bioRxiv Subject Collection: Neuroscience's Journal

A brain-inspired algorithm enhances automatic speech recognitionperformance in multi-talker scenes
Modern automatic speech recognition (ASR) systems are capable of impressive performance recognizing clean speech but struggle in noisy, multi-talker environments, commonly referred to as the "cocktail party problem." In contrast, many human listeners can solve this problem, suggesting the existence of a solution in the brain. Here we present a novel approach that uses a brain inspired sound segregation algorithm (BOSSA) as a preprocessing step for a state-of-the-art ASR system (Whisper). We evaluated BOSSA's impact on ASR accuracy in a spatialized multi-talker scene with one target speaker and two competing maskers, varying the difficulty of the task by changing the target-to-masker ratio. We found that median word error rate improved by up to 54% when the target-to-masker ratio was low. Our results indicate that brain-inspired algorithms have the potential to considerably enhance ASR accuracy in challenging multi-talker scenarios without the need for retraining or fine-tuning existing state-of-the-art ASR systems.

Adult organotypic brain slice cultures recapitulate extracellular matrix remodelling in haemorrhagic stroke
Haemorrhagic stroke is a devastating condition characterised by vessel rupture and free blood within the brain parenchyma or cerebrospinal fluid (CSF) filled spaces. Across the major subtypes of haemorrhagic stroke (subarachnoid, intracerebral, and intraventricular haemorrhages), the presence of blood in the CSF generates significant tissue damage in the first 72 hours after the event, known as early brain injury (EBI). EBI includes neuroinflammation, blood-brain barrier breakdown and dysregulation of extracellular matrix (ECM) dynamics. ECM dysfunction has been shown to trigger fibrosis of the cortical blood vessels, limiting normal CSF circulation and resulting in the buildup of metabolic waste or the development of post-haemorrhagic hydrocephalus. Limiting or preventing this fibrosis may therefore reduce the rate of morbidity experienced by survivors, providing a potential avenue for non-surgical treatment to reduce secondary brain injury post-stroke. Despite this, current in vivo approaches fail to differentiate between the effect of blood products and secondary consequences including intracranial pressure (ICP) elevation and mass effect. Here, we describe an adult rat organotypic brain slice culture (OBSC) model of haemorrhagic stroke which enables the identification of the effect of blood products on ECM dysregulation. We demonstrate the distribution of key cell types across a time course of 0, 3 and 7 days in culture, indicating that such cultures are viable for a minimum of 7 days. Using immunofluorescence staining, Western blotting and RNA sequencing, we show that exposure of OBSCs to lysed blood markedly increases ECM deposition around cortical blood vessels. This is accompanied by dysregulation of ECM regulatory genes and upregulation of inflammation and oxidative stress-related genes, successfully recapitulating the changes seen in human stroke survivors. This versatile ex vivo model provides a translational platform to further understanding of haemorrhagic stroke pathophysiology and develop or trial novel therapeutics prior to progression to in vivo stroke studies.

When one race is not enough: a relay model explains multisensory response times
Humans typically respond faster to multisensory signals than to their unisensory components, a phenomenon known as the redundant signal effect (RSE). One of the earliest and most influential accounts, the race model, attributes the RSE to statistical facilitation, which arises from parallel, independent processing across sensory modalities. While this model captures some key features of the RSE, it frequently underestimates the observed speed-up leading to violations of the race model inequality (RMI), a benchmark used to test the models validity. To reconcile this discrepancy, we introduce the relay model, a minimal extension of the race architecture that incorporates cross-modal initiation. In this model, responses result from two sequential race processes, allowing a signal in one modality to trigger the onset of perceptual decision processing in another. This structure retains statistical facilitation as a core principle while introducing a single free model parameter that divides unisensory processing into gating and decision stages. Through simulations and fits to foundational empirical datasets, we show that the relay model captures both the magnitude and distributional shape of the RSE, including RMI violations. It also accounts for changes in the RSE under asynchronous stimulus onsets, a critical test in multisensory integration research. By extending the classical race model with minimal added complexity, the relay model offers a mechanistically explicit and biologically plausible framework for explaining the dynamics of multisensory decision-making.

