Consciousness remains incompletely explained at the anatomical, biophysical, and functional levels. Although existing theories characterize key functional correlates of conscious experience—including global broadcasting, information integration, attentional selection, working-memory-based awareness, and electromagnetic field dynamics—none has yet identified a specific cellular mechanism embedded within a defined long-range cortical pathway that could support the transmission of integrated sensory, interoceptive, emotional, and cognitive information toward anterior access and control systems. Here, the “Raslan Circuit” is defined as a candidate solution centered on the fronto-insular cortex (FI), the anterior cingulate cortex (ACC), and their von Economo neurons (VENs). Anatomically, VENs form a selectively localized population in FI and ACC, two regions already implicated in salience processing, interoception, emotional evaluation, and conscious awareness. Biophysically, within the Raslan Circuit hypothesis, VENs are proposed to act not merely as histological markers of higher cognition but as elongated, polarity-organizing neuronal elements with biological dipole antenna properties, supporting field-sensitive coupling and potentially resonance-based communication between FI and ACC. Functionally, FI is proposed to act as a deep multimodal collecting and organizing station, while ACC serves as a receiving and redistributive hub that links the transmitted signal to working memory, awareness, emotional valuation, and action selection. Together, this FI–ACC–VEN axis is advanced as a previously unrecognized communication infrastructure that could complement—rather than replace—Global Workspace Theory, Integrated Information Theory, embedded-processes and awareness-buffer models of working memory, salience-network accounts, and electromagnetic field theories of consciousness. The dimensional properties of VENs suggest a theoretically motivated, though speculative, prediction of resonance in the terahertz range, providing a falsifiable spectral hypothesis. More generally, the Raslan Circuit is grounded in four interlocking components: an anatomical foundation in VEN-rich FI and ACC, a biophysical rationale based on field-sensitive coupling, an explicit integration with existing functional theories of consciousness, and a set of testable predictions spanning connectivity, cellular and field measurements, clinical syndromes, and computational modeling.
The increase in mental health disorders in college populations necessitates novel assessment strategies that circumvent the limitations of existing self-report instruments. To address this issue, this paper presents a new deep learning framework for mental health monitoring in academia by integrating multimodal passive sensing data collected from smartphones. A cross-modal attention network, searched by a new metaheuristic algorithm the Addax Optimization Algorithm for neural architecture search, was trained and first evaluated on the StudentLife dataset. To further validate the results due to the extremely limited number of samples in the test set (N = 7), we subsequently performed a zero-shot transfer and fine-tuning evaluation on the College Experience Study (CES) dataset. CES is a large longitudinal dataset of over 200 students collected over 5 years including passively-collected sensor, survey and brain-imaging data. The validation of the model in 140 local participants resulted in classification accuracy for depression of 87.4% on StudentLife (with adjusted 95% CI 62–98%), suggesting considerable uncertainty. On CES the performance of the zero-shot transferred model reached 72.3% of classification accuracy and 83.9% when fine-tuned. This means the initial 87.4% result can be interpreted as overoptimistic. Modality weight analysis showed the importance of survey in predicting depression and the effect of activity on stress predictions. This paper serves as a proof-of-concept of a novel system for screening mental health disorders passively using mobile phones within academia, and suggests it may have a role to play in the early detection of risk.
IntroductionRespiration is increasingly recognized as a key modulator of brain-body interactions, influencing both neural and autonomic dynamics. While prior work has demonstrated respiratory-driven entrainment, it remains unclear whether distinct phases of the breathing cycle give rise to different dynamical regimes of cortical activity.MethodsIn this study, we investigated respiratory-phase-dependent modulation of brain-heart interactions during controlled breathing, focusing on breath-hold periods following inhalation and exhalation. EEG band power time series and heart rate (HR) were analyzed in 15 healthy subjects. In addition to conventional spectral measures, recurrence quantification analysis was employed to characterize non-linear dynamics in cortical activity.ResultsResults revealed clear respiratory-phase-dependent differences. Inhale-hold was associated with increased γ band power, higher recurrence rate, and elevated HR, whereas exhale-hold showed reduced γ activity, lower recurrence, and decreased HR. These findings are consistent with known mechanisms of respiratory sinus arrhythmia and suggest that the two conditions correspond to distinct autonomic states. From a dynamical systems perspective, increased recurrence during inhale-hold suggests more structured and constrained cortical dynamics, while reduced recurrence during exhale-hold reflects comparatively more variable activity.DiscussionTogether, these findings suggest that respiratory phase modulates the organization of brain-body dynamics, potentially reflecting shifts between more stable and more flexible regimes of neural activity. This phase-dependent organization supports the view that respiration may act as a physiological mechanism for dynamically tuning brain-heart interactions, with implications for understanding the role of breathing in the modulation of neural dynamics.
