Biology subjects

Ding, M.

Publications and source records attributed to Ding, M..

At least 19 recordsLinked to original sources

bioRxiv2026

The Aging Brain Matrix: Cross-Species Evidence Linking FREM3 to Brain and Cognitive Aging

The human extracellular-matrix-related FREM3 gene is strongly age-dependent, enriched in age-sensitive excitatory long-range-projecting neurons, and implicated in neurodegenerative diseases. To establish the first translationally relevant causal evidence for the role of FREM3 in brain structure and cognition, we developed an excitatory-neuron-specific Frem3 conditional knockout mouse model (Frem3-cKO) and computed genetically regulated expression of FREM3 (FREM3-GReX) in humans, enabling convergent neurostructural and cognitive characterizations. Frem3-cKO mice showed widespread lower global and regional brain volume in middle age vs young adulthood, consistent with drastic age-dependent volume loss absent in wildtype mice. These effects were strongest in cognitively relevant cortical regions and paralleled by lower cognitive performance. In humans, lower FREM3-GReX was associated with lower total surface area, which further mediated an indirect link to lower cognitive performance across multiple age-sensitive domains. We provide the first translationally relevant cross-species evidence supporting a key role for FREM3/Frem3 in maintaining brain and cognitive health across aging.

neuroscience↗

bioRxiv2026

CIDER: detecting changes in gene regulatory networks that are associated with changes in phenotype

Changes in gene regulatory networks may drive quantitative traits, or may transmit the effects of one trait, such as blood lipid level, on another, such as cardiovascular health. Yet the standard tools, differential correlation and differential network analysis, compare two discrete groups, while the contexts of interest - circulating lipids, inflammation, and blood glucose - vary continuously; applying them forces dichotomization, discarding within-trait variation. We introduce Continuous Interaction-based Differential Edge Regulation (CIDER), which tests whether a gene regulatory network edge, the relationship between a transcription factor and its target gene, varies with a continuous trait: the target gene's expression is modeled as a function of the TF's expression level, the trait, and their interaction, with the interaction coefficient measuring the trait dependence. To limit multiple testing, CIDER tests only the edges of a reference regulatory network. A generalized additive extension detects interactions that change the shape of the relationship, not only its slope, including forms that cannot be expressed as a difference between two correlations. In simulations it outperformed four two-group methods across sample sizes, effect sizes, and noise levels, with most of its advantage from keeping the trait continuous. In whole-blood transcriptomes from four independent human cohorts across ten quantitative health traits, CIDER identified 63 replicated cases in which a TF's regulation of its target varies with the trait, including coupling of the glucocorticoid-receptor (NR3C1) to the granulocyte colony-stimulating-factor receptor (CSF3R) that strengthens as triglycerides rise, and a pair whose regulation reverses direction across the observed range of C-reactive protein.

bioinformatics↗

bioRxiv2026

Sensory cortical disinhibition is required for long-term fear memory in humans

The transformation of transient threats into enduring memories is fundamental to survival, yet how the human brain constructs long-term fear memory remains elusive. Early sensory cortex, traditionally viewed as a passive feature analyzer, is increasingly recognized as an active participant in fear processing. Here, using a multimodal approach combining high-definition -frequency transcranial alternating current stimulation (HD -tACS) with simultaneous EEG-fMRI and transformer-based neural decoding, we tracked fear memory development across early formation, consolidation, and long-term retention stages (assessed at 15 min, 24 h, and 7 days post-conditioning). Spatiotemporal memory representations were evident at all three time points, beginning at an early sensory stage (40-60 ms post-stimulus). Conditioning potentiated occipital -oscillation desynchronization (a marker of visual cortical disinhibition), which predicted the strength of subsequent long-term memory traces. Critically, targeted -tACS over early visual cortex induced V1/V2 inhibition, abolished conditioning-potentiated desynchronization, and selectively disrupted fear-memory consolidation and retention while leaving initial acquisition intact. These findings establish sensory cortical disinhibition as a necessary mechanism for long-term fear memory in humans and identify early sensory cortex as a causal node for interventions targeting persistent maladaptive fear.

