Nature Neuroscience

Challenges in verbal communication are a prominent feature of autism. However, gene regulatory programs in speech-related cortical regions remain poorly characterized. In parallel, it remains unclear whether the heterogeneous genetic factors underlying autism converge on shared neurobiological mechanisms. To address these gaps, we generated paired transcriptomic and epigenomic data from post-mortem human brain tissue across 100 donors. Here, we show that transcriptional differences in the speech-related Brodmann Area 22 in individuals with neurodevelopmental conditions, including autism, are strongest among those with a known genetic diagnosis. A similar but attenuated signature is observed in those without a genetic diagnosis. These transcriptional differences are most pronounced in neurons, with glutamatergic L4/5 intratelencephalic neurons affected across multiple modalities. Finally, multimodal analysis implicates altered RFX3-dependent networks as a central hub in autism, particularly among L4/5 intratelencephalic neurons in non-verbal individuals. Together, our study identifies regulatory architecture linking chromatin state, transcriptional output, and variation in verbal ability in autism.

Axons are the brains wiring, organized into bundles that connect nearby and distant regions. Axon caliber determines signal conduction velocity and varies both within and across bundles, reflecting the brains diverse functional demands. Much of what we know about this organization derives from 2D histology, assuming cylindrical axons whose calibers are described by their radius. Yet, recent 3D histology reveals that the radius varies along an individual axon--with implications for both characterizing axon caliber and potentially conduction velocity predictions. We show in 450,000 3D rat axon reconstructions that--despite this individual variation--axon bundles possess stable radius distributions at the ensemble level, which 2D cross-sections faithfully represent. This representativeness extends to conduction velocity predictions, as along-axon variation has only modest impact. In particular, large axons exhibit especially stable conduction, emphasizing their key role in time-critical signaling. With 2D sampling validated, we leverage 46 million human corpus callosum axons from 2D histology to determine sample size requirements across neuroscience applications. Our findings reinforce decades of 2D histology-based research on axon organization and its functional implications, while guiding future study design.

Experience unfolds as a stream shaped by hidden causes that change over time. Adaptive behavior requires inferring the underlying states and adjusting when they change. Yet, how neural circuits discover and track latent states remains unclear. Here we introduce NeuraGEM, a neural architecture that combines fast transient activity with slow synaptic plasticity to implement an online analogue of Expectation-Maximization. By separating timescales, NeuraGEM clusters sequential experiences, detects context changes, and stabilizes task-specific computations. The model generalizes beyond conventional recurrent networks and reproduces key features of human contextual learning, including curriculum-dependent effects. It also gives rise to population dynamics resembling those observed in brain circuits, including line-attractor structure and transient error responses at change points. Together, these findings provide a mechanistic account of how neural circuits organize experience into latent states that support rapid inference and adaptive behavior.

Neuronal ensembles--groups of neurons that exhibit coordinated activity during behavior--are a fundamental feature of cortical computation. Dendritic branches amplify clustered synaptic inputs through local nonlinearities, suggesting that presynaptic groups might organize their connections in specific spatial patterns to engage these mechanisms. Whether the same axon groups form synaptic clusters with consistent spatial arrangements across different target neurons remains unknown, but nanoscale connectomes would resolve such anatomical motifs if they exist. We analyzed millions of synaptic connections in a connectome of mouse visual cortex and found over 700,000 axon groups that repeatedly cluster their synapses onto dendritic branches of multiple pyramidal cells, with over 500,000 maintaining consistent distal-to-proximal arrangements. These repeated patterns occur far more frequently than expected from spatial proximity or layer-based connectivity rules. Axon groups preferentially target specific dendritic branches and position their synapses in stereotyped spatial configurations across multiple postsynaptic partners, revealing that functional ensembles leave characteristic anatomical signatures in cortical microarchitecture.