ObjectiveTo systematically evaluate the effects of exercise interventions on brain-derived neurotrophic factor (BDNF) in patients with stroke, and to examine the dose–response relationships associated with acute and long-term exercise.MethodsPubMed, Embase, Web of Science, the Cochrane Library, and APA PsycINFO were searched for studies of exercise interventions in patients with stroke that reported BDNF outcomes. Risk of bias was assessed using RoB 2.0 and ROBINS-I. A network dose–response meta-regression was conducted within a Bayesian random-effects framework.ResultsTwenty-three studies involving 903 participants were included, comprising 14 randomized controlled trials, 8 non-randomized studies, and 1 randomized crossover trial. Acute overall exercise showed a modest positive trend at lower dose levels; however, at a dose of 250 METs⋅min/week, the 95% credible interval included zero, indicating substantial uncertainty regarding whether a single bout of low-dose exercise produces a reliable increase in BDNF. For long-term overall exercise, posterior estimates generally favored a positive BDNF response as dose increased. The 95% credible interval first no longer included zero at 1300 METs⋅min/week, and estimates at 1700, 2000, 2300, and 2700 METs⋅min/week remained compatible with positive effects. However, these findings should be interpreted cautiously because the evidence base was limited and precision varied across dose levels. Clear differences were also observed across long-term exercise modalities. Multicomponent exercise showed the most favorable directional pattern across the 1700–2700 METs⋅min/week range, although several estimates were imprecise and the apparent advantage of this modality should be regarded as exploratory. By contrast, walking, cycling, and exercise combined with cognitive training all showed positive directional trends, but the 95% credible intervals included zero across dose levels, indicating substantial uncertainty regarding these modality-specific effects.ConclusionThe effects of exercise on BDNF in patients with stroke are jointly shaped by intervention duration, dose level, and exercise modality. Compared with acute exercise, long-term exercise at moderate-to-high doses, particularly multicomponent exercise, may be associated with a more favorable BDNF response. However, because the certainty of evidence was mostly moderate to low or very low and publication bias was detected, these findings should be considered preliminary and hypothesis-generating rather than definitive exercise prescription recommendations.Systematic review registrationhttps://www.crd.york.ac.uk/PROSPERO/view/CRD420261350793, identifier CRD420261350793.
This paper gives a roadmap for filling in a new integrated worldview, from physics to emergent phenomena, presented at Werbos (2010). That worldview starts from the assumption that the cosmos we live in is a dynamical system, which obeys what I call Hard Core Einsteinian realism (HCER), which can be approximated very well in our level of life by quantum electrodynamics as formulated by Everett, Wheeler and Deutsch (EWD) (Deutsch, 1997). In principle all of the patterns which emerge in such a cosmos are fully, precisely specified by knowing the equilibrium probability distribution (equivalent to the entropy function) for that dynamical system. These functions are known precisely for a wide class of EWD and HCER theories. Thus the understanding of life, mind and other emergent phenomena in our cosmos is equivalent to knowing the parameters of the underlying dynamical law (e.g., H, for the case of EWD), and knowing useful ways to approximate what that probability function looks like. As an example, present theories of Darwinian evolution are basically just crude approximation systems for specific types pof state or pattern or object which we usually call “life.”
