Neuron

The ventrolateral prefrontal cortex (vlPFC) is well known for its involvement in high-level functions such as cognitive control and language. However, vlPFCs role in visual processing is less clear. Here, we investigated how neuronal ensembles in the vlPFC dynamically encode different types of visual information. Using chronic recording of spiking activity, we investigated vlPFCs representational geometry in a macaque monkey passively viewing a large set of naturalistic images, and compared this to representations in deep neural networks (DNNs). We found that the vlPFC processes visual information in two stages. First, an "early" response from 50 to 90 ms after stimulus onset encodes the low spatial frequency component of an image. It contains sufficient information to form a coarse estimate of the position and category of a salient object. Then, from 100 ms on, the representational geometry changes and contains much richer information. This late period contains non-categorical information typically present in conscious experiences such as the orientation of a face and natural scenes in the background. The late window also enables sub-category identification, which is boosted by the low spatial category prior. These results suggest that the vlPFC has a dual role in natural vision: first forming fast low-spatial-frequency-based priors shaping feed-forward visual processing, and subsequently maintaining a detailed and rich representation of a visual scene. SignificanceWhat information does the prefrontal cortex represent during natural vision, and how does this relate to conscious experience? Theories of consciousness differ sharply on whether prefrontal activity reflects detailed perceptual content or only high-level, task-related information. Using dense neural recordings during passive viewing of thousands of naturalistic images, we show that the ventrolateral prefrontal cortex processes visual information in two distinct stages: an early, coarse estimate of the visual scene is followed by a richer, high-dimensional representation that includes sub-category identity and other perceptual details. These findings reveal that prefrontal circuits encode more of the content of visual experience than previously assumed, even in the absence of a task.

Neuronal cell death is a hallmark of many neurodegenerative diseases. Effective detection and clearance of cell debris generated during cell death events is essential to prevent a degenerative cascade. Brain resident microglia are responsible for performing these functions through complex cell-cell signaling involving both "find-me" and "eat-me" cues. To examine microglial responses to neuronal cell death in vivo, we investigated neuron/microglia CX3CL1/CX3CR1 signaling using intravital optical imaging in mouse cortex and a single-cell ablation technique called 2Phatal. We find that CX3CL1 aggregates as puncta on microglia and that this pattern is maintained when microglia engulf dying neurons. Additionally, disruption of this signaling via Cx3cr1 deletion when both few and many neurons are dying leads to delayed cell corpse clearance, partly due to a delay in microglial engagement with the dying cells. Overall, our work uncovers a precise role for CX3CL1/CX3CR1 signaling in regulating the microglial response to dying neocortical neurons.

How does cortical connectivity support decision-making? Behavioral tasks often involve multiple sequential phases that implement different computations. For perceptual decision-making during navigation these can include evidence accumulation, decision commitment, and motor program read out. How are these different phases implemented in circuits with fixed anatomical synaptic connectivity? One potential contribution is that the connectivity of neurons is modulated in the different phases, but this has never been tested. Here we used an all-optical method to probe the causal connectivity of excitatory neurons in layer 2/3 of mouse retrosplenial cortex during different behavioral epochs of a navigation-based decision-making task, as well as in the absence of the task. In-task connectivity was different from no-task connectivity: furthermore, these differences were selective to the cue / decision phase, tapering off in later stages of the task. We propose that fast modulation of connectivity is a prevalent mechanism in neural circuit function.

Many stroke survivors with aphasia struggle to map their thoughts into words or motor plans. Neuroprostheses that decode concept representations could help these individuals communicate by predicting the words, phrases, or sentences that they are struggling to produce. Here we decoded concept representations measured using functional magnetic resonance imaging (fMRI) from participants with different aphasia profiles. The decoders generated continuous word sequences that could describe the concepts that the participants were hearing about, seeing, or imagining. To forecast how this approach would generalize across the heterogeneity of aphasia profiles, we characterized how stroke affects the anatomical organization and information capacity of conceptual processing. Mapping how concepts are organized across the brain, we found that conceptual tuning during non-linguistic processing was largely consistent between the participants with aphasia and neurologically healthy participants. Comparing information processing between the participants with aphasia and neurologically healthy participants, we found that both groups processed similar amounts of non-linguistic information. Our findings indicate that concept representations can be largely spared in individuals with aphasia and demonstrate how these representations can be decoded to support communication.

