by associative learning in the hippocampal CA2 region The hippocampal CA2 region discriminates social threat from social safety A standardized protocol for repeated social defeat stress in mice Sex differences in appetitive and reactive aggression Sexually dimorphic neurons in the ventromedial hypothalamus govern mating in both sexes and aggression in males Sexual behavior regulated (paced) by the female induces conditioned place preference Characterization of copulatory behavior in female mice: evidence for paced mating A direct lateral entorhinal cortex to hippocampal CA2 circuit conveys social information required for social memory Single-cell transcriptomes and whole-brain projections of serotonin neurons in the mouse dorsal and median raphe nuclei Hippocampal cells segregate positive and negative engrams Anxiety cells in a hippocampal-hypothalamic circuit Reduced fear expression after lesions of the ventral hippocampus Direct ventral hippocampal-prefrontal input is required for anxiety-related neural activity and behavior Two opposing hippocampus to prefrontal cortex pathways for the control of approach and avoidance behaviour Dorsal raphe nucleus to anterior cingulate cortex 5-HTergic neural circuit modulates consolation and sociability Orchestrating opiate-associated memories in thalamic circuits 5-HT release in nucleus accumbens rescues social deficits in mouse autism model Expression and distribution of neuropeptide-expressing cells throughout the rodent paraventricular nucleus of the thalamus Circuit and molecular architecture of a ventral hippocampal network Neuronal responses to conspecifics in the ventral CA1 Serotonin1B receptors: from protein to physiological function and behavior Neurotensin depolarizes globus pallidus neurons in rats via neurotensin type-1 receptor Activation of neurotensin receptor 1 facilitates neuronal excitability and spatial learning and memory in the entorhinal cortex: beneficial actions in an Alzheimer’s disease model Hippocampal contributions to social and cognitive deficits in autism spectrum disorder Drapeau, E., Riad, M., Kajiwara, Y. & Buxbaum, J. D. Behavioral phenotyping of an improved mouse model of Phelan-McDermid syndrome with a complete deletion of the Shank3 gene. eNeuro https://doi.org/10.1523/ENEURO.0046-18.2018 (2018) Shank3 mutant mice display autistic-like behaviours and striatal dysfunction Disrupted social memory ensembles in the ventral hippocampus underlie social amnesia in autism-associated Shank3 mutant mice Tuned geometries of hippocampal representations meet the computational demands of social memory Li, L. et al. Social trauma engages lateral septum circuitry to occlude social reward. Nature https://doi.org/10.1038/s41586-022-05484-5 (2022) Dopamine and serotonin: influences on male sexual behavior Spatial gene-expression gradients underlie prominent heterogeneity of CA1 pyramidal neurons seqFISH accurately detects transcripts in single cells and reveals robust spatial organization in the hippocampus Social amnesia in mice lacking the oxytocin gene Noradrenergic consolidation of social recognition memory is mediated by beta-arrestin-biased signaling in the mouse prefrontal cortex Thirst-associated preoptic neurons encode an aversive motivational drive Canine adenovirus vectors: an alternative for adenovirus-mediated gene transfer Drug-induced conditioned place preference and aversion in mice A genetically encoded sensor for measuring serotonin dynamics Compact and highly active next-generation libraries for CRISPR-mediated gene repression and activation and reliable protocol for in-house AAV vector production based on chloroform extraction Neuroligin-1 signaling controls LTP and NMDA receptors by distinct molecular pathways RNAscope: a novel in situ RNA analysis platform for formalin-fixed Wu, X. Raw/processed data. Zenodo https://doi.org/10.5281/zenodo.14814653 (2025) Wu, X. MATLAB code [Data set]. Zenodo https://doi.org/10.5281/zenodo.14814674 (2025) Download references This work was supported by funding from NIH K99/R00 Career Development Award (grant no NARSAD Young Investigator Award and Friedman Brain Institute Scholar Award Russo for providing mouse breeding pairs; K interpreted the results and wrote the paper performed surgeries and behavioural experiments performed CRISPRi validation and associated histology performed ex vivo electrophysiology recordings The authors declare no competing interests Nature thanks the anonymous reviewers for their contribution to the peer review of this work Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations Statistical tests: duration (a – e): two-tailed paired Student’s t-test Source data social memory: duration subjects spent in chamber with familiar mouse (fm) or novel mouse (nm) and discrimination (d) scores (a: F3,34 = 6.693 Negative social valence (sv): duration in chamber containing the non-aggressor with a previous ntrl encounter or in chamber containing the aggressor with a previous neg encounter and d scores (a: F3,34 = 8.018 Positive sv: duration in chamber containing the partner with a previous ntrl encounter or in chamber containing the potential mate with a previous pos encounter and d scores (a: F3,35 = 4.622 duration in chamber with a novel object (no) or nm and d scores (c: F3,34 = 0.8793 Schematic of experiment and duration subjects spent in chamber containing a conspecific male or female Schematic of experiment and the number of times female test mice escaped from males into the separate compartment of the cage within 5 min (t4 = 0.3203 d scores (a – d): two-way ANOVA with Šidák’s post-hoc test Source data Duration that subjects spent in chamber with a novel object (no) or nm and discrimination (d) scores (F3,36 = 0.617 Duration spent on previously CNO or saline paired surfaces (F3,34 = 1.948 Duration that subjects spent in chamber with a nm or familiar mouse (fm) and d-scores (t14 = 4.034 Duration that subjects spent in chamber with a novel object (no) or nm and d scores (t14 = 1.346 Duration that subjects spent in chamber with (e) fm or nm (f) non-aggressor with previous ntrl encounter or aggressor with previous neg encounter and d scores (g) partner with previous ntrl encounter or potential mate with previous pos encounter and d scores (e: t12 = 0.2963 Duration that subjects spent in chamber with a no or nm and d scores (t12 = 0.4988 Duration that subjects spent in chamber with a no or nm and d scores (F3,26 = 2.552 Duration spent on previously CNO or saline paired surfaces (F3,29 = 0.8432 Sample traces of DR (k) and PVT (l) neuron spiking in response to CNO quantification of spiking (k: uninfected control t5 = 1.581 j): two-way ANOVA with Šidák’s post-hoc test Source data Source data recordings of spontaneous EPSCs (sEPSCs) before and after bath-application of PD and cumulative probability plot of sEPSC amplitudes with representative traces (1 of 10 cells) above (t9 = 2.628 cumulative probability plot of sEPSC inter-event intervals with representative traces (1 of 10 cells) above (t9 = 1.333 recordings of sIPSCs before and after bath-application of PD and cumulative probability plot of sIPSC amplitudes with representative traces (1 of 8 cells) above (t7 = 0.6492 cumulative probability plot of sIPSC inter-event intervals with representative traces (1 of 8 cells) above (t7 = 1.823 recordings of spontaneous IPSCs (sIPSCs) before and after bath-application of CP and cumulative probability plot of sIPSC amplitudes with representative traces (1 of 9 cells) above (t8 = 0.1498 Cumulative probability plot of sIPSC inter-event intervals with representative traces (1 of 9 cells) above (t8 = 2.94 recordings of sEPSCs before and after bath-application of CP and cumulative probability plot of sEPSC amplitudes with representative traces (1 of 10 cells) above (t9 = 0.4582 Cumulative probability plot of sEPSC inter-event intervals with representative traces (1 of 10 cells) above (t9 = 0.2687 Schematic and representative images (1 of 4 mice) of in-situ hybridization in the vCA1 pie chart of cells positive and negative for Gad2 and Htr1b along with a smaller pie chart and bar graph showing the proportion of Gad2+ cells positive and negative for Htr1b one-way ANOVA with Dunnett’s post-hoc test Source data Source data Source data DR 5-HT and PVT NT neurons innervate overlapping but not identical areas within the vCA1 During an aversive social interaction NT is released from PVT neurons into the vCA1 and binds to NTR1 receptors on vCA1 pyramidal neurons During an appetitive social interaction 5-HT is released from DR neurons into the vCA1 and binds onto 5-HT1BRs on GABAergic interneurons and inhibits their activity By releasing the local inhibition onto vCA1 pyramidal neurons the net excitatory drive (likely from the dCA24) onto the pyramidal neurons is increased As there is heterogeneity in the vCA1 cell population a subset of cells can be preferentially potentiated by NT due to higher expression of NTR1s or preferentially potentiated by 5-HT as they receive inputs from interneurons with higher expression of 5-HT1BRs the increase in excitability in vCA1 pyramidal neurons in response to either NT or 5-HT balances social valence The right insert highlights the converging circuitries from the PVT and DR to the vCA1 Download citation DOI: https://doi.org/10.1038/s41586-025-08809-2 Anyone you share the following link with will be able to read this content: a shareable link is not currently available for this article Sign up for the Nature Briefing newsletter — what matters in science The dates displayed for an article provide information on when various publication milestones were reached at the journal that has published the article activities on preceding journals at which the article was previously under consideration are not shown (for instance submission All content on this site: Copyright © 2025 Elsevier B.V. Newcity Music by Dylan Weinert | September 5 Joey Valence & Brae/Photo: Joey Valence & Brae Joey Valence & Brae’s highly anticipated sophomore album “No Hands” (2024) debuted on Spotify’s top ten U.S album chart this summer with no shortage of fanfare and serious acclaim all from the confines of Joseph Bertolino (Joey Valence)’s Pokemon-clad childhood bedroom And that’s who this music is for: Anyone who has spent time listening to hip-hop in a basement shamelessly Beastie Boys-influenced brand of rap that faces zero competition in the modern hip-hop sandbox Their production is fueled by an erudite respect for the old school It doesn’t hurt that JVB indexes high on a hang level a rap collective comes around that makes being a misfit cool JVB does it better than anyone since Odd Future Joseph Bertolino and Braedan Lugue are two of the most likable musicians you’ll ever meet I had a chance to catch up with them in between their last European tour and their upcoming set at Outset on September 13 tour before heading back to Europe on another (largely sold-out) overseas trip Two words stuck with me throughout our entire conversation: humility and fun No need for fancy words—these two are selling out in Europe And that love manifests in unadulterated joy in every live performance and fan interaction In January 2024, they tore up Subterranean a show colored by backflips and fans being pulled onstage for selfies and public dialogues about their fan art It’s a civic honor that they’re back in Chicago so soon the nineties-era-rap duo (who graduated from Penn State in 2021) are bringing their timeless hip-hop to Chicago’s newest venue To kick things off I ask the boys how everything got started and Joey sets the tone with a dime of an answer: “We actually met in an online sex chatroom.” but we were introduced through mutual friends at a Red Lobster and found out we had similar tastes in music and senses of humor I’d go over to his place and he was always making beats and as we got closer to graduation things just started taking off as Joey Valence & Brae.” Considering “Punk Tactics” (2023) came out just a year ago I want to know how the pair approached “No Hands” differently Joey: “The first album was just a bunch of singles We’d made them earlier in the year and just like compiled them into an album “‘No Hands’ is our first time making something that is a complete project It was a gamble for us because we’ve never taken time off (from releasing singles) and had something come out as a full collective project we had more to talk about and wanted to make something more meaningful We wanted to talk about what we’ve been able to achieve the title track is called ‘No Hands.’ It’s about proving to our fanbase—most of whom are just like us same age or maybe younger—the growth of what we’ve been able to accomplish and that they can accomplish whatever their niche or hobby is They can propel themselves into whatever they want Everything we do is DIY and you can do that shit too What we’ve been able to do is impressive and is a lot of hard work but it’s also not that hard because we’re good at it.” The fact that the two have been able to build that so quickly and so globally is and how they communicate or connect with their fans outside of their lyrics literally standing outside of venues and just meeting everyone every time We can’t do that anymore because there are too many people and we get sick But that’s how you create these for-life people taking their phones and posting on their socials Brae: “We have kids making just the dopest art for us and when we take that and actually use it because we love it it makes them feel like they are even more connected to us It’s super dope—not only are we creating our art JVB’s ascent has been nothing short of astonishing and I want to know if there is anything associated with their fame touring or lifestyle that has become a nuisance They’ve got a huge following and people like me bugging them for time these guys are living their dream and treating everything that comes with it with gratitude “We’ve never half-assed a show,” Joey responds as soon as I raise the topic of tour fatigue let’s say you had a free Saturday—no music Joey’s answer is firm and clear: “I’d play video games I would play ‘Rocket League’ literally the entire day Recently the duo released some high-profile collaborations including work with Ayesha Erotica and Danny Brown on “The Baddest (Badder)” and “Packapunch,” respectively Joey: “I literally just sent her [Ayesha] a DM I sent her the opening of [‘The Baddest’] and asked if she wanted to do a remix and within forty minutes she came back and was like ‘I got excited’ and sent me her verse “I’ve been a fan of hers and I really wanted a female voice on our music—that’s something Brae and I have been talking about for a while—and she fucking nailed it.” Brae says: “Danny is just as cool and crazy in person as he is online Joey and I saw him perform in Philly and went up to his hotel room after We brought our laptop and mic and Danny spit his verse in one take in five minutes We kicked it with him for a bit but he’s a super kind guy and really willing to work with new and underground artists.” Joey: “He’s really tapped into the underground but he’s supporting a ton of newer artists right now.” They’re enjoying the DIY hustle and are more interested in building as much leverage as possible for any future deal My main takeaway is JVB is an exercise in intelligent chaos saving money as they rack up more than a hundred million Spotify streams and billions of views on TikTok They tour maniacally at 110 percent and still make time to connect with fans on Discord “No Hands” isn’t even four months old yet and they’re already planning their next album JVB will be dropping “No Hands (Deluxe),” which will include a cover of one of the biggest names on the planet in hyper-pop there’s potentially a single dropping with the most exciting name in indie sleaze today Joey Valence & Brae are playing Outset, 1675 North Elston, on Friday, September 13, 6:30pm. Tickets are sold out French cloud computing provider Sesterce is investing €450 million ($471.85m) into developing an AI data center in Valence The data center is set to be located in the Rovaltain business park in Valence Romans Agglo and will house 40,000 GPUs for the training and deployment of AI The data center will use a closed-loop water cooling system which will enable the waste heat to be reused Works will commence in September 2025 with the data center expected to be launched before the end of 2026 The size of the facility and capacity in MW were not shared DCD has contacted Sesterce for more information The €450m is the first tranche of investment eventually expected to reach €1.8bn ($1.89bn) including hardware costs The data center is part of a wider national plan to develop AI infrastructure in France Sesterce has committed to developing 1.5GW of compute power in France including the deployment of 1.2 million GPUs by 2030 “This project represents a crucial step in our strategy to develop high-performance computing capabilities in France,” said Anthony Tchakerian “The choice of Valence was a natural one thanks to its dynamic ecosystem and strategic position.” “The arrival of Sesterce significantly strengthens our position as a territory of excellence for industry and innovation,” added Nicolas Daragon “This structuring project reinforces our recent Industrial Territory label and our position among the three most dynamic medium-sized agglomerations in France It also opens the doors to the establishment of new companies linked to the AI ​​sector.” In addition to the development of the Valence HPC data center Sesterce is aiming to add 600MW of capacity across two data centers in Grand Est with 500,000 GPUs by 2028 and 1.2 GW with more than one million GPUs by 2030 the company has said it will establish a 250MW supercomputer with 200,000 GPUs in southern France Sesterce established its first data center presence in 2018 the company launched an HPC cluster in Paris featuring Dell PowerEdge XE9680 servers equipped with Nvidia H200 Tensor Core GPUs and in December announced a cluster of H100 GPUs housed at a Digital Realty data center in Marseille it currently has 1GW of compute and more than 100,000 GPUs under management Data Centre Dynamics Ltd (DCD), 32-38 Saffron Hill, London, EC1N 8FH Email. [email protected]DCD is a subsidiary of InfraXmedia Metrics details Dual active sites with synergistic valence state regulation under oxidizing and reducing conditions are essential for catalytic reactions with step-wise mechanisms to modulate the complex adsorption sites of reactant molecules on the surfaces of heterogeneous catalysts with maximized catalytic performances uniformly dispersed CuCo alloy and CoO nanosheet composite catalysts with dual active sites are constructed which shows huge boost in activity for catalyzing water-gas shift reaction (WGSR) with a record high reaction rate reaching 204.2 μmolCO gcat.−1 s−1 at 300 °C for Cu1Co9Ox amongst the reported Cu-based and Co-based catalysts A synergistic mechanism is proposed that Coδ+ species can be easily reduced by CO adsorbed on Cu and Co0 can be oxidized by H2O Systematic in situ characterization results reveal that the addition of Cu can regulate the redox properties of Co species and thus modulate the adsorption properties of catalysts doping of Cu0 sites weakens the affinity of the surface to CO or CO2 to a moderate level it also promotes the oxidation of *CO to *COOH and the desorption of the product CO2 reducing the carbon poisoning of the catalyst and thus increasing the reactivity The results would provide guidance for the construction of novel heterogeneous catalyst with dual active sites and clarify its underlying reactivity enhancement mechanism induced by the tunning of valence state of metal centers for heterogeneous catalytic reactions catalysts composed of transition metal alloy and corresponding mechanisms have rarely been reported for WGSR Exploring catalysts with dual active sites and tuned valence state is a great challenge but also full of scientific and technological significance the heteroatom doping could not only be the additional active sites for heterogenous catalysis but also modulate the electronic states of catalysts with tuned valence state and dynamic active sites this work developed Cu/Co-CoO1-x composite oxides with dual active sites for catalyzing WGSR which greatly increased CO conversion below 250 °C The reaction rate of Cu1Co9Ox catalyst reached 204.2 μmolCO gcat.−1 s−1 at 300 °C which was much higher than the rate of the reported Cu-based and Co-based catalysts Systematic in situ characterization and DFT calculation results revealed that the addition of Cu could tune the valence state of Co species and thus modulate the adsorption properties of catalysts which realized highly efficient catalytic performance for WGSR The novel discoveries in this work suggested a practical strategy to construct synergistic dual active sites in heterogeneous catalysts and offered better understanding about the mechanism of dual active sites thereof a Temperature-dependent activities of the catalysts (reaction atmosphere was 2%CO/10%H2O/88%N2) b Comparison of reaction rates with other catalysts of transition metals c The long-term stability at a high GHSV (Cu1Co9Ox: 240 °C e CO reaction order and f H2O reaction order of Cu1Co9Ox and Co3O4 catalysts The aberration-corrected HAADF-STEM images and corresponding EDS elemental maps of a–c Cu1Co9Ox and d–f Co3O4 after WGSR. a XRD patterns of the catalysts after WGSR In situ XRD pattern of b Cu1Co9Ox and c Co3O4 catalysts e H2-TPR after 5%H2/Ar reduction at 300 °C and f H2-TPR after WGSR of CuaCobOx catalysts g CO-TPSR after WGSR of Cu1Co9Ox catalysts h Schematic illustration of structure evolution of Cu1Co9Ox catalysts during WGSR a Quasi in situ XANES and b EXAFS results of the Co3O4 and Cu1Co9Ox catalysts after WGSR d Co 2p and e O 1s of the Co3O4 and Cu1Co9Ox catalysts after WGSR f In situ Raman spectra of the Co3O4 and Cu1Co9Ox catalysts under WGSR condition (250 °C In situ DRIFTS results of a–c Co3O4 and d–f Cu1Co9Ox catalysts under 2%CO/3%H2O/He reaction atmosphere at different temperatures g Gibbs Free energy diagram for COOR processes on the surface of Co-Cu alloy(100) green and purple atoms are corresponding to Co a CO surface reaction results of the Cu1Co9Ox catalyst In situ DRIFTS results of b Co3O4 and c Cu1Co9Ox catalysts after switching gas mixture from 2%CO/3%H2O/He to 3%H2O/He at 240 °C d NAP-XPS of Cu1Co9Ox catalysts under 2 mbar 10%CO/Ar and 2 mbar 5%H2O/Ar at 300 °C In situ Raman spectra of e Cu1Co9Ox and f Co3O4 catalysts under 2%CO/Ar and 3%H2O/Ar at 300 °C g Schematic illustration of the dual active centers for WGSR The spectra at 0 min were the results of the stabilization of the surface species before the gas switching the signals of gaseous CO and adsorbed CO on the Cu sites disappeared rapidly it could be found that only the intensity of formate and carboxylate species decreased with the generation of CO2 while the intensity of all the other carbonate species did not change significantly This suggested that adsorbed CO would react with OH* to form formate and carboxylate as the critical intermediates following the associate mechanism for WGSR Furthermore, to clarify the role and reaction of CO* and OH* on the surface, cyclic CO-TPR experiments for observing the production of CO2 and H2 for the Cu1Co9Ox catalyst were performed. (Between CO-TPR-1 and CO-TPR-2, the catalyst was treated under 3%H2O/Ar atmosphere at 300 °C for hydroxylation). As shown in Supplementary Fig. 28 the broad peak of CO2 was mainly contributed to the reduction of CoOx and the production of H2 was also observed due to the reaction of CO and OH on the surface the hydroxyl species formed during the synthesis process the co-production of CO2 and H2 indicated that the generation of surface hydroxyl was recyclable and WGSR followed the associate mechanism that CO reacted with the hydroxyl the hydroxyl OH* and CO* on the surface could react with each other to form formate and carboxylate intermediates whose decomposition produced H2 and CO2 Wherein the dissociation of H2O to form OH* was essential for the formation and decomposition of formate and carboxylate intermediates the existence of Cu was conducive to the tunning of valence state between Co0 and Coδ+ to facilitate the CO oxidation and the H2O dissociation this work constructed a uniformly dispersed Cu/Co-CoO1-x composite oxides catalyst with dual active sites Based on these investigations mentioned above Cu was the active site to adsorb CO which reduced the surrounding Coδ+ species and H2O could thus oxidized the surrounding Co0 species on the surface Cu enabled Co species with tuned valence state and proper valence states under oxidizing and reducing conditions Cu could stabilize the surface of Co and was conducive to the decomposition of carbon-containing intermediates The favorable CO adsorption capacity of Cu and the tunning of valence state between Co0 and Coδ+ could facilitate CO oxidation and H2O dissociation thus significantly boosting the catalytic performance towards WGSR This work provides a unique approach to construct and investigate chemical states oscillation between synergistic dual active sites for governing surface kinetics and catalytic reactivities over multi-component catalyst systems The copper-cobalt catalysts were prepared by the hydrothermal method a certain amount of Cu(NO3)2·3H2O and Co(NO3)2·6H2O with the different ratios were dissolved in 39 mL distilled water and the pH value of the solution was controlled to ca.10 by NaOH (4 mol L−1) the obtained suspension was transferred to a Teflon-lined steel autoclave and kept at 100 °C for 24 h The obtained precipitates were collected by centrifugation until the pH was 7 and dried at 70 °C overnight the CuaCobOx catalysts were obtained by calcination at 300 °C for 2 h The catalytic performance towards WGSR was evaluated in a continuous-flow fixed-bed reactor at 1 bar A total of 50 mg of the 80–200 mesh catalysts mixed with 1 g of 40–70 mesh quartz sand was first placed in a quartz tube with an interior diameter of 1 cm a thermocouple was inserted into the extra 1 g quartz sand layer to obtain an accurate reaction temperature the samples were reduced in 5%H2/N2 at 300 °C for 30 min 10%H2O and 88%N2 was fed into the reactor at 70 mL/min together with temperature programming from 150 to 300 °C The CO conversion was analyzed using an online GASBOARD gas analyzer The CH4 generation was analyzed by a gas chromatography The CO conversion was calculated according to the following equation: The CH4 selectivity was calculated according to the following equation: The reaction rate (r) was calculated according to the following equation: Other kinetic studies were tested in the same reactor and the CO conversion was controlled below 15% the activation energy was calculated according to the reaction rates of different temperatures in the kinetic range The reaction order was calculated according to the reaction rates of different feed gas compositions the H2O concentration was maintained at 20% and the CO concentration was adjusted from 1.5 to 3% in 0.5% increments the CO concentration was maintained at 3% and the H2O concentration was adjusted from 2.5 to 10% in 2.5% increments The N2 flow rate was adjusted simultaneously to keep the total flow rate of the feed gas at 70 mL/min The ex situ XRD patterns were acquired by the PANalytical X’pert3 powder diffractometer The in situ XRD patterns were obtained from the same machine with an Anton Paar XRK-900 reaction chamber The in situ reaction camber was heated from room temperature to 200 300 and 400 °C under 5%H2/Ar (30 mL min−1) Two rounds of tests lasting for 20 min and 40 min were carried out for each selected temperature The 40 min measurement was obtained to investigate the phase structure of the catalysts H2-TPR with different pretreatments data were obtained by the Quantachrome autosorb-iQ instrument 50 mg of the sample was first purged with a He flow at 150 °C for 1 h 50 mg of the sample was first reduced with a 5% H2/Ar flow at 300 °C for 30 min 50 mg of the sample was first reduced with a 5% H2/Ar flow at 300 °C for 30 min and then exposed with the 2%CO/3%H2O/Ar flow at 300 °C for 1 h the catalysts were reduced with a 5% H2/Ar flow from 50 to 600 °C at a rate of 10 °C/min The signal was collected using a thermal conductivity detector (TCD) TPD data were acquired using a Quantachrome autosorb-iQ instrument 300 mg of the catalysts was first reduced by 5%H2/Ar at 300 °C for 30 min and then purged with a He flow the sample was adsorbed with CO under a 5%CO/He flow for 60 min After removing the physical adsorption of CO by He purging desorption of the chemisorbed CO was achieved by temperature programming from 50 to 800 °C at a rate of 10 °C/min CO (m/z = 28) and CO2 (m/z = 44) were detected by a mass spectrometer the test process was same except that the adsorbed gas was substituted with CO2 An online mass spectrometer (Ametek LC-D200M) was used to analyze the outlet gases for surface reaction 100 mg of sample was reduced with 5% H2/Ar at 300 °C for 30 min and then went through WGSR for 1 h at 300 °C The catalysts were then purged by Ar at 240 °C followed by purging by 3%H2O/Ar for 1 h to observe H2O dissociation 2%CO/Ar was purged under heating from 30 °C to 240 °C and kept reaction for 1 h The low temperature of 240 °C was used to control the low conversion The catalysts were then purged by Ar and cooled down to room temperature CO-TPSR was performed by purging by 2%CO/Ar under heating from 30 °C to 800 °C The ex situ and in situ Raman spectra were tested by the LabRAM HR800 spectrometer (HORLBA Jobin Yvon) using 473 nm laser excitation The in situ reactor (Xiamen TOPS) was used for in situ tests 25 mg of the catalysts were reduced in 5%H2/Ar at 300 °C for 30 min the catalysts were treated at 250 °C under 2%CO/3%H2O/Ar flow the catalysts were exposed with 2%CO/Ar or 3%H2O/Ar at 300 °C in turn Transmission XAS measurements were performed on a laboratory device (easyXAFS300 which is based on Rowland circle geometries with spherically bent crystal analyzers (SBCA) and a silicon drift detector Si (5,5,3) was used for Co K-edge measurement The powder samples were thoroughly ground and mixed with Boron Nitride using an agate mortar and pestle and pressed into Ø = 10 mm pellets The pressed pellets were then sandwiched by Kapton tapes the catalysts were firstly reduced in 5%H2/N2 flow at 300 °C for 30 min and then catalyzed WGSR at 300 °C for 1 h The treated samples were transferred to a glovebox and prepared for XAS measurement without direct exposure to air The electronic structures of catalysts were investigated by using an Axis Ultra XPS spectrometer from Kratos The treated samples were transferred to a glovebox and prepared for XPS measurement without direct exposure to air The NAP-XPS (near-ambient pressure x-ray photoelectron spectroscopy) measurement was performed on a commercial SPECS NAP-XPS chamber equipped with a PHOIBOS 150NAP The CuaCobOx catalysts were conductive during NAP-XPS measurement no charge was accumulated on the surface of the catalysts The catalysts were first reduced in 5%H2/Ar at 300 °C for 30 min 2 mbar 10%CO/Ar and 2 mbar 5%H2O/Ar was introduced In situ DRIFTS was carried out on a Bruker TENSOR 27 spectrometer between 1000 and 4000 cm−1 catalysts were first reduced under 5%H2/He at 300 °C for 30 min and then purged under He to remove any adsorbed species the temperature was changed in the sequence of 50 °C For transient analysis of intermediate species the feeding gas was switched from 2%CO/3%H2O/He to 3%H2O/He at 240 °C for consuming adsorbed intermediate species CO adsorption experiment was performed under 5%CO/He at 240 °C The background spectrum was detected under a He flow at reaction temperature and then subtracted automatically in subsequent scans 3%H2O is the saturated vapor pressure of H2O at room temperature the Brillouin area was sampled by 3 × 3 × 1 Monkhorst–Pack mesh k-points since their surface energy converged slab model had different layers and the convergence criteria of the total energy was set to 10−7 eV a large enough surface supercell (with only one absorbate) was adopted such that the distance between the adsorbates was larger than 10 Å and a 15 Å vacuum layer was used to minimize the artificial interaction between slabs on vertical direction To simulate a disordered Cu-Co solid solution reduced from Cu1Co9Ox we randomly replaced 10% of the Co atoms by Cu atoms in the supercell atoms on the top layer in slabs were relaxed The Gibbs free energy of each intermediate was estimated according to G = E + EZPE + ΔU0→T – TS where E is the electronic total energy at 0 K ΔU0→T is the thermal energy correction at finite temperature The last three terms result from the vibrational frequency calculations at 298.1 K where frequencies less than 50 cm−1 were set to 50 cm−1 to avoid abnormal entropy contribution The data that support the figures within this paper and another finding of this study are available from the corresponding author upon request. Source data are provided as a Source Data file. Source data are provided with this paper The water gas shift reaction: Catalysts and reaction mechanism Hydrogen-storage materials for mobile applications Catalytically efficient Ni-NiOx-Y2O3 interface for medium temperature water-gas shift reaction Promoted Cu-Fe3O4 catalysts for low-temperature water gas shift reaction: Optimization of Cu content Promoting Molecular Exchange on Rare-Earth Oxycarbonate Surfaces to Catalyze the Water-Gas Shift Reaction Metal-support interaction induced ZnO overlayer in Cu@ZnO/Al2O3 catalysts toward low-temperature water-gas shift reaction Influence of Cu foam framework on the physico-chemical properties and catalytic behavior of Cu(Fe)AlO/Cu(Fe)Al ceramometal granules in WGSR Strong metal-support interactions between highly dispersed Cu+ species and ceria via mix-MOF pyrolysis toward promoted water-gas shift reaction Construction of stabilized bulk-nano interfaces for highly promoted inverse CeO2/Cu catalyst Designing Cu0-Cu+ dual sites for improved C-H bond fracture towards methanol steam reforming Selective CO2 reduction to CH3OH over atomic dual-metal sites embedded in a metal-organic framework with high-energy radiation Construction of Pd-Zn dual sites to enhance the performance for ethanol electro-oxidation reaction Lattice oxygen activation and local electric field enhancement by co-doping Fe and F in CoO nanoneedle arrays for industrial electrocatalytic water oxidation Uncovering the reaction mechanism behind CoO as active phase for CO2 hydrogenation Self-supported Pt-CoO networks combining high specific activity with high surface area for oxygen reduction Highly Ordered Mesoporous Cobalt-Containing Oxides: Structure and Active Sites in Oxidation of Carbon Monoxide Boosting reactivity of water-gas shift reaction by synergistic function over CeO2-x/CoO1-x/Co dual interfacial structures Valence oscillation and dynamic active sites in monolayer NiCo hydroxides for water oxidation Low-temperature water-gas shift reaction over Cu- and Ni-loaded cerium oxide catalysts Remarkable Performance of Ir1/FeOx Single-Atom Catalyst in Water Gas Shift Reaction Weakening the Metal-Support Interactions of M/CeO2 (M = Co Ni) Using a NH3-Treated CeO2 Support for an Enhanced Water-Gas Shift Reaction Boron Radicals Identified as the Source of the Unexpected Catalysis by Boron Nitride Nanosheets Atomic-thick metastable phase RhMo nanosheets for hydrogen oxidation catalysis Iridium single atoms incorporated in Co3O4 efficiently catalyze the oxygen evolution in acidic conditions XPS study of the Cu@Cu2O core-shell nanoparticles Surface chemistry of electrodeposited Cu2O films studied by XPS Engineering Crystallinity and Oxygen Vacancies of Co(II) Oxide Nanosheets for High Performance and Robust Rechargeable Zn-Air Batteries Toluene oxidation process and proper mechanism over Co3O4 nanotubes: Investigation through in-situ DRIFTS combined with PTR-TOF-MS and quasi in-situ XPS Thermodynamic CoO-Co3O4 crossover using Raman spectroscopy in magnetic octahedron-shaped nanocrystals On the Mechanism of Low-Temperature Water Gas Shift Reaction on Copper “Redox” vs “associative formate with -OH group regeneration” WGS reaction mechanism on Pt/CeO2: Effect of platinum particle size Importance of the Metal-Oxide Interface in Catalysis: In Situ Studies of the Water-Gas Shift Reaction by Ambient-Pressure X-ray Photoelectron Spectroscopy On the importance of metal-oxide interface sites for the water-gas shift reaction over Pt/CeO2 catalysts Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set From ultrasoft pseudopotentials to the projector augmented-wave method Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set Restoring the Density-Gradient Expansion for Exchange in Solids and Surfaces VESTA 3 for three-dimensional visualization of crystal VASPKIT: A user-friendly interface facilitating high-throughput computing and analysis using VASP code Download references This work was supported by the National Natural Science Foundation of China (Nos the National Key R&D Program of China (Nos and 2023YFA1506800 to Y.-W.Z.) and the Beijing National Laboratory for Molecular Sciences (BNLMSCXXM-202104 to Y.-W.Z.) Computation was performed on the High-performance Computing Platform of Peking University We greatly acknowledged the help from Ming Zhang LingYu Yan and MeiYao Wu from Shandong University for the help on the in situ Raman and in situ XRD tests We thank the Centre of Electron Microscopy Centre at Lanzhou University for the help on the aberration-corrected HAADF-STEM measurements We thank the Centre of Structural Characterizations and Property Measurements at Shandong University for the help on sample characterizations Beijing National Laboratory for Molecular Sciences (BNLMS) State Key Laboratory of Rare Earth Materials Chemistry and Applications PKU-HKU Joint Laboratory in Rare Earth Materials and Bioinorganic Chemistry College of Chemistry and Molecular Engineering State Key Laboratory of Inorganic Synthesis and Preparative Chemistry Key Laboratory for Colloid and Interface Chemistry Key Laboratory of Special Aggregated Materials School of Chemistry and Chemical Engineering analyzed the results and wrote the manuscript; X.-C.L prepared catalysts and performed most characterization tests; J.-H.W performed the aberration-corrected HAADF-STEM measurement; X.-C.L performed the surface reaction experiments; X.-C.L performed the XAFS experiments and analyzed the data Wei-Wei Wang and Kai Xu performed the in situ XRD Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work Download citation DOI: https://doi.org/10.1038/s41467-025-56161-w Metrics details Subjective feelings are thought to arise from conceptual and bodily states We examine whether the valence of feelings may also be decoded directly from objective ecological statistics of the visual environment We train a visual valence (VV) machine learning model of low-level image statistics on nearly 8000 emotionally charged photographs The VV model predicts human valence ratings of images and transfers even more robustly to abstract paintings limiting conceptual analysis of images enhances VV contributions to valence experience increasing correspondence with machine perception of valence VV resides in lower to mid-level visual regions where neural activity submitted to deep generative networks synthesizes new images containing positive versus negative VV There are distinct modes of valence experience and the other embedded in ecological statistics affording direct perception of subjective valence as an apparent objective property of the external world as well as control for the presence of human faces to track potential contributions from higher-level features While the choice of feature sets was heuristic-based on the existing literature they were all chosen such that the same array of feature values could come from numerous images with different or no clear semantic content extraction of this limited set of features acted as a filter transforming the initial image into an abstract space blind to the precise objects in an image and thus its content Prior to any model training, images were transformed from the large pixel space into this substantially smaller space of global visual features. The Visual Valence Model (VVM, Supplementary Fig. 1) was a trained random forest model that used these low-level features to predict NV the average valence rating of each image obtained from the available affective datasets and scaled to 1–9 (extremely negative to extremely positive) the out-of-sample predictions were correlated with ground-truth NV ratings indicating a small to moderate effect size (t(7982) = 31.2 Presence of faces did not appreciably change this relationship (Δr = 0.016 nor was the relationship sensitive to the choice of regression algorithm (see Supplementary Material) Excluding features derived from the first layer of AlexNet (variances and symmetry) while keeping all other features led to a slightly lower yet still significant predictive accuracy (t(7982) = 26.1 was also predictable from visual features following a similar procedure as VVM (t(7982) = 28.6 NV and NA were negatively correlated in the emotional image dataset (t(7982) = −44.1 such that the more negative images were more arousing The VVM’s predictions of valence (here referred to as Visual Valence or VV) were largely unchanged when controlling for NA (t(7982) = 29.6 confirming predictions of the pleasantness and unpleasantness from visual features was distinct from the intensity of feeling which is traditionally associated with attentional salience a Frequency distribution of NV and VV across realistic photos in study 1 b Distribution of NV and VV across abstract paintings in study 2 c Relationship between VV (y-axis) and NV (x-axis) for realistic photos and abstract paintings Source data are provided as a Source Data file machine and human observers demonstrated greater similarity in range and correspondence of valence experience Since abstract art may contain attenuated conceptual content relative to real photos VV may still be regulated by the processing of weak conceptual associations this provides evidence that VV has a causal role in valence experience Use of VV may be limited by human observers’ dependence on conceptual analysis for valence experience a Visual Valence (VV) distribution for stimuli (N = 280) b Normative Valence (NV) distribution for VV+ and VV− stimuli c Proportion of positive ratings for VV+ and VV− sets relative to subjects’ baseline (N = 87 and 92 for full and limited viewing; center shows median) d Best-fitting lines showing the relationship between VV and different methods of subjective valence acquisition across 280 stimuli: normative ratings from the original dataset (r = 0.04 limited viewing across participants (N = 87 and 92 Violin plots show the distribution of data between the 25th and 75th percentiles and the centers represent mean f Differentiation by VV across Response Time (RT) quantile-based bins full-viewing) is shown for trials in each RT bin Violin plots represent the distribution of data between the 25th and 75th percentiles While limited-viewing conditions would reduce access to high-level object or mid-level categories we further accounted for the possible role of salient complex categories (i.e. The greater contributions of VV in limited viewing (z = −5.4 −0.11)) and effect of NV in full-viewing (z = 5.8 95% confidence interval = (0.23,0.47)) held after statistically controlling for these salient content related effects This is consistent with VV as generalized over many categorical features rather than specific high-level objects or scene inferences These effects were again persistent when controlling for higher-level content (z = −3.9 The double dissociation between valence type (visual versus conceptual) and processing mode (brief vs extended exposures and fast vs slow processing) demonstrates that although correlated in normal experience (Study 1 and 2) VV and NV are distinct competing valence signals with NV related to deeper conceptual content and VV perceived directly from visual features processed distinctly from mid-level scene types and conceptual categories 20 participants) and Natural Scene Dataset (NSD); whole-brain corrected at false positive rate of 0.05; Chikazoe+: red; NSD+: yellow; Chikazoe−: dark blue; NSD−: light blue including the lateral occipital cortices (LOC) Mean activity in the bilateral sLOC was significantly coupled with VV (t(2525) = 3.1 This trend was generalized when examining the entire bilateral LOC including both superior and inferior subregions VV: (t(2525) = 2.6 mean activity in the mPFC was predictable from NV (t(2525) = 3.9 These results confirm the dissociation between the visual and frontal anatomical regions linked to VV and NV Psychophysiological interaction (PPI) analysis further revealed that the functional connectivity between mPFC and bilateral sLOC had a significant interaction with VV (t(2523) = 2.7 This effect of VV generalized to the connectivity between mPFC and the entire bilateral LOC (t(2523) = 2.15 The fact that this connectivity is specifically modulated by VV although not a direct test of causal directionality is in line with a potential bottom-up functional flow from VV embedded in mid-level visual processing to the mPFC that integrates a broader form of valence from different sources and modalities the current evidence suggests that the neural basis of VV relies more on decoding the external appearance of visual stimuli rather than internal neural representations of their conceptual attributes a Image generation procedure: A mask of regions correlated with visual valence (VV) in the NSD dataset was fed into NeuroGen (p < 0.001 uncorrected NeuroGen synthesized VV+ and VV− image sets by maximizing or minimizing expected weighted sum of neural activation in the mask b Within-class comparison of VVM’s VV estimate for VV− and VV+ images (N = 150 each; two-sided paired t-test c Behavioral preference between VV+ and VV− images when shown side-by-side to participants Each point shows a participant’s average preference for VV− (blue) or VV+ (orange) synthesized images (N = 128 participants; two-sided paired t-test Preference is defined as the percentage of times images of one category (VV+ or VV−) were chosen as more positive Boxplots show median ± 1.5 interquartile range overlaid by data points corresponding with images; * p < 0.05 VV+ and VV− images were also slightly different in terms of visual arousal such that the VV− set had larger visual arousal (mean VV positive set = 4.78 The effect of image set (VV+ vs VV−) on VV was intact when controlling for visual arousal (effect of being in VV− group: t(297) = −3.6 confirming that the two sets are distinguishable in terms of VV independent of visual arousal including only regions of the early visual system Even though the generative model was optimized for VV the synthesized VV+ and VV− images generated from early visual system were again significantly different in terms visual arousal (t(149) = −2.9 even if weakly embedded in early visual system is entangled with visual salience (arousal) while it is more distinguishable from salience in mid-level visual processing There was no significant interaction with whether a participant was instructed to indicate the more positive or negative image (z = 0.6 favoring that the effect of VV on choice originated from perceiving valence rather than the visual salience driving choices of both positivity and negativity a Distribution of the average neural objective function for the 1000 Alexnet classes The top 50 largest and top 50 smallest classes represent entities evoking the most similar visual activity to the VV+ and VV− masks b Hierarchical tree-structure semantic analysis of the labels of the top 50 positive (orange) and top 50 negative classes (blue) Horizontal gray bars represent the proportion of classes across the ImageNet’s 1000 class list Nodes are expanded as long as they correspond to at least 5% of the list (>20 classes) The number of classes that fall into each node of the tree is shown on arrows (blue with ‘−’ showing the count out of the lowest 50 and orange with ‘+’ showing the count out of the highest 50 classes) ‘Artifact’ versus ‘Living thing’ are the main nodes separating the two label sets these results confirm that neural activity in visual regions is not only correlated with VV but are sufficient for synthesizing VV back as visual inputs with distinct regions in the visual system differentially supporting positive and negative features We provide support for a distinct form of valence that arises from the process of visual perception Study 1 showed that VV accounts for a small but reliable amount of variance in NV Studies 2 and 3 supported a causal role of VV in subjective valence that is distinct from the conceptual aspects of NV; by manipulating conceptual content (from real to abstract) or display condition (from full to limited viewing) subjective valence shifted from association with NV towards VV These results support not only the causality of VV in subjective experience but also the context of behavioral relevance: use of VV increased under conditions that decreased reliance on conceptual analysis and being more reflexive than reflective Study 4 supported neural substrates of VV residing mainly in visual processing regions as opposed to neutral vision followed by conceptual associations While these results alone cannot prove visual neural origin of valence the visual neural correlates were further tested for causality in study 5 showing that synthesized VV+ and VV− images from the brain’s visual system were distinguishable by model predictions as well as human observers more than forming and relying on internal higher-level inferences while originating from real-world visual regularities associated with affective categories and concepts we hypothesize is exploited to shape visual processing in the brain in a distinct manner Just as texture gradients provide visual cues low or mid-level visual features can provide cues to positive or negative feelings particularly when observers have limited access to deep contents on which traditional valence depends The fact that VV was felt even in prolonged viewing of abstract paintings and reinforced in rapid viewing of realistic scenes suggests that valence is being directly computed from coarse analysis Just as the visual system has acquired mechanisms to detect objects rapidly it may directly estimate ‘positive’ and ‘negative’ based on the average visual features of all ‘good’ and ‘bad’ objects and scenes based on ecological statistics where observers learn the transitional probabilities from seeing to feeling supporting long-term rewiring of the visual system it has been thought that such learning alters the salience of associated physical features All participants provided informed consent prior to their participation in these studies We collected participants’ age and gender as reported in each study’s methodology participant demographics were not included as covariates in our analysis as they were not directly relevant to our main hypothesis Duplicate images were detected and excluded Average valence and arousal ratings of images from each dataset were rescaled from their original range to the 1-to-9 scale to be comparable across different datasets (1 indicating most negative and 9 indicating most positive) All analysis codes were implemented in Matlab 9.10.0 Colored images in the RGB space were transformed into HSV color map and average brightness across all pixels were estimated (3 features) which provides a mapping from each point in the RGB space into one of 11 basic colors as perceived and labeled by humans (black The result was 11 features indicating the amount of each color in an image Then the 2-D space was divided into 4 orientations by 4 lines having 0 Each orientation line was then divided into 20 equally spaced points on a logarithmic scale The amplitude in each point was interpolated Symmetry was estimated using the method proposed in ref. 21 based on filters from a Convolutional Neural Network (CNN) that take colors approximating the function of early visual processing in humans and left-right-up-down symmetry (3 features) We adopted a previous approach67 to assess the variances in the response of the first convolutional layer of a CNN This layer comprises 96 filters that detect edges at various frequencies and orientations or different color gradients similar to early visual processing in the brain each image was segmented into n×n equally sized subregions where n assumed 15 distinct values ranging from 2 to 30 (2 culminating in a total of 15 × 3 = 45 features: (1) Richness: The overall variance computed across all filters and subregions indicating the variability of features throughout the image subregions (2) Median Regional Variability: For each filter the variance is calculated across all subregions The median of these values is then taken across all filters reflecting how variance in filter responses differs from one subregion to another the variance is calculated across all filter responses The median of these variances across all subregions then quantifies the diversity in visual statistics within individual subregions These statistical variances provide a measure of both the uniformity and diversity in the visual characteristics of the image’s subregions when the image is segmented into subregions of different sizes (i.e. subregions of the grid get smaller while they would be higher in number) if an image contains a single object at the center with a solid background size of the object in the image would influence variances for different n’s (grid sizes) If an image represents a uniform texture with no clear bounded objects then the between region variances would be low independent of grid size n if the image or a subregion contains a rich and diverse set of visual features then the filters are expected to make diverse responses within that region leading to a high cross-filter variance within that region Three features were extracted related to faces in an image: number of faces size of the largest face relative to the image size total size of all faces relative to image size We used the MTCNN toolbox in Matlab to automatically detect faces and compute the value of these three features for each image We used the random-forest algorithm with 10-fold cross-validation to test the predictability of valence or arousal from each of the feature sets in the real-world image database A random forest is an ensemble of decision trees each trained on a subset of features The random forest’s prediction for a set of input features is then the average prediction of all trees this averaging prevents overfitting to the training data Random forest was implemented using Matlab’s regression tree ensembles with 500 ensemble trees We defined the prediction accuracy of a random forest or linear regression model as the Pearson correlation coefficient between the out-of-sample predictions and ground-truth values With a total of 100 participants (74 females; ages 18–65; mean = 39.87) each painting in the MART database had been rated by 20 individuals in terms of emotional valence on a 1 (highly negative) to 7 (highly positive) scale Raters did not see any information about the creator of the artwork or its title We estimated the average valence rating of each painting as the ground truth Valence ratings were rescaled to the 1–9 range to match the real photo database rating scale The entire stimulus selection procedure was conducted with a Matlab script A total of 180 individuals participated in the study (42 male; age mean = 21.7 Participants included paid Amazon Mechanical Turk workers (N = 25) and Cornell University undergraduate students participating for course credit (N = 155) Amazon Mechanical Turk workers were paid 1$ for rating each 56 images Inclusion criteria for Mechanical Turk workers were having done at least 100 tasks on the Mechanical Turk website with at least a 95% approval rate There were two valence rating conditions: Full-viewing and Limited-viewing Each participant was randomly assigned to either condition (between-subject design) an image was displayed and the participant had to determine whether the image induced more positive versus more negative feeling (2-alternative valence repsonse) the image was displayed until a response key was pressed with no time limit the image was displayed upside-down and only for 100 ms Participants were instructed to enter their responses with ‘A’ and ‘L’ keys on the keyboard corresponding to positive and negative valence All images were resized to have a width of 400 pixels while maintaining the original aspect ratio The 280 image stimuli were randomly divided into 5 blocks of 56 images (28 from the negative and 28 from the positive set) Each participant performed between 1 to 5 blocks Each stimulus was in total rated by 80 individuals in the Full-viewing and 80 individuals in the Limited-viewing condition All conditions and instructions were implemented in Qualtrics online platform (with embedded HTML/JavaScript) Participants performed the task in a remote online mode on a computer browser Mixed-effect regression was used to analyze the results to account for within and between individual variability in responses or response times In order to take mid and high-level conceptual categories into account the set of 280 image stimuli was rated in terms of presence of humans (binary) and representing outdoors (1 = completely indoors Two raters independently performed the ratings Internal consistency of the two ratings was confirmed (Cronbach’s alpha > 0.9 for all categories) To control for the potential effect of these categories on VV in study 3 the interaction of each of these with the viewing condition (limited versus full-viewing) was included in the mixed-effect logistic regression predicting valence response All analysis of study 3 was implemented in Python 3.11.3 and R 4.3.0 in the limited-viewing condition 1.6% of trials were timed out and dropped from further analysis Additional data was collected for 4 individuals following the same procedure resulting in a total of 20 participants (10 male; age mean = 26.1 An activation map was estimated on the un-smoothed data in response to the stimulus in each trial (using the canonical function in SPM) This resulted in one whole-brain t-map for each trial for each individual (20*128 total beta-maps) t-maps were normalized by subtracting the mean Both positive and negative clusters (voxels whose activity increased with either positivity or negativity) were considered as being sensitive to valence while clusters were either positive or negative A similar procedure was conducted to obtain whole-brain correlates of subjective normative valence Structures associated with significant clusters were named based on probability maps from the Harvard-Oxford Cortical Atlas and the Harvard-Oxford subcortical atlas We also performed a second group-level univariate analysis across the 8 subjects NSD27; see next section We first estimated the z-score associated with the correlation between VV estimated for each stimulus and the beta activation in response to it We then performed a one-sample t-test across the 8 subjects’ z-maps The voxel-level significance threshold was set at p < 0.001 and the minimum cluster size was estimated using AFNI’s 3dclustsim routine to correct for multiple comparisons at p < 0.05 ROI extraction: bilateral sLOC and bilateral LOC (superior+inferior) masks were obtained from the probabilistic Harvard-Oxford Cortical Structural Atlas in FSL, including voxels with at least 50% probability. mPFC ROI was created based on NueroSynth meta-analysis maps (https://neurosynth.org/) using the keyword ‘mPFC’ (association test To eliminate scattered voxels across the brain the largest spatially adjacent set of voxels in the neurosynth-generated mask was used as the mPFC ROI ROI and PPI analysis were performed using the mixed-effect linear regression models (‘lmer’ package in R) with fixed coefficients for regressors and random intercept to account for individual differences in baseline activations to compare the results (early visual regions included: V1v we first selected 150 image classes and then generated an optimal VV+ and VV− image within each class The class selection was done by first creating class-representative synthetic images generated from 100 random initializations with the truncation parameter set to 1 for each of the 1000 ImageNet classes Predicted brain activation of each random image was then used to estimate the corresponding objective value The 150 classes that achieved maximal variance of the objective were selected This approach would obtain classes that are capable of evoking both positive and negative VV responses since they span a wider range of objective values We then performed two optimizations for each selected class one maximizing the objective (VV+) and the other minimizing the objective (VV−) the class information was embedded as a one-hot code into the class vector and fixed and the noise vector was initialized from a truncated normal distribution and optimized The truncation parameter here was set to 0.4 to have a good balance of image fidelity and variety this procedure resulted in 150 (classes) x 2 (VV+ or VV−) = 300 images We also performed a between-class analysis examining the top 50 image classes whose random initializations led to the maximum average objective (positive classes) or the minimum average objective (negative classes) obtaining the hierarchical tree of categories associated with positive and negative VV in the brain This was done using the ntlk package in Python interfacing with WordNet lexical database The generated 150 image pairs were fed into VVM to estimate VV for within-class comparison They were also rated for their relative valence in a behavioral experiment the 150 image pairs were randomly divided into 3 blocks of 50 followed by the image pair displayed with 180° rotation and side by side (with random order) for 600 ms participants were instructed to indicate whether the left or right image was perceived as more positive they had to choose the more negative image The aim of this manipulation was to counterbalance any potential confound of salience on choice A trial was missed with a timeout message if the participant did not respond within 5 s Participants were recruited from the Amazon Mechanical Turk with similar inclusion criteria as study 3 Each block was completed by an average of N = 51.3 participants and the experiment included a total of 128 participants (N = 98 male An average of 3.8% of