"Backpropagation and the brain" realized in cortical error neuron microcircuits
Neural responses to mismatches between expected and actual stimuli have been widely reported across different species. How does the brain use such error signals for learning? While global error signals can be useful, their ability to learn complex computation at the scale observed in the brain is lacking. In comparison, more local, neuron-specific error signals enable superior performance, but their computation and propagation remain unclear. Motivated by the breakthrough of deep learning, this has inspired the "backpropagation and the brain" hypothesis, i.e. that the brain implements a form of the error backpropagation algorithm. In this work, we introduce a biologically motivated, multi-area cortical microcircuit model, implementing error backpropagation under consideration of recent physiological evidence. We model populations of cortical pyramidal cells acting as representation and error neurons, with bio-plausible local and inter-area connectivity, guided by experimental observations of connectivity of the primate visual cortex. In our model, all information transfer is biologically motivated, inference and learning occur without phases, and network dynamics demonstrably approximate those of error backpropagation. We show the capabilities of our model on a wide range of benchmarks, and compare to other models, such as dendritic hierarchical predictive coding. In particular, our model addresses shortcomings of other theories in terms of scalability to many cortical areas. Finally, we make concrete predictions, which differentiate it from other theories, and which can be tested in experiment.

Expectation effects on repetition suppression in nociception
Repetition suppression, the reduced neural response upon repeated presentation of a stimulus, can be explained by models focussing on bottom-up (i.e. adaptation) or top-down (i.e. expectation) mechanisms. Predictive coding models fall into the latter category and propose that repetitions are expected and therefore elicit smaller prediction error responses. While studies in the visual and auditory domain provide some support for such models, in nociception evidence remains inconclusive, despite the substantial influence expectations exert on pain perception. To assess expectation effects on repetition suppression in nociception, we developed a paradigm in which healthy volunteers received brief CO2 laser stimuli, while we acquired electroencephalographic (EEG) and peripheral physiological data. Importantly, laser stimuli could be either repeated after one second or not be repeated, with the probability of repetitions manipulated in a block-wise fashion, such that repetitions were either expected or unexpected. We observed repetition suppression in laser-evoked potentials as well as laser-induced gamma band oscillations, but not in laser-induced desynchronisations in the alpha and beta band. Critically, neither these EEG responses, nor the peripheral physiological data showed significant differences between the expectation conditions, with Bayesian analyses mostly providing evidence for an absence of effects. This indicates that repetition suppression to brief nociceptive laser stimuli is not driven by top-down factors, but rather mediated by other adaptation processes. While this does not preclude an influence of predictive coding models in nociception, it suggests that when the nervous system receives highly precise input, its responses are less susceptible to influence from expectations.

Amyloid beta aggregation promoted by iron leads to neuronal loss in an ex vivo model of Alzheimer's disease
Alzheimer's disease (AD) is a progressive neurodegenerative disorder characterized by amyloid beta (A{beta}) plaques and neurofibrillary tangles. Despite well-established iron accumulation in the AD brain, its role in exacerbating A{beta} toxicity is often overlooked in therapeutic research. We developed a 3D ex vivo organotypic brain slice cultures (OBSC) with A{beta} monomers and ferric citrate to mimic A{beta} deposits and iron overload to investigate the impact of excess iron on A{beta} toxicity in pig and human brains. Light and electron microscopy, biochemical assays, and multiple regression modeling were employed to assess iron-mediated A{beta} toxicity in neurons and glial cells. We show that OBSC offer a close approximation of in vivo morphological and physiological properties and can retain both neurons and glial cells for extended periods, and respond to experimental manipulations. We show that iron promotes A{beta} fibrillization into long fibrils, with this process further influenced by temperature. A{beta} selectively accumulated in neurons, leading to their death, sparing glial cells. In contrast, Iron, though generally toxic to neurons, exhibited unspecific cytotoxicity. Notably, the combined presence of A{beta} and iron synergistically increased neuronal death while reducing glial cell loss. Correlation analysis revealed that this synergic interaction enhances the toxicity of each other in a mutual fashion - A{beta} directs the neuronal toxicity while iron promotes A{beta} fibrillization, leading to targeted neuronal loss. In conclusion, our findings emphasize the critical role of excess iron and A{beta} in driving neuronal death in AD, underlining the importance of targeting iron accumulation along with A{beta} clearance but also addressing in future AD therapies, while also supporting our OBSC model as a valuable platform for studying the same.

CK1δ-Dependent SNAPIN Dysregulation Drives Lysosomal Failure in HIV-1 Vpr-Exposed Neurons: A Targetable Mechanism in HAND
HIV-associated neurocognitive disorders (HAND) persist in nearly 40% of virally suppressed individuals despite antiretroviral therapy (ART). Lysosomal dysfunction has emerged as a key contributor to HAND pathogenesis, yet the molecular mechanisms linking chronic HIV exposure to impaired neuronal degradation remain incompletely defined. Here, we identify HIV-1 Viral Protein R (Vpr) as a driver of lysosomal acidification failure, clustering, and degradative impairment in neurons. We uncovered casein kinase 1 delta (CK1{delta}) as a central mediator of this dysfunction, acting via phosphorylation of the adaptor protein SNAPIN. Vpr-induced CK1{delta} activation leads to hyperphosphorylation of SNAPIN, disrupting lysosomal positioning and motility. These defects are rescued by selective CK1{delta} inhibition, which restores lysosomal acidification, positioning, and mitophagy. Our findings define a novel Vpr-CK1{delta}-SNAPIN axis contributing to HAND and highlight lysosomal transport as a targetable mechanism in neurodegeneration.