Hippocampal pyramidal neurons function as place cells, showing location-specific activity during navigation, to form an internal spatial map of the environment. They are hypothesized to be the neural substrate of episodic memory. However, place cell receptive fields tend to drift or have poor tuning in low demand tasks, lacking operant goals such as random foraging, or in sensory context-deprived environments. Through chronic two-photon calcium imaging of hippocampal area CA1, we directly compare stability in a low versus a high demand task within the same animals over the course of learning and recall in the same environment. We find that compared to random foraging, an odor-context based navigational task stabilizes place cell representations and increases place cell quality and quantity. To investigate the circuit mechanism that may support this stability, we manipulated the activity of lateral entorhinal cortex (LEC) excitatory neurons, which provide both indirect and direct multisensory inputs about context, odor, and time to CA1. We chemogenetically suppressed activity of excitatory neurons in LEC during recall of the odor-context based navigation task and found that context discrimination is impaired at both the behavioral and neural level. With LEC silencing, mice had lower behavioral performance, less stable population activity, and greater similarity between opposing trial types. Our study finds that increasing task demand increases CA1 stability and that this stability is partially supported by LEC.

The hippocampus and the prefrontal cortex interact across a range of cognitive functions, yet how these regions represent space and how these representations evolve across learning remains poorly understood. We recorded large-scale neural activity from the hippocampus (CA1) and the medial prefrontal cortex (mPFC) as rats acquired proficiency in radial maze tasks over weeks. We identified a form of trial-by-trial flickering in which one hippocampal place field gradually overtook the other across successive trials, whereas mPFC firing fields switched randomly. While hippocampal flickering decreased as the task was mastered, mPFC flickering remained random and persisted throughout learning. Population-level UMAP embeddings revealed that mPFC states transitioned smoothly across trials, with the within-session representational drift stabilizing only after a two-week period. These findings suggest that while the hippocampus stabilizes its spatial maps during learning, the mPFC maintains a flexible, flickering representation that facilitates the extraction of abstract task structure and the rapid adaptation required for behavioral flexibility.

Phenotype-driven forward screening offers a powerful strategy to discover neuronal substrates underlying physical and behavioral traits without prior anatomical or functional assumptions. Historically, such approaches have identified many important neuronal circuits using invertebrate model organisms, but application to mammals has remained limited by the lack of appropriate strategies. Here we quantitatively profiled 56 neurological phenotypic features across more than 200 adult mice when neurodevelopmentally classified neurons were activated or inhibited by chemogenetics. This screen yielded a wide spectrum of robust neurological phenotypes. Upon analyzing these phenotypes computationally, we narrowed down to the hypothesis that activation of cortical neurons enlarges pupil size. Experimental evidence using optogenetics and in utero electroporation indicated that this hypothesis is true. Our results provide proof of principle that phenotype-driven forward approach in mammals is a powerful alternative as a laboratory approach to uncover brain substrates, and offers a general framework for systematically mapping neural circuits that regulate physical and behavioral phenotypes.

Rule learning is associated with lasting changes in prefrontal activity. However, experiments typically focus on learning a single set of rules or task and there remains a significant gap regarding how related mechanisms may reflect behavioral improvements in different contexts, especially when examining across different tasks and modalities. We therefore recorded single units from chronic electrode arrays implanted in the prefrontal cortex of four monkeys as they were trained to perform spatial and object working memory tasks with the goal of assessing the resulting activity changes that would be induced by rule learning. Progression of training allowed behavioral improvements to be correlated with a variety of neural effects that could be observed across different task contexts, including increases or decreases of both firing rate and decoding, increases in the proportion of firing rate variance that was unexplained by sensory stimuli and motor actions, and the increased separation of population response trajectories in state space. Our results thus reveal how rule learning induces plasticity of prefrontal cortical activity, and the aspects of neural activity changes that were unique to individual tasks and modalities or common across them. Our results ultimately reveal new patterns of training effects, identifying the generalized prefrontal mechanisms that are responsible for rule learning.