trials were missed due to timeout and excluded from the analysis trials with response times shorter than 100 ms (16% of trials) were considered anticipatory and excluded from the analysis Out of the 150 image pairs generated by the Neurogen 16 images were identified as containing large faces with distorted facial features (resembling facial injury or fictional ghosts) These faces were distorted primarily because of the generative model’s inadequacy in synthesizing realistic faces these images were excluded from the analysis due to the potential to elicit fear responses unrelated to the neural substrates of VV Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article Code includes a sample script for extracting visual features and predicting VV for use in future research The theory of constructed emotion: an active inference account of interoception and categorization The nature of feelings: evolutionary and neurobiological origins Toward an objective neural measurement of subjective feeling states Contour features predict valence and threat judgements in scenes Ganalyze: toward visual definitions of cognitive image properties IEEE/CVF International Conference on Computer Vision Cross-Cultural Universals of Affective Meaning The Ecological Approach to Visual Perception Global image properties predict ratings of affective pictures Common and stimulus-type-specific brain representations of negative affect Modality-general and modality-specific audiovisual valence processing Emotion schemas are embedded in the human visual system Modality general and modality specific coding of hedonic valence Representations of modality-specific affective processing for visual and auditory stimuli derived from functional magnetic resonance imaging data Building the gist of a scene: the role of global image features in recognition Dissociating affective and semantic valence From architecture to evolution: multisensory evidence of decentralized emotion Purely Perceptual Machines Robustly Predict Human Visual Arousal Using convolutional neural network filters to measure left-right mirror symmetry in images Brightness differences influence the evaluation of affective pictures In the eye of the beholder: employing statistical analysis and eye tracking for analyzing abstract paintings 20th ACM International Conference on Multimedia Population coding of affect across stimuli The brain basis of emotion: a meta-analytic review Meta‐analyses of object naming: effect of baseline A massive 7T fMRI dataset to bridge cognitive neuroscience and artificial intelligence NeuroGen: activation optimized image synthesis for discovery neuroscience Enhanced extrastriate visual response to bandpass spatial frequency filtered fearful faces: time course and topographic evoked‐potentials mapping Modulation of visual processing by attention and emotion: windows on causal interactions between human brain regions Reentry: a key mechanism for integration of brain function The time course of visual processing: from early perception to decision-making Coding of pleasant touch by unmyelinated afferents in humans The coding of valence and identity in the mammalian taste system Neural activity at the human olfactory epithelium reflects olfactory perception Emotional objectivity: neural representations of emotions and their interaction with cognition An ecological valence theory of human color preference Vision as Bayesian inference: analysis by synthesis Emotionally negative pictures enhance gist memory Neural mechanisms of incentive salience in naturalistic human vision Escape from harm: linking affective vision and motor responses during active avoidance Differential aversive learning enhances orientation discrimination Implicit and explicit memory for new associations in normal and amnesic subjects Feeling and thinking: preferences need no inferences Valid facial cues to cooperation and trust: male facial width and trustworthiness Aesthetic preference for art can be predicted from a mixture of low- and high-level visual features Is the preference of natural versus man-made scenes driven by bottom–up processing of the visual features of nature Opposing influences of affective state valence on visual cortical encoding EmoMadrid: an emotional pictures database for affect research EmoPics: subjektive und psychophysiologische evaluation neuen bildmaterials für die klinisch-bio-psychologische forschung The Geneva affective picture database (GAPED): a new 730-picture database focusing on valence and normative significance International affective picture system (IAPS): technical manual and affective ratings The Nencki Affective Picture System (NAPS): introduction to a novel Erotic subset for the Nencki Affective Picture System (NAPS ERO): cross-sexual comparison study Introducing the open affective standardized image set (OASIS) The Set of Fear Inducing Pictures (SFIP): development and validation in fearful and nonfearful individuals The Socio-Moral Image Database (SMID): a novel stimulus set for the study of social The role of low-level image features in the affective categorization of rapidly presented scenes Using CNN features to better understand what makes visual artworks special Learning color names from real-world images 2007 IEEE Conference on Computer Vision and Pattern Recognition Low-level cues and ultra-fast face detection How do amplitude spectra influence rapid animal detection Equitable thresholding and clustering: a novel method for functional magnetic resonance imaging clustering in AFNI Human brain responses are modulated when exposed to optimized natural images or synthetically generated images Brock, A., Donahue, J. & Simonyan, K. Large scale GAN training for high fidelity natural image synthesis. In International Conference on Learning Representations. https://openreview.net/forum?id=B1xsqj09Fm (2019) Personalized visual encoding model construction with small data Very deep convolutional networks for large-scale image recognition International Conference on Learning Representations (2014) Neural system identification for large populations separating “what” and “where” saeedeh. saeedeh/Visual-Valence-Model: v1. Zenodo https://doi.org/10.5281/zenodo.13345696 (2024) Sadeghi, S. saeedeh/sensoryValence-onlineStudy: v1. Zenodo https://doi.org/10.5281/zenodo.13345691 (2024) Download references and Junichi Chikazoe for their valuable insights and helpful feedback throughout the course of this research Division of Humanities and Social Sciences School of Electrical and Computer Engineering conceptualized the project and performed data analysis and modeling wrote the manuscript with input from all authors Nature Communications thanks Judith Domínguez-Borràs Steven Scholte and the other anonymous reviewer(s) for their contribution to the peer review of this work Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations Reprints and permissions Download citation DOI: https://doi.org/10.1038/s41467-024-53668-6 Metrics details Molecules that can reversibly switch between electronic states under an external stimulus are of interest to numerous applications Complexes of open shell metal ions with redox active ligands undergo valence tautomerism the precise structural changes occurring during valence tautomerism in the solid state are unclear due to the lack of atomic-resolution characterization variable temperature and high-pressure single crystal X-ray diffraction is used to characterize valence tautomerism in two isostructural cobalt complexes [Co(bis(6-methyl-2-pyridylmethyl) amine)(3,5-di-tert-butyl-1,2-dioxolene)]+ and [Co(tris(6-methyl-2-pyridylmethyl)amine)(3,5-di-tert-butyl-1,2-dioxolene)]+ to atomic resolution The less sterically hindered dimethylated complex exhibits two-step thermally-induced interconversion between the high-spin CoII-seminquinonate and low-spin CoIII-catecholate forms (valence tautomerism) at 155 and 95 K due to the presence of two symmetry-independent complexes the more sterically hindered trimethylated complex does not display thermal valence tautomerism Both complexes exhibit unique behaviour under high pressure The dimethylated species undergoes gradual one-step valence tautomerism in both symmetry-independent complexes concurrently between 0.43 GPa and 1.30 GPa pressure is sufficient to overcome steric hindrance leading to one-step valence tautomerism between 2.60 GPa and 3.10 GPa demonstrating pressure-triggered valence tautomerism in a thermally inactive complex This study is among the few investigations using in situ high-pressure single crystal X-ray diffraction to achieve atomic-level structural characterization of valence tautomerism aiding the development of robust structure-property relationships in these types of complexes a Molecular diagram of [Co(Mentpa)(dbdiox)]+ (where n = 2 (Me2tpa) or 3 (Me3tpa) (shaded blue)) as the LS-CoIII-Cat valence tautomer (shaded red) with a photograph of the red single crystal inset (a) or the HS-CoII-SQ tautomer (shaded green) with a photograph of the green single crystal inset (b Molecular structure of [Co(Mentpa)(dbdiox)]+ The yellow sphere represents either a hydrogen atom in [Co(Me2tpa)(dbdiox)]+ or a methyl group in [Co(Me3tpa)(dbdiox)]+ In Me2tpa the Me2tpa ligand is disordered over two binding configurations in a 60:40 ratio (not shown) Atoms are coloured as follows: Co – dark blue d Packed structure of [Co(Mentpa)(dbdiox)]+ viewed along the a-axis showing alternating layers of complex molecules (stick model) and solvate (space-filled model) The lighter and darker coloured molecules distinguish the two symmetry-independent molecules e Intermolecular CH∙∙∙π and π∙∙∙π interactions (labelled I − XIII) within each layer of molecules in Me2tpa and Me3tpa Molecules A and B are coloured with light and dark carbon atoms One t-butyl group on one of the symmetry-independent molecules (A) is also rotationally disordered in both Me2tpa and Me3tpa under ambient conditions have average Co–O/N lengths of 2.1311(21) Å and 2.1263(21) Å and average C–O lengths of 1.286(7) Å and 1.277(7) Å A and B in [Me3tpa] have average Co–O/N lengths of 2.1576(15) Å and 2.1551(16) Å and average C–O lengths of 1.274(6) Å and 1.278(5) Å a Plot of the relative change in unit cell dimensions of [Me2tpa] (circles) and [Me3tpa] (squares) at variable temperature and pressure and C–O bond lengths (shades of orange) in [Me2tpa] (circles) and [Me3tpa] (squares) for each symmetry-independent molecule Black horizontal lines represent the expected values of the Co–O/N and C–O bond lengths for the HS-CoII-SQ tautomer and the red horizontal lines represent the expected values for the LS-CoIII-Cat tautomer c Photographs of [Me2tpa] (upper) and [Me3tpa] (lower) single crystals during compression in a modified Merrill-Bassett diamond anvil cell showing a change in colour from green HS-CoII-SQ to red LS-CoIII-Cat It is likely that these factors lead to valence tautomerism occurring in B at a higher temperature (155 K) than A (95 K) [Me2tpa] also becomes more disordered upon cooling from 298 K to 155 K with the t-butyl group in A becoming rotationally disordered at 155 K followed by the same disorder occurring in B at 95 K This order to disorder transition upon cooling coincides with A and B moving closer together in the asymmetric unit (by ~0.3 Å) between 298 and 95 K The reasoning for this subtle crystallographic behaviour and whether it is linked to valence tautomerism needs further investigation consistent with formation of the LS-CoIII-Cat state in both molecules octahedral distortion values and colour of the single-crystal sample remain relatively constant between 1.30 GPa and 1.90 GPa indicating the cessation of any valence tautomeric interconversion at 1.30 GPa the MOS appears to change almost immediately on application of pressure to 0.14 GPa and increases to a minimum of approximately –2.5 This discrepancy is likely an artifact of the large error on C–C and C–O bond lengths as a result of the low-completeness of the data which is not unusual for high-pressure single-crystal data the pressure-induced valence tautomerism appears to occur simultaneously in both symmetry-independent molecules we cannot rule out a two-step process at pressure as smaller pressure increments may be necessary to observe this which is near impossible in a diamond anvil cell The structural effect of valence tautomerism on the crystal of [Me2tpa] can be followed by evolution of the unit cell dimensions during compression The average Co–O/N lengths shortened to 1.972(9) Å and 2.008(9) and the average C–O bonds lengthened to 1.405(42) Å and 1.39(3) at 3.10 GPa for A and B respectively characteristic of the LS-CoIII-Cat tautomer of both complexes dramatically decreases from ~100° to ~60° during valence tautomerism These changes in molecular geometry are in line with those seen in [Me2tpa] because of valence tautomerism indicating that the conversion has progressed to completion Despite the lack of data of atomic resolution obtained after 3.10 GPa this cessation of valence tautomerism at 3.10 GPa is supported by the decrease in compressibility of the unit cell volume seen between 3.10 GPa and 4.40 GPa The green colour is returned upon full pressure release this complex does not exhibit thermally-induced valence tautomerism due to the additional steric bulk of the trimethylated tpa ligand high-pressure overcomes the steric barrier enabling access to the lower-volume LS-CoIII-Cat tautomer a Plot of the tpa-Δ in [Me3tpa] (squares solid lines) at variable temperature and pressure Values for both symmetry-independent complexes with vertices on the methyl-bearing carbon atom Vertical error bars are calculated from the estimated standard deviations on tpa-Δ from the crystal structures b Plot of relative spin-state energy (ΔE) vs applied stimuli for [Me2tpa] (circles) and [Me3tpa] (squares) for each symmetry-independent molecule On undergoing valence tautomerism, the tpa-Δ dramatically decreased due to the reduction in the entire molecular volume, as indicated in Fig. 3a for [Me2tpa] on cooling to 95 K or increasing pressure to ~1 GPa with the tpa-Δ measuring ~7.3 Å2 under both conditions causing the tpa-Δ to decrease by ~0.4 Å2 until 2.6 GPa at which the tpa-Δ for [Me3tpa] reached approximately the same value as [Me2tpa] under ambient conditions ( ~ 8.2 Å2 for both molecules) Further compression of [Me3tpa] to 3.10 GPa resulted in contraction of the tpa-Δ in both independent molecules to ~7.3 Å2 This demonstrates the capacity for pressure to facilitate valence tautomerism in [Me3tpa] through compression of the tpa ligand overcoming the increased steric bulk due to the additional methyl group Contraction of the unit cell volume in [Me2tpa] and [Me3tpa] under pressure and during valence tautomerism derives from compression of both the molecular and supramolecular structures Compression of the supramolecular structure is facilitated by shortening of intermolecular π•••π and CH•••π distances by between 6.7% to 15.4% (average = −9.7%) and 6.8% to 7.6% (average = −7.3%) between ambient pressure and 1.90 GPa in [Me2tpa] average = −10.4%) between ambient pressure and 3.10 GPa in [Me3tpa] Upon valence tautomerism in both [Me2tpa] and [Me3tpa] many of these interactions show substantial changes in their compressibility This is most apparent in the π•••π interactions; two of the interactions show a substantial increase in compressibility whereas there is a negligible increase in compressibility for interactions I and III 44% of the compression of interaction I occurs during valence tautomerism (0.43 GPa–1.30 GPa) compared to 78% for interaction II This is substantially more pronounced in [Me3tpa] where only 7.7% of the compression of interaction I occurs during valence tautomerism (2.6 GPa–3.10 GPa) compared to 64% for interaction II Similar behaviour is seen across the CH•••π interactions with the majority becoming more compressible upon valence tautomerism X and XIII expand substantially during valence tautomerism in both [Me2tpa] and [Me3tpa] The vast differences in behaviour of these interactions upon valence tautomerism highlights the strong relationship between the molecular rearrangement caused by valence tautomerism and the supramolecular structure To fully understand whether the changes in these interactions aid or hinder the stabilisation of the LS-CoIII-Cat tautomer and the cooperativity of the conversion a hierarchy of energies for each of the intermolecular interactions for each pressure would be needed; a notable point of future work The calculated ΔE with varying pressure highlights that compression assists in surpassing the spin-state energy requirements. As illustrated in Fig. 3b the ground state of [Me2tpa] and [Me3tpa] is the HS-CoII-SQ form (ΔE > 0) Elevating pressure results in a progressive reduction of ΔE to reach equilibrium (ΔE = 0) Further increases in pressure caused ΔE to be negative favouring the formation of the LS-CoIII-Cat tautomer The correlation of relative spin-state energy at varying pressure for both A and B for [Me2tpa] and [Me3tpa] aligns effectively with the tpa-Δ as well as Co-O/N bond distances The energy requirement for interconversion of [Me3tpa] is not met until 3.10 GPa Increasing the pressure causes both A and B to undergo concurrent switching between 2.60 GPa and 3.10 GPa (ΔE = +27 kcal mol−1 at 2.60 GPa for A and B to ΔE = − 31.1 kcal mol−1 for A and ΔE = − 26.3 kcal mol−1 for B at 3.10 GPa) A and B of [Me2tpa] display a simultaneous transition but at relatively smaller applied pressure between 0.43 GPa and 1.3 GPa with a critical pressure of 0.7 GPa (ΔE = 0) [Me2tpa] exhibits a well resolved two-step process wherein the transition of B precedes that of A The calculated switching points and transition features for both [Me2tpa] and [Me3tpa] align remarkably well with both pressure-induced and thermally-induced valence tautomeric traits seen in experimental measurements Here we report pressure-induced valence tautomerism in two cobalt-dioxolene complexes using a combination of single-crystal X-ray diffraction high-pressure diamond anvil cell techniques and density functional theory calculations [Me2tpa] undergoes a two-step transition from the HS-CoII-SQ form to the LS-CoIII-Cat form upon cooling due to the presence of two symmetry-independent molecules [Me3tpa] does not exhibit thermally induced valence tautomerism due to increased steric bulk Applying pressure facilitates valence tautomerism in both complexes [Me2tpa] undergoes a single-step transition at a lower pressure (0.43 GPa−1.30 GPa) compared to [Me3tpa] (2.60 GPa−3.10 GPa) due to its less sterically hindered tpa ligand Valence tautomerism leads to a decrease in unit cell volume and changes in intermolecular interactions These changes highlight the interplay between the molecular rearrangement and the surrounding supramolecular structure while DFT calculations support the experimental observations demonstrating that pressure progressively reduces the energy barrier for the transition to the LS-CoIII-Cat form Such insights are necessary to understand structure-property relationships in VT complexes highlighting the importance of using both variable temperature and high-pressure single crystal X-ray diffraction to elucidate atomic-resolution structural changes during VT it is notable that valence tautomerism in [Me2tpa] can be induced by thermal paving the way for applications that require multi-stimuli responses we aim to investigate the individual energies of intermolecular interactions to understand their role in stabilising the LS-CoIII-Cat tautomer We also aim to further develop methods to accurately simulate pressure effects on quantum-chemically obtained geometries this study provides valuable insights into the interplay between pressure and steric effects in controlling valence tautomerism in cobalt-dioxolene complexes This knowledge could be crucial for designing future functional materials with desired properties Synthesis of the complexes was performed using standard Schlenk line techniques under a nitrogen atmosphere All chemicals used were of reagent grade and used as received Methanol was dried using 3 Å sieves for three days and degassed through a minimum of ten cycles of vigorous shaking under vacuum and subsequent backfilling with nitrogen The 3 Å sieves were activated by heating at 300 °C for 48 h and cooling under vacuum This ligand was synthesized according to modified literature procedure33 8.3 mmol) and 6-methyl-2-pyridine carboxaldehyde (2.00 g 16.5 mmol) were added to a stirred dichloromethane (200 mL) solution of sodium triacetoxyborohydride (5.24 g The reaction mixture was covered with foil and left to stir for 48 h A saturated aqueous solution of sodium hydrogencarbonate (2.08 g 24.7 mmol) was added and the resulting solution was stirred for another 30 min The organic phase was extracted with dichloromethane (20 mL\(\,\times \,\)3) and dried over MgSO4 to give a dark yellow oil The oil residue was extracted several times with hexane and removal of solvent afforded colourless needles of bis((6-methyl-2-pyridyl)methyl)(2-pyridylmethyl)amine in 90–95% yield The synthesized organic ligand Me2tpa (64 mg 0.20 mmol) was added to a dried and degassed methanol solution (10 ml) of CoCl2.6H2O (48 mg The resulting green solution was left stirring for 10 min under a nitrogen atmosphere A dried and degassed methanol solution (10 ml) containing 3,5-di-tert-butylcatechol (37 mg 0.20 mmol) was deprotonated with triethylamine (55 μL The deprotonated 3,5-di-tert-butylcatechol solution was slowly added to the cobalt-Me2tpa mixture and stirred for 10 min to obtain a reddish-brown solution The resulting solution was subjected to aerial oxidation followed by the addition of an aqueous KPF6 solution to afford a black-green precipitate of [Co(Me2tpa)(dbdiox)][PF6] The crude product was washed with diethyl ether (2 \(\times\) 2 mL) and collected by vacuum filtration under ambient conditions Large dark-green crystals were obtained by heating a moderately concentrated solution of toluene in a sand bath with subsequent slow cooling affording [Co(Me2tpa)(dbdiox)][PF6].(toluene) Excess concentration resulted in co-crystallization of byproducts The synthesised organic ligand Me3tpa (67 mg The resulting purple solution was left stirring for 10 minutes under a nitrogen atmosphere The resulting solution was subjected to aerial oxidation followed by the addition of an aqueous KPF6 solution to obtain a black-green precipitate of [Co(Me3tpa)(dbdiox)][PF6] Large dark-green crystals were obtained by heating a moderately concentrated toluene solution in a sand bath with subsequent slow cooling affording Co(Me3tpa)(dbdiox)][PF6].(toluene) Hydrogen atoms were placed geometrically and constrained to ride on their host atoms The toluene C-atoms were refined using only isotropic thermal parameters because the data were collected at room temperature (to compare to the high-pressure data) The [PF6]− counterions were also modelled as rigid fragments and with thermal similarity restraints Absorption effects were corrected using an empirical multi-scan correction using spherical harmonics in the same program Toluene solvate and [PF6]− counterions were modelled as rigid fragments using a mask based on the 40° half-opening angle of the pressure cell to remove the shaded regions of the diffraction frames The structure at each pressure was refined from the starting coordinates of the ambient pressure data The structures were refined in an almost identical fashion to the ambient pressure data except all t-butyl groups pyridylmethyl-amine groups were refined as rigid bodies constructed from the low-temperature coordinates Only the dioxolene fragment was allowed to refine freely P and F-atoms were refined with anisotropic displacement parameters all other non-H atoms were refined with isotropic thermal parameters Thermal similarity restraints were applied for all non-H atoms This parameterisation strategy was used to reduce the number of refined parameters due to the low completeness of the high-pressure data and the large number of independent atoms (Z’ = 2 On increasing pressure to 3.1 GPa in [Me3tpa] a solvent mask was used to model the two toluene solvent molecules The solvent calculated equated to 340 electrons in a volume of 1048 Å3 located in 1 void per unit cell This was consistent with the presence of one toluene molecule per asymmetric unit (which accounts for 400 electrons per unit cell) The one disordered t-butyl group also became ordered at 3.1 GPa All other refinement parametrisation was kept the same Crystallographic Information Files are available from the Cambridge Crystallographic Data Centre repository under accession codes: 2353216-2353234. Copies of the data can be obtained free of charge by visiting https://www.ccdc.cam.ac.uk/structures/ UK (fax +441223336033; deposit@ccdc.cam.ac.uk) structural data and computational data are provided in the Supplementary Information file All additional data are available from the corresponding author upon request Manipulating metal-to-metal charge transfer for materials with switchable functionality Switchable molecule-based materials for micro- and nanoscale actuating applications: achievements and prospects Studies on charge distribution and valence tautomerism in transition metal complexes of catecholate and semiquinonate ligands Valence tautomerism in metal complexes: stimulated and reversible intramolecular electron transfer between metal centers and organic ligands Tuning the charge distribution and photoswitchable properties of cobalt–dioxolene complexes by using molecular techniques Pressure- and temperature-induced valence tautomeric interconversion in a o-Dioxolene adduct of a cobalt–tetraazamacrocycle complex Spin fluctuations in the light-induced high-spin state of cobalt valence tautomers The effects of pressure on valence tautomeric transitions of dinuclear cobalt complexes Pressure-induced valence tautomerism in cobalt o-Quinone complexes: an X-ray absorption study of the low-spin [CoIII(3,5-DTBSQ)(3,5-DTBCat)(phen)] to High-Spin [CoII(3,5-DTBSQ)2(phen)] Interconversion Beni, A., Dei, A., Rizzitano, M. & Sorace, L. Unprecedented optically induced long-lived intramolecular electron transfer in cobalt–dioxolene complexes. Chem. Commun, 2160–2162 https://doi.org/10.1039/B702932B (2007) Complete direct and reverse optically induced valence tautomeric interconversion in a cobalt–dioxolene complex Single-ion anisotropy and exchange coupling in cobalt(II)-radical complexes: insights from magnetic and ab initio studies DFT prediction and experimental investigation of valence tautomerism in cobalt-dioxolene complexes A DFT computational study of the magnetic behaviour of cobalt dioxolene complexes of tetraazamacrocyclic ligands Metrical oxidation states of 2-amidophenoxide and catecholate ligands: structural signatures of metal–ligand π bonding in potentially noninnocent ligands Spin state switching in iron coordination compounds OctaDist: a tool for calculating distortion parameters in spin crossover and coordination complexes Regulation of multistep spin crossover across multiple stimuli in a 2-D framework material Two-step spin-crossover phenomenon under high pressure in the coordination colymer Fe(3-methylpyridine)2[Ni(CN)4] EosFit7-GUI: a new graphical user interface for equation of state calculations Compressibility and high-pressure phase transition of a metalloporphyrin: (5,10,15,20-tetraphenyl-12H,23H-porphinato)cobalt(II) Antagonism between extreme negative linear compression and spin crossover in [Fe(dpp)2(NCS)2]⋅py High pressure study of Ru3(CO)12 by X-ray diffraction Predicting valence tautomerism in diverse cobalt–dioxolene complexes: elucidation of the role of ligands and solvent A convenient DFT-based strategy for predicting transition temperatures of valence tautomeric molecular switches All-electron scalar relativistic basis sets for third-row transition metal atoms Caldeweyher, E. et al. A generally applicable atomic-charge dependent London dispersion correction. J. Chem. Phys. 150 https://doi.org/10.1063/1.5090222 (2019) Relativistic regular two-component hamiltonians triple zeta valence and quadruple zeta valence quality for H to Rn: Design and Assessment of Accuracy Zhao, Y. & Truhlar, D. G. A New Local Density Functional for Nain-Group Thermochemistry, Transition Metal Bonding, Thermochemical kinetics, and Noncovalent Interactions. J. Chem. Phys. 125 https://doi.org/10.1063/1.2370993 (2006) Stauch, T. A mechanochemical model for the simulation of molecules and molecular crystals under hydrostatic pressure. J. Chem. Phys. 153 https://doi.org/10.1063/5.0024671 (2020) Quantum chemical modeling of pressure-induced spin crossover in octahedral metal-ligand complexes Tuning the charge distribution and photoswitchable properties of cobalt-dioxolene complexes by using molecular techniques SHELXT– integrated space‐group and crystal‐structure determination and Ag and calibration of the ruby R1 fluorescence pressure gauge from 0.06 to 1 Mbar Calibration of the ruby pressure gauge to 800 kbar under quasi‐hydrostatic conditions Neese, F. Software Update: the ORCA Program System, version 4.0. Wires Comput. Mol. Sci. 8, https://doi.org/10.1002/wcms.1327 (2018) Download references acknowledges the Australian Government and the University of Western Australia for the provision of a Research Training Program Scholarship acknowledges the Melbourne Research Scholarship is grateful for generous allocations of computational resources from the National Computational Infrastructure (NCI) Facility within the National Computational Merit Allocation Scheme (project fk5) and Research Platform Services (ResPlat) at The University of Melbourne (project punim0094) acknowledges the Australian Research Council (ARC) for the provision of a Future Fellowship (FT200100243) and a Discovery Project (DP220103690) acknowledges the ARC for funding a Discovery Project (DP19010085) Hay, Lars Goerigk & Colette Boskovic Alan Riboldi-Tunnicliffe, Rachel Williamson & Stephanie Bird and all authors offered further analysis and edited the manuscript Nature Communications thanks Jun Tao and the other Download citation DOI: https://doi.org/10.1038/s41467-024-53311-4 Metrics details Quantum-confined nanoclusters can be described by the jellium model which emphasizes closed-shell electron configurations but an open-shell variation with jellium aromaticity has been proposed Such clusters are termed superatoms because they behave like an atom Superatoms feature metal–metal bonding; hence superatoms have exclusively involved main group or transition metals with actinides only considered computationally as dopants owing to actinide–actinide bonding being exceedingly rare Here we report trithorium nanoclusters exhibiting three-centre-one-electron actinide–actinide bonding Experimental and computational analysis demonstrates Robin–Day Class III 6d-orbital valence delocalization in these clusters but in external applied magnetic fields they exhibit exalted diamagnetism evidencing actinide open-shell jellium aromaticity superatom character Exalted diamagnetism is not normally associated with a single unpaired electron the valence delocalization enables exalted diamagnetism we report the preparation and isolation of the crystalline mixed-valence trithorium nanoclusters [M(2.2.2-cryptand)][{(η8-C8H8)Th(μ-Cl)2}3] (4M 4Cs) that exhibit three-centre-one-electron actinide–actinide bonding Experimental and computational analysis demonstrates Robin–Day Class III formalism 6d-orbital valence delocalization in these clusters which should be the dominant physicochemical behaviour they instead exhibit exalted diamagnetic responses experimentally evidencing actinide superatom and open-shell jellium aromaticity In previous work (ref. 21) which contains a three-centre-two-electron trithorium bonding interaction treatment of 1 and 2.2.2-cryptand with MC8 reducing agents affords complexes 4M (M = K The fate of the excess M and 2.2.2-cryptand component was not determined but with the Cs+ cation replaced by K+ or Rb+ cations singly occupied molecular orbital of the [{(η8-C8H8)Th(μ-Cl)2}3]– anion component of 4M (4′) doubly occupied molecular orbital of the model [{Th(η8-C8H8)(μ3-Cl)2}3(K)2] of 3 (3′) with potassium cations omitted for clarity The molecular orbitals are plotted at the 0.05 isosurface level and the molecular graphs are plotted with critical point electron density thresholds of 0.01 a.u showing bond paths (black lines) and bond critical points (blue dots) Source data suggesting more extensive delocalization in 6d than 5d trimetallo clusters DFT EPR calculations on 4′ predict gx = 1.955 gy = 1.958 (g⊥ = 1.957) and gz (g||) = 1.911 in good agreement with experiment and correctly returning g⊥ > g|| All the experimental and computed g-values correspond to effective magnetic moments of 1.7 μB for 4M consistent with the theoretical value of 1.73 μB for an S = 1/2 d1 ion (assuming g = 2) Cooling will therefore primarily shorten and strengthen the Th–COT and Th–Cl bonds the g-values are a direct reporter of excited states that are mixed into the SOMO by spin–orbit coupling under the influence of an applied magnetic field and the SOMO is Th–Th bonding character and largely ‘blind’ to the Th–COT and Th–Cl interactions the EPR data are consistent with the presence of Th–Th bonding although of course it should be acknowledged that many factors can influence g-values we find a net molar diamagnetic susceptibility of −731 × 10−6 cm3 mol−1 (R2 = 0.9998) for 5 that is within 4.8% of the value predicted from Pascal’s constants (−697 × 10−6 cm3 mol−1) and a magnetization per thorium atom of only −16.9 cm3 Gauss mol−1 at 7 Tesla thus confirming that the anomalous diamagnetism of 4M does not just result from having three thorium ions in close proximity GIMIC plot of 4′ showing diatropic (red) and paratropic (blue) components GIMIC plot of 4′ showing only the diatropic (red) components GIMIC plot of 6 showing only the diatropic (red) components Induced current density streamlines for 4′ The AICD plots are rendered at the 0.015 a.u isosurface level and the GIMIC representations are plotted with modulus of j(r) isovalues of 0.025 a.u for diatropic (red) and paratropic (blue) regions The streamlines vector magnitude colour scale bar is in a.u The j(r) of 4′ produce a complex interplay of paratropic and diatropic currents across topologically diverse 4′. The overall j(r) topology of the Th3Cl6 unit is diatropic, and that is bounded by a paratropic region and then the outer sides of the C8H8 rings provide an additional layer of diatropism (Fig. 5c) probing local j(r) effects is deterministic being conditional on integration box selection we computed the net isotropic magnetic susceptibility of 4Cs While the computed value of −729.59 × 10−6 cm3 mol−1 for 4Cs is approximately half the experimental value of 4Cs given that these magnetic susceptibility calculations are limited to scalar (spin–orbit neglected) relativistic effective core potentials and are computed as the derivative of the energy in the limit of zero magnetic field that the calculation returns a substantially negative (diamagnetic) rather than expected positive (paramagnetic) susceptibility for an S = 1/2 molecule is striking especially when considering that diamagnetism is normally weak and paramagnetism is normally a dominant effect the analogous value for hypothetical diamagnetic S = 0 [M(2.2.2-cryptand)]+[6] of −557.65 × 10−6 cm3 mol−1 is less diamagnetic than the value computed for S = 1/2 4Cs emphasizing the otherwise counterintuitive increase in diamagnetic response when an electron is added to the Th3Cl6 core which is consistent with the SOMO of 4′ having the overall appearance of an s orbital This suggests that f-element metal–metal bonding through the superatom concept shares underpinning analogies to other element groups and is thus poised for further elaboration in the periodic table’s ‘third dimension’ Super atomic clusters: design rules and potential for building blocks of materials Electronic shell structure and abundances of sodium clusters A structural jellium model of cluster electronic structure Open-shell jellium aromaticity in metal clusters Formation of Al13I−: evidence for the superhalogen character of Al13 Al cluster superatoms as halogens in polyhalides and as alkaline earths in iodide salts Anomalous diamagnetic susceptibility in 13-atom platinum nanocluster superatoms Evolution of triangular all-metal aromatic complexes from bonding quandaries to powerful catalytic platforms Localization-vs-delocalization of 5f orbitals in superatom systems Hund’s three rules in actinide-containing superatoms with spin–orbit coupling calculations Observing the role of electron delocalization in electronic transport by incorporating actinides into ligated metal-chalcogenide superatoms Mass-spectroscopic investigation of stability of gaseous U2O2 and U2 Detection of the thorium dimer via two-dimensional fluorescence spectroscopy Souter, P. F., Kushto, G. P. & Andrews, L. IR spectra of uranium hydride molecules isolated in solid argon. Chem. Commun. https://doi.org/10.1039/CC9960002401 (1996) Experimental and theoretical evidence for the formation of several uranium hydride molecules U2@Ih(7)-C80: crystallographic characterization of a long-sought dimetallic actinide endohedral fullerene Characterization of a strong covalent Th3+–Th3+ bond inside an Ih(7)-C80 fullerene cage A crystalline tri-thorium cluster with σ-aromatic metal–metal bonding Fluoride clusterfullerenes: tuning metal–metal bonding and magnetic properties via single fluorine atom doping Half-sandwich cyclooctatetraenethorium compounds Rare-earth cyclobutadienyl sandwich complexes: synthesis and triple-bonded molecules and tetrahedrally bonded crystals: a summary Synthesis and characterization of the first sandwich complex of trivalent thorium: a structural comparison with the uranium analogue Crystal and molecular structure of bis(cyclooctatetraenyl)uranium(IV) Optical transitions of symmetrical mixed-valence systems in the Class II–III transition regime Current trends and future challenges in the experimental theoretical and computational analysis of intervalence charge transfer (IVCT) transitions A trinuclear gadolinium cluster with a three-center one-electron bond and an S = 11 ground state an actinide compound with a 6d1 ground state properties and structures of the tris(cyclopentadienyl)thorium(III) complexes [Th(III){η5-C5H3(SiMe2R)2-1,3}3] (R = Me or tBu) Importance of energy level matching for bonding in Th3+–Am3+ actinide metallocene amidinates and reactivity of the sterically crowded Th3+ complex (C5Me5)3Th including formation of the thorium carbonyl Actinide covalency measured by pulsed electron paramagnetic resonance spectroscopy Chemical structure and bonding in a thorium(III)–aluminum heterobimetallic complex J.Isolation of a square-planar Th(III) complex: synthesis and structure of [Th(OC6H2tBu2-2,6-Me-4)4]1− Replacing trimethylsilyl with triisopropylsilyl provides crystalline (C5H4SiR3)3Th complexes of Th(III) and Th(II) Investigating steric and electronic effects in the synthesis of square planar 6d1 Th(III) complexes Electron spin resonance studies of the cyclooctatetraenyl anions The cyclooctatriene-η2-ynyl potassium zwitterionic radical: evidence for a potassium organometallic Metal–metal bonding and antibonding molecular orbitals in triangular skeletons of the type L6M3 Diamagnetic corrections and Pascal’s constants Anomalous magnetism of small metallic clusters in a weak magnetic field in The Chemical Bond: Fundamental Aspects of Chemical Bonding (eds Frenking Electronic structure of small metal clusters: Thomas–Fermi statistical theory Substrate-free copper nanoclusters exhibit super diamagnetism and surface based soft ferromagnetism Superatom chemistry: promising properties of near-spherical noble metal clusters Spherical aromaticity in Ih symmetrical fullerenes: the 2(N+1)2 rule The magnetic susceptibility of the alkali metals Download references the Chinese Scholarship Council for a PhD Scholarship (202206370015 The University of Manchester for a President’s Doctoral Scholarship (J.T and N.K.) and computational resources and associated support services from the Computational Shared Facility and the EPSRC UK EPR National Research Facility (EP/W014521/1 and EP/V035231/1 The Alexander von Humboldt Foundation is thanked for a Friedrich Wilhelm Bessel Research Award (S.T.L.) University of Manchester) for obtaining elemental analysis data on 4M and 5 Department of Chemistry and Photon Science Institute prepared and characterized the compounds and analysed the results conducted and analysed the quantum chemical calculations conducted and analysed the EPR spectroscopy refined and analysed the single-crystal structure data refined and analysed the powder crystal structure data analysed all the data and wrote the paper with contributions from all the authors Nature Chemistry thanks Xuhui Lin and the other Supplementary Materials and Methods (general experimental details Supplementary Text (additional discussional points) Crystallographic data for 4K; CCDC number 2374489 Crystallographic data for 4Rb; CCDC number 2374490 Crystallographic data for 4Cs; CCDC number 2374491 Crystallographic data for 5; CCDC number 2376515 Download citation DOI: https://doi.org/10.1038/s41557-025-01790-3 Metrics details The valence of stimuli is shaped by various factors the mechanisms behind this flexibility remain elusive can elicit opposite chemotaxis responses – attraction vs aversion – depending on NaCl concentration demonstrating the role of environmental factors in altering valence orchestrate this bidirectional ethanol chemotaxis by integrating information from both stimuli – ethanol and NaCl – into its neuronal activity dynamics different calcium dynamics in the ASER neuron differentially activate the signaling molecule CMK-1 thereby engaging different downstream interneurons and leading to opposite chemotaxis directions optogenetic manipulations of the ASER neuron reverse the chemotaxis directions Our findings reveal a mechanism by which a single neuron integrates multisensory inputs to determine context-dependent behavioral valence contributing to our current understanding of valence encoding Elucidating the precise molecular and cellular mechanisms underlying this flexibility in valence assignment remains a significant challenge Despite the importance of this valence reshaping for animal adaptation and survival the causal relationship between these neural alterations and the subsequent impact on valence changes remains to be established we uncover the mechanism behind these controversial reports as a context-dependent reversal of ethanol odor valence in C elegans exhibits robust attraction to ethanol when the NaCl concentration on the assay plate is lower than that of its cultivation environment when exposed to NaCl concentrations higher than those in its cultivation medium the worms are strongly repelled by ethanol is capable of sensing and driving both attraction and aversion behaviors to ethanol Utilizing optogenetic methods to directly activate or inhibit neurons we identified that the key determinant of this valence reversal is the level of calcium in the ASER neuron the initial calcium influx triggered by this switch activates calcium/calmodulin-dependent protein kinase-1 (CMK-1) the ethanol signal results in attraction via the downstream interneuron AIZ when switching to higher salt environments leading to ethanol avoidance through a distinct downstream interneuron Our findings highlight the remarkable flexibility of a single sensory neuron in generating opposite valence behaviors Through dynamic regulation of its calcium signaling the ASER neuron instantaneously integrates the salt context and ethanol odor enabling context-dependent behavioral responses without synaptic remodeling This study demonstrates how sensory information from one modality can profoundly and instantaneously shape an organism’s valence perception and response to another modality thereby contributing to our current understanding of valence modulation mechanisms We noticed that a key difference between these studies might be the salt concentration in the assay plates The first study employed a minimal buffer solution (5 mM KPO4 while the second study used nematode growth medium (NGM) which contains a higher salt concentration (50 mM NaCl suggesting that the NaCl concentration in the growth plates serves as a determining factor for this turning point this setup was utilized for all following chemotaxis assays suggesting avoidance to ethanol at higher NaCl concentrations these locomotion behavioral results confirm our finding regarding bidirectional ethanol chemotaxis suggesting that the concentration of chloride ion plays a key role in how worms perceive ethanol valence suggesting that other sensory neurons are mediating the ethanol attraction chemotaxis behavior These results suggest the TAX-4/TAX-2 cGMP channel signaling pathway plays a crucial role in ethanol chemotaxis prompting us to focus on identifying the specific ethanol-sensing neuron(s) within this pathway To directly assess the involvement of different sensory neurons in ethanol sensing we measured calcium dynamics to monitor neuronal activity in the four sensory neurons that redundantly mediate ethanol chemotaxis: ASER these results suggest that both perceptivity and valence of the ethanol odor are inherently coded within these sensory neurons these findings suggest that ASER neuronal activity is required for ethanol bivalence behavior Given that the basal calcium level prior to ethanol exposure is set by a switch of NaCl concentrations we propose a working model wherein the degree of calcium drop in the ASER neuron upon ethanol exposure determines the directionality of valence when the basal calcium level is set high by switching to a lower NaCl concentration the following ethanol exposure creates a decrease in calcium levels when worms are switched to a higher NaCl concentration and the ethanol exposure induces only a slight decrease in calcium levels these results demonstrate that the basal calcium levels in ASER prior to ethanol exposure determine the directionality of ethanol valence This highlights the important role of CMK-1 in transmitting positive valence signals from ASER particularly when there is a large calcium increase in ASER that drives attraction behavior under the switch to lower NaCl conditions Different salt concentrations in the environment induce distinct basal calcium levels in the ASER neuron triggering signaling pathways that engage specific interneurons to mediate either attraction or aversion to ethanol This intricate interplay highlights the complexity of sensory perception and valence plasticity emphasizing the dynamic interactions between sensory inputs and neuronal pathways we increased the amount of ethanol in our chemotaxis assay By adding 20 or 40 μl of ethanol to one side of the assay plate we observed robust chemotaxis behavior in C indicating that NaCl serves as an indicator of nutrient availability our study may support the evolutional role of ethanol bivalence in C elegans to enhance their survivability in a wild environment with various levels of food sources worms cultivated in high-salt or high-temperature environments prefer the respective high sides of the gradient worms raised in lower-salt or low-temperature conditions favor the low sides and exhibit negative chemotaxis it has been suggested that valence switches are linked to changes in synaptic transmission over an extended period of time involving either synaptic modification or the recruitment of additional neural circuits the brief exposure times to ethanol odor (20 min in chemotaxis assays and less than 1 min in locomotion assays) suggest the valence switch occurs at the signaling level and neuronal activity without synaptic remodeling This represents an interesting paradigm wherein valence switching is driven by calcium signaling dynamics rather than synaptic plasticity Our results demonstrate that CMK-1 primarily functions at 25 mM NaCl where behavioral defects are observed in cmk-1 mutants We propose that CMK-1’s major role at this condition arises from large calcium increases at this concentration Further investigation is needed to understand how this release contributes to ethanol chemotaxis behavior at 25 mM NaCl Interestingly, during a switch to 75 mM NaCl, where aversion to ethanol remains normal in cmk-1 mutants, the AIY interneuron in cmk-1 mutants, important for executing aversion, displays the same calcium response as in wild-type worms (Fig. 7g This suggests the cmk-1 mutation does not disrupt the function of AIY in executing ethanol aversion when switching to 75 mM NaCl supporting our hypothesis that CMK-1 deficiency specifically affects ethanol attraction behavior we discovered an intriguing phenomenon of valence switching wherein distinct neuronal activation patterns within ASER can selectively engage different downstream interneurons This highlights how a simple organism with limited neurons can encode complex valence information through versatile mechanisms The unique properties of ASER neurons may arise from their asymmetric nature All C. elegans strains were cultured on nematode growth medium (NGM) seeded with Escherichia coli strain OP50 at 20 °C according to standard methods82 ten L4 larvae per genotype were picked to 6 cm NGM growth plates and young adults from their progeny were tested in behavioral and calcium imaging assays 4 days later Animals were maintained under well-fed conditions at all times Total RNA was extracted from approximately one thousand synchronized young adult worms according to standard protocols Freshly extracted RNA was subjected to reverse transcription PCR (RT-PCR) using oligo (dT) primer to generate a cDNA library (SweScript RT I First Strand cDNA Synthesis Kit glr-2 and cmk-1 cDNA were amplified from the cDNA library All constructs were generated using standard restriction enzyme cloning or Gibson assembly Promoters used for tissue-specific expression were: odr-7p (4.8 kb) along with the indicated concentrations of the test component assay plates were then left uncovered to dry for 1 h at room temperature A central line and two flanking lines (~1.4 cm apart) were drawn on the plate bottom worms found between the two flanking lines were disregarded in the count ~50 synchronized young adult worms were picked and placed onto the center of the assay plate A piece of 2 cm × 0.5 cm filter paper was positioned on one side of the lid parallel to the lines drawn on the bottom of the plate Sigma-Aldrich) or another indicated component was added After either 20 min or a specified duration and worms were counted from both sides of the assay plate The chemotaxis index was calculated as: (number of animals at the ethanol side - number of animals at the opposite side of ethanol)/total number of animals on both sides Two perpendicular lines were drawn on the plate bottom to divide the plate into four quadrants Approximately 50 synchronized young adult worms were picked and placed onto the center of the assay plate Two pieces of 1 cm × 0.5 cm filter paper were positioned at opposite ends of the lid The chemotaxis index was calculated as: (number of animals on the two ethanol quadrants - number of animals on the other two quadrants)/total number of animals on all four quadrants Chemotaxis assays for other odorants followed a previously described protocol20 and 1 mM magnesium sulfate were prepared and microwaved until agar had completely melted 10 ml of the solution was dispensed into each 9 cm Petri plate which was then left uncovered to dry for 1 h at room temperature Two 1 μl spots of odorant and diluent ethanol were placed at opposite ends of the plate together with 1 μl of 1 M potassium azide at each spot to immobilize worms Freshly diluted odorants included: butanol (undiluted Approximately 50 synchronized young adult worms were picked and placed onto the center of each assay plate worms anesthetized by potassium azide were counted The chemotaxis index was calculated as: (number of animals at the odor spot - number of animals at the counter spot) / total number of animals at the two regions Chemotaxis indexes are presented as mean ± SEM and individual data points in all figures 35 mm assay plates were prepared with 4 ml of solution using the same formula as that for the ethanol chemotaxis assay plates were selected and placed on the assay plate Each worm was allowed to acclimate freely for 2 min before testing 2 μl of ethanol was pipetted and delivered to the area near the worm’s nose Care was taken not to touch the worm or the plate throughout the process The avoidance response was recorded if the worm ceased forward movement and initiated a reversal within the 5-s timeframe The worm reversal rate was calculated by dividing the number of worms exhibiting reversal behavior by the total number of worms tested RNAi through feeding was accomplished by administering E coli clones from the Vidal RNAi library to wild-type animals NGM plates were prepared with 100 μg/ml ampicillin and 3 mM IPTG coli strain carrying the empty pL4440 vector wild-type animals were transferred to the RNAi plates at the L4 stage and allowed to develop for 22–26 h into young adults before testing a 736 bp sequence from both sense and antisense cmk-1 cDNA was cloned downstream to the gcy-5 promoter The resulting two plasmids were injected into wild-type worms to generate transgenic lines Succeed transgenic worm were selected and prepared for ethanol behavior assay Calcium imaging experiments were performed as described with some modifications85 Transgenic worms expressing the calcium sensor GCaMP6s in specific neuron were loaded into the microfluidic chip The animal’s nose was exposed to a stream of liquid that could be manually switched between diluted odor and buffer The buffer consisted of 25 mM potassium phosphate (pH 6.0) the nose tips of worms trapped in the chip were allowed to adapt for 2 min in the exposure of the buffer solution stream Imaging was conducted on a Nikon Ti2-E inverted microscope with a 40X air objective and an sCMOS camera (PCO The processes of AIY and AIZ neurons were imaged due to their robust calcium signals Time-lapse images were recorded at 1 frame per second The fluorescence intensity during the first 15 s before the ethanol stimulus or during the first 30 s at the beginning of the calcium recording was averaged and defined as F0 Fluorescence intensity change ΔF/F0 was calculated as (background-corrected fluorescence - F0)/F0 Calcium responses are presented as average changes were subjected to 2.5 mW/mm2 LED blue light for 1 h to induce cell death the animals were allowed to recover for 24 h at 20 °C Successful neuronal ablation was confirmed by the loss of fluorescence in mNeptune2.5 evidenced by either disappearance or fragmentation of the fluorescence expressing either Channelrhodopsin-2 (ChR2) or Guillardia theta anion channel rhodopsins 2 (GtACR2) in specific neurons were transferred to seeded NGM plates supplied with 100 μM all-trans-retinal (ATR Sigma-Aldrich) and incubated for 24 h in darkness Control worms were grown on seeded NGM plates without ATR For concurrent optogenetic and ethanol treatment worms were exposed to 2.5 mW/mm2 LED blue light while undergoing the 20-min ethanol chemotaxis assay worms were initially exposed to 2.5 mW/mm2 LED blue light for a specified duration (1 followed by placing the lid containing filter paper soaked in ethanol onto the assay plate to initiate the ethanol chemotaxis assay The assay concluded after 10 min of ethanol exposure Stocks of 1 M histamine dihydrochloride (Cat#59964 Sigma-Aldrich) were diluted to a final concentration of 10 mM in the assay plate agar cooled to ~60 °C Transgenic young adult worms expressing HisCl1 in specified neurons were transferred onto the histamine assay plates for behavior assays Histamine-free assay plates served as controls Image stacks from calcium imaging were aligned using the Template Matching plugin in FiJi (http://fiji.sc RRID:SCR_002285) Each stack was then cropped on a smaller region around the cell body The region of interest (ROI) was defined by outlining the desired neuron cell body or neurite The background-subtracted average fluorescence intensity within the ROI was used for subsequent analysis Average fluorescence intensity of the neuron for the first 15 or 30 s in each recording was defined as baseline (F0) Change in fluorescence intensity relative to the baseline Peak amplitude was calculated as maximum change in fluorescence (F-F0) during odor addition or removal Statistical analysis was performed with GraphPad Prism Data are presented as mean ± standard error of mean (SEM) All statistical tests used are indicated in the corresponding figure legends Difference significance are indicated as *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001; and n.s statistically not significant (P > 0.05) Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article All data generated or analyzed during this study are included in this published article and its supplementary information files. Any additional information required to reanalyze the data is available from the corresponding authors upon request. Source data are provided with this paper The code for analyzing calcium imaging data can be accessed in a public repository86 https://github.com/YuBestLab/YuBestLab.github.io/tree/C.