Predicting Developmental Norms from Baseline Cortical Thickness in Longitudinal Studies
Normative models have gained popularity in computational psychiatry for studying individual-level differences relative to population norms in biological data such as brain imaging, where measures like cortical thickness are typically predicted from variables such as age and sex. Nearly all published models to date are based on cross-sectional data, limiting their ability to predict longitudinal change. Here, we used longitudinal brain data from the Adolescent Brain Cognitive Development (ABCD) study, comprising cortical thickness measures from 180 regions per hemisphere in youths at baseline (N=6179; 47% females), 2-year (N=6179; 47% females), and 4-year (N=805; 45% females) follow-up. A training set was established from baseline and 2-year follow-up data (N=5374; 47% females), while data from individuals with all three time points available served as an independent test set (N=805; 45% females). We developed sex-specific Baseline-Integrated Norms (B-Norms) that predict brain region thickness at follow-up based on baseline thickness, baseline age, and follow-up age, and compared them to sex-specific standard Cross-Sectional Norms (C-Norms) based on age alone. Out-of-sample testing in 2-year and 4-year follow-up data showed that B-Norms consistently provided better fits than C-Norms for nearly all cortical regions. Explained variance was higher in B-Norms than in C-Norms. We found no significant differences between time points (p = 0.45). Repeated measures ANOVA revealed differences in higher-order moments (e.g., skewness and kurtosis) for both models; for example, skewness varied by model, sex, time point, and their interactions. While improved fit alone does not necessarily indicate a superior normative model - since normative models aim to capture population variance rather than simply optimize fit - we demonstrated that four regions were associated with pubertal changes in B-Norms but not in C-Norms, suggesting enhanced sensitivity of B-Norms to developmental processes. Together, our findings highlight the potential of B-Norms for capturing normative variation in longitudinal structural brain change.

Human cortex organizes dynamic co-fluctuations along sensation-association axis
The human brain dynamically organizes its activity through coordinated fluctuations, whose spatiotemporal interactions form the foundation of functional networks. While large-scale co-fluctuations are well-studied, the principles governing their amplitude-dependent transitions--particularly across high, intermediate, and low-amplitude regime--remain unknown. We introduce a co-fluctuation score to quantify how instantaneous functional interactions reorganize with global amplitude dynamics. Using resting-state fMRI data, we identified amplitude-dependent co-fluctuation transitions between functional systems hierarchically aligned with the sensorimotor-association (SA) axis: sensorimotor networks dominated high-amplitude co-fluctuations, associative systems prevailed during intermediate amplitudes, and limbic system preferentially engaged in low-amplitude states. This hierarchy underwent developmental refinement from childhood to adulthood and adaptively reconfigured under external stimuli. Replicated across four independent samples including 7T fMRI, these findings establish the SA axis as infrastructure for amplitude-dependent transitions of co-fluctuation states. Our framework bridges transient coordination and stable functional architecture, demonstrating how brain networks balance external processing (high-amplitude states) with internal cognition/emotion (mid-to-low amplitude states) through amplitude-stratified interactions.

High-Resolution EEG Source Reconstruction from PCA-Corrected BEM-FMM Reciprocal Basis Funcions: A Study with Visual Evoked Potentials from Intermittent Photic Stimulation
Modern automated human head segmentations can generate high-resolution computational meshes involving many non-nested tissues. However, most source reconstruction software is limited to 3-4 nested layers of low resolution and a small number of dipolar sources ~10,000. Recently, we introduced modeling techniques for source econstruction of magnetoencephalographic (MEG) signals using the reciprocal approach and the boundary element fast multipole method (BEM-FMM). The technique of BEM-FMM can process both nested and non-nested models with as many as 4 million surface elements. In this paper, we present an analogue technique for source reconstruction of electroencephalographic (EEG) signals based on cortical global basis functions. The present work uses Helmholtz reciprocity to relate the reciprocally-generated lead-field matrices to their direct counterpart, while resolving the issue of possible biases toward the reference electrode. Our methodology is tested with experimental EEG data collected from a cohort of 12, young and healthy, volunteers subjected to intermittent photic stimulation (IPS). Our novel high-resolution source reconstruction models can have impact on mental health screening as well as brain-computer inter- faces.