Human interactions span a range of contexts, from cooperation to competition. Negotiation, in particular, is a complex and extended social process in which individuals must reach mutually acceptable decisions despite conflicting incentives. The neural computations that support strategic behavior in such social dilemmas remain insufficiently understood. Here, we combine cognitive computational modeling, electroencephalography (EEG), functional Magnetic Resonance Imaging (fMRI), and fMRI-guided Transcranial Magnetic Stimulation (TMS) to demonstrate that oscillatory activity anchored in the temporoparietal junction (TPJ) causally shifts social learning during strategic bargaining. We found that TPJ metabolic activity and alpha-band oscillations are associated with the use of a feedback-based learning strategy during bargaining. Causal perturbation with rhythmic alpha-frequency TMS selectively modulates this strategy, increasing endogenous alpha oscillations and shifting behavioral learning parameters. Together, these findings reveal a frequency-specific mechanism within the neural substrates of social cognition that implements adaptive social learning, offering insights into potential neuromodulatory targets for ameliorating social dysfunction in neuropsychiatric conditions. Significance StatementStrategic negotiation requires predicting how others will respond to our actions, yet the neural computations supporting this form of social learning have remained elusive. By integrating computational modeling with EEG, fMRI, and frequency-specific TMS, we identify a mechanistic link between alpha-band activity in the temporoparietal junction (TPJ) and feedback-based learning during social exchange. Trial-by-trial estimates of this learning strategy were tracked by TPJ metabolic and oscillatory signals, and rhythmic alpha TMS causally enhanced both the neural signature and the behavioral expression of this strategy. These findings provide causal evidence for a frequency-specific mechanism within the neural systems that supports adaptive social learning. They also highlight the TPJ-alpha system as a promising target for neuromodulatory interventions to improve social functioning in neuropsychiatric conditions. Key FindingsO_LIModel-based behavioral analyses revealed two distinct strategies during social negotiation: a feedback-based learning mechanism (U-strategy) and a reputation-based updating mechanism (A-strategy). C_LIO_LIBoth strategies robustly predicted participants adaptive behavior across samples and conditions, and their modulation accounted for differences in negotiation outcomes. C_LIO_LIEEG analyses revealed frequency-specific alpha and beta power modulation linked to U-strategy computations during partner anticipation, localized to right temporoparietal regions. C_LIO_LIfMRI analyses revealed that trial-by-trial U-strategy estimates selectively modulated BOLD activity within the temporoparietal network associated with mentalizing. C_LIO_LIRhythmic alpha-frequency TMS over individually localized Theory-of-Mind TPJ sites causally altered negotiation behavior, shifting U-learning parameters toward a more conservative strategy. C_LIO_LITMS-EEG analyses demonstrated that alpha-frequency TMS induced time-locked alpha activity in functionally connected frontal sites, consistent with enhanced anticipatory computations. C_LIO_LITogether, these multimodal findings establish a causal, frequency-specific mechanism in the TPJ that implements social value learning during strategic bargaining. C_LI

To understand human thought and behaviour, it is vital to understand how internal representations in mind interface with external sensations in the outside world. It is well-established that representations in working memory can involuntarily draw attention towards objects in the outside world with corresponding visual features. Here, we establish that this interface involves a two-way stream by demonstrating that external sensations also involuntarily draw attention to corresponding internal representations in working memory. We demonstrate this across four dedicated experiments in which an unpredictive external stimulus selectively colour-matched working-memory contents, while being completely irrelevant to the working-memory task. We provide converging evidence for this "externally driven internal attention" from tracking attentional dynamics over time through spatial biases in microsaccades as well as from memory performance. We further highlight that engagement with the external stimulus critically shapes the strength of this attentional capture from perception to memory. Together, our findings delineate the properties and consequences of this underexplored attentional pathway, opening novel avenues for research at the interface of perception and memory, and their disorders.