neuroscience↗

bioRxiv2026

Adipose GPD2 Couples Glycolytic Signals to Epigenetic Control of Metabolic Homeostasis

The glycerol-3-phosphate shuttle (G3PS) is a classic reducing equivalent transferring system, yet its role in adipose metabolism remains poorly understood. Here, we identify mitochondrial glycerol-3-phosphate dehydrogenase (GPD2), the core enzyme of G3PS, as a critical orchestrator of adipose thermogenesis and systemic metabolic homeostasis. Adipocyte-specific GPD2 knockout mice exhibited impaired thermogenic capacity and energy expenditure, rendering them susceptible to obesity and metabolic dysfunction. Mechanistically, GPD2 deficiency elevated cytosolic NADH, which suppressed glycolysis and decreased the levels of key metabolites, such as acetyl-CoA, thereby diminishing H3K27ac at thermogenic gene loci and silencing these genes. NADH depletion with an NADH oxidase rescued both glycolysis and thermogenic gene expression. In support of this mechanism, we defined a previously undescribed enhancer of Ucp1 whose H3K27ac was tightly controlled by GPD2. Remarkably, replenishing the acetyl-CoA pool or directly restoring H3K27ac at the Ucp1 enhancer reversed UCP1 expression in GPD2-deficient adipocytes. Importantly, we find that human adipose GPD2 expression is inversely correlated with BMI, WHR, and HOMA-IR, underscoring its clinical relevance. Collectively, our findings establish GPD2 as an essential node in the metabolic-epigenetic axis that governs thermogenic activation and energy balance.

biochemistry↗

bioRxiv2026

Netrin-1 inhibition does not attenuate cancer-induced bone pain in three translational models

1.Cancer-induced bone pain (CIBP) occurs in a majority of patients when primary or metastatic cancer develops within the bone. This pain has a significant impact on quality of life, yet there are limited effective treatment options available. Nerve sprouting is a complex mechanism that has been implicated in CIBP. Netrin-1 is a neuronal guidance molecule that is produced by numerous cell types, including cancer cells. In this study we aimed to determine whether netrin-1 inhibition (with NP137 - a humanized IGg1 monoclonal antibody) could ameliorate nerve sprouting, and nociception by extension, in three models of CIBP - osteosarcoma, metastatic breast cancer, and metastatic prostate cancer. Sustained administration of NP137 failed to produce an anti-nociceptive effect in these models, but a delayed onset was observed in the osteosarcoma model. NP137 did not produce a disease-modifying effect, as micro-computed tomography did not reveal reduced bone destruction in the NP137-treated groups. Additionally, there was no nerve fibre density reduction in any of the groups at the late-stage of the disease, suggesting that nerve sprouting occurs in early- to mid-stage CIBP development. Investigation of NP137 exposure indicated that serum levels of NP137 were comparable between the sham and cancer groups. Our study indicates that netrin-1 may play a role in early-stage CIBP development, but inhibition of this mechanism does not produce robust anti-nociception.

pharmacology and toxicology↗

bioRxiv2026

Paradoxical Th1 activation and CTLA-4 regulation is beneficial during latent cryptococcosis