-elegans-calcium-imaging-analysis, and found online at: https://doi.org/10.5281/zenodo.14498577 Affective valence in the brain: modules or modes A framework for studying emotions across species Architectural representation of valence in the limbic system Antagonistic negative and positive neurons of the basolateral amygdala Genetically identified amygdala–striatal circuits for valence-specific behaviors The emergence and influence of internal states Multisensory integration in caenorhabditis elegans in comparison to mammals Neuroecology of alcohol preference in Drosophila Experience-dependent changes in affective valence of taste in male mice Bargmann, C. I. review 2006, Chemosensation in C. elegans. WormBook 1–29. https://doi.org/10.1895/wormbook.1.123.1 (2006) Chemosensory signal transduction in Caenorhabditis elegans Reprogramming chemotaxis responses: sensory neurons define olfactory preferences in C Odorant-selective genes and neurons mediate olfaction in C Context-dependent reversal of odorant preference is driven by inversion of the response in a single sensory neuron type Neuromodulatory state and sex specify alternative behaviors through antagonistic synaptic pathways in C Functional organization of a neural network for aversive olfactory learning in Caenorhabditis elegans Pathogenic bacteria induce aversive olfactory learning in Caenorhabditis elegans Feeding state sculpts a circuit for sensory valence in Caenorhabditis elegans Concentration memory-dependent synaptic plasticity of a taste circuit regulates salt concentration chemotaxis in Caenorhabditis elegans The Insulin/PI 3-Kinase pathway regulates salt chemotaxis learning in Caenorhabditis elegans Odour concentration-dependent olfactory preference change in C A single-neuron chemosensory switch determines the valence of a sexually dimorphic sensory behavior Neuronal basis of innate olfactory attraction to ethanol in Drosophila Ethanol drinking in rodents: is free-choice drinking related to the reinforcing effects of ethanol elegans che-1 gene encodes a zinc finger transcription factor required for specification of the ASE chemosensory neurons The Gα Protein ODR-3 mediates olfactory and nociceptive function and controls cilium morphogenesis in C A putative cyclic nucleotide–gated channel is required for sensory development and function in C Combinatorial expression of TRPV channel proteins defines their sensory functions and subcellular localization in C Mutations in a cyclic nucleotide–gated channel lead to abnormal thermosensation and chemosensation in C A central role of the BK potassium channel in behavioral responses to ethanol in C Olfaction and odor discrimination are mediated by the C Chemosensory neurons with overlapping functions direct chemotaxis to multiple chemicals in C Development of left/right asymmetry in the Caenorhabditis elegans nervous system: from zygote to postmitotic neuron Functional asymmetry in Caenorhabditis elegans taste neurons and its computational role in chemotaxis The homeobox gene lim-6 is required for distinct chemosensory representations in C Searching for neuronal left/right asymmetry: genomewide analysis of nematode receptor-type guanylyl cyclases Highly efficient optogenetic cell ablation in C Photo-inducible cell ablation in Caenorhabditis elegans using the genetically encoded singlet oxygen generating protein miniSOG Genetic screens for caenorhabditis elegans mutants defective in left/right asymmetric neuronal fate specification An improved inverse-type Ca2+ indicator can detect putative neuronal inhibition in Caenorhabditis elegans by increasing signal intensity upon Ca2+ decrease Rhodopsin optogenetic toolbox v2.0 for light-sensitive excitation and inhibition in Caenorhabditis elegans Connectomes across development reveal principles of brain maturation Encoding of both analog- and digital-like behavioral outputs by One C Glutamate signaling from a single sensory neuron mediates experience-dependent bidirectional behavior in Caenorhabditis elegans Molecular topography of an entire nervous system Calmodulin-Kinases: modulators of neuronal development and plasticity Ca2+/Calmodulin-dependent protein kinase cascade in Caenorhabditis elegans: implication in transcriptional activation* CaMKI-dependent regulation of sensory gene expression mediates experience-dependent plasticity in the operating range of a thermosensory neuron Human olfactory detection of homologous n-alcohols measured via concentration–response functions The nutritional requirements of Caenorhabditis elegans Exogenous ethanol induces a metabolic switch that prolongs the survival of Caenorhabditis elegans dauer larva and enhances its resistance to desiccation Ethanol-induced differential gene expression and acetyl-CoA metabolism in a longevity model of the nematode Caenorhabditis elegans Ethanol metabolism and osmolarity modify behavioral responses to ethanol in C Effect of salt concentrations on the growth of heat-stressed and unstressed Escherichia coli Chemotaxis behavior toward an odor is regulated by constant sodium chloride stimulus in Caenorhabditis elegans A differential role for neuropeptides in acute and chronic adaptive responses to alcohol: behavioural and genetic analysis in Caenorhabditis elegans Behavioral deficits following withdrawal from chronic ethanol are influenced by SLO channel function in Caenorhabditis elegans Different modes of stimuli delivery elicit changes in glutamate driven experience-dependent interneuron response in C The structure of the nervous system of the nematode Caenorhabditis elegans Molecular encoding and synaptic decoding of context during salt chemotaxis in C Neural coding in a single sensory neuron controlling opposite seeking behaviours in Caenorhabditis elegans Presynaptic MAST kinase controls opposing postsynaptic responses to convey stimulus valence in Caenorhabditis elegans Integration of plasticity mechanisms within a single sensory neuron of C Transfer characteristics of a thermosensory synapse in Caenorhabditis elegans Dynamics of presynaptic diacylglycerol in a sensory neuron encode differences between past and current stimulus intensity Population coding of valence in the basolateral amygdala Odor mixtures of opposing valence unveil inter-glomerular crosstalk in the Drosophila antennal lobe Brain-wide mapping reveals that engrams for a single memory are distributed across multiple brain regions elegans is explained by circuit-specific normalization in chemosensory pathways elegans arises from asymmetric modulatory effects within single sensory neurons Efficient gene transfer in C.elegans: extrachromosomal maintenance and integration of transforming sequences Powerful and interpretable behavioural features for quantitative phenotyping of Caenorhabditis elegans Dissecting a circuit for olfactory behaviour in Caenorhabditis elegans Download references We are grateful to the Caenorhabditis Genetics Center which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440) which is funded by the Japanese government Pei Chang for assistance with analyses of calcium imaging data Minghai Ge for assistance with behavioral assays and the staff at the core facility of the Medical Research Institute at Wuhan University for their technical support We thank Jianke Gong and Wenxing Yang for critical comments on the manuscript This research was funded by “Technology Innovation 2030—major projects” on brain science and brain-like computing from the Ministry of Science and Technology of China (2021ZD0202603 – Y.V.Y.) the NSFC (32070832 and 32150610476 – Y.N.J.) and the Fundamental Research Funds for the Central Universities (2042022dx0003) These authors contributed equally: Weikang Xue Tissue Engineering and Organ Manufacturing (TEOM) Lab TaiKang Medical School (School of Basic Medical Sciences) TaiKang Center for Life and Medical Sciences Key Laboratory of Molecular Biophysics of Ministry of Education Huazhong University of Science and Technology Frontier Science Center for Immunology and Metabolism and Y.V.Y.; Writing – review and editing: W.X. Nature Communications thanks Alon Zaslaver reviewer for their contribution to the peer review of this work Download citation DOI: https://doi.org/10.1038/s41467-025-57051-x Metrics details Understanding the relationship between the microscopic structure and topology of a material and its macroscopic properties is a fundamental challenge across a wide range of systems Here we investigate the viscoelasticity of DNA nanostar hydrogels—a model system for physical networks with limited valence—by coupling rheology measurements confocal imaging and molecular dynamics simulations We discover that these networks display a large degree of interpenetration and that loops within the network are topologically linked forming a percolating network-within-network structure the fraction of branching points and the pore size determine the high-frequency elasticity of these physical gels we discover that this elastic response is dictated by the abundance of topological links between looped motifs in the gel Our findings highlight the emergence of ‘topological elasticity’ as a previously overlooked mechanism in generic network-forming liquids and gels and inform the design of topologically controllable material behaviours Purple dashed line indicates the overlapping volume fraction ρ* = 0.056 (smaller than symbol size) from five independent replicates Data are well fitted by a power law \({G}_{{\mathrm{p}}}^{{\prime}} \propto {\rho }^{2.5}\) in the whole range of concentrations We couple extensive molecular dynamics (MD) simulations of an oxDNA-inferred29 coarse-grained model of DNAnss with bulk and microrheology (MR) experiments and confocal imaging to elucidate the connection between network structure and we discover that the system undergoes a topological transition around the overlapping concentration c* where loops within the network start to link with each other While for C < c* the fraction of branching points and the mesh size of the network (that is its structure) govern the elasticity of the gel linking of loops within the network (that is its topology) emerges as the main determinant of the gel’s mechanical properties its solid-like response to high-frequency stress scales with the concentration of DNAnss as \({G}_{{\mathrm{p}}}^{{\prime} } \propto {C}^{2.5}\) and scales linearly with the linking number between minimum loops \({G}_{{\mathrm{p}}}^{{\prime} } \propto {\mathcal{L}}\) where \({G}_{{\mathrm{p}}}^{{\prime} }\) is the elastic plateau the value of the elastic modulus at the largest frequency This simple relationship connects the macroscopic mechanical properties of a soft material with a mathematically rigorous topological invariant that is markedly distinct from generic ‘entanglements’ or crosslinks we also show that G′ and G″ from different concentrations superimpose into a master curve that follows a near-Maxwellian behaviour with a characteristic unbinding time \({\tau_{u}\propto} {\omega }_{0}^{-1}\) (with ω0 the cross-over frequency at which G′ = G″) due to the reversible nature of the hybridization between DNAnss To better understand how this scaling emerges from the network structure our simulations suggest that the gels’ elasticity displays a nonlinear scaling with concentration different from predictions of classic models To understand the underlying physical origin of elasticity in these networks we thus now investigate the microscopic properties and topology of our simulated DNAns gels Lines represent a power law fit to the data from which labels with the best-fit exponents are obtained Temporal evolution of the intensity of the yellow signal of the yellow intensity across the image (calculated at a fixed time point) of yellow intensity for each concentration averaged over all the displayed time points We also show representative confocal images from experiments in the respective phases we introduced modified nucleotides bound to fluorophores—FAMK for DNS-A and Cy3 for DNS-B—in the middle of one of the double-stranded DNA arms of the nanostars Upon mixing equimolar amounts of DNS-A and DNS-B, each beyond their own gel binodal, microemulsions are formed: small A-rich droplets form in a large B-rich droplet, and vice versa. We multiply the normalized green Ig and red Ir signal intensities in each pixel to obtain a two-dimensional map of ‘yellow’ pixels with intensity Iy(i, j) = Ir(i, j) × Ig(i, j) (Fig. 3c) A large value of Iy indicates the presence of both DNS-A and DNS-B which we identify as an interpenetrated region The last two quantities have been multiplied by 0.03 and 0.001 obtained from at least 25 independent configurations per volume fraction We then computed the Gauss linking number (Lk) between all possible pairs of minimum loops in the network as follows: thus establishing a network-within-network structure in DNAns gels This topological transition in the linking of (minimum) loops at ρ* explains the counterintuitive results seen above: as the concentration of DNAnss increases not only do more loops appear in the network but these loops become larger concomitantly with the emergence of more tightly packed which in turn favour the formation of links and increase the topological complexity of the network we expect the scaling of \({G}_{{\mathrm{p}}}^{{\prime} }\) to be determined by the linking of the minimum loops we find that the scaling of the Green–Kubo-measured elasticity is better explained by the following simple argument: assuming that linking between minimum loops becomes the dominant entanglement mechanism on timescales shorter than the typical melting time of the sticky ends the high-frequency elastic plateau should be proportional to (1) the density of elastically active polymers ρe (the density of minimum loops) and (2) the number of entanglements per polymer Ze (the average linking valence) where we made use of Nminloop ∝ ρ and 〈Z〉 ∝ ρ1.4 (Fig. 4g). This scaling is in very good agreement with both experiments and simulations (Fig. 4h) The small discrepancy in the predicted exponent (2.4) and the one measured through Green–Kubo calculations (2.5) is within uncertainty This discrepancy could also be attributed to the fact that our topological analysis neglects links between second-order loops (larger than the minimum ones) and it also ignores loop threading In this paper we have tackled a long-standing question: the connection between a material’s microscopic topology and its macroscopic material properties We focused on soft viscoelastic gels made of limited valence building blocks with transient non-covalent crosslinks and realized them in the lab using tri-armed DNAnss The simple linear relationship found (\({G}_{{\mathrm{p}}}^{{\prime} } \propto {\mathcal{L}}\)) is very powerful as it connects the rheology of the network with its topology (in the formal mathematical sense of the term we argue that if we were to design DNAnss with longer arms or to use binary mixtures of DNAnss with each type having arms of uniform length and with the arm lengths differing between the two types despite the effective reduction of crosslinks per unit volume by designing DNAnss with longer sticky ends (and hence longer unbinding times) we would realize semi-irreversible interactions and complex fluids with an elastic behaviour at lower frequencies through DNA ligation we could render DNAns–DNAns bonds irreversible Whether the ensuing low-frequency elastic behaviour could be described by the linkages between minimum loops in the network is to be determined as the gelation kinetics may also contribute to affect the scaling topology is by its nature universally found across physical and biological systems our findings contribute to a better fundamental understanding of how formal topological motifs within a network affect the material properties of the bulk and will also guide the design of topologically controlled structural and mechanical properties of generic network-forming liquids and materials Nanostar samples were prepared at the desired concentration in Nanostar Buffer containing the following reagents: Tris (Invitrogen ethylenediaminetetraacetic acid (Sigma-Aldrich ED2P) and NaCl (Thermo Scientific Chemicals and the geometry of the bead-patch and parameters of the Morse potential were set to ensure one-to-one binding of the simulated nanostars thereby mimicking sticky-end hybridization MD simulations were performed using the large-scale atomic/molecular massively parallel simulator (LAMMPS) and in the NVT ensemble volume (V) and temperature (T) were held constant An implicit solvent approach (Langevin dynamics) was used The integration time step was set to dt = 0.01τBr (τBr = kBT/Γ is the Brownian time and Γ is the friction set to 1 in Lennard-Jones (LJ) units) and the temperature to T = 1ϵ/kB (ϵ is the characteristic interaction energy) Initial configurations were prepared by performing first an equilibration run for 5 × 105τBr while attraction between DNAnss was not allowed we turned on the Morse attraction and performed a production run for 106τBr We identified hybridization between adjacent nanostars when their patches were located at a distance r ≤ 0.2σ (the cut-off distance of the attraction between patches) Scaling Concepts in Polymer Physics (Cornell Univ Topological nature of the liquid–liquid phase transition in tetrahedral liquids Link to densify: topological transitions and origin of hysteresis during the compression and decompression of amorphous ices Rheology and microscopic topology of entangled polymeric liquids Principles of Polymer Chemistry (Cornell Univ Leaving groups as traceless topological modifiers for the synthesis of topologically isomeric polymer networks A (macro)molecular-level understanding of polymer network topology Topological structure of networks formed from symmetric four-arm precursors A topologically driven glass in ring polymers Unexpected power-law stress relaxation of entangled ring polymers Topological tuning of DNA mobility in entangled solutions of supercoiled plasmids Topological linking drives anomalous thickening of ring polymers in weak extensional flows Active DNA Olympic hydrogels driven by topoisomerase activity Single-molecule structure and topology of kinetoplast DNA networks Tying different knots in a molecular strand Extremely stretchable thermosensitive hydrogels by introducing slide-ring polyrotaxane cross-linkers and ionic groups into the polymer network Topological data analysis for particulate gels Liquids more stable than crystals in particles with limited valence and flexible bonds Increasing valence pushes DNA nanostar networks to the isostatic point Strain-controlled criticality governs the nonlinear mechanics of fibre networks Directed self-assembly of a colloidal kagome lattice Colloids with valence and specific directional bonding Phase behavior and critical activated dynamics of limited-valence DNA nanostars Equilibrium gels of low-valence DNA nanostars: a colloidal model for strong glass formers Nanostars planarity modulates the rheology of DNA hydrogels Evidence for equilibrium gels of valence-limited particles Nanoscopic interfacial hydrogel viscoelasticity revealed from comparison of macroscopic and microscopic rheology Coarse-grained simulation of DNA using LAMMPS: an implementation of the oxDNA model and its applications Estimating the viscoelastic moduli of complex fluids using the generalized Stokes-Einstein equation Efficient on the fly calculation of time correlation functions in computer simulations Entangled polymer melts: relation between plateau modulus and stress autocorrelation function Long-range spatial correlations of particle displacements and the emergence of elasticity Determining the mesh size of polymer solutions via the pore size distribution Quantifying the impact of molecular defects on polymer network elasticity The hidden hierarchical nature of soft particulate gels Sequence-controlled adhesion and microemulsification in a two-phase system of DNA liquid droplets Interpenetrating gels in binary suspensions of DNA nanostars Connectivity and entanglement stress contributions in strained polymer networks Statistical properties of networks of flexible chains Effect of chain polydispersity on the elasticity of disordered polymer networks Structure and elasticity of model disordered Is the kinetoplast DNA a percolating network of linked rings at its critical point Equilibrium structure and deformation response of 2D kinetoplast sheets Equilibrium gels of limited valence colloids Thermodynamics and structure of poly[n]catenane melts A self-quenched defect glass in a colloid-nematic liquid crystal composite Emulsion imaging of a DNA nanostar condensate phase diagram reveals valence and electrostatic effects NUPACK: analysis and design of nucleic acid systems Optical measurements of frequency-dependent linear viscoelastic moduli of complex fluids Gutiérrez Fosado, Y. A., Palombo, G., Weir, S. & Michieletto, D. Topological elasticity. Edinburgh DataShare https://doi.org/10.7488/ds/7839 (2024) Download references acknowledges the Royal Society and the European Research Council (grant agreement no We also acknowledge the contribution of the COST Action Eutopia we have applied a Creative Commons Attribution (CCBY) licence to any author accepted manuscript version arising from this submission Davide Michieletto & Yair Augusto Gutiérrez Fosado developed analysis codes and performed analysis from simulations supervised research and wrote the original draft Nature Materials thanks Emanuela Del Gado and Omar Saleh for their contribution to the peer review of this work Download citation DOI: https://doi.org/10.1038/s41563-024-02091-9 Metrics details Late-stage and advanced colorectal cancer (CRC) often prove to be resistant to current treatment regimens due to the evolving tumor microenvironment Chemotherapy-dominated multi-modal therapeutic strategies based on the specific CRC microenvironment open a new horizon for eradicating colorectal tumors in situ valence-transited arsenic nanosheets are developed as a multi-modal therapeutic platform by responding to the H2S-enriched CRC microenvironment Carrier-free pegylated nanosheets of pentavalent arsenic (AsV) and copper ion (Cu2+) are innovatively self-assembled via coordination with high loading content and good stability AsV in pegylated arsenic nanosheets (CAA-PEG NSs) is rapidly released and reduced to trivalent arsenic (AsIII) to exert its chemotherapy in the local tumor the immunosuppressive microenvironment is thoroughly remodeled via H2S depletion of AsV to AsIII conversion and impairment of H2S production by AOAA the in situ produced ultrasmall CuS nanoparticles exhibit photothermal activity against CRC under the guidance of photoacoustic imaging completely inhibits CRC progression and prevents its relapse Multi-modal therapeutic strategies based on the microenvironment-mediated multiple mechanisms offer new therapeutic avenues that may eradicate colorectal tumors suppose AsV could be transformed to AsIII via a reduction reaction in the local tumor microenvironment; the therapeutic window of arsenic compounds could be significantly enlarged to make arsenic a promising chemotherapeutic agent for solid tumor treatment If arsenic is encapsulated in inorganic functional nanoparticles it would be an excellent synergistic therapeutic platform for the combination of chemotherapy with other therapies photodynamic therapy and photothermal therapy Created in BioRender. Jiang, L. (2025) https://BioRender.com/o52q745 c The zeta potential of CAA and CAA-PEG NSs (Data are presented as mean values +/− SD AOAA in CAA-PEG NSs (Data are presented as mean values +/− SD f Molecular dynamics simulation of CAA NSs formation process The hydrogen bonds formed during the CAA NSs formation The dispersion stability (j) and particle size change (k) of CAA NSs and CAA-PEG NSs in PBS (Data are presented as mean values +/− SD Two-way ANOVA analysis and Sídík’s Multiple Comparison Test was used to compare the two groups at the same time point P values were presented in absolute value in the figures) Further binding trajectory videos can reveal that AOAA and the arsenate ions can be stably bound for a limited period on the order of a few nanoseconds; however this binding cannot be sustained continuously We speculate is due to the stronger electrostatic interaction of copper ions with the arsenate ions Our findings suggest that the coordination between copper ions and arsenate ions along with the hydrogen bond interaction with AOAA effectively integrates the individual components into a self-assembled structure All these results suggested that the ultrathin pegylated nanosheets of AsV and Cu2+ were successfully self-assembled with high loading content and good stability a The mimetic H2S-activated decomposition of CAA-PEG NSs at different time intervals b Arsenic release from different release medium (pH 7.4 and pH 7.4 + 3 mM NaHS) (Data are presented as mean values +/− SD c Arsenic release from release medium with escalating NaHS concentrations (Data are presented as mean values +/− SD e Valence transition efficiency of AsV to AsIII of CAA-PEG NSs in different concentrations of NaHS solution f XRD and SAED pattern of ultrasmall CuS nanoparticles g UV–vis absorption intensity and color changes (inserted) of CAA-PEG NSs at 1064 nm in different concentrations of NaHS solutions (Data are presented as mean values +/− SD h Photoacoustic signal intensity of CAA-PEG NSs with various concentrations of NaHS and the corresponding photoacoustic images (inserted) (Data are presented as mean values +/− SD The heating curve (i) and photothermal images (j) of CAA-PEG NSs under 1064 nm laser irradiation with different concentrations of NaHS solution (Data are presented as mean values +/− SD k Temperature variations of the mixture of aqueous CAA-PEG NSs solutions and NaHS under 1064 nm irradiation for seven light on/off cycles i) One-way ANOVA with Dunnett’s Multiple Comparison Test was used to compare each group with the Control groups P values were presented in absolute value in the figures pegylated arsenic nanosheets responsively released AsV and AsV could be reduced to AsIII in sufficient H2S sites The produced ultrasmall CuS nanoparticles had excellent photoacoustic and photothermal capability e Western blot analysis of cell apoptosis-related proteins (P53 P21) after different treatments (One representative gel from three repeated gels) H2S content change in CT-26 cells after treatment with CAA-PEG NSs for different time intervals observed by confocal microscopy (f) and flow cytometry (g) h Western blot analysis of CBS in CT-26 cells after treatment with CAA-PEG NSs (One representative gel from three repeated gels) i CT-26 cell viability treated with different concentrations of CAA-PEG NSs with or without laser irradiation (Data are presented as mean values +/− SD Two-way ANOVA analysis and Sídík’s Multiple Comparison Test was used to compare the Laser +/− groups j Live/dead cell staining assay of CT-26 cells after treatment with CAA-PEG NSs with or without laser irradiation suggesting its considerable inhibition of H2S production Similar results were also observed in flow cytometry which demonstrated that CAA-PEG NSs could deplete H2S via both consumptions of AsV to AsIII conversion and H2S-relating enzymes inhibition/downregulation of AOAA thereby reshaping the CRC microenvironment This property also offers the possibility of multi-modal treatment in mice in vivo For (f) One-way ANOVA with Dunnett’s Multiple Comparison Test was used to compare each group with the Control groups Significant differences in survival (g) were calculated by the Mantel–Cox test and P values were presented in absolute value in the figure) h Immunofluorescent staining assays (P53 and CBS) were performed on tumor sections after different drug treatments These results illustrated that the locally produced ultrasmall CuS nanoparticles with strong NIR-II absorption had excellent photothermal activities under the guidance of photoacoustic imaging One-way ANOVA with Tukey’s Multiple Comparison Test was used to compare the mean of each group with the mean of the Control group These results indicated that CAA-PEG NSs could reverse the immunosuppressive microenvironment by re-polarizing M2 TAMs to M1 TAMs a Proportion of differential protein and non-differential protein Column (b) and volcano maps (c) of upregulated and downregulated proteins in differential proteins d The heat map of differential immune-related proteins The GO enrichment analysis (e) and the KEGG pathway enrichment analysis (f) of differential proteins related to immune processes The data are presented as mean values +/− SD These results indicated that CAA-PEG NSs could induce DC maturation T cell activation and infiltration after Inhibition of endogenous H2S production in the CRC microenvironment we have exploited the strategy of reduced H2S-mediated arsenic valence transition to develop arsenic-based nanosheets Under the H2S-abundant CRC microenvironment the PEGylated arsenic nanosheets rapidly dissociated thereby triggering a cascade of synergistic effects: Low systemic toxicity AsV is reduced to the therapeutic state AsIII to exert its chemotherapy we are able to significantly broaden the therapeutic window of arsenic compounds rendering them promising chemotherapeutic agents for solid tumors; the H2S depletion during the conversion of AsV to AsIII and the AOAA-mediated inhibition and/or downregulation of H2S-producing enzymes completely remodel the immunosuppressive microenvironments by inducing maturation and differentiation of immune cells to generate more tumor-killing T-cells as well as repolarization of M2 into M1 tumor-associated macrophages; As a by-product of nanosheet decomposition the generation of ultrasmall CuS NPs guided by photoacoustic imaging has a strong NIR-II absorption capacity which is a powerful complement to CRC chemotherapy the synergistic effect of chemotherapy with photothermal therapy was maximized in our study they face challenges including potential toxicity to normal tissues and stability issues during storage and transport Our development of in situ valence-transited arsenic nanosheets offers a highly efficient multi-modal therapeutic platform specifically designed for colorectal cancer effectively addressing the multifaceted challenges of CRC treatment Our method demonstrates comprehensive suppression of CRC progression and prevention of recurrence While our study is currently limited to a single murine CRC cell line the successful development of CAA-PEG NSs underscores the potential of valence-transited arsenic-based therapeutic strategies for CRC treatment 18 ± 2 g) were purchased from SLAC Laboratory Animal Co All experiments were performed under experimental protocols approved by the Animal Care and Use Committee of Zhejiang Chinese Medical University (No Animal ethics states that the weight of tumors in mice should not exceed 10% of the mouse’s body weight and the size of tumors should not exceed 15 mm in any dimension It was confirmed that all the maximal tumor size/burden was not exceeded The environmental conditions in the mouse facility were: 12 h light/dark cycle (light on from 8:00 a.m Sodium arsenate dibasic heptahydrate (Na2HAsO4·7H2O) 99%) was purchased from Sigma-Aldrich (USA) 95%) was purchased from Beijing Qisong Biotechnology Co. NH2-PEG-COOH (MW: 3400 Da) and nitric acid (65–68%) were purchased from Aladdin (Shanghai 68–72%) was purchased from Ourchem (Shanghai 3,3′,5,5′-Tetramethylbenzidine (TMB) was purchased from Solarbio Science & Technology Co. cell counting kit (CCK-8) and BCA protein kit was bought from Beyotime (Shanghai China).7-Azido-4-methylcoumarin (AzMC) was bought from Dehang Wuzhou Technology Co. RIPA buffer and Protein Free Rapid blocking buffer (5×) was bought from Yamei Biopharmaceutical Technology Co. and bFGF ELISA kits was bought from Fiya Biotechnology Co. Lot number: 70016788) was purchased from American Type Culture Collection (ATCC) Cells have been genotyped for STR and routinely tested for mycoplasma Relevant supporting documentation can be viewed in the Supplementary Materials Arsenic nanosheets (CAA NSs) were prepared using a one-pot self-assembly method 170 mg CuCl2·H2O and 18 mg AOAA were each dissolved in 25 mL of ultrapure water The three solutions were then stirred violently at 4 °C for 30 min followed by transfer to hydrothermal autoclave for 30 min at 180 °C CAA NSs was collected by washing by centrifugation (8000 × g 20 mL of NH2-PEG-COOH (8.33 mg/mL) was added to the previously prepared CAA NSs solution and the mixture was stirred overnight to obtain CAA-PEG NSs The morphology and elemental mapping spectra of CAA-PEG NSs were observed with transmission electron microscopy (JEM 2100F Atomic force microscope (Bruker Dimension Icon DEU) was used to observe CAA-PEG NSs thickness The zeta potentials of CAA NSs and CAA-PEG NSs were measured by Malvern laser particle size analyzer (Nano-ZS90 AOAA content was determined with ninhydrin colorimetric method As and Cu contents were determined by inductive coupled plasma emission spectroscopy (ICP The RMSD analysis command: gmx rms -s md.tpr -f md.xtc -o rmsd.xvg The H-bond analysis command: gmx hbond -f md.xtc -s md.tpr -n index.ndx -num NaHS was used to simulate endogenous H2S73 the NaHS solid was dissolved in ddH2O and filtered to obtain a fresh 3 mM solution of NaHS CAA-PEG NSs (Cu: 80 μg/mL) were added to NaHS solution for a certain time and then centrifuged to collect at different time points (15 Nanoparticles were characterized by Electronic Differential System (EDS) and TEM After the reaction of CAA-PEG NSs (Cu: 80 μg/mL) with different concentrations of NaHS (0 the Vis-NIR absorption was measured by an ultraviolet/visible/near infrared spectrophotometer (Shimadzu and the photoacoustic image signal was detected with photoacoustic imaging system (InVision256-TF) USA) was used to record the temperature change of CAA-PEG NSs reaction with different concentrations of NaHS under the irradiation of 1064 nm laser to study the photothermal effect Certain concentrations of CAA-PEG NSs (Cu: 80 μg/mL) and NaHS were subjected to multiple rounds of illumination (1064 nm 5 min on/off) to record the temperature changes CAA-PEG NSs (Cu: 80 μg/mL) solution was loaded into a dialysis bag (MW: 3500 Da) with release medium containing NaHS and was oscillated in an air bath at 37 °C ICP was used to determine the arsenic released from CAA-PEG NSs within a specified time interval The transit of arsenic valence in solution was detected by X-ray photoelectron spectroscopy (XPS USA) after CAA-PEG NSs were mixed with different concentrations (0 CT-26 cells were cultured in laser confocal dishes and incubated with free ICG or ICG-labelled CAA-PEG NSs for 4 h and observed by confocal laser scanning microscope (CLSM) (LSM880 Intracellular degradation of CAA-PEG NSs was observed by bio-TEM CT-26 cells (5 × 105 cells/well) were cultured in 6-well plate at 37 °C the medium was replaced with fresh culture medium containing CAA-PEG NSs bio-TEM images of cells were obtained after frozen section at 1 CLSM and flow cytometry (FCM) were used to observe fluorescence intensity CCK-8 was used to evaluate the in vitro cytotoxicity of CAA-PEG NSs CT-26 cells (1 × 104 cells/well) were cultured in 96-well plates overnight and replaced with fresh culture medium containing CAA-PEG NSs with different concentrations for 24 h 5 min) was used to treat irradiated groups CT-26 cells (5 × 105 cells/well) were cultured in laser confocal dishes and incubated with culture medium or CAA-PEG NSs (As: 15 μg/mL) for 6 h 5 min) was used to treat with irradiated groups CLSM was used to observe the Calcein-AM and PI co-staining CT-26 cells (5 × 105 cells/well) were cultured in 6-well plates and incubated with culture medium or CAA-PEG NSs (As: 15 μg/mL) for 6 h FCM was used to observe the result of Annexin-FITC and PI co-staining CT-26 cells (5 × 105 cells/well) were cultured in 6-well plates and incubated with culture medium containing CAA-PEG NSs (As: 15 μg/mL) for 12 h the CT-26 cells were washed 3 times with PBS sonicated for 5 min in 5% HNO3 solution and centrifuged to collect the arsenic-containing solution the arsenic valence was evaluated by atomic fluorescence spectroscopy (AFS) BCA protein kit was used to normalize the protein concentrations The proteins were isolated applying wet electroblotting systems (Bio-Rad) by SDS-PAGE and transferred to PVDF membranes (Beyotime) The PVDF membranes were blocked in protein free rapid blocking buffer for 30 min and incubated with primary antibodies overnight at 4 °C Antibodies against the following proteins were used for the western blot assay: β-actin (Proteintech ECL Substrate was used to perform signal detection QuickChemi 5200 system (Monad) was used to collect and analyze the signals ICG and CAA-PEG-ICG were injected into mice intravenously The ICG fluorescence was observed by in vivo imaging system (excitation: 710 nm and the biodistribution imaging was used to observe the fluorescence intensity in tumors and major organs To establish the in-situ murine model of colon cancer the mice were prepared as previously described and intracecum injections (5 × 105 CT-26 cells suspended in 25 μL of PBS) were performed using an insulin-gauge syringe the mice were injected with 100 μL of saline CA-PEG NSs and CAA-PEG NSs (As: 1 mg/kg) (n = 4) through a tail vein All mice were injected intravenously with the above formulations through the tail vein once a day for 10 days tumor weight and tumor diameter of mice in each group were recorded every week the mice in each group were euthanized for the collection of tumors CT-26 tumor-bearing mice were injected with CAA-PEG NSs (As: 1 mg/kg) (n = 3) The organs were kept in nitric acid overnight and centrifuged at 8000 × g for 15 min The supernatant was mixed with equal n-hexane and the obtained liquid was filtered through 0.45 μm organic membrane AFS and ICP were used to evaluate the arsenic valence in the extracted solution CT-26 tumor-bearing mice were injected with 100 μL CAA-PEG NSs (As: 1 mg/kg) through a tail vein for in vivo PA imaging All of the PA imaging was obtained using the photoacoustic imaging system The photoacoustic signal of the control group was replaced by the pre-injection CT-26 tumor-bearing mice were injected with 100 μL saline mice were anesthetized by intraperitoneal injection of 0.3% pentobarbital sodium (0.25 mL/10 g) The tumor site was irradiated with 1064 nm near-infrared laser at 1 W/cm2 for 10 min The temperature changes in the tumor site were recorded by IR thermal camera CT-26 tumor-bearing mice were injected with saline CAA-PEG NSs+Laser (As: 1 mg/kg) (n = 6) through a tail vein 10 min) was used to treat with irradiated groups after 8 h The treatment process was repeated every other day for 3 cycles Tumor size and the survival of mice were recorded every other day and the mice were sacrificed following three doses Tissue blocks were processed for immunofluorescence (IF) analysis The tumors were stained with primary antibodies against the following proteins: CBS (Proteintech Dylight 594-Goat Anti Mouse IgG (Immunoway These pathological sections were observed by VS120-S6-W system (OLYMPUS Spleens were collected for T cell activation analysis and bFGF) in serum was measured using ELISA kits The biosafety study was implemented by injecting saline 1 mg/kg) into the BALB/c mice (n = 3) intravenously blood from these mice was obtained for routine blood test and biochemical analysis tumor tissues were collected from each treatment group Protein extraction and quantization were performed with BCA kit The protein samples were subjected to SDS-PAGE electrophoresis to examine the consistency of each group 120 μg of protein sample from each group was reduced digested with trypsin (1 μg/μL) at 37 °C overnight the tryptic peptide solution of each sample was labeled according to the TMT labeling reagent instructions The dry samples were re-suspended with 100 μL of buffer A (2% acetonitrile and the reverse phase liquid chromatography (RPLC) was conducted on the L-3000 HPLC (RIGOL The parameters used were a Waters BEH C18 (4.6 × 250 mm Separation was performed at 1 mL/min using a nonlinear binary gradient starting with buffer A and transitioning to buffer B (98% acetonitrile Samples were collected at minute intervals starting from 1 min Each sample was dried in a vacuum freeze dryer for RPLC-MS/MS analysis of the TMT-labeled samples The Easy-nLCTM 1200 System (Thermo Scientific USA) was used for the online RPLC detection The Q ExactiveTM System (Thermo Scientific Peptides were quantitatively identified according to the peak signal intensity and the secondary mass spectrometry information of different reporter ions of the TMT tag reagents Groups were analyzed in triplicate to minimize the effect of individual differences Number of experimental replicates (n) is indicated in figure legend and refers to the number of experimental subjects independently treated in each experimental condition No statistical methods were used to pre-determine sample size One-way or Two-way ANOVA tests were used when comparing two or more than two normally distributed data sets Statistical analyses were carried out using GraphPad Prism (version 10.4.0) A statistical significance threshold was set at 0.05 and significance levels are presented as *P < 0.05 P values less than 0.001 were presented in absolute values in the figures Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article The inflammatory pathogenesis of colorectal cancer The emerging role of nerves and glia in colorectal cancer Cancer treatment and survivorship statistics Arsenic trioxide - An old drug rediscovered Nanodiamond autophagy inhibitor allosterically improves the arsenical-based therapy of solid tumors Current status and future prospects of nanomedicine for arsenic trioxide delivery to solid tumors MMP2-responsive dual-targeting drug delivery system for valence-controlled arsenic trioxide prodrug delivery against hepatic carcinoma Differential effects of trivalent and pentavalent arsenicals on cell proliferation and cytokine secretion in normal human epidermal keratinocytes and measurement: is modulation of hydrogen sulfide a novel therapeutic for cancer Regulation of CyR61 expression and release by 3-mercaptopyruvate sulfurtransferase in colon cancer cells H2S-scavenged and activated iron oxide-hydroxide nanospindles for MRI-guided photothermal therapy and ferroptosis in colon cancer An endogenous H2S-activated nanoplatform for triple synergistic therapy of colorectal cancer Recent advances in the role of endogenous hydrogen sulphide in cancer cells Inhibition of hydrogen sulfide synthesis reverses acquired resistance to 5-FU through miR-215-5p-EREG/TYMS axis in colon cancer cells Hydrogen sulfide dysregulates the immune response by suppressing central carbon metabolism to promote tuberculosis Depletion and downregulation of hydrogen sulfide using an activatable probe for promoting photothermal therapy toward colorectal cancers Selectivity of commonly used pharmacological inhibitors for cystathionine β synthase (CBS) and cystathionine γ lyase (CSE), Implications of hydrogen sulfide in colorectal cancer: mechanistic insights and diagnostic and therapeutic strategies Lipid encapsulation of arsenic trioxide attenuates cytotoxicity and allows for controlled anticancer drug release The osteogenic niche-targeted arsenic nanoparticles prevent colonization of disseminated breast tumor cells in the bone A novel nanoparticulate formulation of arsenic trioxide with enhanced therapeutic efficacy in a murine model of breast cancer Triggered release of pharmacophores from [Ni(HAsO(3))]-loaded polymer-caged nanobin enhances pro-apoptotic activity: a combined experimental and theoretical study Dual oligopeptides modification mediates arsenic trioxide containing nanoparticles to eliminate primitive chronic myeloid leukemia cells inside bone marrow niches Arsenic trioxide liposomes: encapsulation efficiency and in vitro stability Challenges associated with penetration of nanoparticles across cell and tissue barriers: a review of current status and future prospects Two-dimensional transition metal dichalcogenide (TMD) nanosheets Metal-organic framework nanosheets in polymer composite materials for gas separation Role of 3-mercaptopyruvate sulfurtransferase in the regulation of proliferation and bioenergetics in murine colon cancer cells An injectable and adaptable hydrogen sulfide delivery system for modulating neuroregenerative microenvironment Pharmacological induction of mesenchymal-epithelial transition via inhibition of H2S biosynthesis and consequent suppression of ACLY activity in colon cancer cells Novel aryl-substituted pyrimidones as inhibitors of 3-mercaptopyruvate sulfurtransferase with antiproliferative efficacy in colon cancer International union of basic and clinical pharmacology CII: pharmacological modulation of H2S levels: H2S donors and H2S biosynthesis inhibitors Colorectal tumor microenvironment-activated bio-decomposable and metabolizable Cu(2)O@CaCO(3) nanocomposites for synergistic oncotherapy Photothermal therapy and photoacoustic imaging via nanotheranostics in fighting cancer Deletion of cystathionine-γ-lyase in bone marrow-derived cells promotes colitis-associated carcinogenesis Targeting hydrogen sulphide signaling in breast cancer Role of hypoxia in cancer therapy by regulating the tumor microenvironment Angiotensin II regulates cellular immune responses through a calcineurin-dependent pathway Nanoparticles from cuttlefish ink inhibit tumor growth by synergizing immunotherapy and photothermal therapy Global characterization of macrophage polarization mechanisms and identification of M2-type polarization inhibitors Comprehensive exome analysis of immunocompetent metastatic head and neck cancer models reveals patient relevant landscapes Legumain-deficient macrophages promote senescence of tumor cells by sustaining JAK1/STAT1 activation Arginase 1-based immune modulatory vaccines induce anticancer immunity and synergize with anti-PD-1 checkpoint blockade Immune cells within the tumor microenvironment: biological functions and roles in cancer immunotherapy Molecular principles of metastasis: a hallmark of cancer revisited Overcoming resistance to tumor-targeted and immune-targeted therapies Hydrogen sulfide creates a favorable immune microenvironment for colon cancer Engineering 2D arsenic-phosphorus theranostic nanosheets Multi-omics analysis reveals the unexpected immune regulatory effects of arsenene nanosheets in tumor microenvironment Anti-Her2 affibody-decorated arsenene nanosheets induce ferroptosis through depleting intracellular GSH to overcome cisplatin resistance Bringing the MMFF force field to the RDKit: implementation and validation B97-3c: a revised low-cost variant of the B97-D density functional method Lu, T. Sobtop Version [1.0(dev3)]. http://sobereva.com/soft/Sobtop (2023) Non-bonded force field model with advanced restrained electrostatic potential charges (RESP2) Multiwfn: a multifunctional wavefunction analyzer Development and testing of a general amber force field PACKMOL: a package for building initial configurations for molecular dynamics simulations GROMACS 4.5: a high-throughput and highly parallel open source molecular simulation toolkit GROMACS: high performance molecular simulations through multi-level parallelism from laptops to supercomputers Structures of MERS-CoV macro domain in aqueous solution with dynamics: Impacts of parallel tempering simulation techniques and CHARMM36m and AMBER99SB force field parameters Confirmation of the formation of salt bridges in the denatured state of CutA1 protein using molecular dynamics simulations Predictions of water/oil interfacial tension at elevated temperatures and pressures: A molecular dynamics simulation study with biomolecular force fields Molecular dynamics simulation of methane hydrates: Prediction of the phase equilibria using extracted microscopic parameters from SAFT-VR Mie EOS Different force fields give rise to different amyloid aggregation pathways in molecular dynamics simulations On the use of the Verlet neighbor list in molecular dynamics Molecular dynamics study on micelle-small molecule interactions: developing a strategy for an extensive comparison Particle mesh Ewald: an N⋅log(N) method for Ewald sums in large systems Engineering of an endogenous hydrogen sulfide responsive smart agent for photoacoustic imaging-guided combination of photothermal therapy and chemotherapy for colon cancer The PRIDE database resources in 2022: a hub for mass spectrometry-based proteomics evidences Download references This work is supported by the National Science Foundation of China (NO.82374048 to F.Z.L. the Natural Science Foundation of Zhejiang Province (NO the Funds of the Natural Science Foundation of Hangzhou (No.2024SZRYBH160002 to J.G.P.) These authors contributed equally: Hongyue Zheng Libraries of Zhejiang Chinese Medical University Ke Zhang, Jigang Piao, Chaofeng Mu, Xiaowei Xie, Mengying Cheng, Tianxiang Yue, Jiang Sun, Bin Li, Yinghui Wei, Hangsheng Zheng, Lai Jiang, Douae Nihed Habiballah & Fanzhu Li provided critical insights and supervised the revision of manuscripts All authors contributed to the final version of the manuscript The authors declare no conflict of interest Download citation DOI: https://doi.org/10.1038/s41467-025-57376-7 Sign up for the Nature Briefing: Cancer newsletter — what matters in cancer research Metrics details The optimization of the enzyme-like catalytic selectivity of nanozymes for specific reactive oxygen species (ROS)-related applications is significant and meanwhile the real-time monitoring of ROS is really crucial for tracking the therapeutic process we present a mild oxidation valence-engineering strategy to modulate the valence states of Mo in Pluronic F127-coated MoO3-x nanozymes (denoted as MF-x x: oxidation time) in a controlled manner aiming to improve their specificity of H2O2-associated catalytic reactions for specific therapy and monitoring of ROS-related diseases MF-0 (Mo average valence 4.64) and MF-10 (Mo average valence 5.68) exhibit exclusively optimal catalase (CAT)- or peroxidase (POD)-like activity Density functional theory (DFT) calculations verify the most favorable reaction path for both MF-0- and MF-10-catalyzed reaction processes based on free energy diagram and electronic structure analysis disclosing the mechanism of the H2O2 activation pathway on the Mo-based nanozymes MF-0 poses a strong potential in acute kidney injury (AKI) treatment achieving excellent therapeutic outcomes in vitro and in vivo the ROS-responsive photoacoustic imaging (PAI) signal of MF-0 during treatment guarantees real-time monitoring of the therapeutic effect and post-cure assessment in vivo providing a highly desirable non-invasive diagnostic approach for ROS-related diseases single or multiple enzyme-like activities remarkably improved facilitating the treatment of ROS-related models such as tumor inflammatory bowel disease and liver injury few guidelines exist for the design and regulation of reaction selectivity (pathway) of nanozymes with multi-activities through valence-engineering approach it is highly desirable to develop renal activatable NIR optical sensors that specifically respond to the kidney related biomarkers (e.g the theranostic probes for real-time imaging and amelioration of AKI are still lacking and challenging The x refers to the oxidation treatment time (hrs) for MF-x The ROS-responsive switchable PAI circumvented false-positive signals from nonspecific retention and successfully realized real-time monitoring of therapeutic process and post-cure assessment for AKI in vivo This work provides a promising example on tuning the H2O2 activation pathway on Mo-based nanozymes for highly efficient catalytic therapy and post-cure assessment via non-invasive PAI of ROS-related diseases TEM image (a) and HAADF-STEM image and elemental mapping (scale bar: 10 nm) (b) of MoO3-x The experiments were independently repeated three times with similar results Mo5+ and Mo4+ and the average valence of Mo in the corresponding samples e UV-vis-NIR absorption spectra of each sample (inset: photographs of each aqueous solution) with obviously different colors (MF-0 and MF-2: dark grey due to the variation of valence state of Mo MF-5 and MF-10 treated with H2O2 under pH 7.4 b Absorbance intensity difference (at 650 nm) between different time points and 0 min of TMB aqueous solution treated with the corresponding samples in the presence of H2O2 (100 μM) under pH 6.5 (means ± SD c ESR spectra of ·OH trapped by DMPO in different samples treated with H2O2 under pH 6.5 performed in a quartz tube at room temperature e Free energy diagram of the reaction process responsible for the CAT- and POD-like activities of MF-0 and MF-10 (inset: the surface configuration of MF-0 and MF-10 at different stages) f The calculated Mo 4d PDOS of MF-0 and MF-10 g Charge density difference of the 2*OH or *O2 on MF-0 and MF-10 h Schematic illustration of the relationship between the average valence of Mo and catalysis selectivity Small spheres of different colors in (e) and (g) stand for various kinds of atoms the production of ROS was not observed at pH 7.4 the POD-like activity of MF-x enhanced with the increase of the average Mo valance the RDS of the ·OH formation pathway was *OH + *OH → *OH + ·OH (·OH desorption) and the RDS of the O2 generation pathway was *O2 → MF-0 + O2 (O2 desorption) with free energy barriers of 2.24 eV and 1.21 eV The lower barrier for the RDS of the O2 generation pathway indicated that MF-0 was more likely to catalyze H2O2 to generate O2 (CAT-like activity) rather than ·OH (POD-like activity) and the RDS of the O2 generation pathway was *O + *H2O2 → *OH + *OOH (proton transfer) with free energy barriers of 0.89 eV and 1.10 eV suggesting the remarkable selectivity toward POD-like activity Overall, the selectivity of H2O2-associated enzymatic reactions of MF-x could be tuned by a feasible mild oxidation valence-engineering strategy, as summarized in the schematic diagram (Fig. 3h) MF-0 (Mo average valence 4.64) and MF-10 (Mo average valence 5.68) exhibited exclusively efficient CAT- or POD-like activity a UV-vis-NIR absorption spectra of MF-0 (160 μg of Mo per mL) after incubation with H2O2 at different concentrations for 2 h (inset: corresponding photographs) ·OH (c) and ABTS· (d) scavenging activity of MF-0 at different Mo concentrations (n = 3 independent experiments) e Viability of MREpiC cells treated with MF-0 at different concentrations (n = 6 independent experiments) f Viability of MREpiC cells under different treatment conditions (n = 6 independent experiments; *P < 0.1 g Fluorescence images of MREpiC cells stained by DCFH-DA (ROS probe) for various treatment groups Group I: cells treated without H2O2 and MF-0; group II: cells treated with H2O2; group III: cells treated with H2O2 and MF-0 (20 μg/mL); group IV: cells treated with H2O2 and MF-0 (50 μg/mL); group V: cells treated with H2O2 and MF-0 (80 μg/mL) h Quantification analysis of DCF fluorescence intensity for groups I ~ V in (g) (n = 20 cells; **P < 0.01 The concentration unit of MF-0 in (f) and (g) is μg/mL; H2O2 concentrations used in (f) and (g) were 250 and 100 μM Data in (b–f) and (h) are presented as means ± SD Significance was calculated by one-sided Student’s t-test which confirmed that the introduction of MF-0 could scavenge intracellular ROS effectively and protect cells from oxidative stress-induced cell damages These suggested that the nanoprobes were still retained at most in the reticuloendothelial system (RES) organs including liver a certain dose of MF-0 could be targeted to the kidneys of AKI mice via renal metabolic pathways providing a favorable prerequisite for subsequent treatment design a Timeline of AKI modeling and treatment with mice CRE (b) and BUN (c) levels in the blood serum from each group after indicated treatments (n = 3; Data are presented as means ± SD d The survival rate of AKI mice treated with PBS and MF-0 e H&E staining of renal sections from each treatment group Arrows indicated damaged tubules and asterisks indicated the formation of casts (a marker of more severe tubular damage) f DAPI (blue fluorescence indicating cell nuclei) and dihydroethidium (red fluorescence indicating ROS level) staining of kidney tissues from each treatment group The injection dosage of agents in different treatment groups was 200 μL: NAC (800 μg/mL) a Photographs of MF-0 solution before and after incubation with H2O2 (3 mM) b UV-vis-NIR absorption spectra of MF-0 (60 μg of Mo per mL) after incubation with H2O2 at different concentrations c Normalized absorbance intensity (at 730 nm) evolution of MF-0 solution (60 μg of Mo per mL) incubated with H2O2 (1 mM) within 30 min (n = 3 independent experiments) d PA intensity (at 730 nm) of MF-0 solution (60 μg of Mo per mL) incubated with H2O2 at different concentrations (n = 3 independent experiments e Timeline of photoacoustic imaging for AKI mice and AKI-cured mice In vivo PA images (f) and the corresponding quantitative analysis (g) of the kidney regions at 730 nm in healthy mice AKI mice and AKI-cured mice at different post-injection time points (0 ΔI indicated PA intensity difference (at 730 nm) between different time points and 0 min MF-0 was intravenously (i.v.) injected into the mice (800 μg of Mo per mL further suggesting the favorable biosafety of MF-0 we developed a mild oxidation valence-engineering strategy to tune the selectivity of molybdenum oxide nanozyme The results demonstrated that the CAT-like activity decreased and the POD-like activity enhanced along with the average valence of Mo rising from 4.64 to 5.68 in MF-x MF-0 (Mo average valence 4.64) and MF-10 (Mo average valence 5.68) exhibited exclusively optimal CAT- or POD-like activity indicating the high specificity toward different reaction pathways involving H2O2 DFT calculations also provided theoretical evidence for the point that the valence state of Mo played a leading role in the catalytic reaction selectivity the electronic structure of MF-x and the free energy barriers of the RDS from H2O2 to O2 or ·OH on MF-x nanozymes were well tuned by the mild oxidation valence-engineering strategy Both in vitro and in vivo experimental results demonstrated that MF-0 could effectively scavenge ROS such as H2O2 and ·OH protecting cells from oxidative stress damage thereby accomplishing highly efficient treatment of AKI MF-0 also presented reliable ROS-responsive photoacoustic imaging ability to achieve intuitively post-cure assessment of AKI in vivo providing an advantageous deep tissue imaging tool for ROS-related diseases that lack effective non-invasive in vivo diagnostic and evaluation approaches This work offers guidance for the rational design of redox nanozymes with high catalysis selectivity and H2O2 elimination as well as new insights into the development of integrated nanoprobes for the treatment and post-cure imaging assessment of ROS-related diseases Oleylamine (OAm) was bought from Acros Organics (Belgium) Oleic acid (OA) and 11-Chloro-1,1’-di-n-propyl-3,3,3’,3’-tetramethyl-10,12-trimethyleneindatricarbocyanine iodide (IR780) was bought from Alfa Aesar (Shanghai) dimethylsulfoxide (DMSO) and sodium citrate were supplied by Beijing Chemical Reagent Company Ltd Sodium molybdate and 5,5’-dimethyl-1-pyrrolin-N-oxide (DMPO) were purchased from Macklin (Shanghai) Hydrochloric acid (HCl) was obtained from Tianjin Fuyu Fine Chemical Co. Dichloromethane (CH2Cl2) and glycerin were obtained from Damao Chemical Reagent Factory (Tianjin) Tris-hydrochloride (Tris-HCl) and Dulbecco’s modified Eagle medium (DMEM) were supplied by Beijing Kebio Biotechnology Co. methyl thiazolyl tetrazolium (MTT) and 2’,7’-dichlorofluorescin diacetate (DCFH-DA) were bought from Beijing InnoChem Science & Technology Co. N-acetyl cysteine (NAC) was purchased from Beyotime (Shanghai) Phosphate buffer solution (PBS) was from M&C Gene Technology (Beijing) Potassium titanium oxalate (PTO) was obtained from Shanghai Titan Scientific Co. Methylene blue (MB) was acquired from Tianjin Guangfu Fine Chemical Research Institute FeCl2·4H2O was acquired from Xiya Reagent (Shandong) Diammonium 2,2’-azino-bis(3-ethylbenzothiazoline-6-sulfonate) (ABTS) was bought from Shanghai Yuanye Bio-Technology Co. Potassium persulfate was supplied by Fuchen Chemical Reagent Co. Paraformaldehyde was obtained from Tianjin Chemical Reagent Research Institute All of the abovementioned chemicals were of analytical grade and used as received without further purification Ultrapure water used throughout all experiments was obtained from a Millipore Milli-Q purification system (U.S.) Transmission electron microscope (TEM) images were obtained using a JEM-1200EX (JEOL) transmission electron microscope at a voltage of 100 kV and the data of size distribution was analyzed by Nano Measurer 1.2 High-angle annular dark-field scanning transmission election microscope (HAADF-STEM) images and energy-dispersive X-ray (EDX) elemental mapping were collected on a JEM-ARM200F (JEOL) spherical aberration-corrected transmission electron microscope at a voltage of 200 kV The lattice spacing was measured by Digital Micrograph 3.7 Bio-TEM-EDX elemental mapping images were obtained using a HRTEM-JEM-F200 (JEOL) high-resolution field emission transmission electron microscope at a voltage of 100 kV Fourier transform infrared (FT-IR) spectra were obtained on a Nexus 8700 Fourier transform infrared spectrophotometer (Nicolet) Thermogravimetric (TG) analysis was performed on a Mettler Toledo TGA/DSC3+ thermogravimeter X-ray photoelectron spectrum (XPS) was carried out on a K-Alpha spectrometer (Thermo Fisher Scientific) UV-vis-NIR absorption spectra were recorded on a UV-3600 spectrophotometer (Shimadzu) Oxygen production ability was measured using a F4-Standard Kit dissolved oxygen meter (Mettler Toledo) Electron spin resonance (ESR) spectra were acquired on an EMX-500 spectrometer (Bruker) Cell viability was detected using a Tecan Infinite F50 plate reader (Switzerland) Photographs of cell imaging and kidney tissue section staining were obtained using a model eclipse Ti2-U inverted fluorescence microscope (Nikon) The X-ray diffraction (XRD) patterns were acquired on a Bruker AXS D8-Advanced X-ray diffractometer with Cu Kα radiation (λ = 1.5418 Å) Dynamic light scattering (DLS) size distribution and zeta potential were carried out using a Zetasizer Nano-ZS90 zeta potential and size analyzer (Malvern) The content of Mo accumulated in the major organs (heart kidney and intestine) and the urine of AKI mice was detected by an inductively coupled plasma-mass spectrometer (ICP-MS Fluorescence imaging was performed on a IVIS Spectrum (PerkinElmer) Photoacoustic imaging (PAI) was performed on a multispectral optoacoustic tomography scanner (MSOT MoO3-x NPs were prepared through a hydrothermal method oleylamine (2 mL) and oleic acid (1 mL) were dissolved in a mixed solvent containing 6 mL of n-hexane and 4 mL of ethanol 2 mL of sodium molybdate aqueous solution (50 mg/mL) and 1.2 mL of hydrochloric acid (6 M) were added under vigorous stirring and keeping for 30 min the mixture was transferred into a Teflon-lined autoclave (20 mL) and heated at 200 °C for 6 h the product was washed with n-hexane and ethanol and collected by centrifugation (8497 g the obtained MoO3-x NPs were redispersed in dichloromethane for future use MoO3-x NPs (1 mg) dispersed in 200 μL of CH2Cl2 were added into 10 mL of ultrapure water containing amphiphilic Pluronic F127 (50 mg) The mixture was treated by ultrasonication (400 W) in ice bath for 8 min (3 s/3 s causing the evaporation of CH2Cl2 and forming a uniform clear light grey solution the above solution was stirred vigorously at 37 °C in the open container for 0 followed by rotary evaporation to obtain the stock solution (1 mL) with various degrees of oxidation The corresponding samples were named as MF-x respectively (mass concentration: 40 mg/mL; concentration of molybdenum (Mo): 400 μg/mL) CAT-like activity of the samples (MF-x) was indicated by the oxygen generation ability in the presence of H2O2 200 μL) was dispersed into 1.8 mL of Tris-HCl buffer (pH 6.5 or pH 7.4) and then the MF-x samples (200 μL) were added The produced O2 was measured at different time points TMB (3,3,5,5-tetramethylbenzidine) oxidation assay was carried out to evaluate the POD-like activity of samples as the oxidized TMB (oxTMB) showing maximum absorption at 650 nm 50 μL of TMB ethanolic solution (4 mg/mL) was added into 1.78 mL of Tris-HCl buffer (pH 6.5 or pH 7.4) and then treated with 50 μL of H2O2 solution (4 mM) and each of MF-x samples (120 μL) The absorption spectra of oxTMB solutions at different time points were recorded the type of produced ROS at pH 6.5 was further explored by ESR analysis 85 μL of Tris-HCl buffer (pH 6.5) was mixed with 10 μL of DMPO (a spin trapping agent 5 μL of H2O2 solution (4 mM) and 10 μL of each MF-x sample ESR measurements were performed and the ROS was determined to be ·OH Since titanium ion can react with H2O2 to produce a stable orange-yellow complex with a maximum absorption at 381 nm potassium titanium oxalate (PTO) probe was served to assess the H2O2 scavenging capacity of MF-0 MF-0 and H2O2 solution (20 mM) were mixed in ultrapure water and reacted for 10 min and the final concentrations of H2O2 and PTO were 0.5 mM and 1.0 mM The absorbance of H2O2 solution with PTO (AP) and a mixed solution of PTO 20 and 30 μg of Mo per mL) (AQ) was measured at 381 nm Due to a certain of absorption at 381 nm for MF-0 the absorbance was deducted to avoid interference H2O2 inhibition rate (%) = [(AP - AQ)/AP] × 100% Methylene blue (MB) can be bleached by ·OH thus the absorbance of remaining MB was used to evaluate ·OH scavenging capacity of MF-0 600 μL of FeCl2·4H2O (5 mM) mixed with 400 μL of H2O2 (5 mM) was prepared as ·OH working solution based on the Fenton reaction MF-0 and MB (1 mM) was added successively and kept for another 50 min MB solution with ·OH (AN) and a mixed solution of MB 5 and 10 μg of Mo per mL) (AF) was measured at 664 nm ·OH inhibition rate (%) = [(AF - AN)/(AM - AN)] × 100% Diammonium 2,2’-azino-bis(3-ethylbenzothiazoline-6-sulfonate) (ABTS) decolorization assay was carried out to assess ABTS radical scavenging capacity of MF-0 ABTS aqueous solution (7 mM) and potassium persulfate solution (2.45 mM) were mixed in a 1:1 volume ratio and stored in the dark for 12 h to generate ABTS radical the absorbance of pure ABTS· solution (AB) and ABTS· solution mixed with MF-0 (1 5 and 10 μg of Mo per mL) for 20 min (AE) was measured at 730 nm ABTS· inhibition rate (%) = [(AB - AE)/AB] × 100% pyrogallol can undergo autoxidation to produce ·O2- and colored intermediates whose absorption intensity is proportional to the amount of ·O2- pyrogallol was used to verify the ·O2- scavenging activity of MF-0 an antioxidant as the positive control) was dispersed in Tris-HCl buffer (pH 8.0) then pyrogallol (final concentration 250 μM) was added and incubated for 20 min The absorbance of pure pyrogallol solution (AZ) a mixed solution of pyrogallol and MF-0 (AX) as well as a mixed solution of pyrogallol and AA (AY) was measured at 319 nm Due to a certain of absorption at 319 nm for MF-0 ·O2- inhibition rate (%) of MF-0 = [(AZ - AX)/AZ] × 100% ·O2- inhibition rate (%) of AA = [(AZ - AY)/AZ] × 100% Mouse kidney epithelial cells (MREpiC) were obtained from Hunan Fenghui Biotechnology Co. MTT assay was implemented to assess the cytotoxicity of MF-0 MREpiC cells were seeded in a 96-well plate at 2 × 104 cells per well and incubated for 24 h (37 °C MF-0 dispersed in PBS (final concentration: 0 200 and 400 μg/mL) were added into each well and cultured for 24 h 20 μL of MTT (5 mg/mL) was introduced and incubated for an additional 4 h the absorbance of colored formazan produced from MTT and dissolved by DMSO was measured at 492 nm using a microplate reader MREpiC cells were seeded in a 96-well plate followed by