Among individuals with equivalent Alzheimers pathology, cognitive outcomes can diverge by decades, a phenomenon termed cognitive reserve that remains descriptive after thirty years of research. We propose that the [~]109-to-10 bits/s gap between sensory input and behavioral output functions as error-correcting redundancy in the sense of Shannons channel coding theorem. Progressive neuronal loss maps to symbol erasure in a redundant code, and the critical damage fraction at which cognition fails is dc = 1 - k/n, where k {approx} 10 bits/s is the behavioral channel requirement and n is the effective number of coding units. We evaluate this threshold across three channel models (binary erasure, Gaussian, and Erd[o]s-Renyi percolation) and show that all produce a sharp phase transition from reliable to unreliable decoding. The framework makes four testable predictions: (i) dc scales with the measurable redundancy ratio{rho} = n/k, which accounts for clinical heterogeneity; (ii) information-theoretic redundancy from resting-state fMRI should predict time-to-conversion beyond structural atrophy; (iii) the decline trajectory near dc is sharp, consistent with the "cognitive cliff"; and (iv) motor circuits, operating at higher bandwidth, have lower reserve than cognitive circuits. Significance StatementCognitive reserve (why some brains resist dementia pathology better than others) has been described for thirty years but never given a quantitative, information-theoretic foundation. We propose that the roughly hundred-million-fold gap between sensory input ([~]109 bits/s) and behavioral output ([~]10 bits/s) functions as error-correcting redundancy in the Shannon coding-theoretic sense. This yields a closed-form critical damage threshold, dc = 1 - k/n, below which cognitive function is preserved and above which it collapses; this is consistent with the clinically observed plateau-then-cliff pattern of dementia. The framework unifies cognitive reserve with channel coding theory, accounts for individual heterogeneity in disease onset, and generates falsifiable predictions that link information-theoretic redundancy measures to time-to-clinical-conversion.

Stroke is one of the leading causes of global disability, yet the principles governing how focal brain injury disrupts large-scale neural connectivity over time remain poorly understood. Here, we leverage a longitudinal multimodal dataset to track the evolution of individual-specific connectivity patterns, or brain fingerprints, over the first year after stroke. Despite a persistent shift from healthy architecture, we demonstrate that each patients unique functional connectome fingerprint is remarkably resilient and stabilizes within three weeks. This early global stabilization masks a protracted system-specific reorganization of brain circuits, which is characterized by an initial increase in connectivity within sensory and attention systems, followed by a decline across higher-level association networks. A joint structure-function embedding further shows that recovery involves a gradual shift toward the normative healthy range, driven primarily by functional reconfiguration atop a stable structural lesion. Crucially, a multivariate prediction model reveals that early functional signatures selectively forecast long-term impairment in language, executive function, and attention. Together, our results define the post-stroke brain as a shifting but constrained dynamical system, identifying early-stabilized brain patterns as biomarkers for individual recovery profiles and targets for personalized neurorehabilitation.

Large-scale electrical perturbation of the human brain provides a unique model for understanding how multiscale biological constraints shape behaviorally relevant reorganization. Here, we integrate longitudinal neuroimaging coordinates from 148 experiments ({approx}2,300 subjects) with normative connectomics, chemoarchitecture, intrinsic electrophysiology, and transcriptomics to identify cross-scale principles governing human brain reconfiguration under strong perturbation. Convergent hubs of structural and functional plasticity embed within default-mode and salience systems and show complementary coupling to visual networks, linking perturbation-induced change to large-scale circuits supporting affective regulation, memory, interoception, and psychosis-relevant processes. These macroscopic patterns align with intrinsic cortical dynamics and chemoarchitectural gradients dominated by 5-HT1A receptors, with additional contributions from D2, -opioid and GABAA systems, and are enriched for astrocytic and microglial gene expression, implicating glial plasticity in systems-level reorganization. Finally, in a separate intervention dataset, regularized statistical-learning models demonstrate that this multiscale signature tracks behaviorally relevant symptom change specifically under strong electrical perturbation. Together, these results outline general organizing principles linking molecular, cellular and network-level constraints to human behavioral adaptation, providing a computational framework for understanding how large-scale perturbations reshape brain systems across levels of biological organization.

AbstractMacroscale functional connectivity is jointly shaped by structural wiring and non-tract influences. Because these contributions are intrinsically entangled, their distinct roles remain unclear. Here, we introduce a large-scale brain modeling-based framework that disentangles functional synchrony into two components: tract-explainable and underexplained synchrony. Across two independent human cohorts (total n = 1214) and a marmoset dataset (n = 24), both components were highly reproducible and played distinct roles in shaping network architecture. Tract-explainable synchrony closely aligned with tractography and supported global integration, whereas underexplained synchrony was associated with multiscale cortical similarity features, including microstructure, receptor, and gene-expression patterns, and exhibited higher modularity. Crucially, these components dissociated along the sensorimotor-to-association hierarchy. Tract-underexplained synchrony became increasingly prominent in higher-order regions, exhibiting greater individual variability, behavioral relevance, and clinical sensitivity. Ultimately, tract-based and non-tract-mediated mechanisms serve distinct yet complementary roles in driving functional organization.