Bounds on the computational complexity of neurons due to dendritic morphology
The simple linear threshold units used in many artificial neural networks have a limited computational capacity. Famously, a single unit cannot handle non-linearly separable problems like XOR. In contrast, real neurons exhibit complex morphologies as well as active dendritic integration, suggesting that their computational capacities outperform those of simple linear units. Considering specific families of Boolean functions, we empirically examine the computational limits of single units that incorporate more complex dendritic structures. For random Boolean functions, we show that there is a phase transition in learnability as a function of the input dimension, with most random functions below a certain critical dimension being learnable and those above not. This critical dimension is best predicted by the overall size of the dendritic arbor. This demonstrates that real neurons have a far higher computational complexity than is usually considered in neural models, whether in machine learning or computational neuroscience. Furthermore, using architectures that are, respectively, more 'apical' or 'basal', we show that there are non-trivially disjoint sets of learnable functions by each type of neuron. Importantly, these two types of architectures differ in the robustness and generality of the computations they can perform. The basal-like architecture shows a higher probability of function realization, while the apical-like architecture shows an advantage with fast retraining for different functions. Given the cell-type specificity of morphological characteristics, these results suggest both that different components of the dendritic arbor as well as distinct cell types may have distinct computational roles. Our analysis offers new directions for neuron-level inductive biases in NeuroAI models using scalable models for neuronal cell-type specific computation

Diet context gates AgRP neuron involvement in Semaglutide-induced weight loss
Semaglutide, a GLP-1R agonist, is widely used for obesity and type 2 diabetes, but its neural mechanisms remain unclear. AgRP neurons regulate energy balance, yet their role in the effects of Semaglutide is unknown. We show that sustained treatment of female mice with Semaglutide leads to activation rather than inhibition of AgRP neurons. Ablation or hypofunction of AgRP neurons through cell-specific knockout of Sirt1 reduces Semaglutide-induced weight loss and impairs its hypoglycemic effects in female mice under Standard Diet. However, acute or chronic exposure to High-Fat Diet makes AgRP neurons dispensable for weight loss, suggesting that neural substrates for the actions of Semaglutide depends on dietary composition. Re-exposure to Standard Diet recovers the necessity for AgRP neurons, underscoring the influence of nutritional status on GLP-1R pathways. Our findings show the necessity for AgRP neurons in sustaining Semaglutide-induced weight loss in female mice on standard diet in vivo.

Bmal1 expression is minimal or absent in human and mouse cerebral microglia
Microglia orchestrate immunological responses in the brain and play an important role in maintaining homeostatic brain functions. Several studies have reported clock gene expression in microglia and the circadian rhythm they drive has been linked to the modulation of immune responses and neuronal functions. In the current study, complementary approaches, including immunofluorescence, multiplexed fluorescence in situ hybridization, and liquid chromatography-mass spectrometry proteomics of isolated CD11b+ microglia, were combined with publicly available transcriptomic and epigenomic datasets to investigate the expression of the core clock gene BMAL1 in human post-mortem cortical and limbic areas as well as mouse brain. The majority of BMAL1-expressing cells were found to be neurons, with microglia representing a negligeable proportion. We also identified significantly lower chromatin accessibility or ''openness'' for BMAL1 gene regulatory regions (such as promoters and enhancers) in microglia compared to neurons. These regulatory regions in microglia were enriched for ETS domain transcription factor (TF) binding sites. Together, this suggests a strong role of chromatin remodeling factors in suppressing BMAL1 gene expression in microglia. Finally, while we observed a very low expression, BMAL1 TF motifs were accessible in open chromatin landscape of microglia, which may lead to downstream gene-regulatory effects upon binding, even if BMAL1 expression is constitutively low. Overall, our results reveal low or absent expression of BMAL1 in microglia and point towards potential epigenetic mechanisms regulating its expression in these cells.

New Synapse Detection in the Whole-Brain Connectome of Drosophila
The FlyWire Drosophila brain connectome is a graph of roughly 140K neurons and >50 million synaptic connections reconstructed from the FAFB EM dataset4. Challenges in synapse detection were identified for neurons with features such as dark cytosols, axo-axonic synapses, and complex polyadic synapses, due to limitations in ground truth data for these cells and the inherent complexity of these synapse types. To address these issues, we trained new neural networks using iteratively generated ground truth annotations and detected synapses across the entire FAFB dataset, producing what we refer to here as the Princeton synapses. These synapses were evaluated in both control regions, such as subareas of the mushroom body calyx and lateral horn, which were also chosen by Buhmann et al.3 for evaluation, as well as challenging regions, including Johnston Organ neurons (JONs), photoreceptors, and other cell types. The new model shows significant improvements, achieving up to a 0.23 F-score increase in challenging areas, while maintaining performance in control regions. Princeton synapses also show an 8-9% improvement in neuron clustering within cell types and better left/right symmetry scores, especially for photoreceptors. Additionally, neuron type membership can be predicted from connectivity patterns alone with weighted F-scores of 0.93 for Princeton synapses versus 0.91 for Buhmann synapses. The updated Princeton synapses are now accessible via Codex (codex.flywire.ai).