Cryptococcus neoformans is the predominant causative agent of cryptococcal meningitis in immunocompromised individuals. Conversely in immunocompetent individuals, C. neoformans establishes a latent pulmonary infection characterized by a paucity of clinical symptoms. Using a mouse inhalation model of latent C. neoformans infection, we previously showed that CD4 T-cells are necessary for preventing fungal proliferation in the lungs. In the current study, we performed single cell RNA sequencing (scRNAseq) and found that the CD4 T-cell response was both highly heterogenous and dichotomous during pulmonary C. neoformans infection, with concomitant expression of genes related to Th1 polarization (Tbx21, Ifng) and immune regulation (Ctla4). First, we demonstrated that cells with Th1-like phenotypes are necessary and sufficient to control latent infection via adoptive transfer of T-bet positive cells into infection-matched CD4-depleted recipient mice. Second, scRNAseq analysis revealed the subpopulation of effector CD4 T-cells that co-expressed Ctla4 and Gata3 was significantly higher than a subpopulation that co-expressed Ctla4 and Tbx21. Furthermore, our data suggested that CTLA-4 upregulation is beneficial against C. neoformans infection, as CTLA-4 blockade promoted fungal proliferation. Thus, we propose a model wherein Th1 control of latent C. neoformans infection is supported by CTLA-4 suppression of detrimental Th2 activation. IMPORTANCECD4 T-cells are a critical part of the adaptive immune response in many infections. Upon inhalation of the fungal pathogen Cryptococcus neoformans, CD4 T-cells are necessary to control the initial latent infection and prevent disease. In immunocompromised individuals that lack CD4 T-cells, the initial infection in the lungs can spread to the brain and cause lethal meningitis. In this study, we show that the CD4 T-cells present during latent C. neoformans infection are heterogeneous with expression of diverse polarization and immune regulation markers that function in concert to prevent disease. Our studies highlight the previously unappreciated diversity and complex regulation of the CD4 T-cell response that is required to prevent disease in this important fungal pathogen.

genomics↗

bioRxiv2026

Identification and molecular characterization of a novel TYLCV isolate breaking bred-resistance to threaten tomato cultivar

Tomato yellow leaf curl virus (TYLCV) is a devastating viral pathogen threatening agricultural crops globally. In this study, we identified a novel TYLCV isolate (TYLCV-YN6244), which caused viral epidemic in resistant tomato cultivars at Yuanmo county, Yunnan Province of China. We determined the complete genome of TYLCV-YN6244 and found it encoded six viral proteins characteristic of Geminivirus. We identified its V2 protein as a potent viral suppressor of RNA silencing (VSR), and generated infectious clone of wildtype TYLCV-YN6244, or V2-defective TYLCV-YN6244 (TYLCV-YN6244-{Delta}V2) in which V2 was deleted. Both of infectious clones were capable of systemically infecting tobacco and tomato. However, TYLCV-YN6244 but not TYLCV-YN6244-{Delta}V2 could cause disease symptoms in wildtype tobacco or tomato plants, and viral accumulation was drastically reduced in plants infected with TYLCV-YN6244-{Delta}V2 compared to TYLCV-YN6244 while the efficiency of virus-derived small interfering RNAs (vsiRNAs) biogenesis was conversely increased in plants infected with TYLCV-YN6244-{Delta}V2. Surprisingly, small RNA profiling indicated that 21nt and 22nt rather than 24nt vsiRNAs were predominantly produced in tomato plants infected with either TYLCV-YN6244 or TYLCV-YN6244-{Delta}V2. Furthermore, transcriptome analyses revealed that TYLCV-YN6244 or TYLCV-YN6244-{Delta}V2 infection differentially modulated metabolism and defense-related pathways in tomato, probably underlying distinct viral pathogenicity and disease symptoms induced in plants. Overall, our research not only identified a novel pathogenic TYLCV isolate but also characterized molecular biology and host response in tomato with infectious clones firstly developed, with implications in untangling virus-host interaction for developing novel resistance in crop tomato.