Attention has long been thought to enable efficient vision,1-8 yet it requires additional neural machinery and energy. Whether attention yields net energetic benefits--after accounting for the cost of control--has never been demonstrated. Here we show that attentional control can substantially improve whole-system energy efficiency in a model of primate visual processing. Our model, EAN ("Energy-efficient Attention Network"), implements attention as recurrent top-down multiplicative gain over features, space, and time. EAN is optimized using a joint objective combining task performance and neurobiologically grounded energy costs accounting for action potentials and synaptic transmission across all components,9-11 including the attentional control circuitry itself. On a visual-category-search task requiring joint identification and localization of a target, EAN learns to focus its energy dynamically on task-relevant locations and features, reducing total energy use by up to 50% at matched accuracy and enabling flexible trial-by-trial trading of accuracy against energy. The model variant combining feature-based and spatial attention is most efficient and best captures human errors and difficulty judgments. EAN generalizes to classical attention tasks, replicating canonical effects of attention on firing rates, variability, and noise correlations,12 and patterns of V4-to-V1 feedback suppression.13 Our work connects a cognitive function (attention), a neural mechanism (gain modulation), and a neurobiological constraint (metabolic cost) in a single mechanistic model that explains how selection and recurrence enable flexible, energy-efficient vision.

Biological agents flexibly adapt their behavior to novel goals and environmental demands without additional training, yet the computational principles enabling such control remain unclear. Here, we propose that three cognitive constructs constitute minimal computational motifs for such flexible control: relational structure, spotlight attention, and affordance computation. We examine whether these constructs underpin flexible control in an embodied dynamic pursuit task that requires continuous integration of inter-entity relations, reward, and action feasibility, making it a suitable testbed for real-time control. By implementing these constructs within a multi-module graph convolutional network, we show that the model achieves zero-shot transfer across novel pursuit scenarios that vary in physics, target properties, and interaction policies such as fleeing or chasing, without additional training. Although not explicitly trained to do so, the model also exhibits change-of-mind (CoM) behavior, or mid-course target revision, a hallmark of flexible control exhibited by biological agents. Neural recordings from the primate dorsal anterior cingulate cortex revealed population-level signatures that link these constructs to neural dynamics, providing biological support for the proposed computational architecture.

The striatum integrates dopamine (DA) and acetylcholine (ACh) to shape learning and action, yet the principles governing their in vivo coordination remain unclear. Here we combine dual-color fiber photometry with low-dimensional manifold analysis to characterize DA-ACh interactions in the anterior dorsolateral striatum during associative learning. We find that both neuromodulators undergo learning-related plasticity but exhibit distinct temporal signatures. DA responses are fast and event-locked, whereas ACh shows broad, sustained modulation across behavioral epochs. Low-dimensional coordination between DA and ACh during cue-reward events robustly predicted learned trial states. By contrast, spontaneous lick-related activity was dominated by amplitude scaling and showed weaker low-dimensional structure, indicating that manifold organization preferentially encodes learned states rather than generic motor output. Granger causality analysis reveals a robust temporal asymmetry, with DA reliably predicting subsequent ACh fluctuations but not vice versa. These findings suggest that DA acts as a directional temporal scaffold organizing ACh within a shared neuromodulatory manifold, reframing DA and ACh not as independent channels but as a hierarchically coupled system for striatal computation.

Neural representations evolve over time, yet the relative contributions of online experience and offline states such as sleep remain unclear. Here, we recorded single-unit activity in the olfactory cortex of mice across cycles of awake odour exposure and sleep, and developed a low-rank decoder to track representational drift. We identified four orthogonal drift modes operating on distinct timescales, revealing that sleep and wake drive qualitatively different transformations, which indicates that offline reorganisation is not a simple continuation of online learning. Rather, sleep initiates an about-turn in the overall drift trajectory, which is uniquely characterised by a combination of decorrelation and rotation of odour representations. We also provide the first evidence for olfactory replay, occurring at ~2.5x temporal compression and associated with locally generated piriform cortex sharp waves. Together, these findings demonstrate that representational drift comprises state-dependent components, and reveal distinct contributions of wake and sleep to sensory representational change.