Muscle synergies are traditionally viewed as stable, low-dimensional neuromuscular modules reflecting neural constraints. However, growing evidence suggests this mechanistic interpretation is incomplete. We propose a conceptual extension of the framework, arguing that coordination patterns reflected into multichannel EMG are emergent properties of an integrated cognitive-motor system rather than purely motor primitives. We identify three classes of non-motor factors that may shape motor output: (i) Task internalization and learning dynamics: synergy structure evolves with practice and the formation of internal models; (ii) affective and psychological states: emotional conditions (e.g., stress, anxiety) modulate muscle co-activation and stability; (iii) prior experience and sensorimotor memory: embodied history biases action selection and coordination. Consistent with predictive processing and embodied cognition, we argue these factors are constitutive dimensions of motor control rather than mere noise. This perspective implies that inter-subject variability in synergy structure reflects systematic differences in cognitive, affective, and experiential states. To enhance interpretability, we propose integrating minimal assessments into experimental designs: (i) task strategy evaluation, (ii) affective state characterization via physiological proxies or scales, (iii) documentation of prior motor experience, and (iv) analysis of learning trajectories. This integrative view complements existing models, providing a richer theoretical foundation to interpret variability, adaptability, and individual differences in human motor control.
In the hippocampus, slow waves are accompanied by brief population bursts of high-frequency oscillations (150–250 Hz) known as Sharp-Wave Ripples (SWRs), a phenomenon associated with memory consolidation during offline brain states and Non-Rapid Eye Movement (NREM) sleep. Despite the relevance of SWRs, no standardized criterion for their automatic detection has been established. This work introduces a consensus-based algorithm that first identifies sharp waves and then detects ripples occurring within these intervals. Events are designated as true SWRs only when at least two principal methodologies report overlapping detections. Comparative analyses showed that one detector generated more candidate events but with reduced precision, whereas the other was more selective but computationally slower. The consensus strategy improved reliability by emphasizing the concurrence of independent detectors, contributing to efforts toward standardized and reproducible SWR analysis. The algorithm was used within an alcohol administration model to quantify SWR rate, duration, and peak frequency across control, vehicle, and treated groups. Although no significant group-level differences emerged under the short-term exposure protocol, a significant increase in SWR peak frequency was observed in the treated group after the open field test, suggesting the presence of a transient compensation mechanism. These findings shed light on the brain's ability to adapt temporarily to specific behavioral tasks. However, it is essential to emphasize that additional research is crucial to fully understand the long-term implications and associations with alcohol-induced changes in brain structures and SWR generation.
The stress response is a non-specific adaptive physiological reaction that occurs when the organism is exposed to internal or external environmental stimuli, serving to maintain homeostatic balance. Its regulation involves highly complex neural mechanisms. The hippocampus, the bed nucleus of the stria terminalis (BNST), and the paraventricular hypothalamic nucleus (PVN) are key brain regions that play crucial roles in stress regulation. Focusing on the functional connectivity of the hippocampus-BNST-PVN neural circuit, this article systematically elucidates the multilevel effects of stress on synaptic transmission, synaptic plasticity, neural network integration, and related molecular mechanisms from microcosmic neurotransmitters to macroscopic neural circuits. These insights provide an important perspective for advancing our understanding of the neurobiological basis of stress and for developing new intervention strategies.
Systems neuroscience—from Lashley's distributed engrams through Pribram's field-based processing to Freeman's oscillatory dynamics—has long argued that intelligence is a whole-brain property requiring feedback-driven computation. We formalize this tradition using Reinforcement Learning and Approximate Dynamic Programming (RLADP) and propose that vertebrate intelligence falls into four qualitatively distinct levels—rodent, primate, human, cetacean—each defined by a different architecture for generating and propagating backpropagated feedback signals. The transition between levels is not parametric but architectural, and each architecture demands a different energy strategy. A conserved allometric rule for cortical ion channels holds across nine of 10 mammalian species, fixing the biophysical cost of computation per unit volume; human neurons uniquely violate this rule, reducing channel density to redirect energy toward long-range white matter connectivity. We show that white matter is an active communication system whose superlinear scaling creates a geometric cost trap, that the corticothalamic loop provides master timing for forward-backward cortical processing cycles, and that timing degradation causes qualitative intelligence failure. The biological strategies cataloged here—from selective connectivity reduction to cellular energy reallocation to cortical reorganization—have parallels with the communication-energy wall now constraining artificial intelligence.