Plant Biology↗

bioRxiv2026

High resolution spatial transcriptomics decodes the microenvironmental determinants of response to Nr-CWS therapy in cervical precancerous lesions

Immunotherapy with Nocardia rubra cell wall skeleton (Nr-CWS) can clear human papillomavirus (HPV) and induce regression of cervical precancerous lesions, yet many patients do not respond. The cellular and molecular basis for this heterogeneous clinical outcome remains unknown. Using high-resolution spatial transcriptomics, we profiled squamous intraepithelial lesions (SIL) from patients stratified by their subsequent response to Nr-CWS therapy. We discovered that non-responders were characterized by an immunosuppressive epithelial phenotype, defined by aberrant expression of the pro-inflammatory alarmin S100A9 in squamous epithelial cells (SECs), which was associated with poorer overall survival in cervical cancer. Non-responding lesions were enriched in proliferative (MKI67+) and pro-angiogenic (VEGFA+) SEC subpopulations and spatially organized into a pro-tumorigenic cellular neighborhood (CN-10) overexpressing S100A9 and ANXA1. In contrast, responders exhibited protective glandular-cell-enriched neighborhoods (CN-13) and supportive spatial interactions between lymphatic endothelial cells and fibroblasts. The unfavorable microenvironment was further defined by an immunosuppressive cloak of VCAN+ stromal cells and CD163+ M2 macrophages encircling SECs. Thus, response to Nr-CWS is determined by a pre-existing spatial network of epithelial-stromal-immune interactions, providing a roadmap for patient stratification and rational combination therapies.

cancer biology↗

bioRxiv2026

From Attention Control to Stimulus Selection: Neural Mechanisms Revealed by Multivariate Pattern and Functional Connectivity Analyses

Visual spatial attention deployed in advance of sensory stimulation enhances the processing of the stimuli at an attended location. While it is understood that the attention control signals are established even before the stimulus occurs, how these signals help achieve stimulus selection is still not clear. Here, we investigated the neural mechanisms of spatial attention control and subsequent stimulus selection by recording fMRI data from participants performing a cued visual spatial attention task. At the beginning of each trial, participants were cued to covertly attend either the left or the right visual field. Following a random cue-target period, a target stimulus appeared either at the attended location or at the unattended location. Participants discriminated the stimulus appearing at the attended location and ignored the stimulus appearing at the unattended location. Using MVPA decoding and multivariate functional connectivity techniques, we investigated the nature of the information in visual cortex during both the cue and target periods, and further probed how cue-related information was related to target-related sensory processing. The following results were found: (1) attend-left vs. attend-right conditions could be decoded from the cue-period neural activity in visual cortex, (2) target position (target-left vs. target-right) could also be decoded from the target-evoked activity in visual cortex, (3) classifiers built on cue-period neural activity could cross-decode attended target-evoked neural activity in visual cortex and vice versa, (4) higher pattern similarity across cue and target periods, as indexed by cross-decoding accuracy, was correlated with better behavioral performance, (5) the strength of cue-evoked multivariate functional connectivity patterns in visual cortex was positively correlated with behavioral performance, and (6) cue-evoked multivariate functional connectivity patterns were similar to those evoked by the attended targets, and higher connectivity pattern similarity across cue and target periods was correlated with better behavioral performance. These results suggest that top-down attention control enables the formation of (1) a spatial attention template at the level of individual visual cortical areas and (2) an attention network template across visual areas, and these neural patterns support stimulus selection likely via a template matching mechanism at both area and pathway levels.

neuroscience↗

bioRxiv2026

Visual Emotion Perception in a Deep Neural Network Model with Both Bottom-Up and Top-Down Connections