culturing with MF-0 (final concentration: 0 50 and 80 μg/mL) in the presence of H2O2 (250 μM) for 24 h The cell viability was measured by MTT assay to display the ROS scavenging efficiency DCFH-DA was used to intuitively explore the intracellular scavenging effect of MF-0 MREpiC cells were seed in a 6-well plate at 3 × 105 cells per well and cultured for 24 h (37 °C 50 and 80 μg/mL) were added into each well followed by the addition of H2O2 (100 μM) the cells were further incubated with DCFH-DA (10 μM) dispersed in DMEM for 30 min the cells were washed again and observed through a fluorescence microscope (λex = 480 nm The quantification analysis of DCF fluorescence intensity was from ImageJ 1.52 All animal experiments were performed with the approval of the Institutional Animal Care and Use Committee of the China-Japan Friendship Hospital (Beijing) Six weeks old of female BALB/c mice were provided for the animal experiments The findings apply to both male and female animals All mice were housed in a 12/12 h dark/light cycle with the standard conditions: Temperature The mice were deprived of water but able to have food for 15 h then 50% glycerin (8 mL/kg) was intramuscularly (i.m.) injected into both hind legs equally followed by the provision of water and food as normal For the sake of in vivo fluorescence (FL) imaging MF-0 was labeled with a fluorescent dye IR780 to produce IR780@MF-0 MoO3-x NPs (1 mg) and IR780 (200 μg) were dispersed in 200 μL of CH2Cl2 and then added into 10 mL of ultrapure water containing Pluronic F127 (50 mg) The mixture was treated with ultrasonication for 8 min (3 s/3 s on/off) to acquire a clear grey-green solution AKI mice (back hair removed) were intravenously injected with 150 μL of IR780@MF-0 (600 μg of Mo per mL) to perform FL imaging at 10 the amount of MF-0 accumulated in the various organs (heart kidney and intestine) of AKI mice was evaluated Two hours after the establishment of AKI model 200 μL of MF-0 solution was intravenously injected into the mice The mice were sacrificed and the organs were harvested at different post-injection time points (0 the biodistribution of MF-0 nanoparticles in the kidney of AKI mice was further investigated by bio-TEM-EDX elemental mapping After intravenous injection of MF-0 (200 μL) for 30 min the mice were sacrificed and small slices of renal cortex were collected after the perfusion of the heart of mouse with 5 mL of normal saline and 5 mL of 4% paraformaldehyde solution consecutively Then the renal cortex sections were fixed with 2.5% glutaraldehyde for 24 h post-fixed with 1% osmium tetroxide and stained with lead citrate and uranium acetate for bio-TEM-EDX analysis Two hours after the establishment of AKI model (glycerol injection) therapeutic agents (200 μL) were intravenously (i.v.) injected into the mice of various treatment groups: (1) healthy mice treated with PBS; (2) healthy mice treated with MF-0 (800 μg of Mo per mL); (3) AKI mice treated with PBS; (4) AKI mice treated with the antioxidant NAC in PBS (800 μg/mL); (5) AKI mice treated with MF-0 (800 μg of Mo per mL) to assess the therapeutic effect of MF-0 in AKI mice All the mice were sacrificed after 24 h of AKI model implementing In order to evaluate the treatment effect of MF-0 for AKI blood samples were centrifuged at 2000 rpm for 5 min to collect plasma for renal function test (creatinine and blood urea nitrogen) a portion of kidneys were harvested and stored in 4% paraformaldehyde solution for hematoxylin and eosin (H&E) staining and the remaining parts were frozen rapidly by liquid nitrogen and set at –80 °C for subsequent ROS staining experiments The ROS staining images were processed by ImageJ 1.52 The mice used for in vivo PAI were divided into three groups: (1) healthy mice; (2) AKI mice; and (3) AKI-cured mice The back hair of the mice was removed to acquire clear imaging effects 200 μL of MF-0 solution (800 μg of Mo per mL) were intravenously (i.v.) injected into the mice of different groups 60 and 120 min after the injection of MF-0 the MF-0 was injected 2 h after glycerol injection; and for AKI-cured mice the MF-0 was given a second injection 24 h after glycerol injection The corresponding data were obtained from viewMSOT 4.0 100 μL of blood was taken from the mice and dispersed in 5% sodium citrate solution 200 and 400 μg/mL) were mixed with 100 μL of RBCs suspension H2O and PBS were served as positive and negative control groups the supernatants from all groups were collected by centrifugation (543 g The obtained supernatants were transferred to a 96-well plate and then the absorbance at 414 nm was measured Hemolytic degree (%) = [(Asample -Anegative)/(Apositive - Anegative)] × 100% The optimized models for MF-0 and MF-10 in this work correspond to MoO2 and MoO3 mentioned above The corresponding data were processed by VESTA 3.4 The data were analyzed using Origin software (version 2022) All experiments were repeated at least 3 times and presented as means ± SD Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article Matching the kinetics of natural enzymes with a single-atom iron nanozyme Cascaded nanozyme system with high reaction selectivity by substrate screening and channeling in a microfluidic device Atomic-level regulation of cobalt single-atom nanozymes: engineering high-efficiency catalase mimics Intrinsic strain-mediated ultrathin ceria nanoantioxidant Surface ligand engineering ruthenium nanozyme superior to horseradish peroxidase for enhanced immunoassay Bioinspired porous three-coordinated single-atom Fe nanozyme with oxidase-like activity for tumor visual identification via glutathione Biosystem-inspired engineering of nanozymes for biomedical applications Stimuli-responsive manganese single-atom nanozyme for tumor therapy via integrated cascade reactions Chiral metal-organic frameworks incorporating nanozymes as neuroinflammation inhibitors for managing Parkinson’s disease Nitrogen-doped carbon nanomaterials as highly active and specific peroxidase mimics Modulation of the biocatalytic activity and selectivity of CeO2 nanozymes via atomic doping engineering Coordination number regulation of molybdenum single-atom nanozyme peroxidase-like specificity Regulating the H2O2 activation pathway on a well-defined CeO2 nanozyme allows the entire steering of its specificity between associated enzymatic reactions Facet energy versus enzyme-like activities: the unexpected protection of palladium nanocrystals against oxidative damage Differential Pd-nanocrystal facets demonstrate distinct antibacterial activity against Gram-positive and Gram-negative bacteria Reducing valence states of Co active sites in a single-atom nanozyme for boosted tumor therapy A valence-engineered self-cascading antioxidant nanozyme for the therapy of inflammatory bowel disease Ultrasmall ruthenium nanoparticles with boosted antioxidant activity upregulate regulatory T cells for highly efficient liver injury therapy Molybdenum derived from nanomaterials incorporates into molybdenum enzymes and affects their activities in vivo Triggering of low-valence molybdenum in multiphasic MoS2 for effective reactive oxygen species output in catalytic Fenton-like reactions The age of bioinspired molybdenum-involved nanozymes: synthesis Defect engineering enables synergistic action of enzyme-mimicking active centers for high-efficiency tumor therapy Intercalation-activated layered MoO3 nanobelts as biodegradable nanozymes for tumor-specific photo-enhanced catalytic therapy Coordination-driven self-assembly strategy-activated Cu single-atom nanozymes for catalytic tumor-specific therapy One-pot synthesis of PEGylated plasmonic MoO3-X hollow nanospheres for photoacoustic imaging guided chemo-photothermal combinational therapy of cancer Hydrophilic molybdenum oxide nanomaterials with controlled morphology and strong plasmonic absorption for photothermal ablation of cancer cells A unique multifunctional nanoenzyme tailored for triggering tumor microenvironment activated NIR-II photoacoustic imaging and chemodynamic/photothermal combined therapy Atomic-level nanorings (A-NRs) therapeutic agent for photoacoustic imaging and photothermal/photodynamic therapy of cancer Paradigms of acute kidney injury in the intensive care setting Acute kidney injury and chronic kidney disease as interconnected syndromes Epigenetic regulation in AKI and kidney repair: mechanisms and therapeutic implications Ultrasmall copper-based nanoparticles for reactive oxygen species scavenging and alleviation of inflammation related diseases Cerium-luteolin nanocomplexes in managing inflammation-related diseases by antioxidant and immunoregulation Prevention of radiographic-contrast-agent-induced reductions in renal function by acetylcysteine Ruthenium-based metal-organic framework with reactive oxygen and nitrogen species scavenging activities for alleviating inflammation diseases Self-cascade uricase/catalase mimics alleviate acute gout Catalytic activity tunable ceria nanoparticles prevent chemotherapy-induced acute kidney injury without interference with chemotherapeutics Ceria nanozyme-integrated microneedles reshape the perifollicular microenvironment for androgenetic alopecia treatment Se-containing MOF coated dual-Fe-atom nanozymes with multi-enzyme cascade activities protect against cerebral ischemic reperfusion injury Edge-site engineering of defective Fe–N4 nanozymes with boosted catalase-like performance for retinal vasculopathies Fluoro-photoacoustic polymeric renal reporter for real-time dual imaging of acute kidney injury A promising NIR-II fluorescent sensor for peptide-mediated long-term monitoring of kidney dysfunction NIR-II photoacoustic-active DNA origami nanoantenna for early diagnosis and smart therapy of acute kidney injury Rational design of a double-locked photoacoustic probe for precise in vivo imaging of cathepsin B in atherosclerotic plaques Zwitterionic sulfhydryl sulfobetaine stabilized platinum nanoparticles for enhanced dopamine detection and antitumor ability self-limiting growth of glucose oxidase-mimicking gold nanoparticles Intrinsic peroxidase-like activity of ferromagnetic nanoparticles Assessment of the Perdew–Burke–Ernzerhof exchange-correlation functional Generalized gradient approximation made simple and surface reactivity of α-MoO3/γ-Al2O3 composites: DFT and DFT + U calculations DFT studies of selective oxidation of propene on the MoO3(010) surface High-performance Mo2C/MWCNT electrocatalyst for MOR: comparison with MoO2/MWCNT and MoO3/MWCNT A consistent and accurate ab initio parametrization of density functional dispersion correction (DFT-D) for the 94 elements H-Pu Effect of the damping function in dispersion corrected density functional theory Implicit solvation model for density-functional study of nanocrystal surfaces and reaction pathways Electrochemical potential-driven shift of frontier orbitals in M-N-C single-atom catalysts leading to inverted adsorption energies Structure-performance descriptors and the role of the axial oxygen atom on M-N4-C single-atom catalysts for electrochemical CO2 Reduction Theoretical understanding of the electrochemical reaction barrier: a kinetic study of CO2 reduction reaction on copper electrodes Theoretical investigation of 2D conductive microporous coordination polymers as Li-S battery cathode with ultrahigh energy density Computational investigation on hydrodeoxygenation (HDO) of acetone to propylene on α-MoO3 (010) Surface Role of oxygen vacancy defects in the electrocatalytic activity of substoichiometric molybdenum oxide Oxygen healing and CO2/H2/anisole dissociation on reduced molybdenum oxide surfaces studied by density functional theory Download references This research was supported in part by the National Natural Science Foundation of China (22334002 the National Natural Science Foundation of China (22076010 Beijing Municipal Science and Technology Special Project (Z231100002723006 and the Fundamental Research Funds for the Central Universities (JD2308 and XK2023-19 State Key Laboratory of Chemical Resource Engineering All the authors discussed the results and commented on the manuscript Nature Communications thanks Jingxiang Zhao and the other Download citation DOI: https://doi.org/10.1038/s41467-024-53047-1 Metrics details Acidic electrochemical CO2 conversion is a promising alternative to overcome the low CO2 utilization over-reliance on highly concentrated K+ to inhibit the hydrogen evolution reaction also causes (bi)carbonate precipitation to interfere with catalytic performance under the screening and guidance of computational simulations we present a carbon coated tip-like In2O3 electrocatalyst for stable and efficient acidic CO2 conversion to synthesize formic acid (HCOOH) with low K+ concentration The carbon layer protects the oxidized In species with higher intrinsic activity from reductive corrosion and also peripherally formulates a tip-induced electric field to regulate the adverse H+ attraction and desirable K+ enrichment only 0.1 M low K+ is required to achieve a Faradaic efficiency (FE) of 98.9% at 300 mA cm−2 for HCOOH and a long-time stability of over100 h By up-scaling the electrode into a 25 cm2 electrolyzer setup a total current of 7 A is recorded to sustain a durable HCOOH production of 291.6 mmol L−1 h−1 a Schematic illustration of challenges in acidic CO2RR c Potential-dependent concentrations of K+ and H+ in tip-featured catalyst d The effect of tip carbon coating on the distribution of H+ e Diffusion-dependent concentrations of K+ and H+ in tip-featured catalyst with carbon coating f Schematic modeling of the electronic interaction between the carbon layer with In2O3 The blue and yellow regions indicate electron loss and gain g Influence of carbon layer on oxygen vacancy generation h Gibbs free energy diagrams for HCOOH on In and In2O3 The refresh step includes HCOOH desorption and CO2 re-adsorption The catalytic cycle starts from the second step as indicated by the arrows The possibly tip-induced H+ enrichment effect is generally overlooked by relevant studies which may be deviated from the original motivation of designing locally concentrated K+ to restrain HER kinetics an external concentration field is required to cooperate with the tip-induced electric field to selectively permeate K+ and throttle H+ ions flux the concentration field regulation module is also required to ideally protect the oxidation station of catalysts therefore ensuring efficient and durable HCOOH synthesis in low K+-mediated acidic CO2 electrolysis theoretical calculations are performed to screen candidates with suitable concentration field modulation and oxidation state protection capabilities Simulation modeling reveals that applying a tightly coated carbon layer on the tip-like catalyst selectively ensures tip-induced K+ cumulation while eliminating the adverse H+ enrichment The robust anti-reduction property is also endowed to the encapsulated In2O3 catalyst for sustaining high HCOOH activity the carbon coated tip-like In2O3 model catalyst is prepared by vacuum pyrolysis which delivers a superior HCOOH FE of 98.9% at a current density of 300 mA cm−2 in H2SO4 (pH = 0.94) solution with 0.1 M K+ together with long-term catalytic stability over 100 h Despite at a reduced current density of 50 mA cm−2 the carbon coated tip-like In2O3 still manifests significant inhibition of HER with a minimized FE of only 18% in sharp contrast to the 43% of the carbon-free counterpart In situ characterization measurements combined with theoretical calculations demonstrate the dual-field synergy the interfacial concentration field and the tip-induced electric field on regulating the reaction microenvironment and catalyst oxidation state This work provides a technically feasible and economically valuable solution to accelerate the industrialization of CO2RR technology which showcased the considerably mitigated H+ accumulation due to carbon layer confinement and the concentrated K+ was ideally retained the cooperation of tip-like structure and carbon layer confinement led to selective K+ enrichment at the catalyst surface through the synergy of concentration and electric fields which is expected to realize efficient acidic CO2RR at low K+ concentrations The adsorption strength of CO2 on In2O3 was much stronger than that on In favoring the reactant enrichment at the catalyst surface Both In2O3 and In tended to produce HCOOH along two-electron CO2RR and the overall reaction was limited by the second electrochemical proton-coupled electron transfer step The free energy change of this rate-limiting step on In2O3 was significantly lower than that on In implying an efficient conversion of CO2 to HCOOH on In2O3 the carbon layer was anticipated to maintain the high intrinsic catalytic performance of In2O3 by preventing its reduction under the CO2RR working conditions f HRTEM and (g) EDS elemental mapping of Vac Vac-air and Ar-air in 0.05 M H2SO4 with different K+ concentrations Vac-air and Ar-air in 0.05 M H2SO4 electrolyte with 0.1 M K+ g Catalytic stability test at 300 mA cm−2 of Vac Error bars represent the standard deviation of three independent measurements The applied potentials for in situ measurements were without iRΩ correction the designed model catalysts based on computational simulations proposed a feasible approach to improve the catalytic activity and stability in low K+ acidic CO2RR system it is rationalized that the desired electrocatalytic performance of Vac was boosted by simultaneous optimization of intrinsic catalytic activity and local microenvironmental b In 3 d XPS spectra and (c) O 1 s XPS spectra of Vac before and after CO2RR d In K-edge XANES of Vac before and after CO2RR e Calculated valance state by linear combination fitting of Vac before and after CO2RR f FT-EXAFS spectra and (g) WT-EXAFS plots of Vac before and after CO2RR Vac-air and Ar-air in Ar and CO2-saturated electrolyte RHE and (c) corresponding calculated jK and DH+ In situ ATR-SEIRAS spectra with three O-H stretching mode through Gaussian fitting of (d) Vac There is no iRΩ correction for applied potentials One possible explanation is that the electric field acting on K+(H2O)n for Vac-air may be partially shieled by the concurrent cumulation of H+ within the EDL the interfacial K+(H2O)n in Vac featured a larger dipole moment due to the selective permeation of H+ through the carbon layer and demonstrated a fast response to potential switching The above polarization curves and in situ spectroscopic analyzes confirmed the crucial role of microenvironmental regulation in the low K+ acidic CO2RR a Photograph of a scale-up electrolyzer with 25 cm2 b Current-dependent potential plots of Vac d Long-term stability test of Vac at total current of 7 A and (e) corresponding time-dependent HCOOH production rate The cell voltages were compensated by iRΩ correction with an RΩ value of ~6.3 ± 0.1 Ω we demonstrated an efficient electrochemical route for synthesizing HCOOH from acidic CO2 electrolysis in a low K+ environment Based on the cumulation effect of the tip-induced electric field for cationic species and selectively blocked H+ flux by carbon layer modulation a specific enrichment of K+ was achieved to generate a local alkaline buffer favorable for CO2 conversion in a strongly acidic (pH = 0.94) and low K+ (0.1 M) electrolyte the carbon layer also strengthens the stability of oxidized In sites under reductive CO2RR conditions guaranteeing both optimized microenvironment and robust intrinsic activity to deliver superior catalytic performance a high FE of 80.5% for HCOOH was delivered even at a low current density of 50 mA cm−2 and a superior FE of 98.9% for HCOOH was maintained during operation at 300 mA cm−2 over 100 h The performance metrics in the demonstration of a 25 cm2 scale-up reaction device with a total current of 7 A highlight the great promise of deploying the low K+ acidic CO2RR toward industrialization All chemicals including indium nitrate (In(NO3)3 99%) were purchased from Aladdin Chemical Reagent Co. Nafion 115 proton exchange membranes were purchased from Fuel Cell Store 99.99%) were purchased from Hua er Wen Gas Ltd All materials were used without undergoing any additional purification processes 4,5-imidazoledicarboxylicacid (0.42 g) and benzimidazole (2 g) were added sequentially into 120 mL DMF solution the mixture was subjected to 10 mins of magnetic stirring after which it was transferred to an oil bath maintained at 120 °C for 4 h to allow for a white precipitate formation The precipitate was subjected to multiple washes with ethanol and DMF and subsequently dried in a vacuum oven to yield the In-rho-ZMOF powder 200 mg of In-rho-ZMOF was placed in a tube furnace and a vacuum pump was used to maintain a low-pressure atmosphere in the quartz tube followed by raising the temperature (2 °C min−1) to 500 °C and maintaining for 1 h The light-yellow Vac-air powder was obtained by heating the Vac powder at 500 °C for 3 h in a muffle furnace 200 mg of In-rho-ZMOF was placed in a tube furnace where the air inside the tube was purged with Ar gas The In-rho-ZMOF powder was heated at 500 °C for 1 h to obtain a black powder The above black powder was subsequently heated at 500 °C for 3 h in a muffle furnace to obtain a light-yellow Ar-air powder The preparation process for Vac-1 was identical to that of Vac with the exception being a reduction in the vacuum degree The Vac-2 powder was obtained by heating the Vac powder at 300 °C for 1 h in a muffle furnace The bare carbon was prepared by calcining the ligand (4,5-imidazoledicarboxylicacid and benzimidazole) in an Ar-filled tube furnace at 500 °C for 1 h X-ray diffraction (XRD) patterns were recorded using a Bruker D8 ADVANCE instrument with Cu Kα radiation X-ray photoelectron spectroscopy (XPS) was conducted using a ThermoFisher ESCALAB 250XI instrument which is equipped with a 200 W monochromatic Al Kα radiation and a charge neutralizer Scanning electron microscope (SEM) was operated and collected using a Hitachi New Generation SU8010 Transmission electron microscopy (TEM) was performed using a JEOL JEM 2100 F X-ray absorption spectroscopy (XAS) at the In K-edge of samples were performed at the XAFCA beamline of the Shanghai Synchrotron Light Source and the data were processed and analyzed using the Demeter software package Contact angle measurements were performed using a Dataphysics OCA20 All electrochemical measurements were carried out in ambient conditions The acidic CO2RR was evaluated in a three-electrode flow cell configuration with Gamry Reference 3000 and Reference 30k Booster which was constructed with a gas diffusion electrode of Sigracet 28BC (Fuel Cell Store) a KCl saturated (3.5 M) Ag/AgCl and a 1 cm × 3 cm × 1 mm Pt foil serving as the working Fuel Cell Store) was employed for the separation of the cathode and anode chambers after the sequentially soaked for 1 h at 80 °C in 5 wt% hydrogen peroxide The reference electrode can be selected for use by testing the open-circuit potential difference between the reference electrode and the reversible hydrogen electrode in two-electrode mode and the potential difference can be maintained within 3 mV for 1000 s To prepare the working electrode (1 cm × 3 cm) 10 mg of catalyst was homogeneously dispersed in a mixture containing 0.9 mL of ethanol and 100 µL of Nafion solution (5 wt% and then 300 µL of the catalyst ink was pipetted onto the Sigracet 28BC (loading: ~1 mg cm−2) and dried overnight The entire electrolyzer configuration was mounted and secured using polytetrafluoroethylene gaskets and controlled the reactive electrolytic area to 1 cm2 (2 cm × 0.5 cm) the electrochemical tests were performed on a Gamry Reference 3000 with Reference 30k Booster in a homemade two-electrode electrolyzer and the area of the cathode gas diffusion electrode was 5 cm × 5 cm with a catalyst loading of 1 mg cm−2 anode was the porous titanium oxide electrode with iridium oxide/ruthenium oxide and Nafion 115 (6 cm × 6 cm × 127 μm) as the membrane The electrolyte was circulated within the reaction chambers through a BT 100 M pump (Baoding Chuang Rui Precision Pump Co. CO2 was injected into the cathode gas channel using a mass flowmeter (HORIBA The electrolyte contained 0.05 M H2SO4 and various concentrations of K2SO4 at 0.01 M The potential was applied against the Ag/AgCl electrode and all values were transformed into reversible hydrogen electrode potential through the Nernst equation with iRΩ compensation conversion: The electrode potentials for two-electrode electrolyzer were also iRΩ-corrected using the following formula: where the Ecell is the compensated cell voltage Eapplied is the actual applied cell voltage RΩ is the uncompensated resistance quantified by extrapolation of the impedance data (~4.1 ± 0.1 Ω in the three-electrode flow cell and ~6.3 ± 0.1 Ω in homemade two-electrode electrolyzer) pH is ~0.94 ± 0.05 in 0.05 M H2SO4 with various K2SO4 concentrations measured with a pH meter (PHS−3E) The chronopotentiometry curves were collected in the CO2 atmosphere at different current densities The OH− adsorption measurements were conducted in an Ar-saturated 1.0 M KOH electrolyte to record the LSV curves Rf is the roughness factor (Cdl/Cs) and S represents the electrode active geometric area (1 cm2) The value of Cs is a constant of 60 μF cm−2 Cdl was determined by plotting the current verse the scan rate (20 to 100 mV s−1) in CV curves which were obtained in a 0.1 M KHCO3 electrolyte with CO2 saturation EIS was recorded at varying potentials with a frequency range of 0.01 to 105 Hz and an amplitude of 10 mV to determine the resistance and pseudo-capacitance from the fitting result The Shimadzu GC-2014 gas chromatograph was used to qualitatively and quantitatively analyze the gas phase products through the flame ionization detector (FID) and thermal conductivity detector (TCD) Faradaic efficiency (FE) of gas products in this work was calculated with the following formula: n represents the number of moles of transfer electrons for producing a gas product (CO: 2; H2: 2) V denotes the measured volume concentration of gas products v is the flow rate of CO2 fed gas (mL min−1) The liquid product was quantitatively analyzed with a 400 MHz 1H-NMR and 13C-NMR spectrometer using the internal standard method This entailed the mixing of 500 µL of electrolyte with 100 µL of deuterated water (D2O) along with the addition of 50 µL of a mixed solution of dimethyl sulfoxide (5 mM) and phenol (25 mM) The FE for liquid products was determined on the formula: where the z represents the number of moles of transfer electrons for producing a liquid product (HCOOH: 2) and Q is the total electrons consumed during the whole CO2RR The single-pass utilization (SPU) of CO2 towards HCOOH was calculated as follow formula: where the jtotal represents the total current applied to work electrode ATR-SEIRAS spectra (Bruker Vertex70 spectrometer) equipped with a VeeMAX III ATR accessory was used for the in situ ATR-SEIRAS investigation The counter electrode and reference electrode in the in situ ATR-SEIRAS electrochemical cell are a Pt wire and a Ag/AgCl electrode A gold film on the surface of a semicylindrical silicon prism substrate served as the working electrode The electrolyte was identical to that used in the electrochemical evaluation and a continuous CO2 flow was maintained throughout the experiment period The electrochemical cell was linked to a CHI 760E electrochemical workstation and subjected to constant potential electrolysis at varying potentials A confocal Raman microscope (LabRAM HR800) equipped with a 1200 groove mm−1 diffraction grating was used for the in situ Raman measurements The 532 nm air-cooled laser beam was selected as the excitation source The counter electrode and reference electrode in the in situ Raman electrochemical cell are a Pt wire and a Ag/AgCl electrode The gas diffusion electrode sprayed with catalysts ink was used as the working electrode The LSV curves were measured at varying rotating speeds in the CO2 or Ar atmosphere saturated electrolyte The kinetic current density can be obtained from the Koutecký-Levich equation: jK is the measured kinetic-limited current density of H+ reduction jplateau is the plateau current density of H+ reduction ω is the rotating speed of the RDE (rad s−1) where DH+ is the diffusion coefficient of H+ velectrolyte is the kinematic viscosity of electrolyte The value of DH+ was calculated according to Levich equation: The electric potential difference between the bulk electrolyte and the catalyst was changed from −0.01 to −0.05 V More negative potential not only enriche the cations near the electrode surface but also enlarge the concentration gradient requiring denser boundary layer meshes and making calculations more costly We have tested the mesh independence of the current results with the electric potential difference up to −0.05 V and the surface charge was calculated during the simulations K+ and SO42− at the bulk electrolyte were set to 0.1 M corresponding to the electrolyte solution using in the catalytic system The consumption of H+ was modeled by setting a negative flux of −0.01 mol m−2 s−1 at the catalyst which corresponds to a current density of 100 mA cm−2 Considering the current density is defined with the geometric area of the electrode we set the flux boundary condition at the bottom of the catalyst to guarantee the conceptional consistence we set large diffusion coefficients (10 times larger than the electrolyte diffusion coefficients) in the catalyst domain to eliminate the influences of this special treatment K+ and SO42− in the electrolyte were set to 9.31 × 10−9 We employed an isosceles triangle with a height of 500 nm and a width of 400 nm to represent the catalyst which is comparable to our experimental samples The thickness of the carbon layer was set to about 10 nm The triangle was deposed at the bottom of a 1000 nm × 1000 nm square to simulate the electrolyte We set the ion concentrations and the electric potential of the electrolyte at the top edge of the square The left and right sides of the square were set to insulate without flux to mimic the symmetric boundary condition Durable CO2 conversion in the proton-exchange membrane system Nanoscale engineering of p‐block metal‐based catalysts toward industrial‐scale electrochemical reduction of CO2 Recent advances in p-block metal chalcogenide electrocatalysts for high-efficiency CO2 reduction A CO2 utilization framework for liquid fuels and chemical production: techno-economic and environmental analysis CO2 electroreduction to ethylene via hydroxide-mediated copper catalysis at an abrupt interface Effects of electrolyte ionic species on electrocatalytic reactions: Advances Highly selective electrocatalytic CO2 conversion to tailored products through precise regulation of hydrogenation and C-C coupling Nano‐engineering in zinc‐based catalysts for CO2 electroreduction: advances and challenges Advances in cutting‐edge electrode engineering toward CO2 electrolysis at high current density and selectivity: A mini‐review High-resolution neutron imaging of salt precipitation and water transport in zero-gap CO2 electrolysis Recovering carbon losses in CO2 electrolysis using a solid electrolyte reactor Advanced catalyst design and reactor configuration upgrade in electrochemical carbon dioxide conversion Energy- and carbon-efficient CO2/CO electrolysis to multicarbon products via asymmetric ion migration–adsorption CO2 electrolysis to multicarbon products in strong acid Reaction environment regulation for electrocatalytic CO2 reduction in acids High carbon utilization in CO2 reduction to multi-carbon products in acidic media Modulating electric field distribution by alkali cations for CO2 electroreduction in strongly acidic medium Operando cathode activation with alkali metal cations for high current density operation of water-fed zero-gap carbon dioxide electrolyzers Acidic electroreduction of CO2 to multi-carbon products with CO2 recovery and recycling from carbonate A covalent molecular design enabling efficient CO2 reduction in strong acids Immobilized tetraalkylammonium cations enable metal-free CO2 electroreduction in acid and pure water Cationic-group-functionalized electrocatalysts enable stable acidic CO2 electrolysis gold and silver electrodes without metal cations in solution Surface-immobilized cross-linked cationic polyelectrolyte enables CO2 reduction with metal cation-free acidic electrolyte Stabilizing the oxidation state of catalysts for effective electrochemical carbon dioxide conversion Efficient electroconversion of carbon dioxide to formate by a reconstructed amino-functionalized indium-organic framework electrocatalyst Selective increase in CO2 electroreduction to ethanol activity at nanograin-boundary-rich mixed Cu(I)/Cu(0) sites via enriching co-adsorbed CO and hydroxyl species In situ copper faceting enables efficient CO2/CO electrolysis Protecting copper oxidation state via intermediate confinement for selective CO2 electroreduction to C2+ fuels Carbon-confined indium oxides for efficient carbon dioxide reduction in a solid-state electrolyte flow cell Breaking K+ concentration limit on Cu nanoneedles for acidic electrocatalytic CO2 reduction to multi-carbon products Enhanced electrocatalytic CO2 reduction via field-induced reagent concentration Tip-like Fe-N4 sites induced surface microenvironments regulation boosts the oxygen reduction reaction Regulating the tip effect on single-atom and cluster catalysts: Forming reversible oxygen species with high efficiency in chlorine evolution reaction Proton diffusion in polyethylene oxide: relevance to electrochromic device design Nitrogen‐doped carbon‐coated CuO‐In2O3 p–n heterojunction for remarkable photocatalytic hydrogen evolution Gas diffusion enhanced electrode with ultrathin superhydrophobic macropore structure for acidic CO2 electroreduction Enriching metal-oxygen species and phosphate modulating of active sites for robust electrocatalytical CO2 reduction Oxygen‐pinned Ag1In single‐atom alloy for efficient electroreduction CO2 to formate Enhanced electrochemical CO2 reduction to formate over phosphate-modified In: Water activation and active site tuning Local steric hindrance for CO2 electroreduction at a thermodynamic potential and wide working window Size-dependent selectivity of electrochemical CO2 reduction on converted In2O3 nanocrystals Organic additive-derived films on Cu electrodes promote electrochemical CO2 reduction to C2+ products under strongly acidic conditions Modulating adsorbed hydrogen drives electrochemical CO2-to-C2 products Boosting the proton-coupled electron transfer via Fe-P atomic pair for enhanced electrochemical CO2 reduction Modulating water hydrogen bonding within a non-aqueous environment controls its reactivity in electrochemical transformations Strong hydrogen-bonded interfacial water inhibiting hydrogen evolution kinetics to promote electrochemical CO2 reduction to C2+ The solvation environment of molecularly dispersed cobalt phthalocyanine determines methanol selectivity during electrocatalytic CO2 reduction Ab initio molecular dynamics for liquid metals DFT study of CO2 adsorption and hydrogenation on the In2O3 surface Active oxygen vacancy site for methanol synthesis from CO2 hydrogenation on In2O3 (110): A DFT study Methanol synthesis from CO2 hydrogenation over a Pd4/ In2O3 model catalyst: a combined DFT and kinetic study How copper catalyzes the electroreduction of carbon dioxide into hydrocarbon fuels VASPKIT: a user-friendly interface facilitating high-throughput computing and analysis using VASP code The field synergy (coordination) principle and its applications in enhancing single phase convective heat transfer Field synergy principle for enhancing convective heat transfer–its extension and numerical verifications Download references This work was supported by the National Natural Science Foundation of China (22325901 (B.Y.X.) the National Key Research and Development Program of China (2021YFA1600800 (B.Y.X.) and 2021YFA1501000 (B.Y.X.)) the Innovation and Talent Recruitment Base of New Energy Chemistry and Devices (B21003 (B.Y.X.)) Key Research and Development Project of Hainan Province (ZDYF2024SHFZ074 (P.D.)) and Start-up Research Foundation of Hainan University (KYQD(ZR)23035 (Z.W.) the authors acknowledge the support for comprehensive characterizations by Pico Election Microscopy Center of Hainan University and are grateful for resources from the High-Performance Computing Center of Central South University and HPC facilities at HSE University These authors contributed equally: Zhitong Wang State Key Laboratory of Materials Processing and Die & Mould Technology Key Laboratory of Material Chemistry for Energy Conversion and Storage (Ministry of Education) Hubei Key Laboratory of Material Chemistry and Service Failure Wuhan National Laboratory for Optoelectronics Huazhong University of Science and Technology (HUST) State Key Laboratory of Photovoltaic Science and Technology School of Information Science and Technology Hunan Joint International Research Center for Carbon Dioxide Resource Utilization School of Materials Science & Engineering Hunan Provincial Key Laboratory of Electronic Packaging and Advanced Functional Materials of Hunan Province Donostia International Physics Center (DIPC) characterization and performance evaluation of materials carried out and analyzed the DFT calculations performed the XAS measurements and fitting assisted with catalyst synthesis and data analysis performed the TEM characterization and analyzed the data All authors discussed the results and assisted with the paper preparation Download citation DOI: https://doi.org/10.1038/s41467-025-56977-6 Metrics details Homogeneous mixed-valence (MV) behaviour is one of the most intriguing phenomena of f-electron systems a fundamental aspect which remains unsettled is the experimental determination of the limiting cases for which MV emerges a prototypical MV system characterized by two nearly-degenerate Sm2+ and Sm3+ configurations By combining angle-resolved photoemission spectroscopy (ARPES) and x-ray absorption spectroscopy (XAS) we track the evolution of the mean Sm valence Upon substitution of Sm ions with trivalent La we observe a linear decrease of valence fluctuations to an almost complete suppression at x = 0.2 approaching the value of vimp ~ 2.35 in the dilute-impurity limit Such behaviour departs from a monotonic evolution of vSm across the whole series as well as from the expectation of its convergence to an integer value for x → 0 demonstrate an unconventional evolution of the MV character in the SmxLa1−xB6 series paving the way to further theoretical and experimental considerations on the concept of MV itself and its influence on the macroscopic properties of rare-earth compounds in the dilute-to-intermediate impurity regime an experimental study tracking the crossover of the MV character going from a periodic f-electron lattice to a dilute f-impurity system is still lacking we base our experimental strategy on elemental substitution on the rare-earth site This approach provides a powerful chemical control parameter acting directly on the occupation of the f-states allowing the precise tracking of the mean Sm valence across different concentration regimes while the single d band is characterized by a strongly mixed B-Sm character such reduction rate would not be compatible with having Sm ions either purely divalent or trivalent in this concentration regime we track the electronic structure of the SmxLa1−xB6 series over the entire doping range by means of angle-resolved photoemission (ARPES) and x-ray absorption spectroscopy (XAS) and observe a non-monotonic evolution of the mean Sm valence While the strong Sm2+/Sm3+ admixture is quenched in the intermediate substitution regime with a persisting MV behavior all the way into the dilute-impurity limit These results provide experimental evidence of the emergence of the MV phenomenon even in this dilute limit and establish the key role of unconventional behavior of f-electrons in defining the properties of rare-earth compounds also in such extreme regimes a ARPES spectra along the \(\overline{{{{\rm{XM}}}}}\) high-symmetry direction of the Brillouin zone (black dashed line in b b ARPES iso-energy contours close to EF for the same samples shown in a; the integration window in energy is 15 meV about EF All data were acquired at 10 K with hν = 21.2 eV the size of the X-pockets progressively decreases upon increasing x While the interplay between localized 4f-electrons and itinerant 5d electrons is undoubtedly an important defining aspect of the electronic structure of these materials we focus our analysis on the evolution of the bulk X-pockets’ dispersion as a function of x which can be directly linked to changes in the concentration of trivalent ions in the system c are determined as follow: on the x axis are based on energy dispersive x-ray (EDX) measurements; on n5d(x) are derived from the fitting of the ARPES data; on the fractions of Sm2+/Sm3+ are calculated by combining the uncertainties on x and n5d(x) via error propagation rules we compute the fractional percentage of Sm2+ and Sm3+ from n5d(x) as follows (normalized over the total amount of Sm in the system our ARPES results suggest a clear distinction between the low (x ≤ 0.2) and high (x ≥ 0.55) Sm concentration regimes along with a substantial variation of the Sm2+:Sm3+ ratio across the SmxLa1−xB6 series allowing one to circumvent some of the challenges characteristic of ARPES on SmxLa1−xB6 XAS has already been shown to be a powerful technique to explore the physics of MV systems such as SmB6: the absorption spectrum can be described as the first approximation by the sum of two independent components By tuning the incident energy across the Sm M4 and M5 edges (i.e. exciting 3d core electrons into 4f orbitals) each XAS spectrum can be mapped into a specific Sm2+:Sm3+ ratio providing us with a tool to directly determine the mean Sm valence in the SmxLa1−xB6 series a Evolution of the XAS intensity at the Sm M5 edge for x = [0.07 The absorption profiles (gray dots) have been extracted at 640 eV (La fluorescence line) of IPFY spectra for x ≤ 0.3 and at 850 eV (Sm fluorescence line) of PFY spectra for x > 0.3 The total XAS spectral weight is fit (black lines) by the sum of two independent components associated to Sm2+ (red lines and shaded regions) and Sm3+ ions (orange lines and shaded regions) b Intensity evolution of the Sm2+ (top) and Sm3+ (bottom) components normalized to the x = 1 case All data were taken at a base temperature of 20 K Error bars in b are determined as follow: on the x axis are based on EDX measurements; on \(\Delta {{{{\rm{I}}}}}_{{{{{\rm{Sm}}}}}^{2+}/{{{{\rm{Sm}}}}}^{3+}}\) are derived from the fitting of the XAS data We can then express the evolution of the Sm valence in the SmxLa1−xB6 series as: Error bars in b are determined as follow: on the x axis are based on EDX measurements; on vSm are derived from the uncertainties on the fractions of Sm2+/Sm3+ and on \(\Delta {{{{\rm{I}}}}}_{{{{{\rm{Sm}}}}}^{2+}/{{{{\rm{Sm}}}}}^{3+}}\) for ARPES and XAS data As a final note, we emphasize that the model of Eq. (2) does not fully describe the nearly complete suppression of Sm2+/Sm3+ admixture detected at x = 0.2 When performing density functional theory calculations of the doping dependence on the Fermi energy using virtual crystal approximation no anomalous behavior was found which could explain this observation; indeed additional investigations are needed to specifically address the sharp crossover observed around x = 0.2 with the development of more refined theories our combined ARPES and XAS study provides evidence of the realization of a dilute-impurity MV state in the SmxLa1−xB6 series Our results may stimulate further theoretical and experimental considerations on the concept of MV and its influence on the macroscopic electronic and transport properties of rare-earth compounds in the dilute-to-intermediate impurity regime with base pressure and temperature of 5 × 10−10 Torr and 22 K Additional data are available from the corresponding authors upon request Orbital ordering phenomena in d-and f-electron systems Magnetically mediated superconductivity in heavy fermion compounds Antiferromagnetic spin fluctuation and superconductivity Quantum criticality in heavy-fermion metals Heavy-electron metals: new highly correlated states of matter Coleman, P. Heavy fermions and the kondo lattice: a 21st century perspective. arXiv https://arxiv.org/abs/1509.05769 (2015) Resistance minimum in dilute magnetic alloys Molecules into materials: case studies in materials chemistry-mixed valency magnetism and superconductivity (World Scientific Intermetallic compounds of rare earths and non-magnetic metals Valence instabilities and related narrow-band phenomena (Springer Science & Business Media Review of techniques in the large-n expansion for dilute magnetic alloys Non-fermi liquids in the extended hubbard model New model for the mixed-valence phenomenon in rare-earth materials Valence transitions in rare-earth chalcogenides Non-fermi-liquid states of a magnetic ion in a metal Theory of the marginal-Fermi-liquid spectrum and pairing in a local copper oxide model Temperature dependence of the samarium oxidation state in SmB6 and Sm1−xLaxB6 Temperature dependence of Sm valence in SmB6 studied by X-ray absorption spectroscopy Pressure-resistant intermediate valence in the Kondo insulator SmB6 Pressure-induced exotic states in rare earth hexaborides The fluctuating population of Sm 4f configurations in topological Kondo insulator SmB6 explored with high-resolution X-ray absorption and emission spectra Quantum phase transition and destruction of Kondo effect in pressurized SmB6 Temperature and pressure dependences of Sm valence in intermediate valence compound SmB6 Magnetic and semiconducting properties of SmB6 Low-temperature surface conduction in the Kondo insulator SmB6 and surface states in the Kondo insulator SmB6 Surface Hall effect and nonlocal transport in SmB6: evidence for surface conduction Observation of possible topological in-gap surface states in the Kondo insulator SmB6 by photoemission Surface and bulk electronic structure of the strongly correlated system SmB6 and implications for a topological Kondo insulator Surface electronic structure of the topological Kondo-insulator candidate correlated electron system SmB6 Kondo hybridization and the origin of metallic states at the (001) surface of SmB6 Polarity-driven surface metallicity in SmB6 Samarium hexaboride is a trivial surface conductor Two-dimensional Fermi surfaces in Kondo insulator SmB6 Unconventional Fermi surface in an insulating state Quantum oscillations in flux-grown SmB6 with embedded aluminum 4f crystal field ground state of the strongly correlated topological insulator SmB6 Fermi surfaces changes in La1−xSmxB6 and Ce1−xCaxB6 studied using the de Haas–van Alphen effect and magnetic susceptibility Please note that for a quantitative comparison with our work the raw FS volume values presented in this reference should be used and not the percentage volume reductions which were multiplied by an additional factor of 3 Fermi-surface sum rule and its consequences for periodic kondo and mixed-valence systems The fermi surface and Fermi liquid properties of periodic Kondo and mixed valence systems Study of the valence transition in La- and Yb-substituted SmB6 Valence transition of samarium in hexaboride solid solutions Sm1−xMxB6 (M= Yb2+ Magnetic and nonmagnetic doping dependence of the conducting surface states in SmB6 Transport properties of variously doped SmB6 An undulator-based spherical grating monochromator beamline for angle-resolved photoemission spectroscopy An in-vacuum diffractometer for resonant elastic soft x-ray scattering Download references This research was undertaken thanks in part to funding from the Max Planck-UBC-UTokyo Centre for Quantum Materials and the Canada First Research Excellence Fund Quantum Materials and Future Technologies Program and Natural Sciences and Engineering Research Council of Canada’s (NSERC’s) Steacie Memorial Fellowships (A.D.); the Alexander von Humboldt Fellowship (A.D.); the Canada Research Chairs Program (A.D.); NSERC Canada Foundation for Innovation (CFI); the Department of National Defence (DND); British Columbia Knowledge Development Fund (BCKDF); and the CIFAR Quantum Materials Program Part of the research described in this work was performed at the Canadian Light Source a national research facility of the University of Saskatchewan the Canadian Institutes of Health Research (CIHR) The research carried out in Aarhus was supported by the Independent Research Fund Denmark (Grant no 1026-00089B) and the VILLUM FONDEN via the Centre of Excellence for Dirac Materials (Grant no Work at Los Alamos was performed under the auspices of the U.S Division of Materials Science and Engineering Max Planck Institute for Chemical Physics of Solids Centre Énergie Matériaux Télécommunications Institut National de la Recherche Scientifique Department of Physics & Engineering Physics analyzed the ARPES data with input from M.M. Nature Communications thanks Chandra Varma and the other Download citation DOI: https://doi.org/10.1038/s41467-024-51569-2 « Back Metrics details Emotion enhances the subjective experience of recollection Previous research examined associations between memory accuracy and metamemory confidence judgments but these studies have not quantified the relationship between accuracy and metacognitive confidence judgments In this study we utilize signal detection theory frameworks to investigate how memory accuracy (measured by discrimination sensitivity) and the alignment between metamemory confidence judgments and memory accuracy (ie metacognitive efficiency) varies for neutral and negative valence as well as item and associative detail memory types Our results indicate that valence and memory type have different effects on accuracy Negative valence was associated with enhanced accuracy for both items and associated details but its relationship with response bias varied across memory types with conservative recognition responses observed for items and liberal responses for associative details We also observed a double dissociation between metamemory confidence judgements across valence and memory type with negative valence associated with increased confidence for item memory but decreased confidence for associated details Examining the association between memory accuracy and metamemory confidence revealed that metacognitive efficiency was greater for negatively valenced items compared to neutral but this effect did not generalize to details associated with negatively valenced items These findings advance our understanding of how arousing negatively valenced information modulates memory and metacognition These studies typically compared memory accuracy to participants’ self-reported confidence or recollective experience but did not precisely quantify the relationship between memory accuracy of metamemory confidence judgements all past studies either did not specifically require participants to report their subjective confidence or failed to account for participants’ tendencies to favor one type of response over another (response bias) we leverage signal detection theory frameworks to mitigate the potential biases present in earlier research and quantitatively characterize how the quality of metamemory confidence judgments differs for negatively valenced items and associated contextual details high confidence and “remember” judgments were similarly associated with detail memory accuracy however the relationship between remember and high confidence judgments with associated detail accuracy varied depending of type of detail and valence They present evidence that participants can report high confidence for both their remember and know judgements with high confidence remember judgments associated with superior source memory accuracy compared to high confidence know judgments Another inherent limitation of R/K judgments is their reliance on distinguishing between two qualitative responses to assess differences in subjective recollection—which precludes detailed analyses of incremental differences in recollective experience although research using the R/K paradigm hints at potential associations between metamemory confidence and memory accuracy results from this paradigm cannot directly inform us about associations between confidence in memory accuracy and memory accuracy itself but still suffers from lingering analytical confounds including the lack of a robust metric for quantifying the relationship between memory accuracy and metamemory confidence judgments we consider a d’ score of 2 or greater to indicate strong discrimination performance in memory SDT frameworks can be extended to address this limitation and estimate three variables associated with metacognitive confidence: metacognitive bias Metacognitive bias captures differences in subjective confidence unrelated to task performance—like a tendency to report high confidence independent of accuracy Metacognitive sensitivity (meta-d’) captures the efficacy of distinguishing between incorrect and correct judgments when assigning confidence ratings (e.g being accurate for high confidence judgments and inaccurate for low confidence) meta-d’ is the idealized first-order sensitivity (d’) expected from confidence ratings given a metacognitively ideal observer compares idealized sensitivity (meta-d’) to observed sensitivity (d’) to characterize how well metacognitive confidence ratings align with performance Metacognitive efficiency is crucial as it enables meaningful comparisons across individuals with different performance SDT provides a flexible and robust theory-driven computational modeling framework which utilizes knowledge of the generative processes underlying behavior to effectively summarize choice and metacognitive behavior using a small set of parameters research suggests the relationship between metamemory confidence and memory accuracy differs between neutral and negatively valenced items but methodological shortcomings in previous studies limit the robustness and generalizability of these findings To address these confounds and quantitatively characterize the association between metamemory confidence and memory accuracy we leverage SDT frameworks in our experiment: a longitudinal 2-day study investigating how metacognitive confidence judgments are associated with the accuracy of memory for items and their associated details across valence This strategic experimental design enabled us to reliably characterize mnemonic choice behavior precisely quantify how associations between metamemory confidence and memory accuracy change when exposed to negatively valenced information evaluate whether the metacognitive efficiency of metamemory confidence judgments is consistent across memory types and determine the influence of individual-level trait differences on the metacognitive efficiency of item memory Participants completed an incidental encoding session for neutral and negatively valenced images followed 24 h later by a surprise memory test assessing recognition and subjective confidence in recognition for images and an associated contextual detail—a colored border surrounding each image (Yellow During the incidental encoding session participants categorized images and symbols Image categorization always occurred first During image categorization participants viewed 60 scenes with colored borders (30 neutral 30 negative) and responded whether the border color appeared anywhere within each framed image Border colors were equally balanced across images and valence Image categorization trials consisted of a 4000 ms presentation followed by a 1000 ms fixation cross we administered a symbol categorization task using only black symbols to preclude rehearsal practice This distractor task was included to prevent excessive rehearsal associated with the vivid nature of highly arousing negative stimuli During this task participants indicated whether presented stimuli were letters or numbers; each trial consisted of a 4500 ms stimulus presentation We do not discuss results from this task further participants were shown the item paired with the same colored border as seen during encoding This associated detail of frame color was balanced across images Participants were also notified of lures in this associative detail memory test which used the same nested response scale as the item memory test and required them to report their confidence level that a given old item was paired with old or new colored border Presented and novel stimuli were counterbalanced across subjects and border color frequency was equivalent across valence and lure conditions and a shortened practice version of the task was completed to ensure participants understood the task and response procedure Seventy-four participants were recruited from Harvard Universities’ SONA participant recruitment platform (https://husp.sona-systems.com/) and completed both parts of this 2-day study Several participants were excluded from reported analyses due to our preregistered exclusion criteria: not using the complete range of the confidence scale (n = 2) poor item memory accuracy < 70% (n = 3) and below chance (50%) memory accuracy for associated details (n = 2) All participants provided informed consent and ethics approval was obtained in accordance with the guidelines and regulations set forth by the Harvard University Committee on the Use of Human Subjects (IRB19-0789) To validate normative arousal and valence ratings an independent experiment (n = 22) was carried out where participants rated the selected scenes on valence and arousal Ratings from our subjects confirmed the expected trends with a 100% probability of valence being higher for neutrally valenced items (M = 3 and 100% probability of arousal being greater for negatively valenced items (M = 2.26 this practice strongly biases individual-level parameter estimates in situations with limited trial counts Hierarchical estimation applies a shrinkage effect on individual-level estimates by sharing information across individual and group levels but outputs more accurate and generalizable estimates Given the limited number of trials feasible for a memory experiment we extract individual-level estimates from our hierarchical estimation process The H-Meta-d’ toolbox uses Markov Chain Monte Carlo (MCMC) sampling to estimate posterior distributions over model parameters using JAGS See supplementary materials for more information regarding MCMC sampling specifications Hierarchical sampling enables accurate estimation of group-level parameters by allowing group-level estimates to constrain individual-level fits—minimizing the impact of individual estimates with high uncertainty at the group-level (ie Group-level metacognitive performance was compared across neutral and negatively valenced items and for items and their associated details The significance of parameter values was determined by evaluating whether 89% HDIs on the posterior distributions included zero This extension enables inferences about whether individual traits influence metacognitive efficiency at the group-level Our sample performed well on the old/new recognition task—effectively discriminating between previously seen and novel items (M = 2.23 This performance is particularly noteworthy given liberal response biases demonstrated by an increased tendency to claim recognition compared to an unbiased observer (M = −0.36 Alignment between effective discrimination and liberal responding suggests participants were metacognitively aware of their performance we must evaluate confidence judgments and metacognitive efficiency to confirm this association Average confidence in our sample was 2.45 (SD = 0.26) Compared to the median of our confidence scale (2) this indicates participants in our sample were overconfident To evaluate how well metamemory confidence ratings align with accuracy we hierarchically estimated metacognitive efficiency at the group level while taking into consideration uncertainty at the individual level Metacognitively ideal observers align confidence ratings with performance and possess an M-ratio of 1 metamemory confidence judgments were metacognitively inefficient (Median = 0.55 Considering participant’s high discrimination sensitivity this suggests metacognitive inefficiency stemmed from overconfidence in performance This analysis concluded there is insufficient evidence to support an association between visual imagery and metacognitive efficiency with 89% confidence (Median = −0.13 and insufficient evidence to support an association between body awareness and metacognitive efficiency with 89% confidence (Median = 0.11 the evidence indicates that neither visual imagery nor body awareness is positively associated with recognition or metacognitive judgments for items and conservative for negatively valenced items (d’: M = 2.44 SD = 0.27) compared to neutral (d’: M = 2.02 We can be at least 89% certain that discrimination sensitivity meaningfully differs across valence (M = 0.37 0.51]) with a 99.99% probability of being higher for negatively valenced items and response bias meaningfully differs across valence (M = −0.18 HDI89% = [−0.26 −0.09]) with a 99.94% probability of being more conservative for negatively valenced items Differences in behavior across valence conditions suggest that individuals increased discrimination sensitivity and conservative response behavior was due to the arousing negatively valenced nature of emotional stimuli but qualitative differences across valence conditions were still apparent; metamemory confidence judgments for negatively valenced items were higher (M = 2.49 We can be at least 89% certain that confidence ratings meaningfully differ across valence (M = 0.08 0.12]) with a 99.94% probability of being higher for negatively valenced items This conclusion is additionally supported by 75% (30/40) of participants being more confident when recognizing negatively valenced items The metacognitive efficiency \((\frac{meta-{d}^{\prime}}{{d}^{\prime}})\) of metamemory confidence judgments was higher for negatively valenced items (M = 0.74 Comparing metacognitive efficiency across valence conditions in item memory provides evidence that we can be at most 88% certain there is an effect of valence on metacognitive efficiency (Median = 0.13 The evidence therefore indicates that negative valence is associated with improvements in metacognitive efficiency for item memory but further investigations are necessary to replicate this finding and evaluate whether this association generalizes to other features of negatively valenced episodic memories like associated details Recognition judgments for items were highly sensitive (d’: M = 2.22 SD = 0.14) compared to associated details (d’: M = 1.3 We can be at least 89% certain that discrimination sensitivity meaningfully differs across memory type (M = −0.89 −0.75]) with a 100% probability of being lower for associated details and response bias meaningfully differs across memory type (M = 0.71 0.74]) with a 100% probability of being more conservative for associated details Alignment between diminished discrimination and conservative responding for associative details suggests participants were metacognitively aware of their performance Metamemory confidence judgments were higher for items (M = 2.45 SD = 0.26) compared to associated details (M = 1.91 We can be at least 89% certain that confidence ratings meaningfully differ across memory type (M = 0.54 0.63]) with a 100% probability of being higher for items This conclusion is additionally supported by trends demonstrating that 97.5% (39/40) of participants were more confident in their recognition of items than associated details The metacognitive efficiency \(\left(\frac{meta-{d}^{\prime}}{{d}^{\prime}}\right)\) of metamemory confidence judgments for items (Median = 0.52 SD = 0.12) was qualitatively greater than the metacognitive efficiency of associated details (Median = 0.17 Comparing metacognitive efficiency across memory type conditions provides evidence that we can be at least 89% certain there are differences in metacognitive efficiency across memory type (Median = 0.37 This conclusion is additionally supported by 92.5% (37/40) of participants being more metacognitively efficient for item metamemory judgments This provides compelling evidence that the metacognitive efficiency of item metamemory confidence judgments is more than three times greater than associated details The effect of negative valence on confidence judgements varies by memory type increasing confidence for item memory but decreasing confidence for details associated with items The discrimination sensitivity (d’) of recognition judgments was greater for details associated with negatively valenced items (M = 1.35 and recognition was less conservative for details associated with negatively valenced items (M = 0.24 We can be at most 88% certain that discrimination sensitivity for associated details differs across valence (M = −0.19 0.00]) with a 93.75% probability of being higher for details associated with negatively valenced items and at least 89% certain that response bias meaningfully differs across valence (M = 0.18 0.28]) with a 99.97% probability of being less conservative for details associated with negatively valenced items Our results therefore provide evidence that negative valence improves discrimination sensitivity for items and their associated details but differs regarding its effect on response bias—resulting in conservative recognition judgments for items and liberal recognition for associative details The metacognitive efficiency \((\frac{meta-{d}^{\prime}}{{d}^{\prime}})\) of metamemory confidence judgments for details associated with negatively valenced items (M = 0.61 SD = 0.056) was qualitatively lower compared to neutral items (M = 0.83 We can be at least 89% certain that confidence ratings meaningfully differ across valence (M = 0.11 0.16]) with a 98.89% probability of being lower for details associated with negatively valenced items Comparing metacognitive efficiency across valence for associative details provided insufficient evidence to assert that valence effects the metacognitive efficiency of associated details with 89% confidence (Median = −0.17 negative valence improves metacognitive efficiency for item memory but this effect does not extend to associated details emotionally evocative images may have evoked different response strategies Our investigation is the first we are aware of which quantitatively characterizes the metacognitive efficiency of metamemory confidence judgments—the accuracy of metacognitive judgments regarding recognition performance On average metacognitive efficiency was 0.51 generally inefficient compared to an ideal Bayesian observer metamemory confidence judgments were 3.06x greater for items compared to their associated details This is particularly striking because we observed a meaningful average decrease of 0.51 in confidence judgments for details associated with items Decreased metamemory confidence and increased conservative recognition biases for details compared to items suggests participants were metacognitively aware of the inherent uncertainty involved in recalling specific details associated with items; however their poor metacognitive efficiency underscores a failure to effectively monitor this uncertainty By leveraging signal detection frameworks to estimate discrimination sensitivity this article advances the field by providing a more reliable characterization of the accuracy of metacognitive processes in memory Applying these methods in future research can help identify fine-grained distinctions in metacognitive efficiency for different types of mnemonic details the differences we observed in metacognitive efficiency between our two memory types suggest that the accuracy of metacognitive judgments in memory may rely on domain-specific memory processes Our findings implicate behavioral differences but future research investigating the neural correlates of metacognitive efficiency in memory can reveal how distinct neural structures contribute to the calibration of confidence across valence and memory types This would refine our understanding of how negative emotion affects the contributions of different memory systems to the relationship between memory accuracy and confidence they proposed that an enhanced subjective sense of recollection for emotionally charged memories likely stems from recalling a single item or specific details associated with an emotional event we refine this perspective by demonstrating that enhanced recollective experiences for emotional events are only accurate for items This insight highlights a potential evolutionary advantage conferred by metacognition for decision-making involving arousing and motivates future investigations to characterize this association We carefully designed our experiment and utilized signal detection frameworks to address potential confounds such as confidence biases acknowledging and addressing remaining confounds will be crucial for further refining our understanding of the factors influencing the accuracy of memory and metacognitive judgments for negatively valenced stimuli One notable limitation is our reliance on the assumption that negatively valenced stimuli increased physiological arousal based on self-reported ratings from an additional sample of participants Although these ratings offer some insight into the impact of arousal on behavior Future research can improve this design by incorporating objective measures of physiological arousal during task performance which would provide a more precise understanding of how arousal impacts discrimination sensitivity our study did not account for potential confounds arising from affective information elaboration related to the conceptual congruence of border colors surrounding affective images We were unable to investigate this confound in our sample due to an insufficient number of trials to conditionalize signal detection analyses based on frame color This limitation affects the interpretation of our data but future research with larger stimulus sets could address this issue the methodological framework presented in this manuscript demonstrates that changes in the association between confidence and memory accuracy can precisely be delineated across choices offering valuable insights for future research in affect and memory-based decision-making Brown, R. & Kulik, J. Flashbulb memories. Cognition 5(1), 73–99. https://doi.org/10.1016/0010-0277(77)90018-x (1977) Flavell, J. H. Metacognition and cognitive monitoring: A new area of cognitive-developmental inquiry. Am. Psychol. 