Emotional memories are essential for survival, enabling individuals to avoid future threats or seek safety or food. How does the brain link a specific spatiotemporal context with the associated emotional experience? The hippocampus (HPC) encodes spatial information through place cells, which collectively form a cognitive map that adapts to contextual changes. Along its longitudinal axis, the HPC shows marked functional divergence: the dorsal HPC supports fine-tuned spatial representations, whereas the ventral HPC exhibits broader connectivity with limbic circuits and plays a key role in encoding emotional valence, salience, anxiety, and motivational state. Here, we synthesize current evidence comparing spatial and emotional coding across these regions, emphasizing differences in spatial properties and circuit interactions. Finally, we propose a framework describing how dorsal and ventral HPC integrate spatial and emotional information to guide adaptive behavior.
Ketamine responses vary across patients and sessions, suggesting that dose alone is an incomplete organizing principle. We propose a state-first Gate-Amplifier-Reintegration framework in which awake low-dose ketamine acts primarily as an Amplifier of transient network flexibility, whereas autonomic-salience stability is treated as a candidate Gate that may shape whether this flexibility remains steerable. In this framework, cardio-autonomic and interoceptive state may constrain or modulate salience-network gain, interoceptive precision, and thalamocortical selectivity, thereby influencing whether ketamine-associated loosening of default-mode constraints is available for frontoparietal-control-compatible reintegration or drifts toward dysphoric dissociation and vigilance instability. We formalize a three-step sequence: Gate, Amplifier, Reintegration. Gate refers to candidate autonomic-salience stability; Amplifier refers to awake low-dose ketamine delivered under operational invariants that preserve vigilance and behavioral interpretability; Reintegration refers to the organization of ketamine-amplified flexibility into language, joint attention, task context, and action-oriented consolidation. Heart-rate variability (HRV) is used only as a bounded peripheral state-verification proxy. We distinguish observed Autonomic Affirmative Window quality assurance (AAW-QA), a post-sequence quality-assurance signal, from AAW-Gate, a proposed prospective pre-dose criterion. The framework is informed by, but not validated by, observations from a single-center outpatient chronic pain care pathway using awake low-dose ketamine and route-defined cervicothoracic sympathetic modulation. These observations document clinical provenance but do not provide comparative efficacy evidence, causal efficacy, dose-sparing evidence, salience-network mediation, or HRV biomarker validity. The clinical provenance was nonrandomized, chart-based, clinician-directed, and lacked concurrent neural measurement. The framework yields falsifiable predictions: prospective Gate manipulation should reproducibly alter pre-dose autonomic state; prospectively defined AAW-Gate-positive sessions should be tested for convergence with low-burden EEG/fNIRS markers of salience switching and task control; and randomized Gate designs should determine whether autonomic shifts moderate, and in adequately powered designs mediate, session-level tolerability, reintegration, and usable clinical change. Alternative accounts, including analgesia, expectancy, clinician attention, workflow era, documentation bias, respiratory/postural effects, and photoplethysmography (PPG) artifact, are treated as competing explanations that future designs must separate.
Emotion-related states encompass a wide range of affective and behavioral dimensions, including anxiety, fear, pleasure, and aversion, whereas motivation refers to internal drives that energize goal-directed behaviors such as feeding, reward seeking, or avoidance. These states are thought to arise from the brain’s integration of exteroceptive inputs, interoceptive signals reflecting bodily states, and prior experiences. Recent studies have increasingly highlighted the critical role of gut-to-brain communication, particularly vagal and other interoceptive pathways, in shaping emotion-related and motivational behaviors. In parallel, advances in genetic manipulation techniques have enabled the identification of neuronal circuits innervating the gastrointestinal tract and have provided new insights into their links with affective-like behavior. A better understanding of these mechanisms will be essential for elucidating how gastrointestinal-derived signals contribute to emotion-related states and how they are represented and processed in the brain. In this review, we discuss recent advances in gut-to-brain neural pathways, with a particular focus on vagal afferent and interoceptive mechanisms, while also considering spinal afferent pathways and ENS-related mechanisms that may contribute to emotion-related behavioral regulation.