Emotion reshapes perception by modulating sensory processing through top-down feedback--a process referred to as emotional perception. The computational mechanisms by which distinct affective signals influence visual representations however remain poorly understood. Here, we use a deep neural network to simulate this process and test mechanistic hypotheses about how top-down feedback guides emotional peception. Most existing models treat the perception of emotional content as a static, feedforward task, overlooking the dynamic interplay between internal states, external goals, and sensory input that characterizes affective perception in the brain. We introduce EmoFB, a biologically inspired model that integrates an affective system with a visual processing hierarchy through two functionally distinct feedback signals: intrinsic feedback, arising from the models own affective appraisal of perceptual input, and external steering, conveying contextual priors such as task expectations or target categories. We evaluated EmoFB on three tasks varying in perceptual ambiguity (Single Image, Side-by-Side, and Overlay). External steering exerted the strongest influence, not only improving recognition under challenging conditions but also restructuring internal representations by sharpening category-specific clustering in feature space. Crucially, top-down feedback increased brain-model representational similarity, strengthening alignment with human fMRI responses across early visual cortex, ventral visual areas, and the amygdala. EmoFB provides a computational framework for testing neurocognitive theories of emotion appraisal and top-down feedback modulation. It bridges affective neuroscience and artificial intelligence, offering mechanistic insight into how emotional signals shape perception in both brains and machines.

neuroscience↗

bioRxiv2026

A network for self-transcendence derived from patients with brain lesions

Self-transcendence, the reorientation of experience away from the self toward others, nature, or broader meaning, is a fundamental dimension of human psychology, yet its causal neural architecture remains poorly understood. Here we applied lesion network mapping to 88 neurosurgical patients with pre- and post-operative assessments of trait self-transcendence to identify the distributed brain network whose disruption alters this capacity. The resulting network showed significant spatial correspondence with the default mode network and, at a finer parcellation level, with frontoparietal control subnetworks. Leave-one-out analyses identified posterior midline regions as the most stable correlates of increased self-transcendence following brain lesions. Independent validation against fMRI meta-analyses of self-referential processing, compassion, and ketamine administration, alongside a neuromodulation target previously shown to modulate the sense of self, converged on a consistent model. These findings provide causal evidence for a network architecture in which posterior midline hubs constrain, and brainstem and anterior midline regions facilitate, self-transcendent experience.

neuroscience↗

bioRxiv2026

Rapid and Interpretable AMR Diagnostics via Genomics and Cell Painting using Differential Geometry-based Directed-Simplicial Neural Networks on Multimodal Data

Antimicrobial resistance (AMR) remains a critical global health challenge, particularly in high-prevalence regions such as India, where rapid and interpretable diagnostic tools are urgently needed. To address this challenge, we present a computational framework for AMR prediction that integrates genomic and cellular phenotypic data using an in-house developed differential geometry-based Directed Simplicial Neural Network (Dg-Dir-SNNs) applied to multimodal datasets. Using this framework, we analyzed 384 clinically relevant AMR isolates, including Escherichia coli and Klebsiella pneumoniae, integrating 256 genomic k-mer features with 503 cellular morphology descriptors derived from high-content Cell Painting assays. The Dg-Dir-SNNs model constructs an inferred-causal network of top-ranked biomarker-driving features, predicting potential directional dependencies among genomic motifs and phenotypic features. Network analysis identified kmer_TATG as the top-ranked driver associated with predicted resistance, with a local neighborhood including other genomic motifs (kmer_TTTT, kmer_CGTG, kmer_TCAC, kmer_CGTA, kmer_GAAA, kmer_TAAA, kmer_TACA, kmer_TGTG, kmer_TGAG, kmer_AAAA) and a key morphological feature (Cells_correlation_ER_Brightfield). These relationships suggest potential mechanistic associations in which specific genomic motifs may influence cellular phenotypes linked to antimicrobial resistance. Although not yet clinically deployed, this approach demonstrates the potential of multimodal AI-driven modeling for rapid in silico AMR prediction. By providing interpretable, biologically grounded insights, the framework may support future diagnostic development, targeted surveillance strategies, and experimental validation in high-resistance healthcare settings.