34(10), 906–911. https://doi.org/10.1037/0003-066x.34.10.906 (1979) Ochsner, K. N. Are affective events richly recollected or simply familiar? The experience and process of recognizing feelings past. J. Exp. Psychol. Gen. 129(2), 242–261. https://doi.org/10.1037/0096-3445.129.2.242 (2000) Rimmele, U., Davachi, L. & Phelps, E. A. Memory for time and place contributes to enhanced confidence in memories for emotional events. Emotion 12(4), 834–846. https://doi.org/10.1037/a0028003 (2012) Perfect, T. J., Mayes, A. R., Downes, J. J. & Van Eijk, R. Does context discriminate recollection from familiarity in recognition memory. Q. J. Exp. Psychol. 49(3), 797–813. https://doi.org/10.1080/713755644 (1996) Kensinger, E. A. & Corkin, S. Memory enhancement for emotional words: Are emotional words more vividly remembered than neutral words?. Mem. Cognit. 31(8), 1169–1180. https://doi.org/10.3758/bf03195800 (2003) Sharot, T., Delgado, M. R. & Phelps, E. A. How emotion enhances the feeling of remembering. Nat. Neurosci. 7(12), 1376–1380. https://doi.org/10.1038/nn1353 (2004) Phelps, E. A. & Sharot, T. How (and why) emotion enhances the subjective sense of recollection. Curr. Dir. Psychol. Sci. 17(2), 147–152. https://doi.org/10.1111/j.1467-8721.2008.00565.x (2008) Rimmele, U., Davachi, L., Petrov, R., Dougal, S. & Phelps, E. A. Emotion enhances the subjective feeling of remembering, despite lower accuracy for contextual details. Emotion 11(3), 553–562. https://doi.org/10.1037/a0024246 (2011) Yonelinas, A. P. The nature of recollection and familiarity: A review of 30 years of research. J. Mem. Lang. 46(3), 441–517. https://doi.org/10.1006/jmla.2002.2864 (2002) Rimmele, U., Lackovic, S. F., Tobe, R. H., Leventhal, B. L. & Phelps, E. A. Beta-adrenergic blockade at memory encoding, but not retrieval, decreases the subjective sense of recollection. J. Cognit. Neurosci. 28(6), 895–907. https://doi.org/10.1162/jocn_a_00941 (2016) Umanath, S. & Coane, J. H. Face validity of remembering and knowing: Empirical consensus and disagreement between participants and researchers. Perspect. Psychol. Sci. 15(6), 1400–1422. https://doi.org/10.1177/1745691620917672 (2020) Roediger, H. L. & Tekin, E. Recognition memory: Tulving’s contributions and some new findings. Neuropsychologia 139, 107350. https://doi.org/10.1016/j.neuropsychologia.2020.107350 (2020) Guggenmos, M. Measuring metacognitive performance: Type 1 performance dependence and test-retest reliability. Neurosci. Conscious. 2021(1), niab040. https://doi.org/10.1093/nc/niab040 (2021) Anderson, N. D. Teaching signal detection theory with pseudoscience. Front. Psychol. 6, 762. https://doi.org/10.3389/fpsyg.2015.00762 (2015) Stanislaw, H. & Todorov, N. Calculation of signal detection theory measures. Behav. Res. Methods Instrum. Comput. 31(1), 137–149. https://doi.org/10.3758/bf03207704 (1999) Dougal, S. & Rotello, C. M. “Remembering” emotional words is based on response bias, not recollection. Psychon. Bull. Rev. 14(3), 423–429. https://doi.org/10.3758/bf03194083 (2007) Flavell, J. H., Everett, B. A., Croft, K. & Flavell, E. R. Young children’s knowledge about visual perception: Further evidence for the level 1–level 2 distinction. Dev. Psychol. 17(1), 99–103. https://doi.org/10.1037/0012-1649.17.1.99 (1981) Fleming, S. M. & Daw, N. D. Self-evaluation of decision-making: A general Bayesian framework for metacognitive computation. Psychol. Rev. 124(1), 91–114. https://doi.org/10.1037/rev0000045 (2017) Fleming, S. M. & Lau, H. How to measure metacognition. Front. Hum. Neurosc. 8, 443. https://doi.org/10.3389/fnhum.2014.00443 (2014) Fleming, S. M. HMeta-d: Hierarchical Bayesian estimation of metacognitive efficiency from confidence ratings. Neurosc. Conscious. 2017(1), nix007. https://doi.org/10.1093/nc/nix007 (2017) Harrison, O. K. et al. The filter detection task for measurement of breathing-related interoception and metacognition. Biol. Psychol. 165, 108185. https://doi.org/10.1016/j.biopsycho.2021.108185 (2021) Marks, D. F. Visual imagery differences in the recall of pictures. Br. J. Psychol. 64(1), 17–24. https://doi.org/10.1111/j.2044-8295.1973.tb01322.x (1973) Bradley, M. M. & Lang, P. J. Measuring emotion: The self-assessment manikin and the semantic differential. J. Behav. Ther. Exp. Psychiatry 25(1), 49–59. https://doi.org/10.1016/0005-7916(94)90063-9 (1994) Bürkner, P.-C. Brms: An R package for Bayesian multilevel models using stan. J. Stat. Softw. 80(1), 1–28. https://doi.org/10.18637/jss.v080.i01 (2017) Makowski, D., Ben-Shachar, M. S., Chen, S. H. A. & Lüdecke, D. Indices of effect existence and significance in the Bayesian framework. Front. Psychol. 10, 2767. https://doi.org/10.3389/fpsyg.2019.02767 (2019) Doing Bayesian data analysis: A tutorial with R Hautus, M. J. Corrections for extreme proportions and their biasing effects on estimated values of d’. Behav. Res. Methods Instrum. Comput. 27(1), 46–51. https://doi.org/10.3758/bf03203619 (1995) Doerksen, S. & Shimamura, A. P. Source memory enhancement for emotional words. Emotion 1(1), 5–11. https://doi.org/10.1037/1528-3542.1.1.5 (2001) Talarico, J. M. & Rubin, D. C. Confidence, not consistency, characterizes flashbulb memories. Psychol. Sci. 14(5), 455–461. https://doi.org/10.1111/1467-9280.02453 (2003) Mather, M. Emotional arousal and memory binding: An object-based framework. Perspect. Psychol. Sci. J. Assoc. Psychol. Sci. 2(1), 33–52. https://doi.org/10.1111/j.1745-6916.2007.00028.x (2007) Mihaylova, M., Vuilleumier, P. & Rimmele, U. Better memory for intrinsic versus extrinsic details underlies the enhanced recollective experience of negative events. Learn. Mem. 26(11), 455–459. https://doi.org/10.1101/lm.049734.119 (2019) Kruschke, J. K. Bayesian analysis reporting guidelines. Nat. Hum. Behav. 5(10), 1282–1291. https://doi.org/10.1038/s41562-021-01177-7 (2021) Download references Stephen Fleming for perspectives on experimental design and Dr Schacter for insights on an earlier version of this manuscript Ethics approval was obtained from the Harvard University Committee on the Use of Human Subjects (IRB19-0789) and all methods were carried out in accordance with the guidelines and regulations set forth by the institute All participants provided informed consent prior to their involvement in the study Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations Download citation DOI: https://doi.org/10.1038/s41598-024-76208-0 Metrics details Ferroelectrics based on van der Waals semiconductors represent an emergent class of materials for disruptive technologies ranging from neuromorphic computing to low-power electronics many theoretical predictions of their electronic properties have yet to be confirmed experimentally and exploited we use nanoscale angle-resolved photoemission electron spectroscopy and optical transmission in high magnetic fields to reveal the electronic band structure of the van der Waals ferroelectric indium selenide (α-In2Se3) This indirect bandgap semiconductor features a weakly dispersed valence band which is shaped like an inverted Mexican hat Its form changes following an irreversible structural phase transition of α-In2Se3 into β-In2Se3 via a thermal annealing in ultra-high vacuum Density functional theory supports the experiments and reveals the critical contribution of spin orbit coupling to the form of the valence band The measured band structure and its in situ manipulation offer opportunities for precise engineering of ferroelectrics and their functional properties beyond traditional semiconducting systems theoretical predictions for these new band structures and phenomena are yet to be realised experimentally a Side and in-plane view of a single van der Waal layer (α-In2Se3) and the full and projected Brillouin zone (BZ) for bulk 2H-α-In2Se3 b Constant energy ARPES slices taken near the valence band maximum c Constant energy ARPES slices taken near the VBM with high k-resolution d Colour plot of the electron energy (in eV) versus in-plane k-vector as determined by ARPES (top) and DFT (bottom) The DFT plots correspond to different out-of-plane wave vectors kz (at the centre The projected in-plane hexagonal BZ is overlaid on the plots e Colour plot of the electron energy (in eV) versus in-plane k-vector as determined by ARPES (top) and DFT at Γ (bottom) Experimental data in (e) were acquired with greater k-resolution and over a narrower energy range than in part (d) Energies (in eV) are referenced to the Fermi level for ARPES and to the VBM for DFT b ARPES data for bulk 2H-α-In2Se3 taken along the high symmetry (a) \(\bar{\Gamma }\to \bar{{{\rm{K}}}}\) and (b) \(\bar{\Gamma }\to \bar{{{\rm{M}}}}\) directions c Energy dispersion as measured by ARPES for the uppermost VB Energies (in eV) are referenced to the Fermi level e Energy bands as determined by DFT through the centre and edge of the BZ Lower panels: BZ and high symmetry directions f Calculated density of states (DOS) versus energy c Colour plot of the magneto-transmission (normalised to the transmission at B = 0 T) versus B and photon energy (T = 4.2 K) Dashed lines show the calculated exciton (cyan) and electron-hole interband Landau level transitions (red and black lines) Lines correspond to transitions from the VBM to the CBM (black) and from the saddle points in the VB to the CBM (red lines) Right: Line profile of the magneto-transmission spectrum at B = 60 T the energy difference between the two indirect transitions is in agreement with the difference between the VBM and the saddle points measured by ARPES: ΔVBM-SP = (29 ± 10) meV the magneto-optical transmission experiments support the ARPES data and show that 2H-α-In2Se3 is an indirect band gap semiconductor featuring two closely spaced interband optical transitions The LL quantisation of these transitions indicates that the CB electrons are much lighter than the VB holes leading to weakly bound excitons and to an electron cyclotron energy that is approximately 10 times larger than for VB holes a ARPES constant energy slices showing the transition from α to β following an annealing at temperatures Ta above 180 °C The insets show the side view of the single layer for the α and β phases All measurements were conducted at room temperature b Colour plot of the electron energy (in eV) versus k-vectors c ARPES data for bulk β-In2Se3 taken along high symmetry directions b Calculated colour plot of energy (in eV) versus in-plane k-vector for the valence band of bulk 1 T 2H and 3 R β-In2Se3 for kz at Γand at the edge of the BZ The zero energy corresponds to the valence band maximum c Calculated colour plot of energy (in eV) versus in-plane k-vector of the conduction band of bulk 1 T The zero energy corresponds to the conduction band minimum The projected in-plane hexagonal BZ is overlayed on the plots in parts (b) and (c) We assign this discrepancy to a broadening of the measured energy dispersions which arises from the increased disorder following the transition of the crystal from α- to β-In2Se3 implying a strong influence of an applied electric field on the band structure with prospects to generate spin currents and polarisation via SOC guiding further developments of ferroelectrics Samples of commercially available bulk 2H-α-In2Se3 were obtained from 6Carbon Technology grown using chemical vapour transport The mechanical exfoliation of the crystal was conducted using a Loadpoint LP:5660 blue tape with a PVC base and acrylic adhesive The exfoliated flakes of In2Se3 were then stamped directly onto Au substrates (40 nm thick) prepared via thermal evaporation the samples were placed under vacuum ( < 10−10 mbar) Annealing was conducted under vacuum via a resistive heater and monitored with an in-situ thermocouple ARPES and ultraviolet photoelectron spectroscopy (UPS) measurements were conducted using a Scienta Omicron ESCA equipped with a Focus HIS 14 HD UV photon source with energy hν = 21.219 eV which require a small excitation area and high photon flux The ESCA was operated in energy filtered mode with a pass energy of 50 eV and slit width of 0.2 mm photoelectrons were collected from a region of approximately 10 × 20 µm2 The experiments were conducted over several areas of each sample to probe its uniformity Both ARPES and UPS were conducted with the photoemission electron microscopy (PEEM) optics in telescopic mode with an extractor voltage of 12000 V energy resolution and k calibration were confirmed through measurements on a gold reference sample The measured constant energy slices were rotationally symmetrised around the \(\bar{\Gamma }\) point with a six fold symmetry This improved the signal to noise ratio and reduced the effect of non-uniform detector illumination For magneto-optical transmission experiments samples of In2Se3 were placed in a liquid helium cryostat A non-destructive magnet was used to generate magnetic fields up to 60 T with a pulse duration of 36 ms in the Faraday configuration (B⊥E) White light was directed to the sample using a 400 μm diameter optical fibre with a broadband halogen lamp acting as the light source The transmitted light was collected by a 800 μm diameter fibre and guided to a spectrometer equipped with a charge-coupled device (CCD) camera which ensured the transmission spectra were obtained at an essentially fixed magnetic field Raman spectroscopy was conducted using a Horiba LabRAM HR utilising a He-Ne laser (λ = 632.8 nm) and a 1200 g mm−1 diffraction grating Light was directed to and from the sample through a confocal microscope equipped with a 100× objective (laser spot size of ≈ 1 μm) allowing for the flakes in question to be located with the aid of a motorised sample stage The dispersed Raman signal was detected using a CCD camera The Raman spectra were acquired over several spots of each sample to probe its uniformity the Raman spectra are dominated by bulk properties Atomic force microscopy was conducted in air using an Asylum research Cypher-S AFM system operating in non-contact mode The authors confirm that the data supporting the findings of this study are available within the article and its supplementary information 2D materials and van der Waals heterostructures Van der Waals integration before and beyond two-dimensional materials Electronic and thermoelectric properties of van der Waals materials with ring-shaped valence bands Indirect to direct gap crossover in two dimensional InSe revealed by angle-resolved photoemission spectroscopy From graphene to graphite: electronic structure around the K point Tunable electronic and dielectric behavior of GaS and GaSe monolayers Electronic properties of a biased graphene bilayer Tunable magnetism and half-metallicity in hole-doped monolayer GaSe Natural van der Waals heterostructural single crystals with both magnetic and topological properties Two-dimensional materials from high-throughput computational exfoliation of experimentally known compounds Large-grain MBE-grown GaSe on GaAs with a Mexican hat-like valence band dispersion Controlled vapor phase growth of single crystalline two-dimensional GaSe crystals with high photoresponse Valence band inversion and spin-orbit effects in the electronic structure of monolayer GaSe Tunable magnetism in ferroelectric α-In2Se3 by hole-doping Exotic magnetism in As-doped α/β-In2Se3 monolayers with tunable anisotropic carrier mobility Two-dimensional indium selenides compounds: an ab initio study Prediction of intrinsic two-dimensional ferroelectrics in In2Se3 and other III2-VI3 van der Waals materials Intercorrelated in-plane and out-of-plane ferroelectricity in ultrathin two-dimensional layered semiconductor In2Se3 Layer-dependent ferroelectricity in 2H-stacked few-layer α-In2Se3 Atomically resolving polymorphs and crystal structures of In2Se3 Electronic band structure of in-plane ferroelectric van der Waals β’-In2Se3 Characterization of the ferroelectric phase transition in monolayer In2Se3 Thickness-dependent evolutions of surface reconstruction and band structures in epitaxial β–In2Se3 thin films Quantum confinement and electronic structure at the surface of van der Waals ferroelectric α-In2Se3 2D multiferroics in as-substituted bilayer α-In2Se3 with enhanced magnetic moments for next-generation nonvolatile memory device Li, S. et al. Van der Waals ferroelectrics: theories, materials and device applications. Adv. Mater. https://doi.org/10.1002/adma.202301472 (2023) Ferroelectric order in van der Waals layered materials Ferroelectric semiconductor junctions based on graphene/In2Se3/graphene van der Waals heterostructures Preparation of highly oriented α-In2Se3 thin films by a simple technique Large disparity between optical and fundamental band gaps in layered In2Se3 Extraordinary photoresponse in two-dimensional In2Se3 nanosheets Strong quantum confinement effect in the optical properties of ultrathin α‐In2Se3 Interband optical transitions in extremely anisotropic semiconductors coexistence of exciton and the Landau levels Theory of optical magneto-absorption effects in semiconductors Phase and polarization modulation in two-dimensional In2Se3 via in situ transmission electron microscopy Nonvolatile ferroelectric memory with lateral β/α/β In2Se3 heterojunctions Phase-controllable large-area two-dimensional In2Se3 and ferroelectric heterophase junction Phase Instability in van der Waals In2Se3 determined by surface coordination Epitaxial growth of few-layer β-In2Se3 thin films by metalorganic chemical vapor deposition Quantum confinement and photoresponsivity of β-In2Se3 nanosheets grown by physical vapour transport Mexican-hat potential energy surface in two-dimensional III2-VI3 materials and the importance of entropy barrier in ultrafast reversible ferroelectric phase change Crystalline-crystalline phase transformation in two-dimensional In2Se3 thin layers Air-stable atomically encapsulated crystalline-crystalline phase transitions in In2Se3 Epitaxy of GaSe coupled to graphene: from in situ band engineering to photon sensing Observation of flat bands in twisted bilayer graphene Unconventional superconductivity in magic-angle graphene superlattices Antimony-doped p-Type In2Se3 for heterophase homojunction with high-performance reconfigurable broadband photovoltaic effect Subnanometer-wide indium selenide nanoribbons QUANTUM ESPRESSO: a modular and open-source software project for quantum simulations of materials Advanced capabilities for materials modelling with quantum ESPRESSO Many-body perturbation theory calculations using the yambo code Yambo: an ab initio tool for excited state calculations Download references This work was supported by the European Union’s Horizon 2020 research and innovation programme Graphene Flagship Core 3; the Engineering and Physical Sciences Research Council (Grant No EP/T019018/1) and the University of Nottingham Propulsion Futures Beacon The high magnetic field studies were supported by the European Magnetic Field Laboratory (EMFL) and by the EPSRC via the UK membership of the EMFL (Grant No microscopy and Raman studies with input from A.P.; N.A conducted Raman and electron spectroscopy studies of samples of different thickness; Z.Y conducted the magneto-transmission experiments and analysed the data with input from L.E conducted the DFT studies; all authors discussed the results Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work Download citation DOI: https://doi.org/10.1038/s41467-025-56139-8 Jeff Dalton has authored 100+ AI research papers and spent 20 years making search and assistants smarter he is joining Valence to oversee development of their AI-powered leadership coach—building deeper memory and context for hyper-personalized support that democratizes learning and development NEW YORK, Oct. 9, 2024 /PRNewswire-PRWeb/ -- Valence creator of the most widely deployed AI leadership coach for enterprise Jeff Dalton as Head of AI and Chief Scientist Jeff will lead ongoing development of Valence's AI coach Nadia which is already deployed in dozens of Fortune 500 companies and used by tens of thousands of corporate leaders in partnership with the US National Institute of Standards and Technology created one of the world's first benchmarks of conversational search quality Nadia will evolve into a personalized AI coach that you can't work without." "The conversational agents we have today aren't really personal and don't really adapt to us over the long term," Jeff explained but still largely used for simple tasks like writing email or Q&A How do we create next-generation AI agents that know us well enough to proactively help us with all of our tasks?" and it's what sets Valence's Nadia apart as the only high-context AI work coach Nadia has extensive context about each employee and the company and uses that to tailor expert guidance to individual employees and situations Jeff will lead development with a focus on building even deeper context and stronger memory for Nadia "What I love about Jeff's background is that his experience is very practical It's about taking the research and delivering something valuable for users both immediately and for years down the road and Jeff will help us deliver personalized AI-powered support and guidance that creates better leaders and better companies," said Valence CEO Parker Mitchell What career skills have you learned recently Context like this will allow Valence's AI coach to learn about you with each interaction and provide world-class guidance that helps leaders grow Nadia will evolve into a personalized AI coach that you can't work without," said Jeff Valence is on a mission to unlock human potential across the entire workforce of any organization the company began by offering software to support managers to improve team performance and help teams work better together Valence recognized the potential of LLMs to unlock the ultimate goal of personalized support at scale They began offering an AI-powered leadership coach that provides individuals with guidance and support unique to their needs and is affordable enough to scale to every single employee is the most widely deployed AI leadership coach for enterprise Nadia integrates seamlessly with company values and leadership frameworks and builds context about each individual employee for personalized learning and development at scale for every career stage It is technology that puts the world's best executive coach in the pocket of every leader The company raised its Series A in May 2022 Kira Luscher, Valence, 1 4154072994, [email protected], https://www.valence.co/ Do not sell or share my personal information: Metrics details The emotional valence of animals is challenging to assess we attempted to assess emotional valence through memory in 1- and 3-week-old piglets It was hypothesized that piglets would spend less time in a pen where they experienced a negative event (castration) and more time in a pen where they experienced a positive event (enrichment) A testing apparatus was designed with three equally sized pens: two outer sections serving as treatment pens containing unique visual and tactile cues and a center section remaining neutral Piglets received either negative or positive condition in one outer treatment pen and a sham treatment in the opposite Various methods were tested (age of piglets number and length of conditioning sessions piglets did not decrease their time in the pen associated with the negative condition or increase their time in the pen associated with the positive condition results indicate older piglets developed an aversion towards the sham treatment This study provides methodological groundwork for the application of place conditioning in piglets and highlights the nuances important for the use of cognitive tests to assess animal welfare place conditioning has not been used to study the affective states of young piglets We selected two interventions likely to induce changes in affective valence: castration for negative affect We hypothesize that piglets will have a longer latency to enter and spend less time in an environment associated with castration (i.e piglets will have a shorter latency and spend more time in an environment associated with enrichment (i.e This study is exploratory and will investigate several methodologies the study was approved under the University of Pennsylvania IACUC protocol 804656 and reported in accordance with ARRIVE guidelines who enrolled between 12 and 21 piglets for a place conditioning paradigm piglets were pseudo-randomly selected (first ones caught healthy) from 24 litters (4.8 ± 1.1 per trial) All piglets acted as their own control and received both treatment and sham procedures (see next sections for treatment details) Experimental piglets were housed from birth with the rest of their litter in a 2.1 × 2.0 m farrowing pen with their dam Pen flooring was perforated plastic with an embedded 0.6 × 0.6 m heating pad continuously available to the piglets Piglets had ad libitum access to solid feed (PO20PS Kalmbach organic pig starter room temperature was kept at approximately 20 °C Piglets received both treatments (castration or enrichment and sham differentiated by wall colour (blue or white) and tactile cues (door stoppers or drawer knobs) Middle walls could be gated to either restricting piglets to one pen during conditioning or allowing access to all pens during pre-tests and tests Identical parts of the different trials will be described first, then specific sections for each trial will be detailed in separate sections. A summary table is presented below (Table 1) which was an effort to exclude piglets for which maternal separation resulted in immobilization and/or the propensity to freeze Three piglets did not fulfill this criterion 7 piglets were excluded for health reasons or death (4 lame and 3 crushed in their home pen) additional piglets were enrolled to reach a sample size of 22 for each trial piglets were removed from their home pen and brought to the experimental apparatus and number of treatment sessions varied with trial all trials compared two treatments: an affective intervention (positive: enrichment or negative: castration) treatment order and side associated with treatment were balanced by block across piglets The aim was for the piglets to associate one side area with the affective intervention and the other area with the control procedure piglets were tested for place aversion or preference piglets were brought to the experimental apparatus which had open dividers for individual testing Piglets were gently lowered in the middle area and were free to roam between all areas for the same amount of time allotted during pre-test (trial dependent) piglets were brought back to their home pen and returned to routine farm care Trial 1 was a study of negative affective valence using surgical castration as a model of a procedure inducing negative affect 2.5 ± 0.6 kg) had 10 min to roam between pens during pre-test and test Considering the unfeasibility of conducting castration multiple times on the same piglets the number of treatment sessions was limited to two (one for each treatment: surgical and sham castration) piglets were brought to the experimental apparatus then gently lowered in one of the two side areas which were mounted with the full dividers to prevent exploration to other areas Piglets were left in the area for 5 to 10 min before receiving their assigned treatment (castration or sham) and then remained in the area for another 2 h as a pair to avoid isolation stress After being brought back to their home pen and provided 48 h of rest piglets received their second treatment: sham procedure if they had been castrated during their first treatment if they had received their first treatment in the white pen two incisions were made by scalpel on their scrotum and the testicles were pushed out and removed by pull The wound was then sprayed again with iodine and lidocaine and piglets were gently put back in their area a similar procedure with identical handling and pharmaceutical treatments was conducted light pressure was applied with fingers on either side of the scrotum Piglets were tested for place aversion for 10 min 48 h after receiving their second treatment The control procedure took place in a barren pen (no straw toys or sucrose solution) and was identical to Trial 1 (i.e restricting them to the side pen for the remainder of the 3 min experimenters conducted the sham castration and put them back down If they had not entered a side pen after 3 min they were brought back to their home pen and the session was recorded as a “No choice” Due to the potential influence of a partner on choice made After 8 active conditioning sessions (which also acted as preference choice tests) a regular 3 min place conditioning test was conducted at least 1 h after their last training (i.e castration or enrichment) as well as treatment order pen colour associated with treatments and the interaction between sessions (from pre-test to test) Slope for each treatment was compared to a theoretical null slope representing no change between pre-test and test Piglet ID was used as a random factor nested within litter ID Model residuals were checked for assumptions of linearity models on latencies required data transformations (logarithmic for trials 1 and 2 Models analyzing time spent did not require data transformation Significance threshold was set at P < 0.05 (tendency at P < 0.10) 95% confidence intervals (CI) were obtained with R’s confint function and reported for significant results and tendencies Place conditioning results for Trial 1 (Sham vs All piglets received both treatments between pre-test and test Symbols represent significance (*P < 0.05) or tendency (‡P < 0.10) in changes between pre-test and test for each treatment (left: latency to enter Place conditioning results for Trial 2 (Sham vs. Enrichment, young piglets, 2 treatment sessions of 2 h each). All piglets received both treatments between pre-test and test. Symbols represent significance (*P < 0.05) or tendency (‡P < 0.10) in changes between pre-test and test for each treatment (left: latency to enter, right: total time spent in the pen). Place conditioning results for Trial 3 (Sham vs. Enrichment, young piglets, 10 treatment sessions of 3 min each). All piglets received treatments between pre-test and test. Symbols represent significance (*P < 0.05) or tendency (‡P < 0.10) in changes between pre-test and test for each treatment (left: latency to enter, right: total time spent in the pen). Place conditioning results for Trial 4 (Sham vs. Enrichment, older piglets, 10 treatment sessions of 3 min each). All piglets received treatments between pre-test and test. Symbols represent significance (*P < 0.05) or tendency (‡P < 0.10) in changes between pre-test and test for each treatment (left: latency to enter, right: total time spent in the pen). Place conditioning results for Trial 5 (Sham vs All piglets received treatments between pre-test and test No correlation was found between the number of choices made for a pen during active conditioning and the latency to enter that pen during testing (sham: t = −0.55, P = 0.59; enrichment: t = −0.81, P = 0.43). Similarly, no correlation was found between the number of choices made for a pen during active conditioning and the time spent in that pen during testing (sham: t = 1.32, P = 0.20; enrichment: t = 1.30, P = 0.21). Cumulative choices made during the active conditioning sessions If neither Sham or Enrichment pens were chosen after 180 s piglets received one forced choice of each treatment did not significantly decrease their time spent in a pen associated with the presumed negative experience of castration (trial 1) or increased their time spent in a pen associated with the presumed positive experience of enrichment (trial 2) Their time spent in the presumed neutral sham pen also was unchanged in either trial no increase in time spent in the enriched pen was observed (trial 3) we noted a decrease in time spent in the sham pen (trial 4 and 5) These findings suggest that only older piglets exhibited a place conditioning response and benefitted from repeated conditioning (either active or passive) though they interestingly developed an aversion to the sham procedure rather than a preference for enrichment but it remains unclear why this response was not found in trials 2 and 3 Attempts were made to reduce the negative effects of separation and handling by transporting piglets with littermates and treating in pairs when possible Conducting tests in pairs or groups might lessen isolation stress but would also raise interpretative challenges: does a piglet go to a pen simply because another piglet is in it or alternatively is it avoiding an antagonistic partner An interesting application is the use of “social windows” where subjects are tested individually but have visual and auditory access to social partners (Sara Hintze such interpretation remains speculative as the influence of social dynamics on affective processes remains largely unknown in pigs Another reason that could explain the lack of a conditioned aversion or preference was that treatments did not cause enough of an affective contrast with the sham procedure to induce a place conditioning response It is possible that the lack of a conditioned response resulted from the duration of the treatment which allowed for a recovery period that perhaps inadvertently mitigated the memory of the aversive impact of the procedure itself It is also unknown whether the pre-test and test phases which consisted of social isolation in barren compartments induced negative associations with the whole experimental apparatus further reducing the affective contrast between treatment pens feed balls) or negative stimuli (isolation perhaps because their outcomes were more reflective of affective arousal than valence We did not find clear evidence of place conditioned aversion for castration or preference for enrichment our results suggest that older piglets (approximately 3 weeks old) developed an aversion to the sham procedure when paired with enriched conditions This result was consistent for both passive and active conditioning methodologies although active appears to be a more efficient approach We argue that young piglets failed to establish a strong association between pen and treatment for several possible reasons including their young age or low affective contrast between treatment and sham procedures Evaluation of emotional states in young animals is challenging but place conditioning appears in some instances to be a promising avenue for assessment of affective valence This work provides methodological groundwork on the application of place conditioning in piglets and highlights the nuances important for the use of cognitive tests to assess animal welfare Data and R code are accessible in supplementary materials (SM1 and SM2) A scientific conception of animal welfare that reflects ethical concerns An integrative and functional framework for the study of animal emotion and mood Measuring reward with the conditioned place preference paradigm: A comprehensive review of drug effects Exploring time-dependent changes in conditioned place preference for food reward and associated changes in the nucleus accumbens Place conditioning reveals the rewarding aspect of social interaction in juvenile rats Conditioned place preference induced by ∆9-tetrahydrocannabinol: Comparison with cocaine On the rewarding nature of appetitive feeding behaviour in pigs (Sus scrofa): Do domesticated pigs contrafreeload Do pigs distinguish between situations of different emotional valences during anticipation Experimental studies on avoidance behaviour in pigs Effects of oral meloxicam and topical lidocaine on pain associated behaviors of piglets undergoing surgical castration Pain behaviour after castration of piglets; effect of pain relief with lidocaine and/or meloxicam Castration-induced vocalisation in domestic piglets Sus scrofa: Complex and specific alterations of the vocal quality Behavioural responses of piglets to castration: The effect of piglet age Acute physiological responses to castration-related pain in piglets: The effect of two local anesthetics with or without meloxicam Physiology and behavior of pigs before and after castration: Effects of two topical anesthetics Analgesia and/or anaesthesia during piglet castration—part I: Efficacy of farm protocols in pain management Local anesthesia in piglets undergoing castration: A comparative study to investigate the analgesic effects of four local anesthetics on the basis of acute physiological responses and limb movements The effect of environmental stimulation on the development of behaviour in pigs A systematic approach towards developing environmental enrichment for pigs Enrichment satisfying specific behavioural needs in early-weaned pigs An investigation into the causation of chewing behaviour in growing pigs: The role of exploration and feeding motivation Effects of social stress and restraint stress on sucrose preference The effect of pre-weaning housing on the play and agonistic behaviour of domestic pigs The influence of neonatal environment on piglet play behaviour and post-weaning social and cognitive development Influence of environmental enrichment on the behaviour performance and meat quality of domestic pigs The effect of environment enrichment on aggression in newly weaned pigs Chewable materials before weaning reduce tail biting in growing pigs Effect of temporary straw bedding on pigs’ behaviour Global Animal Partnership’s (GAP) 5-step Animal Welfare Standards for Pigs v2.5 Castration induced pain in pigs and other livestock Evaluation of two injection techniques in combination with the Local Anesthetics Lidocaine and Mepivacaine for piglets undergoing Surgical Castration Effect of two methods and two anaesthetics for local anaesthesia of piglets during castration R: A Language and Environment for Statistical Computing (R Foundation for Statistical Computing Fitting Linear mixed-effects models using lme4 lmerTest Package: Tests in Linear mixed effects models and animal welfare: Definitions and essential inquiries to advance animal welfare science The agency domain and behavioral interactions: Assessing positive animal welfare using the five domains model Effects of environmental enrichment on behaviour physiology and performance of pigs: A review Fear assessment in pigs exposed to a novel object test Reducing weaning stress in piglets by pre-weaning socialization and gradual separation from the sow: A review A single exposure to social isolation in domestic piglets activates behavioural arousal and stress-related gene expression in the brain Ontogeny of teat fidelity in pigs and its relation to competition at suckling Social support in pigs with different coping styles Indicators of positive and negative emotions and emotional contagion in pigs Spatial location is critical for conditioning place preference with visual but not tactile stimuli Contextual fear discrimination is impaired by damage to the postrhinal or perirhinal cortex Measuring reward with the conditioned place preference (CPP) paradigm: Update of the last decade The influence of aromatized environmental enrichment objects with changeable aromas on the behaviour of weaned piglets Olfactory signals that modulate pig aggressive and submissive behavior In Social Stress in Domestic Animals (Kluwer Academic Topical wound anaesthesia: Efficacy to mitigate piglet castration pain Topical anaesthesia reduces sensitivity of castration wounds in neonatal piglets Measuring piglet castration pain using linear and non-linear measures of heart rate variability Evidence of pain in piglets subjected to invasive management procedures In Understanding the Behaviour and Improving the Welfare of Pigs (Burleigh Dodds Science Publishing Animal welfare implications of surgical castration and its alternatives in pigs A review of the welfare consequences of surgical castration in piglets and the evaluation of non-surgical methods intradermic needle-free vaccination in piglets: Relevance for Animal Welfare based on an aversion learning test and vocalizations Persistency of the piglet’s reactivity to the handler following a previous positive or negative experience Environmental enrichment improves the performance and behavior of piglets in the nursery phase Rearing piglets in a poor environment: Developmental aspects of social stress in pigs Environmental enrichment induces optimistic cognitive biases in pigs Effects of environmental enrichment on cognitive performance of pigs in a spatial holeboard discrimination task The effect of environmental enrichment on learning in pigs Effects of environmental enrichment on pig welfare—a review A longitudinal study of the effects of providing straw at different stages of life on tail-biting and other behaviour in commercially housed pigs Making decisions under ambiguity: Judgment Bias tasks for assessing emotional state in animals Evaluation of miRNA as biomarkers of emotional valence in pigs Weak General but no specific habituation in anticipating stimuli of presumed negative and positive Valence by Weaned piglets Effects of early life and current housing on sensitivity to reward loss in a successive negative contrast test in pigs The use of garlic oil for olfactory enrichment increases the use of ropes in weaned pigs display rubbing and rolling behaviour when exposed to odours Experience of moderate bedding affects behaviour of growing pigs Coping personality type and environmental enrichment affect aggression at weaning in pigs Low birth weight impairs acquisition of spatial memory task in pigs Sex differences in learning processes of classical and operant conditioning Female and male pigs’ performance in a spatial holeboard and judgment bias task Young pigs exhibit differential exploratory behavior during novelty preference tasks in response to age Effects of age on avoidance learning in pigs Like mother like child: Do fearful sows have fearful piglets Inheriting the sins of their fathers: boar life experiences can shape the emotional responses of their offspring tonic immobility and behaviour in a spatial maze test is affected by gestating sows’ relationship to humans and positive handling at weaning Download references We would like to thank the Swine Center manager Lori Stone and staff Nicole Fox Hanna Lee and Lainey Mikus for the routine care of animals during the study Moore Endowed Fund for Welfare Research and the Wiederhold Foundation University of Pennsylvania School of Veterinary Medicine SI and TDP reviewed and approved the submitted version Download citation DOI: https://doi.org/10.1038/s41598-024-75849-5 All content on this site: Copyright © 2025 Elsevier B.V., its licensors, and contributors. All rights are reserved, including those for text and data mining, AI training, and similar technologies. For all open access content, the relevant licensing terms apply. ShareBy Sharon FlorentineAdobe Stock Startups that delay implementing channel programs may inadvertently limit their growth potential VP of global partnerships at Valence Security in a recent Channel E2E conversation with Sharon Florentine a seasoned veteran in building successful channel partnerships emphasizes that a robust channel strategy enables startups—especially in cybersecurity and managed services—to scale sales effectively without immediately investing heavily in full-time sales staff Early adoption of channel programs not only accelerates brand awareness but also positions startups strategically to engage the right partners who can amplify their market presence "The key is starting early with strong margins," Alosco advises noting that offering attractive margins helps build recognition among influential partners quickly Rather than broadly engaging many partners he recommends focusing deeply on a select few strategic relationships that align closely with the startup's goals and market focus "It's a 'less is more' approach," Alosco says stressing that understanding the businesses of core partners significantly strengthens these crucial relationships and drives long-term success the takeaway is clear: Prioritize your channel strategy early and invest time strategically to build powerful Watch the full conversation below for additional insights on building a high-impact channel strategy from the ground up Sharon manages day-to-day content on ChannelE2E and serves as Editorial Director for CyberRisk Alliance’s Channel Brands She also covers enterprise-class technology companies strategic alliances and channel partner strategies Sharon is a veteran tech journalist and editor with more than 25 years experience in the industry content and leadership positions at Techstrong Group Staci RodriguezMay 5 Channel partnerships can help businesses navigate the challenges of identity management Suparna Chawla BhasinMay 5 Atos Launches NIS2 Compliance Manager App on ServiceNow to Automate EU Cybersecurity Mandates IBM Extends BYOC Deployment Model for Netezza and Db2 SaaS Across AWS and Azure Metrics details The brain’s ability to prioritize behaviorally relevant sensory inputs (i.e. targets) while ignoring irrelevant distractors is crucial for efficient information processing the role of emotional valence in modulating selective attention remains underexplored This study examined how positive and negative emotions alter the spatial scope of visual selective attention using a modified Eriksen Flanker task Participants viewed an emotional face cue (happy or neutral) randomly positioned on the screen and then identified the shape of a subsequent neutral target (bowtie or diamond) at the cued location Adjacent stimuli either matched the target shape (congruent) or differed (incongruent) Results showed that happy faces increased susceptibility to distractors (i.e. suggesting a broadening of attentional scope while angry faces reduced susceptibility (i.e. the magnitude of this emotion-driven attention modulation was negatively correlated with participants’ self-reported levels of psychological distress Participants with higher stress and depression exhibited weaker attention broadening in response to positive cues the findings provide behavioral evidence of how emotional valence influences attention scope offering potential insights into the dynamic interplay between psychological distress Positive emotions are hypothesized to broaden attentional focus and enhance the ability to detect peripheral details positive emotions might increase awareness of surrounding factors such as a barking dog or an approaching cyclist negative emotions are thought to narrow attentional focus This so-called “weapon-focus effect” exemplifies how attention can become concentrated on a perceived threat at the expense of peripheral information (e.g. These emotion-driven shifts in attention can have significant cognitive and practical implications positive emotions may enhance learning and improve the ability to monitor or detect changes in the environment could limit situational awareness and impair performance in dynamic or unpredictable settings there is little direct evidence linking emotional valence to attention scope modulation We examine whether emotional cues modulate attention differently in individuals with self-reported high distress Our hypothesis is that higher distress is associated with weaker emotion-driven modulation Participants were 30 native Thai undergraduates (mean age = 20.97; 16 females; 1 left-handed) from King Mongkut’s University of Technology Thonburi (KMUTT) Each participant had normal or corrected-to-normal vision and received financial compensation for participation Consent forms were obtained from all participants and were signed prior to the experiment The study was approved by the Institutional Review Board (IRB) protocol at King Mongkut’s University of Technology Thonburi and all methods were performed in accordance with the relevant guidelines and regulations To ensure robustness and account for potential variability in participants’ responses we opted to recruit 30 participants for the study Data collection was not influenced by interim analyses or any post-hoc adjustments Emotional cues consisted of seventy-two face images selected from the Nimstim Face Database19 To ensure that these images elicited the desired emotional reactions from our Thai participants an independent group of 30 Thai observers (mean age: 20.20; 14 females; 1 left-handed) who did not take part in the main experiment provided ratings of the emotional content for each image in the Nimstim database Three types of ratings were acquired: (1) emotional valence or the perceived level of positivity or negativity of each facial expression on a 0–9 Likert scale; (2) emotional intensity or the degree of emotion expressed in the image on a 0–9 Likert scale; and (3) emotional category or the interpreted emotional content of each image (i.e. Data from this mini-experiment were used to further screen face images for our study the image selection was performed to ensure that the following three criteria were met: (1) each selected image had a mean categorization score above 70%; (2) the mean categorization accuracies did not statistically differ across selected categories (i.e. respectively); and (3) the emotional intensity ratings were statistically comparable across the happy and angry images (6.72 and 6.93 Given the cultural differences between our participants and those in the original Nimstim validation study (Tottenham et al. we did not use the original ratings for image selection these images were noticeably perceptually dissimilar from the open-mouth face images due to the visibility of teeth or the shape of the mouth Each image was converted to grayscale and manually cropped to remove hair and other non-essential details prior to the experiment Participants were required to discriminate the shape of the target stimulus (depicted in orange) while ignoring the flanking distractors (depicted in grey) Critical experimental manipulations included altering the emotional categories of face cues (angry and neutral) and stimulus congruency (congruent and incongruent) Time delays between cue and stimulus onset (cue-target interval or CSI) were randomized across trials (125 The colors orange and grey are used for illustrative purposes only The trial structures were designed to achieve an even distribution of stimulus conditions Images depicting three facial emotions (happy neutral) were presented in both upright and inverted orientations at one of twelve possible screen locations These stimuli conditions were paired with three cue-stimulus intervals (125 ms and 500 ms) and two congruency conditions (congruent creating a total of 432 stimulus combinations (3 emotions × 2 orientations × 12 locations × 3 CSIs × 2 congruency conditions) these combinations were duplicated to create a total of 864 trials The trial order was randomly shuffled and distributed across 16 blocks The entire experiment lasted approximately 2.5 h Instructions were given in Thai throughout the experiment An omnibus four-way ANOVA was initially performed to analyze the effects of facial emotion (happy Following the identification of a significant three-way interaction (emotion x congruency x orientation) separate two-way ANOVAs were conducted to probe the interaction between facial emotion and stimulus congruency and the nature of this interaction Simple main effects tests with Bonferroni-corrected p-values were subsequently performed to examine the direction of the attention modulation effect for each emotional category Effect sizes in this study were calculated using generalized eta squared (ges) which quantifies the proportion of total variance in the dependent variable that is attributable to each factor we conducted a correlation analysis using the DASS scores and a metric called the “Attention Modulation Score.” This score quantifies emotion-driven changes in attentional focus by measuring the difference in the incongruency effect (i.e. susceptibility to flanker interference) between emotional pairs (e.g. we first calculated the difference in mean accuracy between congruent and incongruent trials (congruent - incongruent) for each emotional category and facial orientation This produced six incongruency effects corresponding to the six possible combinations (i.e. To isolate effects specific to upright faces we subtracted incongruency effects across orientations (upright - inverted) yielding normalized incongruency effects for happy The Attention Modulation Score was then computed by subtracting normalized incongruency effects between specific emotion pairs: happy – neutral Each emotion pair reflects a different aspect of attentional modulation the happy – neutral and angry – neutral comparisons indicate the degree to which happy and angry emotions increased susceptibility to flanker interference (i.e. attention broadening) relative to the neutral baseline Positive values reflect a broadening of attention compared to baseline while negative values indicate a narrowing of attention The happy – angry comparison reflects the extent to which happy faces induced a broader attentional scope compared to angry faces correlation analyses were conducted to explore the relationship between the Attention Modulation Score and the depression and stress subscales of the DASS questionnaire these scores suggest a normal level of depression and a normal level of stress among our subject population A four-way repeated-measures ANOVA was performed to investigate the impact of facial emotion (happy the analysis yielded a significant main effect of stimulus congruency (F(1 with incongruent trials eliciting lower average accuracy scores than congruent trials a significant three-way interaction was also observed among stimulus congruency This interaction suggests that the ability to selectively attend to the target stimulus varied depending on the emotional valence and the orientation of the preceding face cue The analysis revealed no significant main effects of emotion no significant interactions between CSI and other experimental variables were observed (p > 0.1) subsequent analyses combined data across different cue-target time intervals a significant interaction between facial emotions (angry neutral) and stimulus congruency (congruent suggesting that emotional valence modulates the spatial scope of visual selective attention No interaction was observed for inverted faces indicating that the effect cannot be attributed to perceptual dissimilarities across image categories Error bars represent the standard error of the mean (SEM) ***) indicate statistically significant results from simple main effect tests Due to restrictions on the public display of original images from the NimStim dataset the face images shown in this figure were provided by a laboratory