Developmental Stage-Dependent Transcriptomic Responses to Neonatal Intraventricular Hemorrhage
Neonatal intraventricular hemorrhage (IVH) is a major complication of preterm birth, yet how developmental stage influences the brains response to injury remains unclear. We performed single-nucleus RNA sequencing on rat brains 24 hours after IVH at postnatal day 2 (PND2) or day 5 (PND5) to define transcriptional responses across cell types. We identified 42 distinct cell populations and found that PND5 brains exhibited a markedly stronger immune and inflammatory response to IVH, with a threefold increase in differentially expressed genes compared to PND2. Microglia were the most perturbed cell type at both stages, showing increased oxidative stress and polarization toward both pro- and anti-inflammatory phenotypes at PND5. Ligand-receptor and regulon analysis revealed a shift from reparative IGF2 and TGF-beta; signaling at PND2 to proinflammatory Wnt signaling and activation of Runx1 and Stat5 at PND5. These findings highlight the importance of developmental timing in shaping the neuroimmune response to IVH and identify potential stage-specific therapeutic targets.

Motor learning drives region-specific transcriptomic remodeling in the motor cortex and dorsal striatum
Motor learning depends on coordinated activity across the motor cortex (M1) and dorsal striatum (dSTR), yet the molecular mechanisms driving learning-related synaptic and circuit remodeling remain unclear. Here, we combine activity-dependent genetic labeling (TRAP) with single-cell RNA sequencing to generate an unbiased, cell type-resolved transcriptional atlas of behaviorally engaged populations during a forelimb reaching task. We identify diverse activated neurons across M1 and dSTR, including a striking enrichment of Htr3a-expressing interneurons (Htr3a INs) in M1 that are selectively recruited during skilled reaching, as confirmed by two-photon calcium imaging. Corticostriatal projection neurons and striatal spiny projection neurons show subtype- and region-specific transcriptional remodeling involving genes linked to synaptic function, translation, and metabolism. Glial cells, including astrocytes, oligodendrocytes, and microglia, exhibit similarly robust, stage- and region-dependent gene regulation. These findings provide a comprehensive molecular framework for motor learning and highlight coordinated, cell type-specific transcriptional programs in neurons and glia that shape the encoding and retrieval of motor memory. Keywords: Motor learning, transcriptomic remodeling, motor cortex, dorsal striatum, Htr3a-expressing interneurons, single-cell RNA sequencing Highlights: -Motor learning activates interneuron cell types in motor cortex and striatum. -Htr3a-expressing interneurons in motor cortex are specifically activated while performing a learned reaching behavior. -Transcriptome remodeling exhibited distinct patterns between motor cortex and striatum. -Glial cells showed stage- and region-specific transcriptomic alteration patterns that align with those in neurons

The common neural representation in the primary motor area between motor execution and kinesthetic motor imagery
Although motor imagery activates higher-order motor-related areas, the role of the primary motor area (M1) in motor imagery remains unclear. This study aimed to investigate whether motor imagery recruits a neural representation of fingers similar to that of motor execution in the hand M1. Ten healthy right-handed adults executed and kinesthetically imagined tapping using one of four fingers. Using functional magnetic resonance imaging with multi-voxel pattern analysis, we trained the decoder to classify which finger the participants were moving using brain activation during motor execution and tested whether it could predict which finger the participants were imaging to move during motor imagery (cross-classification). We also performed the classification in the reverse direction. The average accuracy of these cross-classifications was significantly higher than chance in the left hemisphere hand M1 (hand-M1). Analysis of the representational geometry showed that the distance of neural representations for the same fingers was statistically shorter than that for different fingers between motor execution and imagery. Furthermore, we conducted a replication study with 14 participants and found results similar to those of the original study. Our results suggest that the neural representation of kinesthetic motor imagery is partially similar to that of motor execution in the contralateral hand M1.

C-LTMRs Regulate Thermosensation and Gate the Transition from Acute to Chronic Pain
C-low threshold mechanoreceptors (C-LTMRs) are traditionally associated with affective touch, yet emerging evidence suggests broader roles in sensory processing and pain modulation. We developed an intersectional genetic approach to selectively ablate C-LTMRs in adult mice by combining Nav1.8IRES-FLPo and THCreER drivers with a conditional DTR reporter. This approach yields robust, tissue-specific deletion of C-LTMRs without off-target effects in non-sensory tissues. C-LTMR-ablated mice exhibit altered thermotaxis behavior, including a sharpened and spatially restricted preference for warmth, while maintaining largely intact responses to gentle touch. Remarkably, following surgical or chemotherapeutic injury, these mice display and persistent mechanical and cold hypersensitivity, implicating C-LTMRs in the resolution of pain. Transcriptomic profiling of dorsal root ganglia (DRG) and dorsal horn of the spinal cord (DHSC) revealed widespread transcriptional dysregulation in pathways related to extracellular matrix remodeling, vascular function, gliogenesis, and inflammation, in naive mice. In C-LTMR-ablated mice, paclitaxel failed to induce pro-recovery transcriptional programs and instead promoted persistent neuroinflammatory signatures. These findings establish C-LTMRs as key modulators of pain recovery, acting through tissue-specific transcriptional programs that suppress inflammation and support sensory homeostasis.