microbiology↗

bioRxiv2025

Neural Mechanisms of Willed Attention Control

Cueing paradigms are commonly used to study the neural mechanisms of visual spatial attention control. In these paradigms, each trial starts with an external cue, which instructs the subject to pay covert attention to a spatial location in anticipation of an impending stimulus (instructed attention). Recent work has introduced a new type of cue which prompts the subject to spontaneously decide which spatial location to attend (willed attention). We studied the neural mechanisms of willed attention control by analyzing fMRI and EEG data recorded at two institutions (UF and UC Davis) using the same willed attention paradigm. The findings include: (1) both instructional cues and the choice cue activated the DAN, (2) the choice cue additionally activated a frontoparietal decision network consisting of dorsal anterior cingulate cortex (dACC), anterior insula (AI), anterior prefrontal cortex (APFC), dorsal lateral prefrontal cortex (DLPFC), and inferior parietal lobule (IPL), (3) the decision about where to attend can be decoded in frontoparietal decision network in choice trials but not in instructional trials, and (4) EEG alpha oscillation patterns immediately preceding the choice cue, but not the instructional cues, predicted the postcue direction of attention and the frontoparietal decision network activity. Based on these findings we proposed a model of willed attention control suggesting how the direction of visual spatial attention was decided upon in the absence of external instructions.

neuroscience↗

bioRxiv2025

ER-to-Golgi transport machinery promotes the excessive cargo-triggered unfolded protein response

Disruptions to ER homeostasis trigger the unfolded protein response (UPR) to restore proteostatic balance. While defects in the secretory machinery are known to induce ER stress, it remains unclear whether specific trafficking components directly modulate UPR signaling. Here, we demonstrate that neuronal overexpression of the gap junction protein UNC-9 activates the IRE1-XBP1 arm of the UPR in C. elegans. Genetic deletion of ERGIC2 or ERGIC3--genes encoding COPII-associated proteins required for UNC-9 transport--suppresses this UPR activation, revealing an unexpected role for these factors beyond cargo trafficking. Mechanistically, ERGIC2 and ERGIC3 interact with the ER chaperone BiP, facilitating its release from IRE1 to enable UPR and alleviate cargo aggregation. Our findings redefine the UPR as a process dynamically regulated by early secretory components and provide novel insights into how cells integrate trafficking demand with stress adaptation, with implications for ER stress-associated diseases such as neurodegeneration.

cell biology↗

bioRxiv2025

Perturbational single-cell profiling of patient tumors defines lineage- and context-specific programs of innate immune resistance

AbstractDespite promise in preclinical models, most immuno-oncology drug candidates fail in clinical trials. These failures reflect limitations in our ability to directly model the response of human tumor and immune cells to immunotherapies. To address this gap and test the effect of innate immune agonists, we developed PERCEPT, an approach that uses ex vivo perturbational single-cell RNA sequencing to compare the response of immunomodulatory treatments with unstimulated controls directly in patient samples. Using PERCEPT, we tested cytokines and innate immune agonists in melanoma and Merkel cell carcinoma (MCC) and identified the dsRNA mimetic, RIG-I agonist, Stem Loop RNA (SLR) 14 as a powerful inducer of anti-viral states and enhancer of T cell activation. We compared transcriptional responder and non-responder patient samples and identified midkine (MDK), a multifunctional cytokine, as a potent repressor of IFN signaling in both tumor and immune cells. MDK expression dampened MHC-I presentation in human tumor cells and reduced activation of antigen-presenting cells, disrupting tumor immunity at multiple levels. In contrast to prior studies, we identified MDK as specifically enriched in neuroendocrine cancers such as MCC and small cell lung cancer compared with melanoma, suggesting the importance of lineage- and context-specific targeting. Our results demonstrate the utility of high-dimensional controlled perturbation of patient samples to identify mechanisms of innate immune response and resistance and demonstrate an actionable path towards clinical development of MDK-inhibiting therapies including FDA-approved ALK inhibitors in neuroendocrine cancers.