member serving as a model and were not used in the actual experiment Post-hoc analyses were conducted to examine the direction of the incongruency effect across facial emotions For positive faces displayed in the upright orientation simple main effects tests revealed higher accuracy on congruent trials than on incongruent trials (mean difference = 4.0% this accuracy difference disappeared for trials with upright negative faces (mean difference = 0.9% a significant incongruency effect was observed for upright neutral faces (mean difference = 2.5% For faces presented in the inverted orientation a significant incongruency effect was observed regardless of emotion type: positive (mean difference = 1.9% these results suggest that the observed valence-induced attention modulation occurred in the expected direction with a broadening of attention scope in response to positive faces and a narrowing in response to negative faces a four-way repeated-measures ANOVA revealed a significant main effect of stimulus congruency on response time in correct trials (F(1,28) = 59.3 indicating that performance was faster on congruent than incongruent trials (mean RT = 568.9 ms and 591.1 ms The analysis also revealed a significant main effect of CSI (F(1.5 no significant congruency × emotion × facial orientation interaction was observed (F(1.56 Further three-way ANOVAs revealed no significant emotion × congruency interaction for upright (F(2,56) = 0.7 p > 0.1) or inverted faces (F(2,56) = 3.1 p > 0.05; see Supplementary Materials S1) We conducted additional correlation analyses to examine the relationships between participants’ psychological states and their attentional modulation An “Attention Modulation Score” was calculated for each participant which quantified the degree to which one emotion induced greater attention broadening relative to another we computed the difference in the incongruency effect across emotional pairs (e.g. which reflected differential susceptibility to flanker interference for each pair Higher Attention Modulation Scores indicate greater attention broadening while lower scores suggest attention narrowing Detailed descriptions of this computation are provided in the Materials and Methods section Figure 3 displays scatterplots of DASS scores (x-axis) against Attention Modulation Scores (y-axis) for three emotion pairs: happy–neutral (3a) An outlier analysis based on scores averaged across the three emotion pairs revealed no significant outliers a significant negative correlation between stress and attention modulation was observed (r = -0.41 suggesting that participants with higher stress exhibited reduced attention broadening when cued by happy faces No significant correlations were found for the angry–neutral pair and Stress Scores (DASS) and Modified Attention Modulation Scores for each emotion pair: (a) Happy-Neutral The present study examines whether the affective valence of facial cues alters the spatial scope of selective attention in a way that impacts the subsequent processing of neutral target stimuli we found that emotional facial cues modulate attentional focus by influencing the ability to suppress flanking stimuli positive facial cues broadened the spatial scope of attention leading to greater susceptibility to flanker interference (i.e. negative facial cues narrowed attentional focus reducing susceptibility to interference (i.e. these valence-driven changes in attention disappeared when the emotional faces were inverted suggesting that the observed effects were not simply due to perceptual differences in the images correlation analyses revealed that the magnitude of attentional modulation was negatively associated with participants’ psychological states Participants who reported higher levels of stress or depression exhibited weaker attention broadening in response to emotional facial cues This suggests that the typical emotion-driven changes in attentional scope may be disrupted in individuals experiencing high psychological distress it is difficult to determine whether the observed effects were due to emotional valence or the attention demands involved in target selection which posits that attentional resource allocation reflects an interaction between bottom-up perceptual competition and top-down cognitive control processes different components may engage these two levels of competition Positive and negative face cues may have influenced bottom-up perceptual competition by reducing (for positive cues) or increasing (for negative cues) competition for the central target stimulus resulting in broader or narrower attentional scope the cognitive demands of identifying the target shape while suppressing distracting flankers likely engaged top-down cognitive control redirecting attention from emotional stimuli to task-relevant goals The patterns we observed may reflect the dynamic interaction between these bottom-up and top-down processes our findings could be explained by a general attentional mechanism unrelated to emotional processing exposure to negative cues might have caused longer disengagement or slower saccadic reaction times making it harder to shift attention away from the cues our analysis of reaction time (RT) data found no evidence of significant differences in engagement times across emotional categories (mean RTs: happy = 593.5 ms we would expect a greater incongruency effect for negative faces compared to positive or neutral faces as prolonged engagement would likely lead to increased encoding of peripheral flanking stimuli along with the target This is in contrast to the observed pattern (a larger incongruency effect for positive compared to negative emotions) suggesting that these alternative explanations are unlikely in a task where participants were required to detect visual stimuli within one or two of four rectangles positioned around a central fixation point (left highly anxious individuals took longer to complete the task when the rectangles were further from the central fixation [26] Given the limited number of studies on this topic future research is necessary to better understand the relationships between psychological distress We also examined the duration of the observed attention modulation effect by analyzing its impact on next-trial performance (n + 1) The analysis showed no significant emotion × congruency interaction for either upright or inverted faces (p-values > 0.05; see Supplementary Figure S3) suggesting that the emotion-induced attentional modulation did not carry over to the next trial (approximately 5.5–6 s later) given the length of our experimental sessions (1.5–2 h; 16 blocks) we investigated whether the effects persisted across blocks or diminished due to fatigue While accuracy did not significantly decline a progressive improvement in response speed was observed stabilizing around blocks 5–6 (F(3,84) = 0.08 indicating an early learning or adaptation period when the trials were divided into the first and second halves of the experiment significant attention modulation effects were present in both halves (first half: F(2,56) = 3.4 these results suggest that the observed effects were robust over time and were not influenced by learning or fatigue leaving it unclear whether the observed effects reflected solely bottom-up processes or also involved top-down mechanisms We speculate that the observed effects arose from the interaction of bottom-up perceptual competition (driven by the emotional face cue) and top-down cognitive control (required for target identification amidst distractors) This interplay likely explains why modulation occurred across all time delays the present results may be influenced by inter-subject variability in sensitivity to timing with some participants experiencing faster or slower impacts of emotional valence on attention Data from the current study is insufficient to determine whythe effects appeared generalized across time intervals Future research would benefit from an in-depth exploration of the complex temporal dynamics of emotion-induced attention scope modulation the present study provides valuable insights into how emotional valence modulates the spatial scope of visual selective attention broaden attentional focus and increase susceptibility to distractors narrow attentional focus and reduce interference these emotion-driven changes were moderated by participants’ levels of psychological distress with those experiencing higher stress or depression exhibiting weaker attention modulation in response to emotional cues These findings underscore the complex interaction between emotion highlighting the potential utility of attentional scope modulation as a behavioral marker for psychological distress The data and source codes will be available on the Open Data Framework once the paper is accepted and published Visual search for faces with emotional expressions Emotion drives attention: detecting the snake in the grass Finding the face in the crowd: an anger superiority effect How brains beware: neural mechanisms of emotional attention Lesions of the human amygdala impair enhanced perception of emotionally salient events Negative facial expression captures attention and disrupts performance Do threatening stimuli draw or hold visual attention in subclinical anxiety Temporary suppression of visual processing in an RSVP task: an attentional blink A normalization framework for emotional attention Emotion facilitates perception and potentiates the perceptual benefits of attention Changing the spatial scope of attention alters patterns of neural gain in human cortex Threat-related attentional bias in anxious and nonanxious individuals: a meta-analytic study Components of attentional biases in contamination fear: evidence for difficulty in disengagement Modulation of focused attention by faces expressing emotion: evidence from flanker tasks R: A language and environment for statistical computing The NimStim set of facial expressions: judgments from untrained research participants The structure of negative emotional states: comparison of the Depression anxiety stress scales (DASS) with the Beck Depression and anxiety inventories Using the depression anxiety stress scale 21 (DASS-21) across cultures and stress scales (DASS-21) among Thai nursing students in an online learning environment during the COVID-19 outbreak: a multi-center study Positive affect increases the breadth of attentional selection Proceedings of the National Academy of Sciences How do emotion and motivation direct executive control Najmi, S., Kuckertz, J. M. & Amir, N. Attentional impairment in anxiety: Inefficiency in expanding the scope of attention: attention scope in anxiety. Depress. Anxiety. 29 (3), 243–249. https://doi.org/10.1002/da.20900 (2012) Social anxiety and narrowed attentional breadth toward faces Download references We thank the members of the Neuroscience Center for Research and Innovation (NX) at KMUTT for their helpful feedback and assistance on this work Special thanks to Kitnipat Boonyadhammakul for allowing us to use his photograph as an example in our manuscript This work was funded by the IBRO Rising Stars Awards to TC the Young Researcher Grant at King Mongkut’s University of Technology Thonburi (KMUTT) to TC the research grant from the Research & Innovation for Sustainability Center Magnolia Quality Development Corporation Limited Thailand as well as the KMUTT’s Frontier Research Unit Grant for Neuroscience Center for Research and Innovation to TC and SI This project was also supported by the National Research Council of Thailand grant (fiscal year 2021–2024) to SI as well as the Thailand Science Research and Innovation (TSRI) Basic Research Fund: fiscal year 2023 under project number FRB660073/0164 fiscal year 2022 under project number FRB650048/0164 fiscal year 2021 under project number FRB640008 and fiscal year 2020 under project number 62W1501 to SI Neuroscience Center for Research and Innovation King Mongkut’s University of Technology Thonburi Thitaporn Chaisilprungraung & Sirawaj Itthipuripat Research and Innovation for Sustainability Center Magnolia Quality Development Cooperation Limited (MQDC) performed the data collection and analysis under the supervision of T.C and approved the final version for submission Below is the link to the electronic supplementary material Download citation DOI: https://doi.org/10.1038/s41598-024-80666-x Metrics details a promising low global warming potential (GWP < 1) alternative to traditional perfluorocarbon etching gases for advanced integrated circuit manufacturing particularly for high-aspect-ratio SiO2/SiN stacked layers in 3D flash memory This study investigates the dissociative photoionization dynamics of C3HF5 across 10.0–26.0 eV Ion yield curves and breakdown diagrams reveal that C3HF5 primarily fragments into C3HF5+ indicate fragmentation pathways at low electronic transitions These findings underscore C3HF5’s potential as an environmentally friendly etching gas with excellent performance characteristics particularly in etching processes for manufacturing advanced devices such as 3D flash memory The environmental impacts of etching have driven the search for more sustainable solutions no information is available on the dissociation energy thresholds of fragments from previous studies By constructing breakdown diagrams that integrate PEPICO signals for specific ion species we visualized the fractional abundance of the precursor and fragment ions as a function of photon energy Analysis of the fragmentation patterns and appearance energies revealed the underlying pathways governing the dissociation of C3HF5 These insights are crucial for optimizing plasma etching processes in semiconductor manufacturing By strategically tailoring the gas composition desired etch profiles and material selectivity can be achieved this research paves the way for more sustainable and efficient etching processes by enabling precise control of active species through controlled molecular dissociation and by informing future gas design strategies Photon Energy resolutions of BL3B evaluated by appearance energy (AE) measurements of Xe+ ions shown in the Supplementary Information The monochromatized VUV radiation was focused using Kirkpatrick-Baez mirrors bypassing traditional glass capillaries to minimize photon scattering and preserve flux integrity Beam intensity was monitored in real time using a calibrated AXUV-100 photodiode (IRD Corporation) to ensure stable illumination during measurements C3HF5 gas was introduced into the high-vacuum chamber maintained at a pressure below 3 × 10⁻⁴ Pa and irradiated with monochromatic VUV light in the energy range from 10 to 26 eV in 0.1 eV increment Static biasing voltages were applied to the spectrometer plates to extract the photoelectron–ion pairs formed in the interaction region A three-element lens system focused on the electrons and the ions were analyzed using a time-of-flight (TOF) mass spectrometer Schematic of the experimental setup for photoelectron–photoion coincidence measurements (MCPs: microchannel plates; MCA: multichannel analyzer; TOF: time-of-flight; dimension unit: mm) Typical mass spectra obtained from the photoionization of C3HF5 at photon energies of 20 eV Ion yield curves of the fragment ions produced from the photoionization of C3HF5. A photon energy resolution of 4.2 meV. Relative abundance curves of the fragment ions produced from the photoionization of C3HF5 Plasma processing typically uses electron energies of 1 ~ 5 eV. To maximize the yield of target ions, such as C3HF4+, C3F4+, CF3+, C2F3+, and C3F5+, we focused on photoionization processes using lower photon energies, closer to the ionization threshold. These fragments were considered the primary dissociation fragments of C3HF5+, as they were directly formed from the parent ion. Threshold photoion yield curves for main fragment ions produced from C3HF5 particularly for 3D NAND flash memory devices requires precise control over etching processes to ensure high selectivity and accurate feature definition C3HF5 gas has emerged as a promising alternative to traditional PFC gases owing to its low global warming potential The dissociation of C–H and C–F bonds in HFC molecules plays a crucial role in determining the etching and deposition behaviors Fluorine (F) atoms generated by C–F bond dissociation are highly reactive and can etch silicon (Si) surfaces Hydrogen (H) atoms produced by C–H bond dissociation can promote polymer deposition on the substrate surface which can enhance selectivity and improve process accuracy the CxFy and CxHyFz radicals formed from C–F and C–H bond dissociation can contribute to polymer deposition particularly those formed via F abstraction a delicate balance between the etching and deposition processes is necessary This balance is influenced by the relative concentrations of the reactive species The dissociative photoionization of C3HF5 provides valuable insights into its fragmentation pathways and the potential for generating reactive species suitable for plasma etching processes The relative abundances of CxFy+ and CxHyFz+ ions produced from C3HF5 as a function of photon energy Model for etching SiO2 and SiN films in C3HF5 plasma chemistry The dissociative ionization of C3HF5 was examined using the PEPICO technique in the energy range of 10–26 eV The main fragment ions observed were C3HF4⁺ The AEs of these ions were determined from their ion yield curves generated through dissociative ionization with a threshold energy of approximately 10.7 eV The findings of this study provide valuable insights into the dissociation pathways of C3HF5 and their potential roles in the etching of SiN and SiO2 films A deeper understanding of these fragmentation processes can aid in optimizing the selectivity control in plasma etching processes leading to improved HAR etching performance This information may also be useful for controlling the generation of fluorocarbon ions and designing novel HFC molecules The raw data supporting the findings of this study are available from the corresponding author upon reasonable request Bit cost scalable (BiCS) technology for future ultra-high-density memories Optimal integration and characteristics of vertical array devices for ultra-high density Bit cost scalable technology with punch and plug process for ultra-high density flash memory Future of plasma etching for microelectronics: challenges and opportunities Effects of mask and necking deformation on bowing and twisting in high-aspect-ratio contact hole etching Formation mechanism of sidewall striation in high-aspect-ratio hole etching Investigation of bowing reduction in SiO2 etching taking into account radical sticking in a hole Study on contact distortion during high aspect ratio contact SiO2 etching Progress in nanoscale dry processes for fabrication of high-aspect-ratio features: how can we control critical dimension uniformity at the bottom Influence of hydrogen in silicon nitride films on the surface reactions during hydrofluorocarbon plasma etching A method for high selective etch of Si3N4 and SiC with ion modification and chemical removal Etch selectivity during plasma-assisted etching of SiO2 and SiNx: transitioning from reactive ion etching to atomic layer etching Hydrofluorocarbon ion density of argon- or krypton-diluted CH2F2 plasmas: generation of CH2F+ and CHF2+ by dissociative-ionization in charge exchange collisions Exploring oxide-nitride-oxide scalloping behavior with small gap structure and chemical analysis after fluorocarbon or hydrofluorocarbon plasma processing Highly selective etching of SiO2 over Si3N4 and Si in capacitively coupled plasma employing C5HF7 gas Effect of hydrofluorocarbon structure of C3H2F6 isomers on high aspect ratio etching of silicon oxide Anthropogenic and natural radiative forcing In: Climate Change 2013: the Physical Science Basis Contribution of Working Group I To the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (eds Stocker High aspect ratio SiO2/SiN (ON) stacked layer etching using C3HF5 Dissociative properties of C4F6 obtained using computational chemistry Dissociation channels of c-C4F8 to C2F4 in reactive plasma Electronic properties and primarily dissociation channels of fluoromethane compounds Electronic properties and primarily dissociation channels of hydrofluoroethane compounds Dissociative properties of 1,1,1,2-tetrafluoroethane (HFC-134a) obtained using computational chemistry Electro impact ionization of perfluoro-methyl-vinyl ether C3F6O Fragmentation of valence electronic states of CHF2CF3+ studied by threshold photoelectron–photoion coincidence (TPEPICO) techniques in the photon energy range 12–25 eV Threshold photoelectron photoion coincidence study of the fragmentation of valence states of CF3-CH3+ and CHF2-CH2F+ in the range 12–24 eV Photoelectron photoion coincidence study of the fragmentation of valence states of CHF2–CH3+ in the range 12–25 eV Download references The authors thank the members of the Center for Low-Temperature Plasma Sciences at the Nagoya University This study was partly supported by JSPS-KAKENHI 21H01073 JST ASPIRE JPMJAP2321 and was conducted at the BL3B of the UVSOR Synchrotron Facility Institute for Molecular Science (IMS program 23IMS6015 The authors are grateful to KANTO DENKA KOGYO Co. Tran Trung Nguyen: Writing – review & editing review & resources.Kenji Ishikawa: Supervision Tran Trung Nguyen reports financial support was provided by Nagoya University Tran Trung Nguyen reports a relationship with Nagoya University that includes: employment All the remaining authors declare no conflict of interest Download citation DOI: https://doi.org/10.1038/s41598-025-94119-6 Sign up for the Nature Briefing: Anthropocene newsletter — what matters in anthropocene research Metrics details This study focuses on optimizing catalytic activity in photocatalytic hydrogen evolution reaction by precisely designing and modulating the electronic structure of metal single atoms integrates low-valence platinum single atoms into sulfur-containing covalent organic frameworks The robust asymmetric four-coordination between sulfur and platinum within the framework enables a high platinum loading of 12.1 wt% resulting in efficient photocatalytic hydrogen production activity of 11.4 mmol g−1 h−1 and stable performance under visible light These outcomes are attributed to a reduced hydrogen desorption barrier and enhanced photogenerated charge separation as indicated by density functional theory calculations and dynamic carrier analysis This work challenges traditional notions and opens an avenue for developing low-valence metal single atom-loaded covalent organic framework catalysts to advance photocatalytic hydrogen evolution This complexity impedes a comprehensive understanding of HER mechanism presenting an ongoing obstacle in the quest to develop universally applicable and straightforward methodologies for fabricating high-loading and precisely defined SACs in the low-valence state the ongoing pursuit of highly efficient and stable low-valence SACs integrated into COFs for HER remains a critical area of research focus we present the development of a catalyst designated as PtSA@S-TFPT showcasing both high catalytic efficiency and remarkable stability This achievement is realized through the incorporation of low-valence platinum (Pt) single atoms into sulfur-containing COFs Utilizing methanol as a mild reducing agent facilitated the reduction of Pt to a low valence state establishing robust asymmetric four-coordination with sulfur/oxygen moieties without inducing particle aggregation Contrary to conventional assumptions regarding sulfur coordination that typically pose challenges to the catalytic activity of noble metals PtSA@S-TFPT demonstrates exceptional photocatalytic hydrogen evolution performance achieving a notable rate of 11.4 mmol g−1 h−1 and sustained stability over a duration of 3 × 3 h under visible light The superior catalytic performance of PtSA@S-TFPT is further elucidated by density functional theory (DFT) calculations and dynamic carrier analysis attributing this outcome to the presence of Pt single atoms which effectively reduce the hydrogen desorption barrier and enhance photogenerated charge separation and simulated PXRD patterns of S-TFPT and PtSA@S-TFPT c Reconstructed structure of 14-layer hexagonal channel These results demonstrate the high porosity of S-TFPT and provide further validation of its crystal structure a A transmission electron microscopy image of PtSA@S-TFPT Insert: fast fourier transform (FFT) from white box and lattice spacing viewed along the [001] direction c AC HAADF-STEM image and (d–h) corresponding EDX mapping of PtSA@S-TFPT a Experimental Pt L-edge XANES and (b) Pt L-edge EXAFS spectra of PtSA@S-TFPT Fourier transforms of EXAFS spectra in (c) k-space and (d) R-space with their corresponding best fitting results for PtSA@S-TFPT excluding phase correction e Schematic model and calculated bond lengths of PtSA@S-TFPT (top: front view Diffuse reflectance UV/Vis spectra (a) and energy band positions (b) of PtSA@S-TFPT c Time course and cycling stability of photocatalytic H2 productions d Diffuse reflectance UV/Vis spectrum of PtSA@S-TFPT and wavelength-dependent AQE of photocatalytic H2 production for PtSA@S-TFPT e H2 evolution rate for state-of-the-art representative COFs f Adsorption strength of hydrogen atoms on PtNPs@S-TFPT and PtSA@S-TFPT at the active site Pt Adsorption model of hydrogen atoms on (g) PtNPs@S-TFPT and (h) PtSA@S-TFPT The lower energy barriers for H* generation observed in PtSA@S-TFPT are conducive to subsequent H2 evolution a Transient photocurrent responses and (b) EIS Nyquist plots of S-TFPT i) distributions for the model molecules: S-TFPT (d (The yellow region indicates charge accumulation while the cyan region indicates charge depletion we have successfully engineered a catalyst with significantly enhanced catalytic activity for the photocatalytic HER By incorporating low-valence platinum single atoms into sulfur-enriched COFs using methanol as a mild reducing agent the resultant SAC demonstrates notable improvements in both photocatalytic efficiency and stability The exceptional performance of PtSA@S-TFPT is primarily attributed to the presence of platinum single atoms which reduce the hydrogen desorption barrier and promote the separation of photogenerated carriers These findings are supported by transient photocurrent measurements This study challenges conventional paradigms and paves the way for the synthesis of low-valence metal single atom-loaded COFs thus propelling the field of photocatalytic hydrogen evolution forward A 10-mL Pyrex tube was initially loaded with 4,4′,4″-(1,3,5-Triazine-2,4,6-triyl)tribenzaldehyde (TFPT 2,5-bis(2-(methylthio)ethoxy)benzene-1,4-dihydrazide (METPH The resulting mixture was sonicated for 5 min to achieve a clear solution followed by the addition of 0.40 mL of trifluoroacetic acid (TFA) the tube was rapidly cooled to 77 K using a liquid nitrogen bath and subjected to three freeze-pump-thaw cycles for degassing The sealed tube was then evacuated under vacuum and heated at 120 °C for 3 days a yellow precipitate was obtained through centrifugation at 6000 rpm for 2 min followed by washing with anhydrous acetone The collected sample underwent a solvent exchange with anhydrous tetrahydrofuran thrice and was dried at 80 °C under vacuum for 12 h to yield a yellow powder (28 mg The infrared spectroscopy (IR) analysis (powder cm−1) revealed characteristic peaks at 1663(m) 33 mg of chloroplatinate hexahydrate (H2PtCl6•6H2O) was dissolved in 8 mL of 25 vt% methanol/aqueous solution using ultrasound 60 mg of S-TFPT was immersed in the solution under light-free conditions the resulting product was filtered and washed with methanol The obtained powder was then immersed in 5 mL of methanol and cooled using liquid nitrogen at 77 K followed by subjecting the mixture to three freeze-pump-thaw cycles the mixture was heated at 60 °C for 1.5 h to enable in-situ Pt reduction the resulting PtSA@S-TFPT product was collected by filtration and washed with anhydrous methanol 33 mg of H2PtCl6•6H2O was dissolved in 8 mL of 25 vt% methanol/aqueous solution 60 mg of S-TFPT was immersed in the solution without exposure to light the resultant product was filtered and rinsed with methanol it was soaked in 2 mL of a 1 M NaBH4/methanol solution for 2 min resulting in the formation of a dark yellow solid powder The photocatalytic H2 evolution test was conducted using a photocatalytic reaction setup (Labsolar 6 A; Beijing Perfect light Technology Co. 20 mg of PtSA@S-TFPT was dispersed in a solution containing 35 mL of deionized water and 5 mL of methanol The suspension was sonicated for 1 h to ensure uniform dispersion before adding 5 mL of triethanolamine as the sacrificial agent The mixture was stirred at 500 rpm and degassed using a vacuum pump Xenon lamps (300 W) emitting light with wavelengths greater than 420 nm and an optical power density of 1 W/cm2 served as the illumination source a water circulation system maintained the samples at room temperature The evolved H2 gas in the reactor headspace was collected automatically using a gas-tight syringe connected to a photocatalytic system and quantified using a gas chromatograph equipped with a thermal conductivity detector the system was degassed for 30 min after each cycle of illumination and the optical power density was recalibrated before the subsequent cycle of illumination All data are available in the main text or the supplementary materials. Correspondence and requests for materials should be addressed to Yongwu Peng (ywpeng@zjut.edu.cn) or Gao Li (gaoli@dicp.ac.cn). Source data are provided with this paper Transition metal nitride coated with atomic layers of Pt as a low-cost highly stable electrocatalyst for the oxygen reduction reaction High-performance core-shell catalyst with nitride nanoparticles as a core: well-defined titanium copper nitride coated with an atomic Pt layer for the oxygen reduction reaction Hierarchically open-porous carbon networks enriched with exclusive Fe-Nx active sites as efficient oxygen reduction catalysts towards acidic H2-O2 PEM fuel cell and alkaline Zn-air battery Advanced atomically dispersed metal-nitrogen-carbon catalysts toward cathodic oxygen reduction in PEM fuel cells Leveraging metal nodes in metal-organic frameworks for advanced anodic hydrazine oxidation assisted seawater splitting Single-atom catalysis of CO oxidation using Pt1/FeOx MoS2 monolayer catalyst doped with isolated Co atoms for the hydrodeoxygenation reaction Low-temperature hydrogen production from water and methanol using Pt/α-MoC catalysts Cation vacancy stabilization of single-atomic-site Pt1/Ni(OH)x catalyst for diboration of alkynes and alkenes Reversed charge transfer and enhanced hydrogen spillover in platinum nanoclusters anchored on titanium oxide with rich oxygen vacancies boost hydrogen evolution reaction Pt/Fe2O3 with Pt-Fe pair sites as a catalyst for oxygen reduction with ultralow Pt loading Confining zero‐valent platinum single atoms in α‐MoC1−x for pH‐universal hydrogen evolution reaction Volcano-type relationship between oxidation states and catalytic activity of single-atom catalysts towards hydrogen evolution Stabilizing low‐valence single atoms by constructing metalloid tungsten carbide supports for efficient hydrogen oxidation and evolution Low‐valence Znδ+(0<δ<2) single‐atom material as highly efficient electrocatalyst for CO2 reduction Controlling the oxidation state of Pt single atoms for maximizing catalytic activity Reticular chemistry: molecular precision in infinite 2D and 3D Covalent organic frameworks: opportunities of covalent organic frameworks for advanced applications Light‐controllable ionic conductivity in a polymeric ionic liquid and defogging wearable devices enabled by plasmonic silver nanoparticle-embedded covalent-organic framework nanosheets Engineering single-atom active sites on covalent organic frameworks for boosting CO2 photoreduction Covalent-organic-framework-based composite materials Emerging covalent organic frameworks tailored materials for electrocatalysis Salt-assisted fabrication of a water-based covalent organic framework ink and its hybrid films for photothermal actuators Direct transformation of bulk copper into copper single sites via emitting and trapping of atoms Atomically dispersed Fe3+ sites catalyze efficient CO2 electroreduction to CO Single platinum atoms immobilized on an MXene as an efficient catalyst for the hydrogen evolution reaction Cascade anchoring strategy for general mass production of high-loading single-atomic metal-nitrogen catalysts Zero‐valent palladium single‐atoms catalysts confined in black phosphorus for efficient semi‐hydrogenation Crystalline lattice‐confined atomic Pt in metal carbides to match electronic structures and hydrogen evolution behaviors of platinum Uncovering near-free platinum single-atom dynamics during electrochemical hydrogen evolution reaction Electron configuration modulation of nickel single atoms for elevated photocatalytic hydrogen evolution Breaking the scaling relationship via thermally stable Pt/Cu single atom alloys for catalytic dehydrogenation A hydrazone-based covalent organic framework for photocatalytic hydrogen production A tunable azine covalent organic framework platform for visible light-induced hydrogen generation Sulfone-containing covalent organic frameworks for photocatalytic hydrogen evolution from water Diacetylene functionalized covalent organic framework (COF) for photocatalytic hydrogen generation Single-site photocatalytic H2 evolution from covalent organic frameworks with molecular cobaloxime co-catalysts Identification of prime factors to maximize the photocatalytic hydrogen evolution of covalent organic frameworks Sulfur-doped covalent triazine-based frameworks for enhanced photocatalytic hydrogen evolution from water under visible light Fast tuning of covalent triazine frameworks for photocatalytic hydrogen evolution Covalent triazine frameworks via a low‐temperature polycondensation approach Efficient visible light-driven water oxidation and proton reduction by an ordered covalent triazine-based framework Crystalline covalent triazine frameworks by in situ oxidation of alcohols to aldehyde monomers Structure-property-activity relationships in a pyridine containing azine-linked covalent organic framework for photocatalytic hydrogen evolution Sustained solar H2 evolution from a thiazolo [5 4-d] thiazole-bridged covalent organic framework and nickel-thiolate cluster in water Facile construction of fully sp2-carbon conjugated two-dimensional covalent organic frameworks containing benzobisthiazole units Covalent organic frameworks enabling site isolation of viologen‐derived electron‐transfer mediators for stable photocatalytic hydrogen evolution Integrated interfacial design of covalent organic framework photocatalysts to promote hydrogen evolution from water Lateral functionalization of a one-dimensional covalent organic framework for efficient photocatalytic hydrogen evolution from water Modulated connection modes of redox units in molecular junction covalent organic frameworks for artificial photosynthetic overall reaction In situ photodeposition of platinum clusters on a covalent organic framework for photocatalytic hydrogen production In situ construction of single‐atom electronic bridge on COF to enhance photocatalytic H2 production Optimizing Pt electronic states through formation of a schottky junction on non‐reducible metal-organic frameworks for enhanced photocatalysis Download references thanks the financial support from the National Key Research and Development Project of China (2022YFE0113800) the National Natural Science Foundation of China (22375179) the start-up grant of Zhejiang University of Technology (2019125016829) and National Natural Science Foundation of China Basic Research Program for Young Students (Doctoral Candidates) (223B2117) acknowledge the financial support from National Natural Science Foundation of China (51907173 and BL14W) beamlines for experimental data collection These authors contributed equally: Liangjun Chen College of Materials Science and Engineering Institute of Catalysis for Energy and Environment College of Chemistry and Chemical Engineering College of Chemical and Environmental Engineering conceived and designed the research project and all data are reported in the main text and supplemental materials reviewers for their contribution to the peer review of this work Download citation DOI: https://doi.org/10.1038/s41467-024-54959-8 Metrics details Reactive oxygen species with evoked immunotherapy holds tremendous promise for cancer treatment but has limitations due to its dependence on exogenous excitation and/or endogenous H2O2 and O2 Here we report a versatile oxidizing pentavalent bismuth(V) nanoplatform (NaBiVO3-PEG) can generate reactive oxygen species in an excitation-free and H2O2- and O2-independent manner Upon exposure to the tumor microenvironment NaBiVO3-PEG undergoes continuous H+-accelerated hydrolysis with •OH and 1O2 generation through electron transfer-mediated BiV-to-BiIII conversion and lattice oxygen transformation The simultaneous release of sodium counterions after endocytosis triggers caspase-1-mediated pyroptosis NaBiVO3-PEG intratumorally administered initiates robust therapeutic efficacies against both primary and distant tumors and activates systemic immune responses to combat tumor metastasis NaBiVO3-PEG intravenously administered can efficiently accumulate at the tumor site for further real-time computed tomography monitoring or alternative synergistic immune-radiotherapy this work offers a nanomedicine based on high-valence bismuth(V) nanoplatform and underscores its great potential for cancer immunotherapy alternative approaches with less or no dependence on exogenous excitation and endogenous H2O2 and O2 are urgently needed to produce abundant ROS and improve cancer immunotherapy because of their instability and the lack of facile structural engineering approaches BiV-based nanomedicines have not been reported or systematically investigated on ROS generation efficiency or direct tumor removal neither to say ROS-activated immunotherapy or imaging-guided combined cancer treatments The pentavalent bismuth(V) nanoplatform undergoes H+-accelerated hydrolysis when accumulated in tumor tissues or endocytosed by tumor cells with spontaneous •OH and 1O2 generation to induce apoptosis simultaneous release of sodium counterions to induce pyroptosis and the formation of insoluble hydrolysates to perform optional radiotherapy These features endow NaBiVO3-PEG with intrinsic capabilities to evoke immune responses to defeat primary and distant tumors and combat tumor metastasis a Schematic diagram of the template-assisted synthesis of NaBiVO3 and surface modification for NaBiVO3-PEG b Transmission electron microscopy (TEM) image of BiIIIOx d XRD pattern of BiIIIOx and NaBiVO3 (standard patterns for NaBiO3·2H2O with PDF No g High-resolution XPS spectra of Bi 4 f (f) h N2 adsorption-desorption isotherm curves of NaBiVO3-PEG Inset is the calculated pore size distribution by classical density functional theory (DFT) k TEM images (j) and the corresponding BF image High-angle annular dark-field (HADDF) image and elemental mapping images (k) of NaBiVO3-PEG and DSPE-mPEG5k (n = 4 independent experiments) and k) were independently repeated at least three times with similar results a Time-dependent TEM images of NaBiVO3-PEG when dispersed in SBF at different pH values b Release profiles of Na+ counterions in sodium-free potassium phosphate buffer at different pH values (n = 3 independent experiments) c Time-dependent Raman spectra of NaBiVO3-PEG when dispersed in SBF at pH 5.5 d XRD pattern of NaBiVO3-PEG after dispersed in SBF (pH 5.5) for 0 and 48 h e High-resolution XPS spectra of O 1 s of NaBiVO3-PEG after dispersed in SBF (pH 5.5) for 0 and 48 h (OA Adsorbed oxygen f EPR spectra of TEMP (1O2) and DMPO (•OH) mixed with NaBiVO3-PEG in PBS at pH 5.5 and pH 7.4 g Schematic diagram for fluorogenic detection of •OH by APF h Time-dependent fluorescence intensity of NaBiVO3-PEG recorded at 525 nm after incubation with APF in PBS at different pH values (n = 4 independent experiments) i Schematic diagram for fluorogenic detection of 1O2 by SOSG j Time-dependent fluorescence intensity of NaBiVO3-PEG recorded at 525 nm after incubation with SOSG in PBS at different pH values (n = 4 independent experiments) k Mechanism diagram of ROS formation of NaBiVO3 m Time-dependent fluorescence intensity of NaBiVO3-PEG recorded at 525 nm after incubation with APF (l) or SOSG (m) in PBS (pH 5.5) with or without H2O2 (1 mM) (n = 4 independent experiments) n EPR spectra of DMPO mixed with NaBiVO3-PEG in methanol-contained PBS (pH 5.5) with or without H2O2 (1 mM) o Mechanism diagram of redox reaction between NaBiVO3-PEG with H2O2 to generate •OOH and its impacts on •OH and 1O2 generation The data of (a) were independently repeated at least three times with similar results and the reaction rates with the values of 0.11 and 7.4 further confirmed the H+-accelerated hydrolytic behavior of NaBiVO3-PEG leading to the consumption of •OH and generation of 1O2 whereas Bi3+ would transfer into corresponding hydrolysates these data well exhibited an endogenous H2O2- and O2-independent and H+-accelerated ROS generation feature of NaBiVO3-PEG through a two-electron transfer associated hydrolytic under no exogenous stimulus for activation neither such systematic studies on the hydrolysis behavior of pentavalent bismuth(V) materials nor ROS generation without exogenous and endogenous stimulus have been reported in the literature yet b Cell viabilities of BIIIOx-PEG and NaBiVO3-PEG on L929 (a) (n = 5 independent experiments) and 4T1 (b) cells (n = 5 independent experiments) c Fluorescence images of intracellular •OH and 1O2 generation measured by APF and SOSG under different treatments d Fluorescence images of γ-H2AX immunofluorescence analysis of 4T1 cells under different treatments e Fluorescence image of 4T1 cells after different treatments detected by Annexin V-FITC/PI double staining g Fluorescence images of intracellular Na+ contents (f) and quantitative analysis of fluorescence intensities (g) detected by sodium-binding benzofuran isophthalate acetoxymethyl ester (SBFI-AM) probe under different treatments (n = 3 independent experiments) h Fluorescence images of JC−1 staining assays of 4T1 cells under different treatments j Fluorescence images of active caspase−1 activity of 4T1 cells under different treatments using the FAM-FLICA probe k Morphology of 4T1 cells under different treatments l Fluorescence image of CRT exposure in 4T1 cells under different treatments n ATP (m) and HMGB1 (n) expressions from 4T1 cells after different treatments (n = 3 independent experiments) o Schematic diagram of the transwell experiment q CD80 and CD86 expressions by flow cytometry (gating on CD11c+) (n = 3 independent experiments) and quantification of mature DCs (q) r Secretion of anti- and pro-inflammatory cytokines measured by ELISA assay (n = 3 independent experiments) and r) were analyzed by one-way ANOVA with Tukey’s multiple comparison test the minimal incubation concentration of NaBiVO3-PEG was 100 μg mL−1 with the concentration of Na+ determined to be 27.63 mM and the free Na+ concentration of 11.38 mM an extra Na+ release from NaBiVO3-PEG to induce pyroptosis would be about 16.25 mM all the results indicated a promising immunotherapeutic approach with pentavalent bismuth(V) nanoplatform a Treatment diagram of the bilateral tumor model in vivo Tumors on the right side were defined as primary tumors and on the left side as distant tumors c Individual (b) and average (c) volume curves of primary tumors from mice with different treatments (n = 6 mice each group) d Photograph of primary tumors collected at day 14 after different treatments Black circles indicate the primary tumors in mice completely disappeared after treatment f Individual (e) and average (f) volume curves of distant tumors from mice with different treatments (n = 6 mice each group) g Photograph of distant tumors collected at day 14 after different treatments Black circles indicate the distant tumors in mice completely disappeared after treatment h Representative ROS staining images of primary tumor slices collected on day 2 from the mice after receiving different treatments and Ki-67 staining images of primary tumor slices collected at day 14 from mice with different treatments and f) were analyzed by one-way ANOVA and Tukey multiple comparison test The P-values in (j) were analyzed by the Kruskal-Wallis and Dunn multiple-comparison test The data of (h and i) was independently repeated at least three times with similar results a–f Flow cytometric analyses of the percentage of major immune cells f) in distant tumor (n = 3 mice each group) g–j Secretion of cytokines in sera measured by ELISA assay (n = 3 mice each group) The P-values were analyzed by one-way ANOVA with Tukey’s multiple comparison test a Schematic diagram of lung metastasis model and treatment protocols b Photographs and H&E-stained images of lungs after different treatments on day 28 (n = 5 mice in each group) Metastatic nodules are marked with blue circles in the lungs and blue arrows in H&E staining images c Statistics and analysis of the number of tumor nodules on day 28 after different treatments (n = 5 independent experiments) d Representative immunohistochemistry images of CD4 and CD8 cell proliferation in the lungs of mice treated with different treatments e Quantification of CD4 and CD8 cell proliferation in (d) (n = 3 independent experiments) Data in (c and e) were shown as mean ± s.d The P-values in (c) were analyzed by the Kruskal-Wallis and Dunn multiple-comparison test The P-values in (e) were analyzed by one-way ANOVA with Tukey’s multiple comparison test a In vivo CT images of subcutaneous tumor-bearing mice after i.v Blue circles indicate the subcutaneous tumor section in mice c Relative intensities of APF incubated with PBS or primed NaBiVO3-PEG hydrolysate (n = 6 independent experiments) e Western blot results (d) and quantitative analysis (e) of DNA damage repair proteins of 53BP1 and RAD51 in cells after different treatments (n = 3 independent experiments) f Cell cycle states of 4T1 cells under different treatments (n = 3 independent experiments) g Viabilities of 4T1 cells after different treatments (n = 4 independent experiments) h Experimental diagram of model establishment and antitumor treatments j Average (i) and individual (j) volume curves of tumors after different treatments (n = 5 independent experiments) l Representative immunohistochemistry images (k) and quantification (l) of CD4 and CD8 cell proliferation in the lungs of mice treated with different treatments (n  =  3 mice each group) and l) were analyzed by one-way ANOVA with Tukey’s multiple comparison test b and k) were independently repeated at least three times with similar results thus confirming the robust therapeutic feature multi-facet therapeutic feature of proposed high-valent bismuth(V) nanoplatform a Long-term biodistribution of NaBiVO3-PEG in the main organs of mice at each time point after systematic administration (n = 3 mice in each group) b Bismuth amounts measured in urine and feces collected at different time points (n = 3 mice each group) d Representative H&E staining images (c) and Hematology (d) at days 1 and 14 after mice treated with NaBiVO3-PEG (n = 4 mice each group) The data of (c) was independently repeated three times with similar results there exists a blank in the field of inorganic chemistry involving the fine fabrication and applications of BiV-based materials we envisage the proposed template-assisted method and robust immunotherapeutic capabilities might proceed more designs and use of bismuth or bismuth(V) in the field of inorganic chemistry this work provides a promising paradigm for achieving ROS regardless of either exogenous stimulus or endogenous H2O2 and O2 based on high-valence bismuth(V) nanomedicine and presents its powerful potential of tumor-specific but independent cancer immunotherapy The research presented here complies with all relevant ethical regulations All experiments involving animals were reviewed and approved (O_A2024061) by the guidelines of the Institutional Animal Care and Use Committee of Shanghai Jiao Tong University The maximal tumor burden permitted by the Institutional Animal Care and Use Committee of Shanghai Jiao Tong University was the weight of the tumor should not exceed 10% of body weight In any of the animal experiments described in this article the maximal tumor size of the mouse was never reached All chemicals were provided by commercial sources without further purification before use Bismuth trinitrate pentahydrate (Bi(NO3)3·5H2O) and Anti-CTLA4 were purchased from Macklin and manganese sulfate tetrahydrate (MnCl2·4H2O) were purchased from General Reagent 1,2-dioleoyl-sn-glycero-3-phosphate sodium salt (DOPA) and mitochondrial membrane potential detection kit (JC-1) were purchased from Aladdin Simulated body fluid was purchased from Phygene 2′,7′-dichlorofluorescein diacetate (DCFH-DA and fluorescent Na+ indicator were purchased from Abcam 2-Distearoyl-sn-glycero-3-phosphoethanolamine-Poly(ethylene glycol) 5000 (DSPE-mPEG5k) was purchased from Yuanye Bio-Technology TEM images and energy-dispersive X-ray spectrometer (EDS) mapping were captured using a scanning / transmission electron microscope (STEM X-ray diffraction patterns were performed using an X-ray diffractometer (D8 DaVinci Bruker) with Cu Kα radiation (λ = 1.5418 Å) Ultraviolet-visible-near-infrared (UV-vis-NIR) absorption spectra were measured using a UV-vis-NIR spectrophotometer (Lamda 950 Fourier transform infrared (FTIR) spectra were obtained by the FTIR spectrometer (Nicolet 6700 Dynamic light scattering (DLS) and Zeta potential were analyzed with Zetasizer Nano S (omni N2 adsorption-desorption isotherm curve was analyzed by surface area and micropore size analyzer (Autosorb-IQ3 Thermogravimetric analysis curves were obtained by thermogravimetric analyzer (TGA8000 0.375 mmol) was dissolved in HNO3 solution (5 mL The mixture was then transferred into a stainless-steel autoclave with a Teflon liner The autoclave was sealed and maintained at 150 °C for 3 h The obtained products were centrifuged and washed with deionized water 3 times and finally dried in a desiccator for further characterization 1 mL BiIIIOx in NaClO solution (20 mg mL-1) was added dropwise to a Teflon tube containing 5 mL NaOH solution (10 M) and stirred vigorously for 10 min at room temperature and finally dried in a desiccator for further characterization 1 mL NaBiVO3 nanoparticles solution (20 mg mL−1 in chloroform) and 1 mL DOPA solution (2 mg mL−1 in chloroform) were mixed under ultrasonication for 20 min The obtained turbid solution was centrifuged to remove unbounded free DOPA and redispersed in chloroform PEGylation of NaBiVO3-PEG was then conducted by mixing a chloroform solution of DSPE-mPEG5k (2 mg mL−1) under vigorous stirring for 8 h and the obtained NaBiVO3-PEG nanoplatform was restored for further use Similar protocols were used for the preparation of BiIIIOx-PEG The prepared NaBiVO3-PEG and BiIIIOx-PEG were freeze-dried and stored at 2 ~ 8 °C in an inert gas atmosphere and low humidity environment 2 mg NaBiVO3-PEG NPs was dispersed in 1 mL SBF or sodium-free potassium phosphate buffer at different pH values (5.5 and 7.4) and gently shaken at 100 rpm at room temperature The solution was centrifuged to collect 100 μL of supernatant at predetermined time points (1 The released amounts of Na+ in sodium-free potassium phosphate buffer were quantified by ICP-MS The hydrolysates were collected for characterization using TEM 2 mg NaBiVO3-PEG was dispersed in 1 mL sodium-free potassium phosphate buffer at different pH values (5.5 and gently shaken at 100 rpm at room temperature The release of Na+ in sodium-free potassium phosphate buffer was quantified by ICP-MS The NaBiVO3-PEG concentration at the corresponding time point was calculated by the Na+ content: M[Na+] = M[BiIII] = 0.633 M – M[BiV] where 0.633 M is the initial concentration of NaBiVO3-PEG To further quantify and compare the degradation rates of NaBiVO3-PEG at different pH values the relationship between the hydrolysis times at each pH can be further fitted with the following formulas (Ep Ct is the concentration of NaBiVO3-PEG at time t (mg L−1) and C0 is the initial concentration of NaBiVO3-PEG (mg L−1) 5,5-dimethyll-pyrroline N-oxide (DMPO) and 2,2,6,6-Tetramethylpiperidine (TEMP) were used as •OH and 1O2 trapping agents An equal volume of 100 μg mL−1 NaBiVO3-PEG solution in PBS (pH 7.4 or pH 5.5) was mixed with 2 mM DMPO or 10 mM TEMP aqueous solution for 1 min and then tested by ESR APF and SOSG were used as 1O2 and •OH sensors NaBiVO3-PEG solution (final concentration 200 μg mL−1) in PBS (pH 5.5 followed by adding APF or SOSG with a final concentration of 10 μM Fluorescence signals (Ex: 488 nm and Em: 525 nm) were collected every 10 min using a multifunctional microscope (BioTek) NaBiVO3-PEG (final concentration 200 μg mL−1) was dispersed in PBS (pH 5.5) with different H2O2 concentrations (final concentration 0 The concentration of NaBiVO3-PEG and ROS sensors in solution is the same as above deoxygenate PBS was using nitrogen bubbling NaBiVO3-PEG and SOSG (or APF) were dispersed in 3 mL deoxygenated PBS and quickly sealed with a sealing film to prevent oxygen from entering then the fluorescence spectra of the sample at 525 nm (510 nm) under 488 nm excitation were monitored every ten minutes using a fluorescence spectrometer (Edinbuegh FLS1000) L929 (SCSP-5039) mouse fibroblasts cell line MC3T3-E1 (SCSP-5218) mouse embryonic osteoblasts cell line and 4T1 (SCSP-5056) mouse breast cancer cells line was obtained from Stem Cell Bank HUVEC (KCB2012087YJ) human umbilical vein endothelial cell line was obtained from Stem Cell Bank MB49 (IM-M012) mouse bladder cancer cell line was purchased from IMMOCELL (Xiamen and MB49 were cultured in RPMI 1640 media with 10% FBS and 1% penicillin/streptomycin 4T1 were cultured in DMEM media with 10% FBS and 1% penicillin/streptomycin Cells were incubated in a humidified atmosphere containing 5% CO2 at 37 °C and MB49 cells were cultured in 96-well plates (37 °C 5% CO2) for 24 h at a density of 10,000 cells per well different concentrations of BiIIIOx-PEG or NaBiVO3-PEG in DMED (or RPMI) medium (0 cells were washed three times with neutral PBS and cell viability was tested by Cell Counting Kit-8 assay kit (Yeasen) 4T1 cells were cultured in 6-well plates (37 °C 5% CO2) for 24 h at a density of 5*105 cells per well NaBiVO3 was modified with rhodamine B (RhB) by using DOPA and DSPE-mPEG5k-RhB was dispersed in the cell culture medium solution (100 μg mL−1) and incubated for another 12 h Fluorescence images were acquired on a Leica DMi8 fluorescence microscope (ex/em: 460-500/512-542 nm; 540-580/592-668 nm) BiIIIOx-PEG and NaBiVO3-PEG in a cell culture medium (100 μg mL−1) were added for another 12 h incubation APF or SOSG (final concentration: 10 μM) was added and ROS generation was estimated by fluorescence microscope (EX: 488 nm and EM: 505–545 nm) after being co-cultured with BiIIIOx-PEG and NaBiVO3-PEG were analyzed by DNA Damage Assay Kit by γ-H2AX immunofluorescence after being co-cultured with BiIIIOx-PEG and NaBiVO3-PEG were analyzed by annexin V-APC/PI dual-staining kit Cellular apoptosis and necrosis were analyzed using a laser confocal microscope (TCS SP8 STED 3X BiIIIOx-PEG and NaBiVO3-PEG in medium (final concentration 100 μg mL−1) were added for another 24 h incubation 4T1 cells were further incubated with 10 µM Na+ indicator sodium binding benzofuran isophthalate acetoxymethyl ester (SBFI-AM) in 0.04% Pluronic F-127 (Beyotime Biotechnology) and the fluorescence signal was measured by fluorescence microscope (EX: 488 nm and EM: 505-545 nm) 4T1 cells were further incubated with JC-1 for 20 min and then washed by JC-1 buffer solution before being measured by fluorescence microscope (EM: 510–540 nm (JC-1 monomer) and EM: 570-620 nm (JC-1 aggregate) 4T1 cells were further incubated with FAM-FLICA®Caspase-1 following the manufacturer’s protocols and the fluorescence signal was measured by fluorescence microscope (EX: 488 nm and EM: 505–545 nm) (Leica DMi8) For the experiment of calculating the intracellular Na+ content 4T1 cells were co-cultured with different concentrations of NaBiVO3-PEG (0 4T1 cells were detected using SBFI AM after different treatments and the fluorescent signal of the cells was detected using a microplate reader (Ex: 488 nm and Em: 525 nm) (BioTek) the concentration of Na+ in cells treated with different concentrations of NaBiVO3-PEG was calculated by using the linear-fit standard curve of the Na+ concentration-fluorescence signal The linear fitting standard curve of the Na+ concentration-fluorescence signal was obtained by incubating SBFI AM with different Na+ concentrations (0 BiIIIOx-PEG and NaBiVO3-PEG in a medium (100 μg mL−1) were added for another 24 h incubation the treated 4T1 cells were further incubated with anti-calreticulin antibody (absin) for 12 h at 4 °C and then incubated with Goat anti-Rabbit IgG-cy5 (absin) for 2 h before being measured by fluorescence microscope (EX: 638 nm and EM: 670 nm) the supernatants of the treated 4T1 cells were collected and then the release of HMGB1 and ATP was detected by the HMGB1 ELISA Kit (Servicebio Biological) and ATP Assay Kit according to the manufacturer’s protocols with 4T1 cells contained in the upper chamber and DCs contained in the lower chamber 4T1 cells were incubated overnight and then added with BiIIIOx-PEG and NaBiVO3-PEG NPs in the medium DCs were collected and stained with anti-mouse CD11C PE (Thermo 1:1000 dilution) antibodies and then analyzed by flow cytometry (Amnis® ImageStream®XMk II and TNF-α was measured by ELISA Kit (Servicebio Biological) A total of 140 female Balb/c mice (3 weeks 14–16 g) were provided by Shanghai Lingchang Biotechnology Co. All mice were acclimatized for 1 week and then used for experiments All mice were housed in SPF-grade facilities cages with standard conditions (50% relative humidity and 12/12 h light-dark cycle) at 25 °C animal euthanasia was performed by CO2 inhalation followed by cervical dislocation All animal procedures were in agreement with the guidelines of the Institutional Animal Care and Use Committee of Shanghai Jiao Tong University (Approved ID: O_A2024061) 4T1 tumor cells were inoculated subcutaneously on the left and right sides of Balb/c mice The tumor on the right side was designed as a primary tumor for the synergistic therapy of ROS and the surge of osmolarity of NaBiVO3-PEG NPs and the left tumor was used as a distant tumor to investigate the antitumor immunological effect The tumor-bearing mice were randomly divided into five groups: (a) PBS group and (e) NaBiVO3-PEG NPs + Anti-CTLA4 group Mice were intratumorally injected with BiIIIOx-PEG NPs or NaBiVO3-PEG NPs (16 mg kg−1) on 0 days Then 20 µg of Anti-CTLA4 was injected intravenously on day 1 The body weight and tumor volume of mice were recorded every two days Tumor volume was measured with a vernier caliper and calculated as (length × width2)/2 Tumor sections of each group were further studied by caspase-1 To systematically investigate the in vivo antitumor immune responses lymph nodes and distant tumors of mice were harvested The lymph nodes and distant tumors were made into a single-cell suspension according to the specified procedure tumor cells were stained with anti-mouse CD11b PE-Cy7 (Thermo Fisher) and anti-mouse CD4 FITC antibodies (Thermo Fisher) according to the manufacturer’s protocols the content of CD8+ and CD4+ T cells was detected by flow cytometry analysis (Fortessa X20) the maturity of DC cells in the lymph nodes was checked by anti-mouse CD11C PE (Thermo Fisher) and anti-CD80- APC antibody (Thermo Fisher) staining assay according to the manufacturer’s protocols Serum samples were isolated from mice after different treatments and diluted for analysis and IL-6 were measured by ELISA Kit (Servicebio Biological) and membrane is assigned as 0 = no positive staining The extent of stained cells is assigned as 0 = 0–5%; 1 = 6–25%; 2 = 26–50%; 3 = 51– 75%; 4 = 76–100% The h-score was calculated by the following formulas (Ep pi = percentage of weak/moderate/strong intensity; i = 1 A micro-CT (Bruker SkyScan 1276) was used for in vivo CT imaging Imaging parameters were as follows: field of view The reconstruction was done by using the filtered back projection method tumor-bearing mice were intravenously injected with 200 μl of NaBiVO3-PEG (8 mg mL−1) before imaging tumor-bearing mice were imaged by a small animal X-ray CT The mice whole-body 360° scan lasted about 20 min under isoflurane anesthesia and 72 h from parallel experimental mice and then dehydrated Three times each experiment was repeated independently with similar results followed by adding NaBiVO3-PEG in medium (100 μg mL−1) for another 24 h incubation and the protein concentration of the cell lysates was determined by bicinchoninic acid (BCA) Protein Assay Equal amounts of protein were loaded onto SDS-PAGE gel (10% gel) (Beyotime) and transferred to PVDF membranes Western blot was performed using primary antibodies and secondary antibodies conjugated with HRP the following antibodies were used: anti-β-actin (Beyotime and anti-53BP1 antibody (Cell SignalingTechnology BIO-RAD GelDoc XR was used to detect the blot bands and HMGB1 of 4T1 cells co-cultured with PBS and NaBiVO3-PEG were tested by similar methods above with the following antibodies: Caspase-1 Rat mAb (ABclonal NaBiVO3-PEG was dissolved in PBS at pH 5.5 for 48 h to ensure complete degradation to form hydrolysates NaBiVO3-PEG and hydrolysates were collected and redispersed in neutral PBS with a final concentration of 200 μg mL−1 in a 96-well plate APF was then added with a final concentration of 10 μM and the solutions were irradiated with X-ray at a dose of 0 The fluorescence signals (Ex: 488 nm and Em: 525 nm) were immediately collected using a multifunctional microscope 5% CO2) for 24 h at a density of 5 × 105 cells per well BiIIIOx-PEG and NaBiVO3-PEG dissolved in DMED medium (final concentration 200 μg mL−1) 4T1 cells were collected and washed twice with cold PBS v/v) was added to the harvested cells for fixation Cell cycle distributions were determined by flow cytometry using the Cell Cycle and Apoptosis Analysis Kit (Beyotime Biotechnology) and all data were processed by ModFit LT 5.0 software 4T1 cells were cultured in 96-well plates (37 °C NaBiVO3-PEG and hydrolysates in a medium (final concentration 200 μg mL−1) were