Dopamine neurons are thought to signal reward prediction errors phasically and the opportunity cost of time tonically, while also displaying ramping activity during goal approach, and coupling with movement. These are often treated as distinct modes of dopamine function, each requiring its own computational explanation. Here we show that all can be unified by considering temporal difference learning within the context of continuous time, combined with the assumption that the brain computes value changes through a fast model-based process while maintaining a slower model-free cache. Together, the inclusion of these two ingredients explains phasic responses, tonic modulation between reward contexts, navigation ramps, speed scaling, and the fading of ramps with learning, without invoking separate mechanisms. We confirm these predictions across two independent datasets of dopamine recordings in rodents spanning freely-moving and head-fixed paradigms. Continuous temporal difference learning may thus provide a unified theory of dopamine function.

Hippocampal sharp wave-ripples (SPW-Rs), neocortical slow oscillations, and thalamocortical sleep spindles are hypothesized to provide a temporal framework for coordinated information transfer during memory consolidation. Hippocampal replay supports this process, yet replayed sequences often unfold across multiple SPW-Rs, suggesting that individual ripples may not constitute the fundamental unit of hippocampal output. Here, using large-scale electrophysiological recordings from the hippocampus and retrosplenial cortex, we show that hippocampal output is organized into clusters of SPW-Rs (cSPW-Rs) during UP states, which are often phase-locked to spindle troughs. Extending this approach with wide-field imaging and unsupervised latent-variable modeling, we found that cSPW-Rs enhanced segregation between the default mode and somatomotor networks and preferentially replayed spatially extended maze trajectories following learning. We propose that SPW-R clusters enable reverberating hippocampal-cortical spike exchange and the concatenation of sequential experiences, establishing ripple clusters as a previously unrecognized syntactic unit of hippocampal-neocortical dialogue.

Perineuronal nets (PNNs) preferentially enwrap parvalbumin (PV) interneurons, regulating plas-ticity and circuit function. The molecular differences between PNN-positive and PNN-negative PV neurons remain unknown. We combined Xenium spatial transcriptomics with PNN-labeling in adult mouse cortex (378,349 cells) and found that 97% of PNNs enwrap PV neurons; sub-stantially higher than previous estimates. PNN status reflected a transcriptional continuum rather than discrete subtypes. A classifier trained on Xenium data (AUC = 0.87) applied to Allen Brain scRNA-seq data, enabled genome-wide analysis of 34,326 cortical PV neurons. PNN-positive PV neurons expressed mature fast-spiking markers: Kv3 channels, mature NMDA subunits (Grin2a), fast-kinetic GABA-A receptors (Gabra1), oxidative phosphorylation genes, and gap junctions (Gjd2). PNN-negative PV neurons expressed neuropeptides and GABA-A subunits typical of Sst interneurons, suggesting a transcriptomic boundary between cell types and potentially ele-vated plasticity. This establishes PNN status as a molecular correlate of PV cell specialization, with implications for therapeutic strategies targeting cortical plasticity.

Remembering events in the correct order, and generating ordered sequences of actions, are fundamental abilities across species. Behavioral studies, and theoretical work, raise the possibility that the brain represents serial order by a domain-general neural code, following the principle of Competitive Queuing. However, direct neurophysiological evidence for Competitive Queuing exists only in the motor domain. When humans and non-human primates prepare for a series of movements, several of the upcoming movements are represented in parallel, with their representational strength reflecting ordinal position in the sequence. We test the generalizability of this so-called primacy gradient across motor sequences and memorized auditory sequences. Using a multivariate decoding approach, Experiment 1 replicated the presence of a Competitive Queuing primacy gradient in magnetoencephalography (MEG) data of young healthy adults (n = 23) when they prepared a sequence of finger movements from memory. Importantly, we observed a similar primacy gradient when participants anticipated a sequence of tones they had learned before, in the absence of any movement. In Experiment 2 (n = 23; naive cohort), we rule out the possibility that this primacy gradient in auditory memory is explained by any learnt association between tones and movements, or by MEG signal fluctuations that are unrelated to discrete sequential events. In sum, we find a similar neural signature of serial order coding when humans prepare a sequence of movements, and when they anticipate a sequence of sounds. This lends support to the generalizability of Competitive Queuing.