Cortical neural landscape captures mouse-to-mouse variability in anticipatory vs. inattentive decision making
Understanding individual variability in behavior is crucial for both basic and clinical neuroscience, yet it remains challenging to study in traditional single laboratory experiments with small sample size. Leveraging standardized behavioral and neural datasets from the International Brain Laboratory, comprising approximately 100 mice trained on a visual decision-making task, we investigated the structure and neural correlates of inter-animal behavioral variability. Using reaction time analysis and a deep learning-based embedding of individual animals, we uncovered large but low-dimensional differences in behavioral traits. Some mice consistently exhibited anticipatory responses, marked by fast reaction times, while others showed slower, more disengaged behavior. These behavioral profiles were consistent across sessions, with female mice tending to show more anticipatory behavior than males. We hypothesized that this behavioral spectrum reflects differences in the depth of underlying cortical states, reflected in the temporal dynamics of neural activity. Supporting this idea, we found that the characteristic timescale of population activity, measured during both inter-trial intervals and passive periods, correlated with an animal's anticipatory tendency across cortical areas, especially in medial visual areas. These findings suggest that individual differences in the cortical dynamics may underlie distinct decision-making strategies.

A Complexity-Science Framework for Studying Flow: Using Media to Probe Brain-Phenomenology Dynamics
Consciousness spans a range of phenomenological experiences, from effortless immersion to disengaged monotony, yet how such phenomenology emerges from brain activity is not well understood. Flow, in particular, has drawn attention for its links to performance and wellbeing, but existing neural accounts rely on single-region or small-network analyses that overlook the brain's distributed and dynamic nature. Complexity science offers tools that capture whole-brain dynamics, but this approach has rarely been applied to flow or to its natural comparison states of boredom and frustration. Consequently, it remains unclear whether tools drawn from complexity science can objectively separate phenomenological experiences while also clarifying their neural basis. Here we show that a complexity science approach distinguishes flow from boredom and frustration. We induced each phenomenological experience with a difficulty-titrated video game during functional magnetic resonance imaging and collected concurrent behavioral and self-report data. Whole-brain analyses revealed that flow is marked by lower global entropy, higher dynamical agility, and decreased dynamical complexity, whereas boredom and frustration exhibit different configurations of brain-dynamics metrics. Notably, these findings integrate previously separate prefrontal, network-synchrony, and cerebellar observations within a single dynamical systems framework and identify complexity-based markers that map the neural correlates of media-related benefits.

One of these things is not like the others: Theta, beta, & ERP dynamics of mismatch detection
Working memory (WM) enables the detection of mistakes by permitting one to notice when sensory input is mismatched to their internal prediction. Prior studies support the role of frontal midline theta activity, with an overlapping N200 event-related potential (ERP), as a mechanism for comparing incoming sensory stimuli to the internal model. Additionally, posterior low-beta activity has been proposed as a mechanism for processing incoming sensory stimuli in WM. However, it is unknown whether frontal midline theta activity and the N200 support mismatch detection, or whether posterior low-beta activity extends from sensory processing to detecting a mismatch between sensory input and the internal model. Here, we reveal that frontal midline theta supports mismatch detection and explains individual WM performance. Unexpectedly, instead of the N200, results show a positive slow wave ERP overlapping with the frontal midline theta mismatch response. Results additionally indicate a late posterior low-beta response persisting from stimulus presentation into the post-stimulus delay. Our findings establish frontal midline theta as a marker of successful mismatch detection, challenge the domain-general role of the N200 in error detection, and support theories linking posterior low-beta to processing incoming sensory stimuli.