added the cells were washed three times with neutral PBS and cell viability was tested by Cell Counting Kit-8 assay kit (Yeasen) Female Balb/c mice (5-6 weeks old) were chosen for the establishment of the 4T1 tumor model When the tumor volumes reached about 100 mm3 4T1 tumor-bearing mice were randomly divided into six groups (n = 5) where PBS or NaBiVO3-PEG (10 mg kg−1) was i.v and groups that needed radiotherapy were treated with X-ray (4 Gy) Tumor sizes and body weights were recorded every two days all the mice were sacrificed for the exfoliation of tumors Healthy female Balb/c (4-week) mice were divided into two groups NaBiVO3-PEG (10 mg kg−1) was perfused into mice by i.v blood samples and major organs were dissected from mice for blood routine tests and H&E staining analysis Blood routine test parameters included white blood cell (WBC) coefficient variation of red blood cell volume distribution width (RDW-CV) mean corpuscular hemoglobin concentration (MCHC) standard deviation in red cell distribution width (RDW-SD) Four female Balb/c mice (4-weeks) were intravenously injected with 100 µL of NaBiVO3-PEG solution (10 mg kg−1) 8 μL of blood was collected from the eye sockets at different time points (20 min and 48 h) and diluted in 0.2 mL of aqua regia the time-dependent concentration of bismuth ions in the blood was determined by ICP-MS the data were analyzed using Origin software and the half-life was calculated based on the one-component pharmacokinetic model the biodistribution and metabolic behavior of NaBiVO3-PEG after i.v 200 μL) was also carried out by using the above method with individual female Balb/c mice (4-weeks) and urine of mice were collected and weighed at different times the concentration of bismuth ions was analyzed by ICP-MS OriginPro (2024) was used for all data analysis Quantitative results were expressed as mean ± s.d and Kruskal-Wallis and Dunn multiple-comparison test were used to determine statistical significance Exact P-values were provided accordingly in the figures or captions P < 0.05 was used as the threshold for statistical significance; (*) indicates P < 0.05 No statistical method was used to predetermine the sample size The exact number of replicates and statistical tests are indicated in the figure legends n represents the number of independent experimental replicates Flow cytometry data were analyzed with CytExpert 2.3 (Cell cycle distribution data were analyzed with ModFit LT 5.0) Western blot data were analyzed with ImageJ Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article The authors declare that all relevant data that support the findings of this work are presented in the Article, and all remaining data can be found in the Supplementary Information and Source Data files. Source data are provided in this paper Patients with melanoma treated with an anti-PD-1 antibody beyond RECIST progression: a US food and drug administration pooled analysis Responsive biomaterials: optimizing control of cancer immunotherapy Immunological off-target effects of imatinib Enhancing cancer immunotherapy with nanomedicine Tumour hypoxia promotes tolerance and angiogenesis via CCL28 and Treg cells Tumor endothelium FasL establishes a selective immune barrier promoting tolerance in tumors and acquired resistance to cancer immunotherapy Reactive oxygen species (ROS)-based nanomedicine Reactive oxygen species-powered cancer immunotherapy: Current status and challenges Unraveling mitochondria-targeting reactive oxygen species modulation and their implementations in cancer therapy by nanomaterials Recent progress and trends in X-ray-induced photodynamic therapy with low radiation doses Antiangiogenesis combined with inhibition of the hypoxia pathway facilitates low-dose CT/MRI-guided synergistic radiotherapy and X-ray inducible photodynamic therapy using Tb-doped Gd-w-nanoscintillators Clinical development and potential of photothermal and photodynamic therapies for cancer Advanced biotechnology-assisted precise sonodynamic therapy Solutions to the drawbacks of photothermal and photodynamic cancer therapy Na2S2O8 nanoparticles trigger antitumor immunotherapy through reactive oxygen species storm and surge of tumor osmolarity HXV2O5 nanocatalysts combined with ultrasound for triple amplification of oxidative stress to enhance cancer catalytic therapy Liquid nanoparticles for nanocatalytic cancer therapy Hollow MnO2 as a tumor-microenvironment-responsive biodegradable nano-platform for combination therapy favoring antitumor immune responses Facile synthesis of sodium bismuthate dihydrate and its efficient visible-light photocatalytic activity characterization and photocatalytic activity evaluation of NaBiO3·2H2O and NaBiO3·xH2O nanosheets Preparation of NaBiO3 and the electrochemical characteristic of manganese dioxide doped with NaBiO3 Metallic artifacts-free spectral computed tomography angiography based on renal clearable bismuth chelate Biomolecules-assisted synthesis of degradable bismuth nanoparticles for dual-modal imaging-guided chemo-photothermal therapy Tetramodal imaging and synergistic cancer radio-chemotherapy enabled by multiple component-encapsulated zeolitic imidazolate frameworks Phase-transitional bismuth-based metals enable rapid embolotherapy An X-ray computed tomography imaging agent based on long-circulating bismuth sulphide nanoparticles The versatile biomedical applications of bismuth-based nanoparticles and composites: therapeutic Nanocellulose templated growth of ultra-small bismuth nanoparticles for enhanced radiation therapy Spatiotemporally synchronous oxygen self-supply and reactive oxygen species production on Z-scheme heterostructures for hypoxic tumor therapy Enhanced sonodynamic cancer therapy through iron-doping and oxygen-vacancy engineering of piezoelectric bismuth tungstate nanosheets Oxygen-vacancy-rich piezoelectric BiO2−x nanosheets for augmented piezocatalytic Self-triggered thermoelectric nanoheterojunction for cancer catalytic and immunotherapy Ultrasmall semimetal nanoparticles of bismuth for dual-modal computed tomography/photoacoustic imaging and synergistic thermoradiotherapy Edge modification facilitated heterogenization and exfoliation of two-dimensional nanomaterials for cancer catalytic therapy Bi3+ self doped NaBiO3 nanosheets: Facile controlled synthesis and enhanced visible light photocatalytic activity Generation of singlet oxygen over Bi(V)/Bi(III) composite and its use for oxidative degradation of organic pollutants Removal of Cl(−I) from strongly acidic wastewater using NaBiO3: A process of simultaneous oxidation and precipitation In situ H+-mediated formation of singlet oxygen from NaBiO3 for oxidative degradation of bisphenol A without light irradiation: Efficiency Bioorthogonal coordination polymer nanoparticles with aggregation-induced emission for deep tumor-penetrating radio- and radiodynamic therapy Zinc cyclic di-AMP nanoparticles target and suppress tumours via endothelial STING activation and tumour-associated macrophage reinvigoration Activation of periodate by freezing for the degradation of aqueous organic pollutants Toxic DNA damage by hydrogen peroxide through the fenton reaction in vivo and in vitro Mechanism of the ferric ion catalyzed decomposition of hydrogen peroxide Electrode potentials of partially reduced oxygen species Simultaneous fenton-like ion delivery and glutathione depletion by MnO2-based nanoagent to enhance chemodynamic therapy An AND logic gate for magnetic-resonance-imaging-guided ferroptosis therapy of tumors Hydroxyl radical scavengers inhibit human natural killer cell activity BiFeO3/MoS2 nanocomposites with the synergistic effect between ≡MoVI/≡MoIV and ≡FeIII/≡FeII redox cycles for enhanced Fenton-like activity Dual Active centers linked by a reversible electron station as a multifunctional nanozyme to induce synergetically enhanced cascade catalysis for tumor-specific therapy Calcium-overload-mediated tumor therapy by calcium peroxide nanoparticles NaCl nanoparticles as a cancer therapeutic Checkpoint blockade and nanosonosensitizer-augmented noninvasive sonodynamic therapy combination reduces tumour growth and metastases in mice MnOx nanospikes as nanoadjuvants and immunogenic cell death drugs with enhanced antitumor immunity and antimetastatic effect Phototheranostic metal-phenolic networks with antiexosomal PD-L1 enhanced ferroptosis for synergistic immunotherapy Heterostructural nanoadjuvant CuSe/CoSe2 for potentiating ferroptosis and photoimmunotherapy through intratumoral blocked lactate Efflux Sodium citrate nanoparticles induce dual-path pyroptosis for enhanced antitumor immunotherapy through synergistic ion overload and metabolic disturbance Therapeutically reprogrammed nutrient signalling enhances nanoparticulate albumin bound drug uptake and efficacy in KRAS-mutant cancer A transistor-like pH-sensitive nanodetergent for selective cancer therapy in Drug Discovery and Evaluation: Safety and Pharmacokinetic Assays (eds Hock F The covalent modifier Nedd8 is critical for the activation of Smurf1 ubiquitin ligase in tumorigenesis Prostate-specific membrane antigen heterogeneity and DNA repair defects in prostate cancer Full-course NIR-II imaging-navigated fractionated photodynamic therapy of bladder tumours with X-ray-activated nanotransducers Bi)-based nanomaterials for cancer imaging and therapy: a materials perspective Full-process radiosensitization based on nanoscale metal–organic frameworks Catalytic MnWO4 nanorods for chemodynamic therapy synergized radiotherapy of triple negative breast cancer Renal albumin filtration: alternative models to the standard physical barriers Structure of the active pharmaceutical ingredient bismuth subsalicylate Chemodynamic terapy: tmour mcroenvironment-mdiated Fenton and Fenton-like ractions Fluorination of arylboronic esters enabled by bismuth redox catalysis Download references This work was financially supported by the National Natural Science Foundation of China the National Key Research and Development Program of China Inner Mongolia Autonomous Region-Shanghai Jiao Tong University Science and Technology Cooperation Special Project and Translational Medicine Research Fund of National Facility for Translational Medicine (Shanghai) Natural Science Foundation of Shanghai (23ZR1434600) (X.Y.) We thank the Instrumental Analysis Center of SJTU for its assistance with optical and TEM characterizations We also thank Zhejiang Orient Gene Biotech Co. State Key Lab of Metal Matrix Composites School of Materials Science and Engineering Shanghai Jiao Tong University 800 Dongchuan Road Future Material Innovation Center Zhangjiang Institute for Advanced Study Shanghai Jiao Tong University 429 Zhangheng Road Department of Comprehensive Oncology National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital Chinese Academy of Medical Sciences and Peking Union Medical College 17 Panjiayuan South Lane Department of Urology Shanghai General Hospital Shanghai Jiao Tong University School of Medicine 85 Wujin Road conceived the idea and designed the experiments performed pentavalent bismuth(V) nanoplatform synthesis and characterization and conducted in vitro and in vivo experiments with help from X.Y. All authors participated in the discussion and analysis of experimental results and manuscript Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work Download citation DOI: https://doi.org/10.1038/s41467-025-56110-7 Metrics details Negative bias is an essential characteristic of depressive episodes leading patients to attribute more negative valence to environmental cues This negative bias affects all levels of information processing including emotional response leading to the development and maintenance of depressive symptoms pleasant stimuli become less attractive and unpleasant ones more aversive yet the related neural circuits underlying this bias remain largely unknown By studying a mice model for depression chronically receiving corticosterone (CORT) we showed a negative bias in valence attribution to olfactory stimuli that responds to antidepressant drug This result paralleled the alterations in odor value assignment we observed in bipolar depressed patients Given the crucial role of amygdala in valence coding and its strong link with depression we hypothesized that basolateral amygdala (BLA) circuits alterations might support negative shift associated with depressive states where limits in spatial resolution of imaging tools impair easy amygdala segmentation recently unravelled specific BLA circuits implicated in negative and positive valence attribution could be studied in mice Combining CTB and rabies-based tracing with ex vivo measurements of neuronal activity we demonstrated that negative valence bias is supported by disrupted activity of specific BLA circuits during depressive states Chronic CORT administration induced decreased recruitment of BLA-to-NAc neurons preferentially involved in positive valence encoding while increasing recruitment of BLA-to-CeA neurons preferentially involved in negative valence encoding this dysfunction was dampened by chemogenetic hyperactivation of BLA-to-NAc neurons altered BLA activity correlated with durable presynaptic connectivity changes coming from the paraventricular nucleus of the thalamus recently demonstrated as orchestrating valence assignment in the amygdala our findings suggest that specific BLA circuits alterations might support negative bias in depressive states and provide new avenues for translational research to understand the mechanisms underlying depression and treatment efficacy Here we propose to study how alteration in valence attribution towards emotional stimuli underly negative bias in depression we analyzed the valence attributed to olfactory stimuli in both humans and mice As we observed a similar olfactory negative bias in depressed patients and mouse model for depression we explored the underlying mechanisms in animal We hypothesized that disrupted neuronal activity of specific BLA circuits supports negative bias associated with depressive states All euthymic and depressive patients were on a mood stabilizer All animal care and experimental procedures followed national and European (2010/63/EU) guidelines and were approved by the French Ministry of Research (APAFiS: #16380-2018080217358599_v1) n = 181) were socially housed (4–6 per cage) and maintained under standard conditions (23 ± 1 °C; humidity 40%) on a 14/10 h light/dark cycle with food and water ad libitum Clean housing cages with regular bedding material were used as testing arenas A petri dish with hooled cover was placed at one side of the arena The behavior was recorded by video camera for 15 min and the Noldus Ethovision 11.0 (Netherlands) system was used to track position and locomotion of mice only a Whatman paper filter (GE Healthcare Life Sciences 2 days were dedicated to each odor with the odorant placed on paper filter in the following order: peanut oil (pure 400 ul) and trimethylthiazoline (Sigma-Aldrich 400 ul) (see Supplemental information for more details) To label neurons of the BLA projecting to NAc or CeA the retrograde tracer Cholera Toxin Subunit B conjugated with Alexa Fluor 555 or 674 (CTB 555 or CTB 647 USA) was bilaterally injected with a micropipette connected to Nanoject III microinjector (Drummond Scientific) in the NAc (AP : +1.4 mm ; ML : ±0.87 mm ; DV : −4.35 mm For labelling presynaptic inputs to BLA-to-NAc and BLA-to-CeA neurons AAV-FLEX-G-TVA-GFP was injected bilaterally in the BLA (AP: −1.75 mm 100 nl) and AAVretro-CRE either in the NAc (100 nl) or in the CeA (50 nl) using previously mentioned coordinates mice were injected with EnvA-RVΔG-mCherry (150 nl) in the BLA and sacrificed one week later AAV5-hSyn-DIO-mCherry and AAVretro-PGK-Cre recombinant adeno-associated virus were used to manipulate BLA neuronal activity Viral vectors were bilaterally injected into the BLA (100–150 nl) and the NAc (100–150 nl) or the CeA (50 nl) using previously mentioned coordinates All AAV were provided by Addgene (USA) and EnvA-RVΔG-mCherry was generated by Karl-Klaus Conzelmann lab (University of Munich Mice were anesthetized with ketamine/xylazine mix (intraperitoneal (i.p). 100-125 mg/kg and 50 mg/kg respectively) and perfused transcardially with 0.9% NaCl followed by 4% paraformaldehyde in phosphate buffer Forty-micrometer-thick coronal brain sections were obtained using a sliding microtome (Leica SM 2010 R) Immunostaining was performed on free-floating sections using the following antibodies: guinea pig anti-c-Fos (1:2000 Abcam) and secondary antibodies (Alexa-conjugated secondary antibodies at 1:1000 Statistical analyses were performed with GraphPad Prism v9 software (USA) Two-sided parametric or non-parametric tests were used according to the normality of datasets and variances homoscedasticity Following post-hoc analyses were applied with the False Rate Discovery (FDR) correction method All datasets were described using the mean; error bars in the figures represent standard error mean (SEM), except for Figure S4E–G where box-and-whiskers plots were used (boxes represent the 25th median and 75th percentiles while whiskers represent the 1st and 99th percentiles) Differences were considered statistically significant for p < 0.05 Detailed methods and materials can be found in the Supplemental information A Depressed bipolar patients (n = 23) classified less odors as pleasant ones than control subjects (n = 11) and remitted bipolar patients (n = 25 ; **p < 0.01) and more as unpleasant ones compared to remitted bipolar depresssed patients and control subjects (*p < 0.05) (Group: F(2,56) = 0.00 B–D The number of classified neutral odors did not change with depression severity (B) Bipolar depressive patients with higher depression severity (i.e higher MADRS score) classified less odors as pleasant (C) F CORT mice presented anxiety-like phenotypes in the open field (OF and depressive-like behaviors in the splash test (ST p = 0.001) but not in the tail suspension test (TST G Global emotionality score is increased in CORT mice (t(32) = 4.23) H Scheme of the olfactory preference test using neutral (grey) appetitive (pink) and aversive (blue) odors I Representative mouse tracks colored by the density of position points TMA and TMT than Veh controls (Group: F(1,20) = 10.63 K Principal Component Analysis on the 23 behavioral parameters evaluated in 87 mice from both CORT (red) and Veh (grey) mice L The olfactory preference test (OP) standed for 86% of contributions to PC1 (blue bar parts) The OF (32%) and TST (28%) contributed mostly to PC2 PC2 and PC2:PC3 significantly predicted the Veh/CORT status N Mice received chronic CORT to model depression (grey or CORT for 4 weeks then fluoxetine in addition to CORT for the following weeks (green O The reduction of the global emotionality score shows an improvement of the depressed- and anxious-like phenotype in responsive mice (U = 11.78 Fluoxetine-responsive mice are defined by an emotionality score lower than −0.5 P CORT + FLX-Responsive (n = 5) mice explored more peanut oil and ♀ urine than CORT (n = 7) and CORT + FLX-Non Responsive (n = 9) groups (Group: F(2,19) = 9.629 To assess valence assignment in mice, we set up an olfactory preference test (Fig. 1H) Each odor was presented on two consecutive days and its attributed valence was defined by increased decreased or similar odor zone exploration time compared to habituation in Vehicle (Veh)-treated control mice (One-way repeated measures (RM) ANOVA: F(4,40) = 44.86 Peanut oil was neutral in our experimental conditions (t(40) = 1.36 whereas female urine was appetitive (t(40) = 3.85 q < 0.001) and synthetic compounds trimethylamine (TMA) and 2,4,5-trimethylthiazole (TMT) were aversive (TMA: t(40) = 5.68 Predictors significantly associated with the group were PC1 PC3:PC6 and PC5:PC6 showed a statistical trend to association with the group these results suggest that olfactory valence assignment is an independent and suitable behavioral measurement to predict differences between control and depressive-like states in mice relief of depressive state in remitted patients and responsive mice was associated with partial improvement of negative olfacory valence bias B Retrograde fluorescent CTB647 and CTB555 dyes were injected in the NAc (green) and CeA (orange) respectively Odors were presented to trigger the immediate-early gene c-Fos expression in CORT (red) and Veh (grey) mice C Representative image of BLA c-Fos expression colocalized with CTB647 and/or CTB555 c-Fos + /CTB555+ (F) and c-Fos + /CTB647 + /CTB555+ cell number (G) in the BLA BLA c-Fos+ cell density was similar between groups (D BLA c-Fos + /CTB647+ cell density decreased in CORT mice (E whereas c-Fos + /CTB555+ cell density increased (F BLA c-Fos + /CTB647 + /CTB555+ cell density tended to decrease in CORT mice (G CORT n = 15) (H) Distribution of CTB647 and/or CTB555 colocalization among the total number of c-Fos+ cells in the BLA (χ²(3) = 340.0 chronic CORT administration induced decreased recruitment of BLA-to-NAc neurons and increased recruitment of BLA-to-CeA neurons circuits already implicated in opposite valence encoding H–I Representative images (right) and percentage of inputs neurons (left) to BLA-to-NAc (H) and BLA-to-CeA (I) arriving from subregions of PVT posterior/anterior or BLA/LA respectively in both CORT and Veh groups (relative to the total) (H Half of these local inputs arrive from CamkII+ cells impinging both BLA-to-NAc (53.3% +/− 4.07; n = 298 cells) and BLA-to-CeA (51.8% +/− 5.10; n = 108 cells) neuronal populations but also with changes from local BLA inputs A To mimic chronic CORT phenotype were injected AAVr-Pgk-Cre in the CeA and AAV-hSyn-DIO-hM3Dq-mCherry (orange n = 11) in the BLA to activate BLA-to-CeA cells (B C) CNO intra-peritoneal (i.p.) injection in the hM3Dq mice had no effect on global emotionality score (t(20) = 0.49 D Representative mouse tracks in the olfactory preference test E Chemogenetic activation of BLA-to-CeA cells did not modify olfactory valence compared to mCherry controls (Group: F(1,18) = 0.10 F Representative images of BLA c-Fos expression mCherry (left) and hM3Dq mice (right) G Percentage of c-Fos expression among mCherry+ cells after CNO injection (mCherry n = 5 Total number of mCherry+ cell number analysed dot not diffet between groups (t(8) = 0.37 H AAVr-Pgk-Cre in the NAc and AAV-hSyn-DIO-hM3Dq-mCherry (green n = 10 or AAV-hSyn-DIO-mCherry for the controls n = 12) in the BLA were injected to activate BLA-to-NAc cells J CNO intra-peritoneal (i.p.) injection in the hM3Dq mice reduce the global emotionality score (t(20) = 3.40) K Representative mouse tracks in the olfactory preference test L CORT-hM3Dq mice explored more peanut oil and ♀ urine relative to CORT-mCherry controls but not aversive odors (Group: F(1.20) = 6.56 M Representative images and magnifications of BLA c-Fos expression CORT-mCherry (left) and CORT-hM3Dq mice (right) N Percentage of c-Fos expression among mCherry+ cells after CNO injection (CORT-mCherry n = 12 Total number of mCherry+ cell number analysed dot not diffet between groups (U = 36 BLA-to-NAc neuronal activation is sufficient to improve negative olfactory bias and potentiates positive valence assignment of neutral or appetitive odors in a mouse model for depression Depressive state in both BD patients and CORT-induced model for depression induce a negative olfactory bias (right) respect to healthy state (left) Patients present decreased number of odors rated as pleasant associated with increased number of rated unpleasant odors respect to control subject while CORT-treated mice exhibit decreased approach behavior towards appetitive odors and increased avoidance behaviour towards aversive odors respect to control group The chronic CORT administration elevates the activity of BLA projecting neurons to the CeA which preferentially encode negative valence and reduces the activity of BLA projecting neurons to the NAc We found that chronic CORT treatment induces negative shift of valence assignment on both positive (that became less attractive) and negative (that became more aversive) odors without effect on neutral odor We observed similar bias in bipolar depression such that patients rated less odors as pleasant and more odors as unpleasant than both control subjects and remitted patients Although the low number of control subjects in our study could be a limitation the fact that this negative bias correlated with depression severity in bipolar patients clearly links mood state with biased valence attribution no difference was found in the valence assignment of neutral odors suggesting bias specific towards salient emotion-triggering stimuli this emotional bias regresses after recovery of the depressive episode in patients (i.e in remitted bipolar patients) and the alleviation of depressive-like phenotype in mice responsive to FLX These results suggest a major link between depressive states efficiency of antidepressant drug and the expression of negative emotional bias we only analyzed spontaneous responses to individual highly concentrated odors the increased response to aversive odor in CORT-treated animals reinforces the assumption that mice effectively detected these stimuli showing that the negative emotional bias affects both positive and negative odors supporting the idea that it represents common hallmark of depressive state independently of the type of mood disorder To avoid as much as possible any cognitive and motivational confounding factor we evaluated in mice innate attractive and aversive odor responses such that no previous learning is required in a more suitable way for an animal model of depression To investigate the mechanisms underlying negative bias we explored circuits involved in valence assignment We found that BLA-to-NAc neurons expressed less c-Fos in CORT-treated animals compared to the Veh-treated control group the density of c-Fos+ BLA-to-CeA neurons was higher in depressed-like mice Some differences were found in BLA vs LA projecting neurons suggesting both regions could be differently affected in depression These changes could be due to alterations in the intrinsic electrophysiological properties of different BLA principal neurons and/or modification at the circuit levels Although our experiments do not allow to untangle the precise mechanism underlying these disruptions these modifications would promote information flow from BLA to CeA while reducing the output to NAc in depressive states Further experiments are required to decipher the time course of these activity dysregulations during depressive-like phenotype instatement and how are they restore by antidepressant treatments in our experiment we cannot decipher if the altered presynaptic connectivity of BLA circuits reported in CORT-treated mice relies on structural and/or functional changes Further experiments are required to disentangle these alternatives We observed increased activity of BLA-to-CeA neurons in CORT-treated mice suggesting that depressive-like phenotype and negative bias could be mimicked by over-stimulation of this circuit chemogenetically stimulating the BLA-to-CeA pathway in control animals induced neither anxiety- or depressive-like phenotypes These data confirm that the anxiety/depressive-like phenotype is highly related to emotional bias activating the BLA-to-NAc circuit reduced the anxiety/depressive-like phenotype in CORT-treated mice and increased attractiveness of both neutral and positive odors BLA-to-NAc neurons are primarily suggested to be involved in reward processing and positive valence assignment activating the BLA-to-NAc neurons did not modify the aversion towards negative odors indicating that other neuronal pathways could be required to restore the bias regarding aversive odors Negative stimuli predicting danger might rely on multiple evolutionally selected and redundant mechanisms more difficult to hijack the restricted activation obtain in BLA neurons might be consequence of circuit-dependent regulations of their neuronal activity BLA-to-CeA neurons in control mice as well as BLA-to-NAc in CORT-treated mice might be submitted to over-inhibition at baseline Our gain-of-function approach reveals of the role of BLA circuits on valence assignment bias and depressive symptoms in particular BLA-to-CeA neurons in CORT-treated mice might be essential to investigate their involvement in these phenotypes our results suggest BLA-to-NAc neurons might represent a common pathway for antidepressant action and a potential therapeutic target for future drug development including both positive and negative emotional evaluation and the development of integrative cross-species models will be crucial to disentangle the psychopathology of depression and ultimately contribute to alleviate suffering patients The datasets and codes that support the findings of this study are available from the corresponding author upon reasonable request and national burden of 12 mental disorders in 204 countries and territories 1990–2019: a systematic analysis for the Global Burden of Disease Study 2019 Emotional information processing in mood disorders: a review of behavioral and neuroimaging findings Neurocognitive Mechanisms in Depression: Implications for Treatment Effect of Acute Antidepressant Administration on Negative Affective Bias in Depressed Patients An emotional-response model of bipolar disorders integrating recent findings on amygdala circuits A Unified Model of Depression: Integrating Clinical Modelling cognitive affective biases in major depressive disorder using rodents Further validation of the affective bias test for predicting antidepressant and pro-depressant risk: effects of pharmacological and social manipulations in male and female rats [no A neural model of voluntary and automatic emotion regulation: implications for understanding the pathophysiology and neurodevelopment of bipolar disorder Emotional valence modulates brain functional abnormalities in depression: Evidence from a meta-analysis of fMRI studies Functional connectivity of the amygdala and the antidepressant and antisuicidal effects of repeated ketamine infusions in major depressive disorder Unbalanced amygdala communication in major depressive disorder Association between depression severity and amygdala reactivity during sad face viewing in depressed preschoolers: an fMRI study Short-term SSRI treatment normalises amygdala hyperactivity in depressed patients The Amygdala and Depression: A Sober Reconsideration Extending the amygdala in theories of threat processing Divergent Routing of Positive and Negative Information from the Amygdala during Memory Retrieval Shen C-J, Zheng D, Li K-X, Yang J-M, Pan H-Q, Yu X-D, et al. Cannabinoid CB1 receptors in the amygdalar cholecystokinin glutamatergic afferents to nucleus accumbens modulate depressive-like behavior. Nat Med. 2019 https://doi.org/10.1038/s41591-018-0299-9 Reward and aversion processing by input-defined parallel nucleus accumbens circuits in mice [no American Psychiatric Association (2000): DSM-IV-TR: Diagnostic and Statistical Manual of Mental Disorders Gustatory and olfactory function in patients with unipolar and bipolar depression Prog Neuro-Psychopharmacol Biol Psychiatry Neurogenesis-Dependent and -Independent Effects of Fluoxetine in an Animal Model of Anxiety/Depression Innate Predator Odor Aversion Driven by Parallel Olfactory Subsystems that Converge in the Ventromedial Hypothalamus Normative data for the “Sniffin’ Sticks” including tests of odor identification and olfactory thresholds: an upgrade based on a group of more than 3000 subjects How to assess olfactory performance with the Sniffin’Sticks test® Anxiety- and Depression-Like States Lead to Pronounced Olfactory Deficits and Impaired Adult Neurogenesis in Mice Differential Peripheral Proteomic Biosignature of Fluoxetine Response in a Mouse Model of Anxiety/Depression Definitions of antidepressant treatment response and other relevant outcomes: a focus on treatment-resistant depression Adult-born neurons boost odor-reward association Multidimensional processing in the amygdala The Amygdaloid Complex: Anatomy and Physiology (n.d.): https://doi.org/10.1152/physrev.00002.2003 Amygdala Circuit Substrates for Stress Adaptation and Adversity Brain‐wide mapping of presynaptic inputs to basolateral amygdala neurons Connectivity characterization of the mouse basolateral amygdalar complex anatomically and functionally distinct cell types segregate across anteroposterior axis of paraventricular thalamus Placing the paraventricular nucleus of the thalamus within the brain circuits that control behavior The olfactory deficits of depressed patients are restored after remission with venlafaxine treatment The Association Between Olfaction and Depression: A Systematic Review Olfactory markers for depression: Differences between bipolar and unipolar patients Olfactory anhedonia and negative olfactory alliesthesia in depressed patients State and Trait Olfactory Markers of Major Depression Toward a Valid Animal Model of Bipolar Disorder: How the Research Domain Criteria Help Bridge the Clinical-Basic Science Divide [no Assessing positive and negative valence systems to refine animal models of bipolar disorders: the example of GBR 12909-induced manic phenotype Separate neural networks of implicit emotional processing between pictures and words: A coordinate-based meta-analysis of brain imaging studies The amygdala: historical and functional analysis Behavioral and Neurobiological Convergence of Odor Influence of chronic stress on network states governing valence processing: Potential relevance to the risk for psychiatric illnesses neurons mediate anxiogenic effects of the glucocorticoid receptor Central amygdala glucocorticoid receptor action promotes fear-associated CRH activation and conditioning Revisiting the Stress Concept: Implications for Affective Disorders Corticosterone can act at the posterior paraventricular thalamus to inhibit hypothalamic-pituitary-adrenal activity in animals that habituate to repeated stress Functional Connectivity of the Human Paraventricular Thalamic Nucleus: Insights From High Field Functional MRI Decreased plasma neuropeptides in first-episode schizophrenia major depressive disorder: associations with clinical symptoms and cognitive function Functional neuroanatomy of the basolateral amygdala: Neurons Organization of Valence-Encoding and Projection-Defined Neurons in the Basolateral Amygdala Acute Food Deprivation Rapidly Modifies Valence-Coding Microcircuits in the Amygdala Holographic stimulation of opposing amygdala ensembles bidirectionally modulates valence-specific behavior via mutual inhibition The basolateral amygdala γ-aminobutyric acidergic system in health and disease: BLA GABAergic System in Health and Disease A Student’s Guide to Neural Circuit Tracing Can transsynaptic viral strategies be used to reveal functional aspects of neural circuitry Viruses in connectomics: Viral transneuronal tracers and genetically modified recombinants as neuroscience research tools Clozapine N-Oxide Administration Produces Behavioral Effects in Long-Evans Rats: Implications for Designing DREADD Experiments Considerations for the Use of DREADDs in Behavioral Neuroscience New perspectives for refining future treatment approaches Activating positive memory engrams suppresses depression-like behaviour Activation of Basolateral Amygdala to Nucleus Accumbens Projection Neurons Attenuates Chronic Corticosterone-Induced Behavioral Deficits in Male Mice Depression researchers rethink popular mouse swim tests Download references We thank Denis David for advices on the corticosterone-induced model for depression the Institut Pasteur Central animal facility and especially Noémi Dominique for her assistance with behavioral experiments Corentin Guérinot for his advices in the PCA analyses and the Institut Pasteur Image analysis hub for help in developing the Icy protocol K Sailor for critical reading of the manuscript and all Perception and Action lab members their constructive inputs This work was supported by Agence Nationale de la Recherche Grants ANR-15-NEUC-0004-02 “Circuit-OPL” Laboratory for Excellence Programme “Revive” Grant ANR-10-LABX-73 the Life Insurance Company AG2R-La Mondiale These authors contributed equally: Chantal Henry Centre National de la Recherche Scientifique Funding acquisition; CH: Conceptualization Download citation DOI: https://doi.org/10.1038/s41398-024-03085-6 Metrics details Emotion classification using electroencephalographic (EEG) data is a challenging task in the field of Artificial Intelligence While many researchers have focused on finding the best model or feature extraction technique to achieve optimal results few have attempted to select the best methodological steps for working with the dataset we applied two different theoretical approaches based on the noise of the dataset: curriculum learning and confident learning Curriculum learning involves presenting training examples to the model in a specific order starting with easier examples and gradually increasing in difficulty This approach has been shown to improve model performance Confident learning is a method for identifying and correcting label errors in datasets By identifying and correcting these errors confident learning can improve the performance of machine learning models trained on noisy datasets We then applied the Integrated Gradient technique in order to assess the explainability of each model Our aim was to explore the impact of different models and methods on emotion classification performance using EEG data We collected and used an EEG dataset in which participants rated the emotional valence of positive and negative pictures while performing an emotion regulation (ER) task comparing a control condition (Look) with two ER strategies: cognitive reappraisal and expressive suppression We performed a multilabel classification to identify emotional neutrality or polarization of emotional valence (both positive and negative) rated by participants and the emotion regulation strategy adopted during the task We compared the performance of models trained on three datasets selected based on label noise and evaluated their suitability for this task Our results suggest different patterns based on the architecture used for feature importance highlighting both advantages and criticisms a tool that can translate emotions into feedback that is understandable by therapists or parents could be extremely helpful is represented by the struggle to have good results with machine or deep learning techniques researchers reached an accuracy above 70% for the emotional valence Another promising approach that has been scarcely used in emotion classification could be using a CNN + GRU (Gated Recurrent Unit) network to combine the advantages of both methods using the raw signal produces less accurate results but this method could be more useful in this context in which signal preprocessing is not possible we aim to employ emotion recognition within a unique context where participants were trying to modify their emotions this is the first study exploring the field of emotion regulation we used an EEG emotion regulation (ER) task to classify the perceived emotional valence of emotional pictures observed by participants and the ER strategy they adopted We asked participants to assess the emotional valence of 60 images participants were instructed to adopt one out of two possible ER strategies (and a control condition) We used a novel approach based on identifying the optimal methodology to classify EEG data after minimal preprocessing We included this manipulation in our study for two reasons this represents the first effort in the literature to employ an AI model for classifying ER strategies ER strategies are employed by individuals on a daily basis a tool that can discriminate these strategies may be exceedingly valuable in research and clinical settings it could facilitate investigations about how and when individuals tend to regulate their emotions or about comprehending typical and atypical coping mechanisms emotion classification is still a challenging task for machine learning models and it could become even more difficult considering the attempt to classify the ER strategies as well we believe that the most appropriate strategy for these tasks is focusing on the data quality especially considering the intrinsic noise of EEG data already discussed above Curriculum learning can help models focus on simpler concepts first and gradually progress to more complex ones without being overwhelmed by noise or ambiguity we suggest that even the categorisation of emotional valence clusters (e.g neutral and negative) should be based on a data-driven approach a promising way to approach this problem is to study not only the EEG variation but also the noisiness of the data and the label assigned we believe that confident learning can be very beneficial for our work the first aim of our study is to use a data-centric approach to feed the proposed models the best possible data based on input EEG data Our hypothesis is that adopting those two novel approaches (namely data-centric approach and XAI) for emotion recognition while employing different emotion regulation strategies could lead to less biased networks and to understand how the predictions are made Curriculum learning could help models focus on simpler concepts first and gradually progress to more complex ones Confident learning could help models avoid learning from unreliable or inconsistent labels which can degrade their accuracy and generalization IG could enhance the understanding of how the predictions are made by different models we compared four different architectures: CNN we propose a novel approach that combines curriculum learning confident learning and IG for emotion classification using different models while adopting two ER strategies expressive suppression and cognitive reappraisal and a neutral strategy that implies just looking at the stimuli Average and standard deviation of emotional valence ratings in the different ER strategies in red those that were considered positive The quality of data labels was evaluated using Random Forest (RF) as a benchmark model for the confident learning approach Separate RF models were trained for emotional valence and ER strategy as target classes All results reported refer to the dataset after applying the SMOTE algorithm When classifying participants’ ratings (from 1 to 7) in seven classes, the model performed with an F1-score of 0.15 (random chance = 0.14). (see Table 2) The labels that were best predicted were 4 and 5 followed by label 3 with an F1-score of 0.15 The labels that were least accurately predicted were the most positive ones In order to simplify the classification problem and to balance the sample of each class, we then classified the emotional valence dividing it into three classes, i.e., negative (i.e., labels 1–2), neutral (3–4-5) and positive (6–7). The results reported in Table 3 of the random forest on the validation set show that the class with a higher F1-score was the neutral one with an F1-score of 0.64 and a general accuracy of 0.48 This may be the result of the unbalanced dataset having a very represented neutral class and two other classes with fewer examples Given the unbalanced dataset, we then tried to classify data into two classes by merging positive (i.e., 6 and 7) and negative samples (i.e., 1 and 2) in a “polarized” class. Table 4 reports the results of a binary classifier that showed a more balanced result having an F1-score of 0.39 for the polarised class and 0.64 for the neutral one having respectively 0.40 and 0.63 for the polarised and neutral classes For what concerns the ER strategy classification, results can be observed in Table 5 The precision and F1-score were also similar for all classes indicating that the model had a balanced performance across the classes The model was slightly better at predicting the class Reappraise The model was slightly worse at predicting the class Suppress The N.examples column shows that the validation set was balanced we began our analysis by training deep neural networks with different architectures using a curriculum learning approach we divided the dataset into three parts based on label noise computed with Cleanlab and for each deep learning architecture we created three deep learning models We then used an ensemble learning approach to combine the predictions of these three models to obtain the final result Table 6 shows the classification report for the MINI-VGG (Visual Geometry Group) a convolutive neural network with a shallow architecture When classifying valence and ER strategy the model obtained a weighted average F1-score of 0.42 on the test set The model shows better performance in classifying emotional valence with the highest F1-score for the Neutral class at 0.55 followed by the Polarised class with an F1-score of 0.51 The model performs poorly in recognizing ER strategies while Reappraise and Suppress classes showed an F1-score of 0.27 and 0.29 respectively Table 7 shows the results of the MINI-VGG with GRU layers The model showed a weighted average F1-score of 0.43 the highest F1-score is for the Neutral class at 0.57 while the Polarised class showed an F1-score of 0.41 The ER classes totalized an F1-score of 0.31 Table 8 shows results of the MINI-VGG with bidirectional GRU layers the best performance concerned the Polarised class The ER strategies performed better than every other model tested Table 9 shows the results of the Transformer architecture The model showed a weighted average F1-score of 0.28 the best performance regards the Polarised class Each of the ER classes showed an F1-score of 0.02 Graphic representation of IG attribution score for each model tested Here we report as an example the condition when each model correctly classified the Neutral class (True Positive) The attribution value represents the relation intensity between each feature (time x channel) and the target class The colour scale ranges from dark blue to yellow representing positive and negative attribution scores (feature importance) respectively All figures for other classes and classifications (i.e. False Negative) are reported in the supplementary materials To test the observed differences in the IG data on a quantitative basis, we employed four random-intercept linear mixed-effect models (LMMs). The results of these analyses are summarized in Table 10 and the data are represented in Fig. 3. Graphic representation of results of the statistical comparisons of the IG attribution score for each model tested Time bins of 50 ms are represented on the x-axis while the mean of the IG absolute value is represented on the y-axis The three colored lines represent three channel clusters (i.e. Error bars represent 95% confidence intervals post-hoc comparisons performed on the Cluster main effect showed that the central cluster presented significantly higher IG scores than the posterior cluster (z = 8.18 which in turn presented greater values than the anterior cluster (z = 2.7 independently of different channel clusters 150-200 ms was the time bin with the highest IG value as it was significantly higher than 0-50 ms The 450-500 ms time bin also showed greater IG values than the 50-100 ms time bin (z = 3.9 The unidirectional GRU showed the greatest importance of the central cluster again as it showed higher values than both the anterior (z = 4.7 all ps < 0.001) and 450-500 ms time bins (all zs > 220 all ps < 0.001) presented significantly higher values than any other time bins The post-hoc comparisons performed on the interaction effect showed that differences between clusters were present only in the 450-500 ms time bin with significantly higher values in the central cluster than in the posterior one (z = 8.2 which in turn was higher than the anterior one (z = 6.2 The bidirectional GRU showed the same pattern across clusters as the unidirectional GRU with the central cluster showing higher values than both the anterior (z = 5.9 The time main effect showed that 0-50 ms (all zs > 155 when considered separately for each channel cluster all ps < 0.001) and 450-500 ms time bins (all zs > 97 all ps < 0.001) showed significantly higher values than any other time bins The comparisons based on the cluster showed that the central cluster presented greater IG values than both the posterior (z = 7.8 p < 0.001) and anterior clusters (z = 5.9 while these latter two did not differ significantly between each other (z = 1.9 the central cluster presented greater IG values than the posterior one (z = 9.0 which in turn was significantly greater than the anterior one (z = 3.4 The only statistically significant effect in the Transformer model was the Cluster main effect This showed that the central cluster presented greater IG values than the anterior (z = 32.4 p < 0.001) and posterior clusters (z = 14.0 the latest being also significantly higher than the anterior one (z = 18.4 we selected the proper labels based on the confident learning approach thus resulting in two classes for emotional valence classification (i.e. polarised) and three classes for ER strategy (i.e. we used an ensemble approach with three models having the same architecture giving the three different models different training based on the dataset’s difficulties which is a Convolutional Neural Network (CNN) designed to process data -in this case EEG data- by applying multiple convolutional layers to extract spatial features The MINI-VGG with GRU model combines the MINI-VGG with a Gated Recurrent Unit (GRU) which is a type of Recurrent Neural Network (RNN) that processes sequential data by maintaining hidden states to capture temporal dependencies The MINI-VGG with bidirectional GRU model further enhances the GRU by using bidirectional GRUs which process the input sequence in both forward and backward directions to capture context from both past and future time steps The Transformer model utilizes self-attention mechanisms to handle sequential data by allowing each position in the sequence to attend to all other positions effectively capturing long-range dependencies without relying on recurrent structures the best classes’ performance was neutral and polarised the models still cannot predict those strategies properly Our intuition was that the neurons of the long-term memory recurrent neural network have the benefit of memorizing both the long-term and short-term emotional information present in the EEG signal Considering the architectures’ interpretability studying the attributions of the gradients of each model) we realised that RNN tends to memorize shorter sequences and that an array with 64 × 500 time points is too long to be learned In the statistical comparisons it can be seen that unidirectional GRU tend to consider only the last 100 ms of the input while the bidirectional GRU only the first and last 100 ms Further studies combining Convolutional and Recurrent architectures should therefore test shorter time series compared to our dataset (e.g. the pattern of the importance of the feature is scattered and it makes the interpretation very difficult It looks like there are no specific channels or time points to influence the outcome of the classification the Transformer is the model showing the most interpretable patterns This model showed patterns of specific channels that can be positively or negatively correlated with the prediction of the model over time thus showing that this architecture can learn from the whole epoch It is also interesting to notice that the Transformer model has the best score for the emotional valence domain and the worst in the ER strategy one these findings provide information about the validity of the models used It appears that they focused on the electrodes showing the activity of areas most commonly involved in emotional experience which are widely used in the literature to evaluate deep learning models do not reveal how these black box models make decisions we believe that it is always appropriate to adopt an XAI technique to turn models into explainable white boxes rather than solely discussing their performance This approach is valuable not only in choosing the best model for a classification problem but also in refining the architecture accordingly we observed a temporal limit in the time series memorization only thanks to the XAI technique data-centric approach and XAI) in our methodological framework could be useful to researchers and practitioners to study a specific dataset in other fields and build a personalised model that could be used for a real-time BCI it should be considered that there are many reasons for this result despite the strict methodological and theoretical background of our work which are intrinsic to their perceptual representations splitting the data based on subjects ensures that the training and test sets are independent and reduces the risk of overfitting Randomly selecting epochs from all participants implies that data from the same subject could appear in both the training and test sets as the model may learn to recognize specific characteristics of individual subjects rather than generalizing to new data splitting the data based on subjects allows us to evaluate the generalizability of our model to new subjects By training our model on data from one set of subjects and testing it on data from a different set of subjects we can assess how well our model can classify emotional valence in new individuals the main limitation of this work is that our models do not outperform the current state-of-the-art but we have adopted a challenging and specific dataset (EEG raw data) without any real benchmark in model performance Since each example is potentially affected by artifacts it is reasonable to obtain lower performance compared to a pre-processed EEG dataset we obtained fairly good performance in emotional valence classification and this research line is potentially more useful in creating models classifying real-time data with minimal preprocessing (i.e. Future directions could consist of applying and comparing the effectiveness of this data-centric and XAI approach to different classification problems using EEG raw data this work could be a seminal step in finding solutions when dealing with large artifacts such as movement It must also be considered that a model’s performance can surpass the state-of-the-art literature if there is no way to understand why or what it is classifying This is especially true in fields like healthcare representing a possible application for BCIs where decisions have significant consequences Interpretable models allow us to gain insights into the model’s decision-making process and ensure that the model aligns with our expectations and requirements this work represents an important advance in the field of emotion classification using EEG allowing a new methodological approach to disentangle the common issues of label noise and overestimated performance We collected data from thirty paid participants (17 females Our exclusion criteria comprised the presence of a history of any neurological or psychiatric disease abuse of any drugs (including nicotine within 2 h preceding the study and alcohol within 24 h preceding the study) as well as being informed about the aim of the study Written informed consent was obtained from all participants according to the declaration of Helsinki; the IMT Ethical Committee approved the project We selected sixty stimuli from the Oasis database65 Every image was rated via a Likert scale from 1 to 7 according to their valence and arousal Our aim was to have a small dataset with positive emotional valence for half of the stimuli and negative for the other half We selected images that respected three criteria: (i) choosing images with no sensitive contents (i.e. sexual or violent scenes) based on the indications of the Ethical Committee; (ii) selecting images with high arousal values (> 4) to clearly elicit emotions; (iii) including images with the highest or lowest validation valence values to use strongly polarized positive and negative stimuli 30 positive images with mean ± sd valence and arousal respectively of 5.7 ± 1.06 and 4.7 ± 1.6 and 30 negative images with mean ± sd valence and arousal values of 2.1 ± 1.1 and 4.7 ± 1.8 The mean valence of each stimulus assessed by our subjects during the experiment is in red compared with the mean valence rated in the Oasis dataset participants were asked to sit on a comfortable chair in an electrically shielded and soundproof room while not crossing their legs or arms A 19″ monitor was positioned in front of them The EEG cap was then prepared on the participant’s head Resting state activity was then recorded for five minutes Experimenters explained to the subjects the experimental procedure before the experiment started written instructions were shown on the monitor to give them the information they needed The experimental design is shown in Fig. 5. The experiment consisted of three blocks in which participants were asked to assess the emotional valence of 60 pictures shown on the monitor while adopting three different emotion regulation strategies: ‘Look’, ‘Reappraise’ and ‘Suppress’, for a total of 180 stimuli seen during the experiment. Experimental Procedure: After being instructed about the ER strategies and their meaning participants were asked to adopt two possible ER strategies (i.e. expressive suppression or cognitive reappraisal) “Look”) in three separate counterbalanced blocks a fixation cross was displayed on the monitor for 1000 ms followed by either a positive or a negative stimulus for 1000 ms Participants then rated the emotional valence of the stimulus on a Likert scale from 1 (absolutely negative) to 7 (absolutely positive) with their mouse Each mindset was explained by a short training by experimenters and with written instructions at the beginning of each of the three blocks which were shown in a counterbalanced order Participants were free to ask the experimenters information about the different ER strategies at any moment via a microphone Once participants understood what mindset they had to adopt they could press a key to begin the visualization and rating of the stimuli The mindset was also reminded during the whole experimental block by showing a small text (i.e. ‘REAPPRAISE’ and ‘SUPPRESS’) above the pictures was preceded by a fixation cross of 1000 ms participants were asked to rate the emotional valence of the picture according to the mindset they adopted in that block on a Likert scale ranging from 1 (absolutely negative) to 7 (absolutely positive) The dataset generated during the current study is available in the Emotion Regulation Task repository, https://osf.io/yv468/ The neurophysiological activity of each participant was recorded with a 64-channel EGI EEG system Electrodes were positioned according to the 10–20 International System The online reference was Cz and the sampling frequency was 1000 Hz with an impedance below 50KΩ All the pre-processing was performed using the MNE-Python package The raw signal was filtered (1–48 Hz bandpass with a 50 Hz notch filter). We used the RANSAC (random sample consensus) method66 to detect bad channels The signal was then epoched: each epoch began 500 ms before the stimulus onset lasting 1000 ms We then normalized the epoched signal considering a baseline from 500 to 200 ms before the stimulus onset and cropped the signal before the appearance of the stimulus We set the consensus from 0.2 to 0.5 in steps of 0.1 The consensus corresponds to the proportion of bad channels that are allowed in order to accept the epoch If the number of bad channels allowed exceeds the real number of bad channels it can interpolate the signal from 1 to 8 bad electrodes The algorithm chooses the best value possible for each subject The average number of rejected epochs was 9.16 ± 9.97 per participant Then the signal was re-referenced offline to the average of all channels we included 64 channels and a time window of 500 ms In order to adopt a data-centric approach with confident and curriculum Learning A random forest (RF) is an ensemble learning method that consists of a collection of decision trees: it is called a “forest” because it is made up of many decision trees78 the random forest could split the data for each electrode or time point The model incorporates randomness by randomly choosing the splitter and each tree from various random subsamples of data Every node in the tree indicates a split in the data and the leaf node is where the ultimate prediction is formed To create different datasets based on both the difficulty of the samples (curriculum learning) and the correct classification classes (confident learning) we trained two kinds of random forests (RFs) One type of RF was used to classify the emotional valence of the images and the other to classify the ER strategies We used three criteria to prepare our dataset in the most methodologically sound way: (i) ensuring that the dataset was as balanced as possible (ii) making theoretical assumptions about both emotional valence and emotion regulation strategy and (iii) simplifying the classification problem we trained the first RF for emotional valence with 7 classes corresponding to the 7-point Likert scale used for assessment and negative) and the third RF had just two classes (neutral and polarised) The other type of RF was used to classify the ER strategy of the participants into 3 classes corresponding to the 3 emotion regulation strategies We then calculated the class overlap for each model to determine the best labels for our dataset for multilabel classification of both emotional valence and emotion regulation We first selected the best label match for emotional valence and ER strategy based on the above-mentioned criteria we decided to use a binary classification for emotional valence (polarized and neutral) while maintaining the three original classes for emotion regulation strategy we measured the label quality for each example and averaged it with emotional valence and mindset We divided the dataset into three parts based on the 33rd and 66th percentiles of the distribution: easy we used four different ensemble deep learning architectures for our multilabel classification with five classes (Neutral (3) MINI-VGG with bidirectional GRU and (4) A Transformer model Multilabel classification involves N non-exclusive labels (in our case Each label is considered as a binary classification problem whose predicted probability is independent with respect to the other classes This multilabel classification approach allows us to avoid using three separate models (one for each ER strategy) and training our data on emotional valence on a dataset split into three parts