Brain state modulates sensory processing across visual cortex, yet how it relates to the organization of functional circuitry at the level of individual neurons and cell types remains largely unknown. To address this, we constructed one of the largest mi-croscale directed functional circuit maps in mouse visual cortex from calcium imaging of more than 57,000 neurons across four visual areas and five cortical layers. Using a time-aware causal inference framework, we found that intra-areal connections dom-inate across arousal states, consistent with experimental findings on the local bias of cortical anatomy. Among intra-areal connections, anterolateral area (AL) had the highest density, and among inter-areal connections, the AL{leftrightarrow}rostrolateral area (RL) axis formed the strongest pathway. Laminar circuit organization was dominated by layer 6 recurrence within-layer, while the most prominent between-layer pathway was layer 5-to-layer 6 in low arousal and layer 4-to-layer 5 in high arousal. Spatial extent was selectively greater for excitatory-to-inhibitory connections in high arousal, but not for excitatory-to-excitatory connections. Across 6,597 electron-microscopy recon-structions of neuron pairs, synapse count predicted functional connection strength in both arousal states, but structure-function coupling was weaker in high arousal. In stimulus-driven response prediction, neuron pairs with stronger functional connections exhibited more similar predictive performance in both states, with performance vary-ing by layer and cell type. Overall, our findings map, at single-neuron resolution, the multi-scale organization of directed functional circuitry in mouse visual cortex across brain states.

Sleep transforms fragile experiences into lasting memories, but the neuronal basis of this process in humans has remained elusive. In rodents, hippocampal ripples orchestrate the replay of place cell sequences, establishing a cellular mechanism for consolidation - though with limited generalizability to human memory. In humans, neuroimaging has revealed large-scale offline reactivation, but these coarse signals leave open whether individual neurons are reactivated and how ripples might mediate this process. Here, we bridge this gap by directly recording 1,466 medial temporal lobe (MTL) neurons and intracranial electroencephalography during learning, post-learning wakefulness, and sleep. We show that ripples robustly drive neuronal firing, with sleep ripples eliciting stronger activation than wake ripples. Critically, neurons tuned to items that were later remembered fired more strongly during ripples than those coding for forgotten items, and this memory-linked reactivation was selectively observed during sleep. Finally, ripple-associated neuronal MTL bursts were detectable across widespread cortical activity, pointing to a mechanism for systems-level consolidation. Together, these findings provide the first direct evidence that ripple-driven single-neuron reactivation supports human episodic memory consolidation and reveal why sleep -- compared to wakefulness -- offers a privileged window for stabilizing memories.

The thalamus plays a central role in cortical development, organisation and function. Thalamic nuclei acquire distinct molecular identities during gestation, with first-order relays maturing before higher-order nuclei. Thalamic afferents innervate the cortical plate with a precise order, disruptions to which alter cortical function. Recent models propose that thalamic input to primary sensory cortex constrains the development of wider cortical networks, promoting the formation of highly-connected hubs in association cortex. Here, we combine neuroimaging, post mortem gene expression data and network modelling to examine how the timing and spatial distribution of thalamocortical innervation influences the formation of cortical networks during gestation. We find that the maturation rates of thalamic nuclei align with predicted timing and distribution of afferent outgrowth. While higher order nuclei connect widely across the cortex, they do not preferentially target high-degree hubs. Instead, hubs emerge from interdependent spatiotemporal constraints imposed by both wiring distance and thalamocortical maturation.

Humans and other animals process sensory uncertainty by integrating stimulus information with prior knowledge and expectations. Predictive coding conceptualizes perception as a form of Bayesian inference wherein hierarchical brain circuits update internal models to reconcile bottom-up sensory input with top-down predictions. Whereas predictive coding is a leading theory, the extent to which it is implemented in primary sensory cortices remains a matter of debate. To further investigate this issue, we examined single-neuron spiking activity in macaque primary auditory cortex (A1) to expected versus unexpected stimulus repetitions and to expected versus unexpected omissions. On average, we found that A1 neurons did not show enhanced responses to unexpected stimulus repetitions, contrary to predictive-coding theory. However, they did show enhanced responses to unexpected stimulus omissions. Taken together, these mixed results place empirical restraints on how PC is implemented in A1. Significance StatementPerception depends on the brains ability to infer the causes of sensory inputs by integrating new information with prior knowledge under uncertainty. Our results reveal nuanced evidence for predictive coding within the primary auditory cortex (A1). Specifically, spiking activity during unexpected stimuli and unexpected stimulus omissions provide conflicting and supporting, respectively, data for this Bayesian framework. These findings refine our understanding of neural mechanisms underlying perception and provide empirical constraints on the neurobiological implementation of predictive processing.