Twice as nice: Boosts in adolescent reinforcement learning from Pavlovian bias and age-related prioritization of reward-motivated incidental memory.
Adolescence is a period marked by profound changes in both capacities for learning and the motivational drives that guide behavior. Motivated learning, including the ability to associate cues with actions that lead to positive or negative outcomes, is a fundamental component of adaptive behavior and is essential for survival. Equally important is the encoding of events during learning, which may be influenced by the valence of outcomes. Given the substantial neurocognitive changes in motivated learning and memory that occur from childhood to adulthood, adolescence provides a unique window to investigate mechanisms of these adaptive behaviors. Yet, we know surprisingly little about the development of these behaviors, with sparse extant research fraught with inconsistent findings. In this study, we examined motivated learning and incidental memory using a validated affective learning task in a sample of 174 participants aged 8 to 25 years. The task orthogonalized action and outcome valence and included incidental encoding of trial-unique images presented during feedback, followed by a delayed memory test. We show that adolescents outperform both children and adults in learning by leveraging Pavlovian response biases. In contrast, children exhibit enhanced memory for stimuli associated with positive outcomes compared to adolescents and adults. These findings point to distinct developmental advantages: enhanced learning performance in adolescence and enhanced memory for rewarding events in childhood, each potentially adaptive at their respective developmental stages. Together, these findings suggest opportunities to leverage learning and memory in youth for practical applications, such as education and policy setting.

Fast efficient coding and sensory adaptation in gain-adaptive recurrent networks
As the statistics of sensory environments often change, neural sensory systems must adapt to maintain useful representations. Efficient coding prescribes that neuronal tuning curves should be optimized to the prior, but whether they can adapt rapidly is unclear. Empirically, tuning curves after repeated stimulus presentations exhibit "adapter repulsion", whose underlying mechanism remains uncertain, and which contrasts with the "prior attraction" expected under many efficient-coding models. We propose a gain-adaptive, recurrent sensory network model in which gains optimize a novel efficient-coding objective balancing accuracy and spiking cost. From the propagation of modulated gains throughout the network emerge quickly adaptive tuning curves. The model accounts for subtle adapter-repulsion effects under peaked priors and predicts fast prior attraction under broader distributions, for which we provide supporting behavioral evidence. Our framework reconciles seemingly contradictory adaptive phenomena, under a unified theoretical and mechanistic model of efficient coding mediated by gain modulation in recurrent circuits.

Chronic pain is linked to a resting-state neural archetype that optimizes learning from punishments
Chronic pain is a leading cause of disability, yet its underlying susceptibility traits remain unclear. Disorders like chronic pain may stem from extreme neural types, or archetypes, optimized for specific cognitive strategies and reflected in patterns of resting-state networks. Here, we examined a sample from the general population (n = 892) and three clinical samples with subacute back pain (n = 76), chronic back pain (n = 30), and treatment-resistant depression (n = 24). Using the sample from the general population, we found three neural archetypes that prioritize different cognitive strategies. Clinical pain samples, compared to the sample from the general population, mapped close to an archetype optimized for punishment learning (Archetype P). We replicated these results by recomputing the archetypes starting from the clinical pain samples, additionally revealing an association between Archetype P and pain severity. These findings suggest a neural-cognitive trait underlying susceptibility to chronic pain.

Complementary cortical and thalamic contributions to cell-type-specific striatal activity dynamics during movement
Coordinated motor behavior emerges from information flow across brain regions. How long-range inputs drive cell-type-specific activity within motor circuits remains unclear. The dorsolateral striatum (DLS) contains direct- and indirect-pathway medium spiny neurons (dMSNs and iMSNs) with distinct roles in movement control. In mice performing skilled locomotion, we recorded from dMSNs, iMSNs, and their cortical and thalamic inputs identified by monosynaptic rabies tracing. An RNN classifier and clustering analysis revealed functionally heterogeneous subpopulations in each population, with dMSNs preferentially activated at movement onset and offset, and iMSNs during execution. Cortical and thalamic inputs were preferentially activated during onset/offset and execution, respectively, though dMSN- and iMSN-projecting neurons in each region showed similar patterns. Locomotion phase-specific rhythmic activity was found in a subset of thalamic dMSN-projecting neurons and dMSNs. Cortex or thalamus inactivation reduced MSN activity. These findings suggest that corticostriatal and thalamostriatal inputs convey complementary motor signals via shared and cell-type-specific pathways.

Targeting Neuromuscular Junction Regeneration is a Therapeutic Strategy in ALS
Instability and denervation of the neuromuscular junction (NMJ) are early events in Amyotrophic Lateral Sclerosis (ALS), likely reflecting a progressive decline in the regenerative capacity of motor neurons (MNs) and their environment. To investigate this, we evaluated NMJ regeneration throughout disease progression in SOD1G93A mice following reversible axon terminal degeneration induced by -Latrotoxin. In parallel, we monitored the expression of CXCR4, a GPCR upregulated during axonal regeneration, and tested whether its pharmacological activation could mitigate ALS-related functional decline. We found that NMJ regenerative capacity is largely preserved during pre- and early symptomatic stages, and remains active in subsets of NMJs even at later stages. CXCR4 is expressed at axon terminals from early disease stages, declining only at end stage. Its expression is conserved across ALS models, including SOD1G93A pigs, hiPSC-derived MN with ALS mutations, and biopsies from sporadic ALS patients. CXCR4 stimulation improved motor function, NMJ innervation, MN survival, and respiratory performance in ALS mice, and axon outgrowth in iPSC-derived MN. These findings identify the NMJ and CXCR4 as viable therapeutic targets in ALS.