it was necessary for two labels to be present simultaneously (i.e. one label for emotional valence and one for ER strategy) because easy data are more easily learned by the model by training three models on datasets of increasing difficulty we aimed to provide the ensemble models with a broader view of EEG signals the prediction with the highest number of votes was selected thus a prediction was confirmed if at least two out of the three classifiers voted for it We used a Sigmoid function in the last layer of each model thus obtaining in output a prediction for each label independently we set a threshold of 0.5 to classify the presence (i.e. probability > 0.5) or absence of the variable The MINI-VGG we used consisted of two 1D convolutional blocks followed by a classification block The convolutional blocks were composed of two convolutional layers the convolutional layers used 32 filters with a kernel size of 3 64 filters were used with a kernel size of 3 At the end of each block a pooling layer performed max pooling operation over a window of 2 There was always a batch normalization after the convolutional one After each pooling a dropout layer was used the output was passed through a Dense layer with sigmoid activation function to produce the final predictions With regards to the MINI-VGG + GRU and MINI-VGG + bidirectional GRU we refer to the modification of the MINI-VGG by including three GRU (unidirectional or bidirectional) layers with 128 units and a tanh activation function after the second convolutional block The output of the GRU blocks was then passed through a Batch Normalization layer and another MCDropout with a dropout percentage of 0.15 the model consisted of two encoders and a classification block Each encoder block applies layer normalization and multi-head attention with 64 head size and 2 heads to the inputs followed by a dropout with a rate of 0.4 and a normalization two convolutional layers with 64 filters with a kernel with size of 1 are applied Only the first convolutional layer used a ReLU activation function a dropout with a dropout percentage of 0.15 was applied The output of the final transformer block is then passed through a global average pooling layer and a series of dense layers with 128 units The final output is produced by a dense layer with 5 units and a sigmoid activation function For every model, we used Adam optimizer80 with a learning rate of 1e-06 For a graphical representation of each architecture employed We trained the model with a default of 20,000 epochs, but then we also used the early stopping callback from Keras81 (with patience of 20 epochs) taking as a benchmark the loss functions of the validation set The main component of the CNN are the convolutional layers The convolutional layer applies each filter to the input data by sliding it over the input data and performing an element-wise multiplication with the values at each position It then sums the results and stores the result in a new feature map The output of the convolutional layer is a multi-dimensional array called a feature map which has the same number of dimensions as the input data (e.g. and depth) but the size of the feature map is usually smaller than the input data because the filters are smaller and do not overlap The convolutional layer may also have parameters called stride and padding which control how filters are applied to the input data Stride controls the step size with which the filters are applied and padding controls how the input data is padded with additional values around the border before the filters are applied One of the main issues with convolutional layers is that the feature map generated by the filter is dependent on its location convolutional neural networks learn to associate the presence of specific features with their location in the input features GRUs can be unidirectional or bidirectional: a unidirectional GRU processes the input sequence in one direction while a bidirectional GRU processes the input sequence in both forward and backward directions This allows the model to capture both past and future context when making predictions Unlike recurrent neural networks (RNNs) and LSTMs Transformer networks do not have an inherent way to capture the relative positions of the input elements positional encoding is used in conjunction with each input vector Positional encoding is not part of the model architecture itself but rather a pre-processing step A positional encoding vector is generated for each input element and added to its corresponding embedding vector This allows the model to learn spatial information from the “injected” pattern in the embedding vector integrated gradients define an attribution value for each feature by considering the integral of the gradients taken along a straight path from a baseline instance x’ to the input instance x the gradient usually refers to the output corresponding to the true class or to the class predicted by the model We used Alibi92 We used a null baseline and the number of steps we used was 25 We used the “explain” method of this instance to compute the attributions for our selected examples The attributions were then obtained from the “attributions” attribute of the returned explanation object one for the bidirectional GRU and one for the Transformer) Each model included the main effects of time and channels and their interaction as fixed factors When encountering a statistically significant fixed effect it was probed through post-hoc comparisons with Tukey’s correction for multiple comparisons where R2AB represents the coefficient of determination of the full model (i.e. with the effect of interest) and R2A represents the coefficient of determination of the null model (i.e. LMMs present two possible methods to compute R2 marginal R2 (without the contribution of random effects) and conditional R2 (with the contribution of random effects) two effect sizes were computed for each effect The dataset generated during the current study is available in the Emotion Regulation Task repository, https://osf.io/yv468/ The influence of visual and auditory stimuli 56 EEG-based detection of emotional valence towards a reproducible measurement of emotions Artifacts and noise removal for electroencephalogram (EEG): A literature review in 2018 IEEE Symposium on Computer Applications & Industrial Electronics (ISCAIE) 326–332 (IEEE Everything you wanted to ask about EEG but were afraid to get the right answer Brain-Computer Interfaces 1: Methods and Perspectives Feature extraction and selection for emotion recognition from EEG Toward emotion aware computing: an integrated approach using multichannel neurophysiological recordings and affective visual stimuli Emotion processing by ERP combined with development and plasticity Improving BCI-based emotion recognition by combining EEG feature selection and kernel classifiers Emotion recognition related to stock trading using machine learning algorithms with feature selection Predicting exact valence and arousal values from EEG Electroencephalogram emotion recognition based on empirical mode decomposition and optimal feature selection IEEE Transactions on Cognitive and Developmental Systems 11 Emotion detection from EEG recordings based on supervised and unsupervised dimension reduction Concurrency and Computation: Practice and Experience 30 Cross-subject EEG feature selection for emotion recognition using transfer recursive feature elimination ScalingNet: extracting features from raw EEG data for emotion recognition Understanding of a convolutional neural network in 2017 international conference on engineering and technology (ICET) 1–6 (Ieee Automatic emotion recognition using temporal multimodal deep learning Deep physiological affect network for the recognition of human emotions IEEE Transactions on Affective Computing 11 ICA-evolution based data augmentation with ensemble deep neural networks using time and frequency kernels for emotion recognition from EEG-data IEEE Transactions on Affective Computing 13 Human emotion recognition with electroencephalographic multidimensional features by hybrid deep neural networks Supervised sequence labelling with recurrent neural networks 37–45 (2012) Advances in neural information processing systems 30 Transformers:“the end of history” for natural language processing in Machine Learning and Knowledge Discovery in Databases Liu, M. et al. Gated transformer networks for multivariate time series classification. arXiv preprint arXiv:2103.14438 (2021) TC-Net: A Transformer Capsule Network for EEG-based emotion recognition A Step towards EEG-based brain computer interface for autism intervention in 2015 37th annual international conference of the IEEE engineering in medicine and biology society (EMBC) 3767–3770 (IEEE Misinterpretation of facial expressions of emotion in verbal adults with autism spectrum disorder Emotion regulation: A theme in search of definition Monographs of the society for research in child development 25–52 (1994) Emotion regulation in adulthood: Timing is everything Current directions in psychological science 10 Healthy and unhealthy emotion regulation: Personality processes Emotion regulation and decision making under risk and uncertainty Emotion regulation difficulties in trauma survivors: The role of trauma type and PTSD symptom severity Emotion regulation mediates the relationship between ADHD and depressive symptoms in youth Cognitive reappraisal and expressive suppression strategies role in the emotion regulation: an overview on their modulatory effects and neural correlates Frontiers in systems neuroscience 175 (2014) Short-circuiting of threat by experimentally altering cognitive appraisal The Journal of Abnormal and Social Psychology 69 Journal of personality and social psychology 64 Confident learning: Estimating uncertainty in dataset labels Journal of Artificial Intelligence Research 70 in Proceedings of the 26th annual international conference on machine learning 41–48 (2009) International Journal of Computer Vision 130 Curriculum learning for speech emotion recognition from crowdsourced labels in International conference on machine learning 3319–3328 (PMLR Effective deep learning approach based on VGG-mini architecture for iris recognition Annals of the Romanian Society for Cell Biology 4718–4726 (2021) Simonyan, K. & Zisserman, A. Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556 (2014) Deep learning-based electroencephalography analysis: a systematic review Anterior insular cortex and emotional awareness EEG microstate analysis of emotion regulation reveals no sequential processing of valence and emotional arousal Decoding naturalistic affective behaviour from spectro-spatial features in multiday human iEEG Functional imaging studies of emotion regulation: a synthetic review and evolving model of the cognitive control of emotion Annals of the New York Academy of Sciences 1251 The selective processing of briefly presented affective pictures: An ERP analysis How context influences the interpretation of facial expressions: a source localization high-density EEG study on the “Kuleshov effect” EEG-based Positive-Negative Emotion Classification Using Machine Learning Techniques in 2021 IEEE 14th International Symposium on Embedded Multicore/Many-core Systems-on-Chip (MCSoC) 135–139 (IEEE EEG-based classification of positive and negative affective states Review of machine learning techniques for EEG based brain computer interface Archives of Computational Methods in Engineering 1–20 (2022) Micro-valences: perceiving affective valence in everyday objects and the importance of data partitioning decisions for training and testing of EEG cross-participant models Analyzing Neural Time Series Data: Theory and Practice Diversity-Washing and Consent Circumvention Affective picture processing: an integrative review of ERP findings Autoreject: Automated artifact rejection for MEG and EEG data Chemometrics and intelligent laboratory systems 2 On relationships between the Pearson and the distance correlation coefficients SMOTE: synthetic minority over-sampling technique Journal of artificial intelligence research 16 On the relative value of data resampling approaches for software defect prediction The Tukey multiple comparison test: 1953–1976 A survey of decision tree classifier methodology Kingma, D. P. & Ba, J. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980 (2014) O’Shea, K. & Nash, R. An introduction to convolutional neural networks. arXiv preprint arXiv:1511.08458 (2015) Rethinking the inception architecture for computer vision in Proceedings of the IEEE conference on computer vision and pattern recognition 2818–2826 (2016) Bailer, C., Habtegebrial, T. & Stricker, D. Fast feature extraction with CNNs with pooling layers. arXiv preprint arXiv:1805.03096 (2018) Learning long-term dependencies with gradient descent is difficult A review of recurrent neural networks: LSTM cells and network architectures Cho, K. et al. Learning phrase representations using RNN encoder-decoder for statistical machine translation. arXiv preprint arXiv:1406.1078 (2014) Comparing LSTM and GRU models to predict the condition of a pulp paper press Siddhad, G., Gupta, A., Dogra, D. P. & Roy, P. P. Efficacy of transformer networks for classification of raw EEG data. arXiv preprint arXiv:2202.05170 (2022) Rojat, T. et al. Explainable artificial intelligence (xai) on timeseries data: A survey. arXiv preprint arXiv:2104.00950 (2021) Alibi explain: Algorithms for explaining machine learning models The Journal of Machine Learning Research 22 Deap: A database for emotion analysis; using physiological signals IEEE transactions on affective computing 3 Download references This work was financially supported by Intesa Sanpaolo Innovation Center S.p.A Linda Fiorini and Francesco Bossi contributed equally to this work Francesco Bossi & Francesco Di Gruttola Department of Psychology ‘Renzo Canestrari’ Download citation DOI: https://doi.org/10.1038/s41598-024-75263-x Metrics details The oxygen reduction reaction (ORR) catalyzed by efficient and economical catalysts is critical for sustainable energy devices Although the newly-emerging atomically dispersed platinum catalysts are highly attractive for maximizing atomic utilization their catalytic selectivity and durability are severely limited by the inflexible valence transformation between Pt and supports we present a structure by anchoring Pt atoms onto valence-adjustable CuOx/Cu hybrid nanoparticle supports (Pt1-CuOx/Cu) in which the high-valence Cu (+2) in CuOx combined with zero-valent Cu (0) serves as a wide-range valence electron reservoir (0‒2e) to dynamically adjust the Pt 5d valence states during the ORR In situ spectroscopic characterizations demonstrate that the dynamic evolution of the Pt 5d valence electron configurations could optimize the adsorption strength of *OOH intermediate and further accelerate the dissociation of O = O bonds for the four-electron ORR the Pt1-CuOx/Cu catalysts deliver superior ORR performance with a significantly enhanced four-electron selectivity of over 97% and long-term durability few investigations seek to develop a coherent design strategy to precisely tune the valence states of the loaded metal atoms on the supports under working conditions and a fundamental understanding of the critical mechanism that govern the performance of the catalytic sites is yet to be elucidated the MOx/M hybrids is promise to undergo valence evolution under reaction conditions which is crucial to realize a dynamic electron supply from supports to catalytic metal atoms taking advantage of the easily oxidizable property of small Cu particles we successfully designed a type of catalysts with Pt atomically anchored on the valence-flexible CuOx/Cu hybrid nanoparticles (Pt1-CuOx/Cu) via an in situ partial oxidation strategy The wide adjustable range of valence states (Cu0 ↔ Cu+2) renders CuOx/Cu an ideal VER for sustainable electron exchange with Pt 5d states By using in situ X-ray absorption fine structure (XAFS) and synchrotron fourier transform infrared (SR-FTIR) techniques we followed the catalytic process and unraveled that the 5d electron occupancy of single Pt sites are effectively increased by 0.53 units at the initial ORR stage based on the electron flow from Cu 3d to Pt 5d as the elevated oxidation state of CuOx/Cu supports the key intermediate *OOH rapidly dissociates to form *O species due to the optimized adsorption on the highly 5d electron-occupied Pt sites the Pt1-CuOx/C exhibits improved durability and electrocatalytic activity with a high half-wave potential of 0.92 V vs reversible hydrogen electrode (RHE) and a significantly enhanced selectivity of four-electron pathway (>97%) Our strategy contributes to the design of efficient electrocatalysts and addressing the catalytic stability by manipulating the valence evolutions of active sites and supports The Pt1-CuOx/Cu dispersed on carbon nanosheets was synthesized through sequential hydrothermal and pyrolytic processes a TEM image of Pt1-CuOx/Cu; inset is the corresponding size distribution; (b) HRTEM image of Pt1-CuOx/Cu; (c–e) AC-HAADF-STEM image (c) EDS elemental mapping (d) and line-scan analysis f XRD results of Pt1-CuOx/Cu and the control samples these above characterization results clearly demonstrate that the catalysts composed of the evenly dispersed Pt on hybrid CuOx/Cu nanoparticles were successfully obtained a XANES spectra at O K-edge of Pt1-CuOx/Cu and CN; (b) Cu K-edge k3-weighted FT-EXAFS spectra of Pt1-CuOx/Cu Cu foil and CuO; (c) Pt L3-edge k2-weighted FT-EXAFS spectra of Pt1-CuOx/Cu Pt foil and PtO2; Cu 2p (d) and Pt 4 f (e) XPS spectra of Pt1-CuOx/Cu; (f) XANES spectra at Pt L3-edge of different samples These results clearly suggest a modified valence state of Pt 5d in the Pt1-CuOx/Cu hybrid nanoparticles no iR correction was applied in the measurements b The comparison of kinetic current density (Jk) and the inset is the locally amplified polarization plots c Tafel slopes for Pt1-CuOx/Cu and the reference catalysts (The Tafel slopes were obtained at the scan rate of 10 mV s–1 without iR correction) d Turnover frequency and mass activity of Pt1-CuOx/Cu and the reference catalysts at 0.9 V vs e The selectivity parameters of ORR pathway for the catalysts involved The error bars were estimated by the standard deviations of three individual calculations In situ Pt L3-edge XANES spectra of Pt1-CuOx/Cu (a) and PtSA-CN (b) c The d-band electrons at different conditions of Pt1-CuOx/Cu and PtSA-CN d In situ XANES spectra at Cu K-edge of Pt1-CuOx/Cu e In situ Pt L3-edge EXAFS spectra of Pt1-CuOx/Cu the in situ Cu K-edge XANES spectra of CuOx/Cu show essentially no changes in the absorption edge at the initial reaction stage (before 0.80 V vs and the slightly positive shift during the reaction process may be related to the adsorption of oxygen species on the surface of CuOx/Cu The above results prove that the changes in the valence electron quantity for Pt and Cu in Pt1-CuOx/Cu are originated from the electron interaction between the Pt sites and CuOx/Cu supports at the initial reaction stage The increased oxidation state of Cu corresponds to a decrease in the valence electron quantity suggesting that the valence electron transformation is characterized by the CuOx/Cu support serving as a VER to rapidly increase the quantity of Pt 5d valence electrons at the initial reaction stage which is beneficial for the orbital hybridization with O 2p and increasing the adsorption strength of oxygen species at the Pt sites during the ORR process which can prevent the dissolution of the supports and leaching of active sites in the process of ORR the dynamic valence evolution during the catalytic reactions promotes the evolution of reactive oxygen species and maintains the structural integrity for boosting activity and durability of the catalysts In situ SR-FTIR spectra in the range of 700–1500 cm–1 of Pt1-CuOx/Cu (a) and PtSA-CN (b) c The corresponding infrared signal fluctuations In situ EIS measurements of Pt1-CuOx/Cu (d) and PtSA-CN (e) f The corresponding adsorption resistance (Rct) of oxygen species and it can be concluded that the valence electron transport from the VER (CuOx/Cu) to Pt active sites at the initial reaction stage (open circuit → 1.00 V vs RHE) promotes the proton coupling and electron transfer process of intermediate evolution thereby accelerating the kinetics of the four-electron ORR the Rct values of Pt1-CuOx/Cu are smaller than that of PtSA-CN in the whole reaction process and decrease rapidly at the kinetic control region suggesting significantly accelerated adsorption dynamics of oxygen-related intermediates at the electrode/electrolyte interfaces the higher Cd values at various ORR potentials of Pt1-CuOx/Cu further demonstrates that the optimum adsorption of oxygen species on valence-reduced Pt sites dominates a highly selective four-electron ORR pathway the above in situ SR-FTIR and EIS results reveal that the dynamic valence transformation between Pt and CuOx/Cu at the reaction stage facilitates the formation and conversion of oxygen intermediates a strategy of using valence-adjustable metal oxide/metal (CuOx/Cu) hybrid nanoparticles as an VER was proposed to tune the Pt 5d valence states for efficient ORR Benefiting from the flexible valence electron exchange between the dynamically valence-reconstructed CuOx/Cu supports and single Pt center the obtained Pt1-CuOx/Cu catalyst delivers improved four-electron selectivity and long-term stability of ORR Using combined in situ synchrotron radiation characterization technologies we uncovered that the increase of oxidation state in CuOx/Cu nanoparticles can effectively elevate the Pt 5d electron quantity by 0.53 units at the initial reaction stage which effectively tunes the adsorption strength of the *OOH intermediates thereby improving the selectivity of ORR Our work provides a universal wide-range valence-regulatory strategy for the development of advanced electrocatalysts All chemicals were used without further purifications The chloroplatinic acid hexahydrate (H2PtCl6·6H2O ≥85%) and urea (99%) were acquired from Sinopharm Reagent Co. and Nafion (5%) solution were acquired from Aladdin Biochemical Technology Co. Nitrated pyrene) dissolved in deionized water (55 mL) under ultrasonic conditions Then add 5 mL ammonia water and mix thoroughly After hydrothermal reaction at 200 °C for 10 h The amino-functionalized CN support was obtained The Pt1-CuOx/Cu nanoparticles supported on CN support were prepared by wet-impregnation of aqueous solution containing H2PtCl6·6H2O and Cu(acac)2 two metal salts of 100 μL H2PtCl6·6H2O (20 mg mL-1) and 15 mg Cu(acac)2 were added to the concentrated 20 mL CN support solution Ultrasonically dispersed for 30 minutes and then freeze-dried The power was ground with urea (mass ratio of 1:10) and heated in a tubular furnace (5 °C/min rate of heating the product was heated in muffle furnace at 200 °C for 60 min to obtain the final products of Pt1-CuOx/Cu The synthesis processes for Pt single-atoms and Cu nanoparticles supported on C-N (PtSA-CN CuNP-CN) catalyst were similar to that of the preparation of the Pt1-CuOx/Cu electrocatalyst The notable difference is the use of the H2PtCl6·6H2O for PtSA-CN and the Cu(acac)2 for CuNP-CN and CuOx/Cu control samples The CuNP-CN was heated in muffle furnace for 60 minutes to obtain the CuOx/Cu-CN samples And the Pt1-CuOx/Cu was heated in muffle furnace for 200 minutes to obtain the Pt-CuO samples The AC-HAADF-STEM was conducted on a JEM-ARM200F instrument at 200 kV TEM images were conducted on JEM−2100F at 200 kV The XRD analysis was performed on miniflex600 of multifunctional rotating-anode X-ray diffractometer The surface chemical state analysis by XPS were performed on thermo scientific ESCALAB 250Xi with Al kα 150 W ICP-MS analysis was employed at iCAP RQ of inductively coupled plasma-mass spectrometry The soft X-ray absorption spectra was collected at soft X-ray imaging line station (BL07W) in Hefei and the mass loading on the electrode is 20 μgPt cm-2 The electron transfer number (n) was calculated according to the equation below: the kinetics and the diffusion-limited current densities ω is the rotational speed of the RDE (rad·s−1) and the Faraday constant (F) is 96485 C·mol−1 which conducted as bulk concentration (1.2 × 10–6 mol·cm−3) and diffusion coefficient (1.9 × 10–5 cm2·s−1) of oxygen %) and the n based on RRDE measurements can be calculated according to the equation below: where the disk and ring current are marked as Id and Ir N is the Pt ring current collection efficiency The mass activity of the noble metal-based catalysts is usually measured by the amount of noble metal used A gmetal-1) was evaluated by the equation: the kinetic current density is marked as jk m is the noble metal loading mass on the electrode The turnover frequency (TOF) of the Pt1-CuOx/Cu and control samples are calculated by the formula of TOF = (jk × A × Ne)/(4 × Mmetal × NA/M) mA cm-2; A is the surface of the electrode cm-2; Ne is the electron number per coulomb 6.24 × 1018 C-1; Mmetal is the mass of active metal leading on the electrode mg; The Avogadro constant is 6.02 × 1023 mol-1; and M is the Pt molar mass Typically, the in situ SR-FTIR measurements were conducted at BL01B of the National Synchrotron Radiation Laboratory (NSRL, China). The reflection mode of the SR-FTIR light was adopted. The three-electrode system was applied in the homemade FTIR electrochemical cell with three electrodes in one chamber (~3 mL) (Supplementary Fig. 24a) The infrared spectra at different potential conditions were measured after a certain potential applied for 5 min to obtain the steady signal The catalyst ink was dissolved as follows: Sample (5 mg) was dissolved in a 1 mL solution (0.9 mL isopropyl alcohol The cleaned carbon cloth with area of 1 × 1.5 cm−2 was employed as the working electrode a certain amount of the catalyst ink was coated onto carbon cloths evenly the carbon cloth was pressed into the ZnSe window of the electrochemical cell The O2-saturated 0.1 M KOH electrolyte was circulated into the electrochemical cell through a peristaltic pump The in situ XAFS measurements were carried out at the 1W1B station in the Beijing Synchrotron Radiation Facility (BSRF, China). The in situ XAFS tests for electrochemical process were measured in a self-designed in situ cell with three electrodes in one chamber (~150 mL) (Supplementary Fig. 28) The in situ XAFS spectra at different electrochemical conditions were recorded using a Lytle detector (fluorescence mode) the working electrode is close to the X-ray window 0.1 M KOH solution (O2-saturated) was used as electrolytes To prepare the catalyst electrode for use as working electrode 5 mg sample was dispersed in a 1 mL mixed solution (0.9 mL isopropyl alcohol A certain amount of the catalyst ink was uniformly coated onto the carbon cloths (1 × 1.5 cm–2) as the working electrode The spectra labeled “open-circuit” were collected under open-circuit condition with the catalyst immersed in electrolyte In situ spectra were collected by chronoamperometry measurements at a series of representative potentials (1.0–0.75 V vs To analyze the local structure information of Pt atoms in the Pt1-CuOx/Cu samples the Athena and Artemis program was used to fit the EXAFS data The k2-weighted χ(k) data (2.5‒11.5 Å−1 of k-range) were Fourier-transformed to R-space under the hanning windows of dk = 1.0 Å–1 the number of independent points (Nipt) can be evaluated by the formula: Nipt = 2Δk × ΔR/π = 2 × (11.5 − 2.5) × (2.3 − 1.0)/π = 7 the amplitude reduction factor S02 for Pt foil was first obtained by fitting the Pt foil data and the value was conducted as a fixed parameter in the fitting procedure of R-space data for Pt1-CuOx/Cu Based on structural characterization information the Pt–O scattering path was considered to fit the R-space data under open-circuit and 1.00 V vs Debye–Waller factors (σ2) and energy shift (ΔE0) were conducted as adjustable parameters (Npara) according to the fitting criterion and the Npara was Npara = 4 <Nipt in this condition As for the in situ spectrum under 0.90 V vs the intensity of characteristic peak increased obviously compared to the 1.00 V vs Taking into account the possible evolution of oxygen-related intermediates on the Pt sites an additional Pt–O path was incorporated in the fitting procedure And the energy shift (ΔE0) for different paths was set equal to reduce the Npara and the Npara was Npara = 7 ≤ Nipt in this condition which is reasonable in the fitting procedure The data that support the findings of this work are available within the manuscript, Supplementary information files, and Source Data File. Source data are provided with this paper Electrocatalyst approaches and challenges for automotive fuel cells PdMo bimetallene for oxygen reduction catalysis Challenges in applying highly active Pt-based nanostructured catalysts for oxygen reduction reactions to fuel cell vehicles Experimental Sabatier plot for predictive design of active and stable Pt-alloy oxygen reduction reaction catalysts Modulation of electronic states in bimetallic-doped nitrogen-carbon based nanoparticles for enhanced oxygen reduction kinetics High performance platinum single atom electrocatalyst for oxygen reduction reaction The progress and outlook of metal single-atom-site catalysis Intrinsic electrocatalytic activity regulation of M-N-C single-atom catalysts for the oxygen reduction reaction Palladium single‐atom catalysts supported on C@C3N4 for electrochemical reactions Catalyst with a low load of platinum and high activity for oxygen reduction derived from strong adsorption of Pt−N4 moieties on a carbon surface Theoretical insights into the activation of O2 by Pt single atom and Pt4 nanocluster on functionalized graphene support: Critical role of Pt positive polarized charges Simultaneously achieving high activity and selectivity toward two-electron O2 electroreduction: The power of single-atom catalysts Tuning selectivity of electrochemical reactions by atomically dispersed platinum catalyst Improved oxygen reduction activity on Pt 3 Ni(111) via increased surface site availability Regulating the scaling relationship for high catalytic kinetics and selectivity of the oxygen reduction reaction Striding the threshold of an atom era of organic synthesis by single-atom catalysis Electronic regulation of Pt single-atom catalysts via local coordination state adjustment for enhanced photocatalytic performance In-situ spectroscopic observation of dynamic-coupling oxygen on atomically dispersed iridium electrocatalyst for acidic water oxidation Stabilizing Pt electrocatalysts via introducing reducible oxide support as reservoir of electrons and oxygen species Single-atom catalysts based on the metal-oxide interaction An atomic-scale view of single-site Pt catalysis for low-temperature CO oxidation Dispersion and support dictated properties and activities of Pt/metal oxide catalysts in heterogeneous CO oxidation Reconstructing the coordination environment of platinum single-atom active sites for boosting oxygen reduction reaction Unveiling role of sulfate ion in nickel‐iron (oxy)hydroxide with enhanced oxygen‐evolving performance Cationic oxidative leaching engineering modulated in situ self-reconstruction of nickel sulfide for superior water oxidation Reconstructed Ir‒O‒Mo species with strong Bronsted acidity for acidic water oxidation Generation of oxide surface patches promoting H-spillover in Ru/(TiOx)MnO catalysts enables CO2 reduction to CO The role of the metal core in the performance of WOx inverse catalysts Double perovskites as a family of highly active catalysts for oxygen evolution in alkaline solution Electrical pulse induced one-step formation of atomically dispersed Pt on oxide clusters for ultra-low-temperature zinc-air battery Modulating the electronic metal-support interactions in single-atom Pt(1) -CuO catalyst for boosting acetone oxidation Efficient Cu/FeOx catalyst with developed structure for catalytic transfer hydrogenation of furfural Cobalt single atom site isolated Pt nanoparticles for efficient ORR and HER in acid media Review of metal catalysts for oxygen reduction reaction: From nanoscale engineering to atomic design Oxyanion engineering suppressed iron segregation in nickel-iron catalysts toward stable water oxidation In situ modulating coordination fields of single-atom cobalt catalyst for enhanced oxygen reduction reaction Tensile straining of iridium sites in manganese oxides for proton-exchange membrane water electrolysers Tracking the oxygen dynamics of solid-liquid electrochemical interfaces by correlative in situ synchrotron spectroscopies Determination of photoelectrochemical water oxidation intermediates on haematite electrode surfaces using operando infrared spectroscopy Operando infrared spectroscopic insights into the dynamic evolution of liquid-solid (photo)electrochemical interfaces In-situ synchrotron radiation infrared spectroscopic identification of reactive intermediates over multiphase electrocatalytic interfaces Tailoring acidic oxygen reduction selectivity on single-atom catalysts via modification of first and second coordination spheres Adsorbed intermediates in oxygen reduction on platinum nanoparticles observed by in situ IR spectroscopy Operando identification of the dynamic behavior of oxygen vacancy-rich Co(3)O(4) for oxygen evolution reaction Symbiotic synergy enabling moderate oxo-hydroxy adsorption capacity for high-selectivity oxygen reduction Download references This work was supported by the National Key R&D Program of China (2022YFA1502903 (Q.L.)) the National Natural Science Foundation of China (22241202 (Q.L.) the Natural Science Foundation of Anhui Province (2208085J01 (Q.L.)) the Russian Science Foundation (project No the Postdoctoral Science Foundation of China (BX20230345 (W.Z.) and 2023M743366 (W.Z.)) and the Key Program of Research and Development of Hefei Science Center Part of this work was supported by the Instruments Center for Physical Science University of Science and Technology of China These authors contributed equally: Wanlin Zhou Experimental Center of Engineering and Material Science designed the experimental scheme and conducted the in situ experiments prepared the samples and conducted the electrochemical experiments provided constructive ideas to the experiment results Jun Pan and Gabor Samjeske for their contribution to the peer review of this work Download citation DOI: https://doi.org/10.1038/s41467-024-51157-4 Sorry, a shareable link is not currently available for this article. Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily. Volume 16 - 2022 | https://doi.org/10.3389/fnsys.2022.1014745 This article is part of the Research TopicNegative Valence SystemsView all 6 articles Editorial on the Research Topic Negative valence systems it is imperative that mental health research—particularly negative valence systems—include sex as a biological variable all constructs are disrupted in individuals with affective dysfunction although not necessarily at the same levels These constructs should be considered within the context of the organisms environmental and social factors (denoted by blue background) This timely Research Topic provides an opportunity for clinical and preclinical researchers to discuss the importance of investigating negative valence systems at multiple levels of analysis in humans and non-human models in a series of reviews and experimental reports Herein, Hu et al. highlight the importance of early life adversity (ELA) to pathology discussing the development of negative valence circuits and caregiver psychosocial regulation in human infants and rodent models following typical rearing vs The authors focus on the interface between the amygdala and mesolimbic dopamine system as a locus of dysfunction after ELA Considering the importance of these circuits in social behavior and recruitment following ELA the authors propose targeting these neural systems and upstream regions (i.e. habenula) to develop age-appropriate interventions following adversity Expanding on putative mechanisms underlying connectivity between negative experience and disease (Waters and Gould) review various ELA paradigms and describe different The authors discuss the growing literature supporting adversity type likely contributing to mental health outcomes They conclude that comparing brain changes resulting from different rodent ELA models will help clarify how specific subtypes of ELA influence discrete neural circuitry this knowledge could reveal specific targets for prevention/treatment of ELA-related neuropsychiatric disorders Specific to the construct of acute threat, du Plessis et al. investigated sex-differences in the neural networks that underlie threat conditioning in adult mice Their work quantifies neural activity (via c-fos) across multiple brain regions after cued threat conditioning and found that trained females engaged prelimbic cortex and subparafasicular nucleus more than—and subparaventricular zone less than—trained males they found sex differences in functional coordination of the threat conditioning network the globus pallidus and ventral lateral septum were the most robust hubs for trained males and females These findings suggest the existence of sex differences in threat conditioning neural circuitry which may at least partially explain sex differences in vulnerability to threat-related psychiatric disease Another important mental health issue is the persistence of pathological fear memories. Thus, Maren more explicitly reflects on their durability and how their resistance precludes success of therapeutic interventions for disorders of fear and anxiety relying on extinction of conditioned responses Maren reviews the literature regarding neural mechanisms underlying resistance to fear extinction particularly when extinction procedures are administered soon after fear conditioning (Immediate Extinction Deficit Several studies reveal that IED is mediated by recruitment of a stress-related neural network that facilitates encoding and consolidation of fear memory even when the threat has passed He further emphasizes the modulatory role that locus coeruleus norepinephrine exerts in amygdala-prefrontal cortical circuits and proposes specific neural mechanisms that balance excitation and inhibition in brain areas critical for extinction the reviews and studies presented here have important implications for understanding disorders of fear and anxiety in humans and using systems models synaptic and behavioral mechanisms underlying negative valence systems remains critical for developing treatments to prevent/treat emotional problems in anxiety and stress-related disorders researchers must harness the increased research validity that the RDoC Negative Valence Systems constructs afford to translational research models to better characterize and predict underlying neurobiological drivers of affective disorders All authors listed have made a substantial and intellectual contribution to the work and approved it for publication JH was supported by the Maine IDeA Network for Biomedical Excellence (INBRE) subaward which is supported by the National Institute of General Medical Sciences of the National Institute of Health (Award P20GM103423) AP was supported by the National Center for Complementary and Integrative Health (Award K99AT010903) The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher Animal models in psychiatric research: the RDoC system as a new framework for endophenotype-oriented translational neuroscience CrossRef Full Text | Google Scholar Challenges in the use of animal models and perspectives for a translational view of stress and psychopathologies Sex bias in neuroscience and biomedical research PubMed Abstract | CrossRef Full Text | Google Scholar A modern learning theory perspective on the etiology of panic disorder Toward the future of psychiatric diagnosis: the seven pillars of RDoC PubMed Abstract | CrossRef Full Text | Google Scholar A functional behavioristic approach to aversively motivated behavior: Predator imminence as a determinant of the topography of defensive behavior Changes in feeding and foraging patterns as an antipredator defensive strategy: a laboratory simulation using aversive stimulation in a closed economy Research Domain Criteria (RDoC): toward a new classification framework for research on mental disorders Sex and depression in the National Comorbidity Survey I: lifetime prevalence The ecology of human fear: survival optimization and the nervous system When fear is near: threat imminence elicits prefrontal-periaqueductal gray shifts in humans Animal models of neuropsychiatric disorders PubMed Abstract | CrossRef Full Text | Google Scholar PubMed Abstract | CrossRef Full Text | Google Scholar CrossRef Full Text | Google Scholar CrossRef Full Text | Google Scholar The psychological impact of COVID-19 on the mental health in the General population Are hormones a “female problem” for animal research PubMed Abstract | CrossRef Full Text | Google Scholar Factors affecting sex-related reporting in medical researh: a cross disciplinary bibliometric analysis CrossRef Full Text | Google Scholar Vera-Villarroel Psychology and COVID-19: an analysis from the basic psychological processes Sex differencs in rates of depression: cross-national perspectives Research domain criteria versus DSM V: how does this debate affect attempts to model corticostriatal dysfunction in animals Porcu A and Sabariego M (2022) Editorial: Negative valence systems Received: 08 August 2022; Accepted: 31 August 2022; Published: 21 September 2022 Copyright © 2022 Honeycutt, Young, Porcu and Sabariego. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) distribution or reproduction in other forums is permitted provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited in accordance with accepted academic practice distribution or reproduction is permitted which does not comply with these terms *Correspondence: Marta Sabariego, bXNhYmFyaWVAbXRob2x5b2tlLmVkdQ== Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher 94% of researchers rate our articles as excellent or goodLearn more about the work of our research integrity team to safeguard the quality of each article we publish Metrics details Information about the concordance between dynamic emotional experiences and objective signals is practically useful Previous studies have shown that valence dynamics can be estimated by recording electrical activity from the muscles in the brows and cheeks whether facial actions based on video data and analyzed without electrodes can be used for sensing emotion dynamics remains unknown We investigated this issue by recording video of participants’ faces and obtaining dynamic valence and arousal ratings while they observed emotional films detected through an automated analysis of the video data were negatively and positively correlated with dynamic ratings of subjective valence Several other AUs were also correlated with dynamic valence or arousal ratings interpreted using the SHapley Additive exPlanation tool revealed non-linear associations between the AUs and dynamic ratings of valence or arousal These results suggest that an automated analysis of facial expression video data can be used to estimate dynamic emotional states which could be applied in various fields including mental health diagnosis dynamic sensing of emotions using objective signals can be useful laying the foundation for practical applications in fields ranging from mental health diagnostics to enhanced interpersonal communication The results showed that the EMG activity of the corrugator supercilii and zygomatic major muscles was negatively and positively correlated with the dynamic valence ratings Such results suggest associations between facial EMG signals and the dynamic experiences of emotional valence AU 12) associated with memory-based emotional experiences are also negatively and positively associated with valence detected through an automated analysis of video data would be negatively and positively associated with dynamic ratings of subjective valence revealed non-linear associations between subjective valence ratings and facial EMG activity we hypothesized that ML modeling and SHAP analysis would reveal the non-linear associations between dynamic ratings of valence or arousal and automated AU data we explored 18 other AUs that could be automatically coded by the software We also performed RF regression modeling and SHAP analysis of the relationships between the dynamic ratings and the AUs The mean second-by-second subjective dynamic ratings of valence (left) and arousal (right) solid-colored lines represent the mean values and shaded areas of the same color indicate the standard error redder hues in the spectrograms correspond to higher levels of valence/arousal and more purple hues correspond to lower levels of valence/arousal To illustrate the differences in facial expressions elicited by different emotional stimuli, two different visualization methods are used in Fig. 2 to show how the mean values and standard errors (SEs) of specific facial AUs varied over time during the viewing of the five films. Group mean action unit (AU) intensities associated with the viewing of five films The upper panels show line charts highlighting the differences in activation levels of the same AUs across different films emphasizing how emotional responses varied with the content viewed and the shaded area of the same color indicates the standard error The lower panels show spectrograms highlighting the differences in activation levels of the different AUs while watching the same film revealing the complexity of the emotional reactions elicited by a single film A redder hue indicates higher AU intensity and a more purple hue indicates lower intensity Mean (standard error) Pearson’s correlation coefficients between second-by-second dynamic valence (left) or arousal ratings (right) and action units across time (concatenated for all film conditions) Coefficients indicating positive and negative associations are shown as red and blue bars The significance of the correlations is denoted by asterisks: *p < 0.05 we conducted exploratory analyses of the associations between the valence or arousal ratings and all 20 analyzed AUs using Hotelling’s one-sample T2 test and follow-up univariate one-sample t-tests Hotelling’s one-sample T2 test revealed a significant association between the valence ratings and AUs (T2[3 Follow-up univariate tests indicated significant negative correlations for AUs 01 and significant positive correlations for AUs 06 Hotelling’s one-sample T2 test revealed a significant association between arousal ratings and AUs (T2[3 Follow-up univariate tests indicated a significant negative correlation for AU 43 and significant positive correlations for AUs 02 Figure 4 illustrates the AUs showing significant correlations with valence or arousal ratings. To evaluate the predictive performance of the RF and linear models we calculated correlation coefficients between the actual and predicted values Leave-one-out cross-validation was employed where the data from one participant served as the evaluation dataset and the data of the other participants comprised the training dataset a correlation coefficient was computed for each individual participant serving as the evaluation dataset The mean ± SE correlation coefficients between the actual and predicted valence and arousal ratings in the RF model were 0.42 ± 0.05 One-sample t-tests revealed that all models had correlation coefficients significantly greater than zero (t[22] = 6.56 Paired t-tests indicated no significant differences between the RF and linear models in valence (t[22] = 0.21 SHAP tools were applied to quantify and visually depict nonlinear associations in the RF model. The absolute mean SHAP values, indicating feature importance in the RF model, are shown in Fig. 5. The results suggested that AUs 06, 12, 09, and 04, and AUs 43, 09, 06, and 07, were the four most important features for predicting valence and arousal, respectively. Mean absolute SHapley Additive exPlanation (SHAP) values for each action unit (AU) in the random forest regression model showing the effect of each feature on the subjective valence (left) and arousal (right) rating predictions The SHAP dependency plots in Fig. 6 show representative relationships between the AUs and SHAP values for changes in valence/arousal ratings. Both simple linear and more complex non-linear associations are shown (e.g., step-like and gradual changes in the valence—AU 12 and valence—AU 04 relationships, respectively). SHapley Additive exPlanation (SHAP) dependency plots showing relationships between predicted valence (left) or arousal (right) ratings and facial action units (AUs) The AUs shown are the four most important features for predicting valence and arousal The SHAP values demonstrate how the valence and arousal ratings change as the AUs are inputted although dynamic ratings of subjective valence were not obtained the present study is the first to report that automated video analysis of AUs 04 and 12 is associated with subjective valence ratings Our study replicated these results using a film presentation paradigm to induce spontaneous emotional responses our findings suggest that changes in AU activation patterns may reflect dynamic subjective experiences our data suggest that facial actions may reflect the dynamics of arousal experiences The results suggest that the relationships between emotional valence/arousal ratings and AUs may primarily be linear our sample size was relatively small and may not have been sufficient for ML modeling; thus Mental health professionals are occasionally asked to monitor patients’ emotional states remotely for early identification of a disturbance and timely intervention Precise and non-invasive facial expression analysis enhances the capability to identify suspicious behavior in public and sensitive environments thereby supporting safety protocols without impinging on personal privacy This innovative approach enables instructors to evaluate student engagement and emotional state in real time facilitating the development of teaching approaches that are responsive to the emotional and educational requirements of learners Future research should include participants from different cultural backgrounds we did not assess the participants’ default emotional states which may affect emotional experience and AU activation Studies with a more detailed assessment process would deepen understanding of the relationships between subjective emotional dynamics and facial actions this study demonstrated associations between emotional valence/arousal ratings and facial AUs through automated video analysis and provided the first evidence that automated video analysis can reveal the dynamics of subjective emotional valence/arousal Facial AU activity exhibited both simple linear and complex non-linear relationships Different AUs vary in their sensitivity to emotional stimuli and thus also in their response to changes in emotions and 12 (lip-corner pulling) are extremely sensitive to slight changes in emotion showing significant activation in response to subtle changes in facial expressions when emotions are finely tuned other AUs may only show significant activation when there is a stronger emotional stimulus when analyzing and interpreting AU activity data to accurately characterize people’s emotional responses Despite several methodological limitations our study provides support for facial expression-based emotional monitoring and personalized interventions Future research should improve the experimental methods and explore non-linear relationships between dynamic emotional states and AUs to more accurately understand and predict these states This approach will further advance the application and development of facial expression analysis technology for use in various critical sectors Analysis of subjective facial associations using a two-step procedure with one-sample t-tests (two-tailed) was planned An effect size d of 0.55 was estimated based on the weak subjective-facial association; with an α level of 0.05 and a power (1 – β) of 0.8 the power analysis showed that > 22 participants were needed were excluded due to technical problems with the video acquisition system The participants were recruited via advertisements at Kyoto University and were compensated in book coupons corresponding to a value of 4000 Japanese yen The inclusion criteria were as follows: willingness to participate in subjective and physiological measurements; normal or corrected-to-normal vision without the use of glasses; Japanese as the first language; and no neurological or psychiatric issues The exclusion criterion was previous experience of participating in experiments using the emotional film clips employed in our study All participants provided written informed consent after a thorough explanation of the procedures This study was approved by the RIKEN Ethics Committee The experiment was conducted following the institutional ethical guidelines and the Declaration of Helsinki experimental events were managed using Presentation software (Neurobehavioral Systems USA) running on an HP Z200 SFF Windows computer (Hewlett-Packard Japan The software presented films and response displays and provided digital trigger pulses synchronized with film onset Visual stimuli were displayed on a 19-inch cathode ray tube monitor (HM903D-A; Iiyama Japan) with a 100-Hz refresh rate and 1024 \(\times \) 768-pixel resolution Japan) was placed above the monitor for video recording A655sc infrared thermal imaging cameras (FLIR Systems USA) were used to acquire the facial thermal images; these data are not reported here For the cued-recall dynamic valence and arousal ratings we used PsychoPy software (v2023.2.3; Open Science Tools Ltd. UK) running on a MacBook Air laptop (M2; Apple The stimulus resolution was 640 \(\times \) 480 pixels corresponding to visual angles of approximately 25.5\(^\circ \) and 11\(^\circ \) The experiments were completed on an individual basis participants were briefed on the overall procedure The study commenced with a one-shot rating session for the acquisition of video data we set the indoor temperature to 24 °C using central air conditioning We also asked the participants to remove their outerwear and to indicate their level of comfort with the room temperature while they completed the study forms Adjustments were made based on the participants’ feedback to ensure that the temperature did not become too high or too low during the experiment All participants reported feeling comfortable at around 24 °C leading us to designate 23.5–24.5 °C as the neutral temperature range in our experiments Each participant was seated comfortably on a chair that was fixed such that the participant’s face was approximately 0.77 m from the monitor The digital web camera was located above the monitor in alignment with the front edge of the monitor Response display for the one-shot ratings of valence and arousal Response displays for the dynamic ratings of valence (left) and arousal (right) by using an artificial neural network trained on a large database of AUs the software quantified the intensities of 20 AUs (01 The start and end times for each film stimulus were determined based on digital triggers synchronized with the film onset and stimulus duration The video data for each participant were segmented into five clips based on the start and end times One-shot ratings were exported directly from the log files of Presentation software Dynamic ratings were exported from the log files of PsychoPy software where they were originally recorded on a frame-by-frame basis We used the codeOnlineRating function in PsychoPy software to record dynamic valence and arousal ratings while participants watched the films component setup was as follows: codeOnlineRating has fixed parameters including “Before Experiment,” “Begin Experiment,” “Begin Routine,” “Each Frame,” “End Routine,” and “End Experiment” It also featured frame-by-frame recording: we wrote code for the “Each Frame” parameter to ensure that whenever participants updated their rating (i.e. Each data entry event included the current rating and the timestamp (in milliseconds) as well as the monitor refresh rate; the recording of frames is tied to the monitor’s refresh rate such that there were 60 frame updates per second To ensure alignment with the AU intensity data the ratings were converted to a second-by-second format by calculating the mean values of all frames within each second along with their continuous valence and arousal ratings were arranged in a fixed film stimulus order (anger and amusement) and stored in CSV files for subsequent analysis Results were considered significant at p < 0.05 the linear and quadratic natures of the valence and arousal ratings were assessed across films These data suggest that almost none of the reported dynamic rating–AU correlations were restricted to specific types of stimulus valence We calculated the absolute mean SHAP values for the obtained hyperquantities; this allowed us to assess the significance of each input feature we plotted scatterplots of the relationships between input features and their SHAP values for each instance The datasets used and/or analysed during the current study available from the corresponding author on reasonable request The feeling of music past: How listeners remember musical affect How we remember the emotional intensity of past musical experiences From experience to memory: On the robustness of the peak-and-end-rule for complex When the parts of the sum are greater than the whole: Assessing the peak-and-end-theory for a heterogeneous The relation between short-term emotion dynamics and psychological well-being: A meta-analysis Advancements in sensors and analyses for emotion sensing Physiological correlates of subjective emotional valence and arousal dynamics while viewing films Emotional valence sensing using a wearable facial EMG device Exploration of emotion dynamics sensing using trapezius EMG and fingertip temperature Altered emotional mind-body coherence in older adults The review of applications and measurements in facial electromyography Electromyographic validation of spontaneous facial mimicry detection using automated facial action coding Development of the RIKEN database for dynamic facial expressions with multiple angles Facial action coding system (FACS): A Technique for the Measurement of Facial Action (Consulting Psychologists Press Machine learning-based interpretable modeling for subjective emotional dynamics sensing using facial EMG A unified approach to interpreting model predictions In Proceedings of the 31st International Conference on Neural Information Processing Systems (NIPS’17) Emotion elicitation effect of films in a Japanese sample Automated facial coding: Validation of basic emotions and FACS AUs in FaceReader Continuous measurement of emotion: The affect rating dial In Handbook of Emotion Elicitation and Assessment (eds Coherence between expressive and experiential systems in emotion and methodological issues in inferring subjective emotion experience: Recommendations for researchers In Handbook of Emotion Elicitation and Assessment FACSGen: A tool to synthesize emotional facial expressions through systematic manipulation of facial action units FACSGen 2.0 animation software: Generating three-dimensional FACS-valid facial expressions for emotion research Coherence between emotion and facial expression: Evidence from laboratory experiments Coherence between emotions and facial expressions: A research synthesis Facial expressions of basic emotions in Japanese laypeople Components and recognition of facial expression in the communication of emotion by actors Pleasure-arousal-dominance driven facial expression simulation In 3rd International Conference on Affective Computer Intelligence on Interacting Workshops with special emphasis on somatovisceral “illusions” and anxiety: Brain mechanisms and psychophysiology Dynamics of facial expression: Recognition of facial actions and their temporal segments from face profile image sequences Quantifying human sensitivity to spatio-temporal information in dynamic faces A facial-expression monitoring system for improved healthcare in smart cities Deep convolution network based emotion analysis towards mental health care Diagnosis of depressive disorder model on facial expression based on fast R-CNN Facial expressions in adolescent–parent interactions and mental health: A proof-of-concept study Human behavior prediction using facial expression analysis In International Conference on Computing Communication on Automation (ICCCA) Raspberry Pi assisted facial expression recognition framework for smart security in law-enforcement services Towards emotion detection in educational scenarios from facial expressions and body movements through multimodal approaches Towards emotion-sensitive learning cognitive state analysis of big data in education: Deep learning-based facial expression analysis using ordinal information Recognizing students’ emotions based on facial expression analysis In 11th International Conference on Information Technology Medical Education (ITME) Naturalistic emotion decoding from facial action sets Facial displays are tools for social influence Unmasking the Face: A Guide to Recognizing Emotions from Facial Clues (Prentice-Hall Inc Audience effects on solitary faces during imagery: Displaying to the people in your head G*Power 3: A flexible statistical power analysis program for the social A controlled study on evaluation of thermal stimulation influence on affective measures of uninformed individuals Study on winter comfort temperature in mixed mode and HVAC office buildings in Japan Affect grid: A single-item scale of pleasure and arousal A performance comparison of eight commercially available automatic classifiers for facial affect recognition Rapid object detection using a boosted cascade of simple features Statistical Models of Appearance for Computer Vision Recognizing Semantic Features in Faces Using Deep Learning JASP (Version 0.14.1) (computer software) (2020) Trujillo-Ortiz, A. Hotelling T2. MATLAB Central File Exchange. https://www.mathworks.com/matlabcentral/fileexchange/2844-hotellingt2 (2024) Duration neglect in retrospective evaluations of affective episodes Empirical comparison of univariate and multivariate analysis of variance procedures Download references The authors thank Masaru Usami for his technical support Academic Center for Computing and Media Studies Conceived and designed the experiments: JZ Download citation DOI: https://doi.org/10.1038/s41598-024-70563-8