IntroductionVolatile anesthetics, such as isoflurane, generate a state of unconsciousness and analgesia across the animal kingdom and are widely used in clinical settings. Yet, anesthetic mechanisms are poorly understood: the volatile anesthetics are so profligate in their potential effects that it has proven difficult to determine which actions are most causal at the systems level.MethodsTo test if specific cellular targets mediate the anesthetic effect across a complete, intact nervous system, we imaged neuron activity in the Caenorhabditis elegans head ganglia at cellular resolution. We measured the effect of increasing anesthetic concentrations across a range of identified neurons within the C. elegans nervous system.ResultsHowever, rather than dramatic effects on any particular neuronal class, we measured uniform suppression of both neuron activity and connectivity with increasing isoflurane across the nervous system. We find the degree of activity suppression to be proportional to the baseline activity of the neuron in the awake state. Within this context, highly connected neurons, specifically neurons with high in-degree connectivity, are inherently active and display large activity suppression. These include hub interneurons within the C. elegans command locomotory circuit that control behavioral crawling states and contribute to system-wide coherence of neuron dynamics. By analyzing the effect of isoflurane on the activity of two specific hub interneuron classes, AVA and AVE, we show that the large degree of suppression observed in these neurons corresponds to high baseline activity.DiscussionExploiting the small size, simplicity and optical accessibility of C. elegans, our results demonstrate that isoflurane anesthesia globally suppresses activity and connectivity across a wide range of neuron types, and suggest a model of anesthesia in which proportional suppression of activity results in disruption of highly connected, highly active, hub loci that are critical to nervous system coordination and state dynamics.
ObjectiveThe aim of this study is to investigate whether treadmill exercise alleviates motor dysfunction in a mouse model of Parkinson’s disease (PD) by modulating the excitability of striatal medium spiny neurons expressing dopamine type 2 receptors (D2-MSNs).MethodsA unilateral 6-hydroxydopamine (6-OHDA) injection was performed in the right striatum of D2-Cre mice to establish a hemi-lesioned PD model, with sham-operated mice serving as controls. PD mice were subjected to treadmill exercise (18 m/min, 40 min/day, 5 days/week for 4 weeks). Motor function was evaluated using the open-field test, rotarod, and negative geotaxis test. The excitability of D2-MSNs was assessed via fiber-photometric calcium imaging (ΔF/F, AUC, and peak amplitude), Western blotting for c-Fos protein expression, and double immunofluorescence labeling of D2R and c-Fos. Furthermore, chemogenetic approaches (hM4Di-Gi for inhibition and hM3Dq-Gq for activation) were employed to validate the causal role of D2-MSN excitability in exercise-mediated motor recovery.ResultsPD mice exhibited significant motor deficits, characterized by reduced locomotor activity, shortened latency to fall on the rotarod, and increased turning latency in the negative geotaxis test. Calcium imaging and c-Fos expression analyses revealed a marked hyperexcitability of striatal D2-MSNs in PD mice compared to controls (p < 0.01). Treadmill exercise significantly attenuated this D2-MSN hyperexcitability and concurrently improved all motor performance metrics (p < 0.01). Chemogenetic inhibition of D2-MSNs mimicked the beneficial effects of exercise in PD mice, whereas chemogenetic activation of these neurons abolished the exercise-induced motor improvements and reversed the reduction in neuronal excitability (p < 0.01).ConclusionOur findings demonstrate that striatal D2-MSN hyperexcitability is a critical pathological feature of motor dysfunction in PD mice. Treadmill exercise rescues motor deficits by suppressing this hyperexcitability. These results provide novel insights into the neurobiological mechanisms underlying the therapeutic benefits of physical exercise in Parkinson’s disease.