Neuroligin-3 interaction with CSPG4 regulates normal and malignant glial precursors through PIEZO1
Glioma pathophysiology is robustly regulated by interactions with neurons. Key to these interactions is the role of neuroligin-3 (NLGN3), a synaptic adhesion molecule shed in response to neuronal activity1-5 that functions as a paracrine factor crucial for glioma growth. Here, we elucidate the mechanistic pathway whereby shed NLGN3 interacts with glioma and their normal glial counterpart. NLGN3 interacts with Chondroitin Sulfate Proteoglycan 4 (CSPG4) on both glioma and healthy oligodendrocyte precursor cells (OPCs)6-9, facilitating CSPG4 shedding by ADAM10. NLGN3-CSPG4 interactions and consequent shedding alter membrane tension, thereby activating PIEZO1 mechanosensitive channels and causing membrane depolarization. The NLGN3-CSPG4-PIEZO1 axis maintains OPCs in an undifferentiated, stem-like state and promotes glioma proliferation, underscoring important functional roles for the NLGN3-CSPG4-PIEZO1 axis in both healthy and malignant glial precursors.

Perceived time drives physical fatigue
Recent studies suggested that fundamental physiological processes, such as physical fatigue, rely on the perceived rather than the actual time. However, the neural correlates underlying this effect and its disentanglement from motivational factors (i.e., performance goals) remain unknown. To investigate the time deception effect on fatigue, we developed a novel EEG design in which participants performed 100 isometric contractions at a fixed pace and resistance per session. The actual contraction duration (short or long) and the calibration of the displayed clock (normal or biased to an acceleration or deceleration) were independently manipulated between sessions to examine whether fatigue and its neural correlates evolved based on perceived or actual time. Our results show an accumulation of physical fatigue that follows the perceived time, irrespective of motivational factors. This effect was consistently observed only when the clock was slowed down. This time deception effect was mediated by an oscillatory dynamic that followed perceived time in frontal (theta- and beta-bands) but not motor areas (beta-band). Further analyses highlighted the key role of the frontal oscillatory dynamics in the effectiveness of the time deception effect on physical fatigue.

Sensory neurostimulation promotes stress resilience with frequency-specificity
Chronic stress is a major risk factor for neuropsychiatric disorders, acting via increased neuroinflammation and disrupted synaptic plasticity. While non-invasive visual or audiovisual neurostimulation (AV flicker) at 40Hz has been shown to modulate brain immune signaling and improve cognitive performance in mouse models of Alzheimer's disease, its effects in the context of stress remain unknown. Here we show that AV flicker protects against stress-induced behavioral, microglial, astrocytic, and synaptic changes in a sex- and frequency-specific manner. Male and female mice underwent 28 days of chronic unpredictable stress with concomitant daily AV flicker exposure at 10Hz, 20Hz, or 40Hz. Stress-induced behaviors were most effectively mitigated by 10Hz AV flicker in males and 40Hz AV flicker in females. In the medial prefrontal cortex, AV flicker normalized the balance of mature and immature dendritic spines and counteracted stress-induced molecular changes in neurons, microglia, and astrocytes, including in key neuropsychiatric risk genes. These findings show that frequency optimized AV flicker induces resilience to chronic stress.

The neural mechanisms supporting the rise and fall of maternal aggression
To protect the helpless young, lactating females dramatically increase aggression towards intruders, known as maternal aggression. However, attack is costly and risky. When pups no longer exist, maternal aggression rapidly declines. Our study reveals the critical role of the pathway from posterior amygdala cells expressing estrogen receptor alpha (PAEsr1) to the ventrolateral part of ventromedial hypothalamus cells expressing neuropeptide Y receptor Y2 (VMHvlNpy2r) in the rise and fall of maternal aggression. Functional manipulations and recordings demonstrate that VMHvl-projecting PAEsr1 (PAEsr1[->]VMHvl) cells are naturally active and required for maternal aggression. During lactation, PA-VMHvlNpy2r connection strengthens, and VMHvlNpy2r excitability increases to facilitate attack. Furthermore, PAEsr1 expresses abundant oxytocin receptors, enabling oxytocin to boost PAEsr1 cell output. After pup separation, the oxytocin level drops, causing decreased maternal aggression, which can be restored by optogenetically increasing the oxytocin level. Thus, diverse forms of plasticity occur at the PAEsr1-VMHvlNpy2r circuit to support need-based maternal aggression.