cancer biology↗

bioRxiv2025

The Rise and Fall of SARM1 Base-Exchange Inhibitors

The sterile alpha and TIR motif containing 1 (SARM1) enzyme is a key driver of axonal degeneration in response to injury, making it an attractive target for treating chemotherapy-induced peripheral neuropathy (CIPN) and other nervous system diseases. In this study, we identified and optimised a new class of base-exchange inhibitors (BEXi) targeting SARM1 and explored their molecular interactions and conformational effects using cryo-EM, HDX-MS and SAXS. Although BEXi produced robust inhibition across all biochemical and cellular assay formats, application at sub-inhibitory concentrations consistently led to paradoxical SARM1 activation, and in neuronal assays, accelerated neurite degeneration. Further analysis showed that BEXi only delayed, rather than prevented, neurite degeneration when applied to primary neuronal cells, even at exceedingly high inhibitor concentrations. These results prompted us to discontinue BEXi development in favour of alternative strategies, underscoring the complexity of SARM1 as a therapeutic target and the need for comprehensive, mechanistically informed screening cascades.

neuroscience↗

bioRxiv2025

Emotional Responses to Naturalistic and AI-generated Affective Pictures: A Systematic Comparison

Pictures depicting naturalistic scenes are widely used in studies of human emotion. However, the practical use of affective pictures is limited by several factors, including the difficulty of obtaining content-diverse, high-quality, openly accessible, and standardized stimuli that are necessary for specific research questions. The use of artificially generated (AI) pictures could address this limitation, but it is unclear if AI-generated pictures evoke reliable emotional responses. The present study sought to address these challenges by comparing emotional responses to AI-generated pictures with responses to original, standardized, pictures. In two study iterations, standardized pictures containing pleasant, neutral, and unpleasant content were selected from the International Affective Picture System (IAPS) and other sources. Then, a matched AI-counterpart was created for each original picture using generative deep neural networks. A total of 109 participants viewed the picture sets while pupil diameter and electroencephalogram (EEG) were recorded. Evaluative ratings of hedonic valence and emotional arousal were also collected. For both AI and original exemplars, pictures depicting emotional content elicited stronger responses than neutral content for ratings and EEG-derived variables, with weaker effect sizes for the AI-generated pictures. Furthermore, picture-level analyses found that ratings and EEG measures were strongly correlated between matched AI and original pictures. Pupil data also showed the expected content effects in study iteration 2, but not in iteration 1. Together, this initial study suggests that AI-generated picture sets can effectively elicit well-established self-reported affect and physiological responses, presenting a promising avenue for future studies of human emotion.

neuroscience↗

bioRxiv2025

Biologically Inspired Deep Neural Network Models for Visual Emotion Processing

The perception of opportunities and threats in complex visual scenes represents one of the main functions of the human visual system. The underlying neurophysiology is often studied by having observers view pictures varying in affective content. While deep neural networks (DNNs) have shown promise in modeling visual recognition of objects, their capacity to model visual affective processing remains to be better understood. In this study, we proposed a biologically inspired deep neural network model, referred to as the Visual Cortex Amygdala (VCA) model, that integrates a vision transformer with an amygdala-mimetic module designed to reflect both the anatomical hierarchy and self-attention-based computational mechanisms of the human brain. We evaluated the model along three dimensions: (1) predictive accuracy for emotional valence and arousal, (2) representational alignment with human amygdala activity, and (3)neural mechanisms of emotion representation. Our results showed that (1) the model can predict with high accuracy human emotion ratings on 1,182 images from the International Affective Picture System (IAPS) dataset (valence: r {approx} 0.9; arousal: r {approx} 0.7), (2) the models internal representations aligned with functional Magnetic Resonance Imaging (fMRI) data from the human amygdala, and (3) at the single neuron level, the amygdala module evolved emotion selectivity, and at the neuronal population level, the representational geometry became progressively more aligned with psychological models of emotion. Additional issues explored included (1) the effect of visual encoding and (2) the effect of structure and computational mechanisms on emotional assessment.

neuroscience↗