BackgroundThis paper addresses a critical challenge in developing practical EEG-based brain-computer interfaces (BCIs): enhancing cross-subject generalization by mitigating individual differences in brain signals. How can we effectively leverage data from existing subjects to improve performance for a new user with minimal subject-specific calibration?MethodsWe systematically compare and optimize three prominent data alignment techniques, Riemannian Procrustes Analysis (RPA), Euclidean Alignment (EA), and Correlation Alignment (CORAL), designed to transform EEG data from multiple source subjects and a target subject into a common representation space, mitigating variability.EvaluationWe employed leave-one-subject-out cross-validation (LOSO-CV) framework on EEG-based attention decoding data to empirically evaluate the effectiveness of each alignment method compared to a baseline condition with no alignment. Key parameters, specifically the regularization parameter α for EA, were optimized to maximize cross-subject transfer performance.ResultsThe study demonstrates that alignment methods improve classification accuracy compared to the baseline. Notably, EA evaluated at α = 100 the scaling value at which the largest fraction of subjects attained their best accuracy in our parameter sweep yielded the largest mean improvement, increasing classification accuracy by 3.44% over the no alignment baseline (paired t(17)≈2.48, p≈0.024; Cohen's dz≈0.59; 95% confidence interval for the mean improvement [0.52%, 6.36%]). Because this α value was identified from the same sweep that produced the per-subject accuracies, this estimate together with the per-subject “best-parameter” results should be interpreted as an oracle sensitivity-analysis upper bound on subject-specific tuning rather than as a leakage-free LOSO estimate. While optimized EA showed the best mean performance, the analysis also demonstrated subject-specific differences in the most ideal alignment strategy.ConclusionThis comparison framework quantifies the benefits of different alignment approaches and highlights the valuable contribution of parameter optimization, particularly for EA.SignificanceThese results indicate the potential of optimized alignment techniques, EA in particular, to significantly enhance cross-subject transfer learning in EEG-based BCIs. This has practical ramifications for methodology selection and tuning, and maps a path toward more robust and generalizable BCI systems requiring less subject-specific calibration for real-world applications.
Introduction:Affective disorders (ADs) are characterized by profound emotional processing deficits involving disrupted neural network activity and connectivity, particularly within the default mode network and fronto-temporal circuits, with abnormalities in theta and alpha oscillatory patterns. While current treatments primarily target mood symptoms, emotional processing impairments often persist and predict relapses. Awe, a complex self-transcendent emotion, may counteract such deficits through its capacity to reduce rumination and enhance positive affect. However, the neural correlates of awe experiences in clinical populations remain unexplored. Objective:For the first time, this exploratory study investigated the electroencephalographic (EEG) correlates of awe induced by validated virtual reality (VR) scenarios in individuals with ADs compared to healthy controls (HCs). Methods:Participants were exposed to immersive VR scenarios designed to elicit different awe experiences (mountains, waterfall, Earth) and a reference (awe-neutral) scenario. EEG activity was recorded during VR exposure and at baseline, followed by emotional state questionnaires. Power spectral density and graph-theoretical connectivity indices - Nodal Positive Strength and Global Efficiency - were computed across theta, alpha, and beta bands. Results:Healthy controls showed high awe responses in awe-inducing scenarios with selective, scenario-specific modulations in alpha and theta band activity and connectivity, reflecting preserved cognitive flexibility. Conversely, ADs reported similar awe responses across all VR scenarios with reduced environmental differentiation. With respect to HCs, ADs showed elevated theta power in bilateral frontal and temporal regions, suggesting compensatory activity related to emotional processing alterations. Both groups exhibited VR-induced reductions in alpha-band global efficiency, more pronounced in ADs, suggesting compromised neural integration during complex emotional processing. Discussion:Taken together, the results suggest that the emotional processing deficits inherent to ADs may limit the capacity to engage differentially with emotionally complex stimuli such as awe, while nonetheless providing initial evidence that VR-based awe exposure combined with neurophysiological recording represents a valuable approach for discriminating differential cerebral emotional responses in clinical populations. This proof-of-concept work warrants further investigation in larger cohorts to evaluate the therapeutic potential of awe-based interventions for affective disorders.
IntroductionUnderstanding how artificial neural networks (ANNs) can capture biologically meaningful dynamics is a central challenge in systems neuroscience. In this work, we investigate whether spiking neural networks (SNNs) can function not only as machine-learning tools but also as biologically inspired computational analogs and tractable testbeds for studying pathological neural dynamics.MethodsWe implemented a spiking autoencoder composed of Leaky Integrate-and-Fire and Synaptic neuron models to create a controlled framework for analyzing how biologically related parametric changes to neuronal and synaptic dynamics influence learning and information transfer. By tuning model parameters to induce persistent overfiring-like behavior, we emulated a hyperexcitability-like regime conceptually analogous to NaV channel dysfunction in hippocampal circuits. Reconstruction performance and network activity were evaluated under both noiseless and noisy conditions.ResultsThe induced hyperexcitability-like regime degraded image reconstruction performance and disrupted stable information propagation, consistent with impaired processing in hyperexcitable neural systems. Layer-wise firing-rate analysis revealed that the altered regime was characterized by unstable activity redistribution rather than sustained global overactivation. Importantly, introducing controlled Gaussian noise into the input stream partially restored reconstruction quality and improved learning performance, suggesting that stochastic perturbations can partially compensate for instability in dysfunctional network regimes.DiscussionThese findings demonstrate that specific SNN parameter regimes can reproduce key signatures of pathological excitability while also providing a platform for investigating compensatory mechanisms. Overall, this work positions spiking autoencoders as scalable, biologically grounded frameworks for hypothesis-driven studies of neural dysfunction and candidate interventions, supporting the integration of ANN methodologies with mechanistic models in systems neuroscience.
BackgroundHuman conversation involves moment-to-moment reciprocal adjustments between interlocutors, expressed through both emotional cues and autonomic physiology.ObjectivesTo quantify how physiological synchrony continuously builds and subsides between debate partners during speaker-listener turn-taking, and to test whether the direction of this coupling (speaker-leading vs listener-leading) is associated with (i) self- versus partner-perceived arousal/valence and (ii) autonomic and complexity-based heart rate variability (HRV) characteristics.MethodsMultimodal data from the K-EmoCon database were analyzed, comprising HRV-derived cardiac activity, speech timing, and multi-perspective emotion ratings from 32 individuals engaged in a structured dyadic debate. Interactions were segmented into speaking and listening phases, and a phase-based bidirectional coupling framework was applied to quantify both the strength and polarity of physiological synchrony. Associations between emotional states and HRV features were examined using correlation analysis across coupling segments, followed by principal component analysis (PCA), to reduce dimensionality and cluster emotions and features based on their shared variance.ResultsPositive coupling segments, corresponding to speaker-leading dynamics, were characterized by strong associations between partner-related emotional states and parasympathetic HRV indices, including RMSSD and SD1, with correlations reaching up to 0.63 (p < 0.001). In contrast, negative coupling segments, reflecting listener-leading dynamics, showed stronger associations with sample entropy, Rényi entropy, and low-frequency power, with correlations reaching 0.71 (p < 0.001). Diffusion entropy exhibited a polarity-dependent pattern consisting of positively correlated self-reported emotions during positive coupling, whereas during negative coupling it was negatively correlated with partner-related emotions, with correlations reaching 0.71 (p < 0.001) for the complexity index μr at scale 1. PCA showed that positive coupling was characterized by a clear separation of arousal, with self-related emotions aligning with diffusion entropy features and partner-related emotions clustering with HRV and Rényi entropy measures. In contrast, negative coupling exhibited a pattern in which partner-related emotions formed more compact clusters across power- and entropy-based features.ConclusionsThese findings demonstrate that bidirectional physiological coupling provides a sensitive framework for disentangling leadership, responsiveness, and emotional exchange during conversation. By revealing distinct autonomic and complexity-based signatures of self- and partner-related affect, this work advances understanding of interpersonal emotional regulation. It has implications for therapeutic, educational, and collaborative communication contexts.