Sensory cortices filter repeated inputs through rapid adaptation over seconds and experience-driven learning over days. Although these forms of plasticity occur simultaneously, it is not known how they interact within cortical circuits. We combined two-photon calcium imaging, data-driven circuit modelling and optogenetics to investigate how short-term adaptation in layer 2/3 of mouse V1 is shaped by stimulus familiarity and reward association. Habituation reduced the fraction of pyramidal cells responsive to a visual stimulus, whereas reward association maintained overall responsivity. In contrast, both forms of learning shifted pyramidal cell adaptation from depression toward sensitization, but through distinct circuit mechanisms. Habituation reduced disinhibition through the VIP[->]SST[->]PC pathway by weakening feedback activation of VIPs and VIP[->]SST connections. Reward association counteracted this effect by increasing disinhibition through the SST[->]PV[->]PC pathway, strengthening SST[->]PV connections while reducing SST[->]PC inputs. Despite engaging distinct disinhibitory circuits and producing divergent effects on pyramidal cell responsivity, both forms of learning converged on a reduced PV:SST input ratio to pyramidal cells, thereby biasing V1 toward sensitizing adaptation. These results identify changes in cortical circuits that link the plasticity of fast adaptation to simple forms of learning.

NPAS4 is an activity-dependent transcription factor that, in CA1 of the hippocampus, regulates inhibitory synapses made onto the active pyramidal neuron. In principle, NPAS4 thereby allows the past activity of a neuron to influence how it encodes information, although this has not yet been demonstrated. Here, we generated a sparse, CA1-specific knockout (KO) of NPAS4 in the mouse hippocampus and used optogenetic tagging to identify KO neurons in vivo. Recordings from intermingled wild-type (WT) and KO neurons in awake behaving animals revealed that NPAS4 deletion degrades spatial representations and temporal precision of spiking: KO neurons exhibited larger place fields with reduced in-field firing and increased out-of-field firing, less stable place fields, reduced coupling to local field potential theta oscillations, and diminished phase precession. These findings demonstrate that NPAS4 plays a crucial role in refining the spatial and temporal properties of CA1 pyramidal neuron spikes, which themselves are thought to be fundamental building blocks of more complex processes such as learning and memory.

Dexterous hand function underlies many essential human activities, from tool use to expression through gestures. Coordinated digit movements are enabled by the intricate musculature of the hand and forearm, which also imposes mechanical coupling between the digits and wrist, constraining their independent control. It remains unclear whether motor cortex inherits these constraints in its activity or encodes digit and wrist independently. To address this problem, we asked individuals with intracortical microelectrode arrays implanted in motor cortex to attempt flexion and extension of individual digits, either in isolation or in combination with attempted wrist movements. We could accurately decode which digit was moving based on cortical recordings, and channels selective for digit identity were arranged somatotopically across the recording arrays. Nevertheless, the activity during flexion or extension overlapped between digits, and movement direction of a given digit could be reliably inferred by a decoder trained on movements of other digits. This directional signal was largely invariant to the digits initial posture. The population axis describing digit movement direction was aligned with the axes associated with wrist flexion-extension or pronation-supination. This alignment persisted during simultaneous wrist and digit movements, which complicated efforts to control them individually. However, by decoding wrist and digit motion from activity orthogonal to the shared direction axis, a participant was able to achieve continuous control of virtual hand movements with improved speed and reduced unintended movements. Together, the results identify both a code for digit identity and a low-dimensional flexion-extension signal which is shared across the digits and wrist. This arrangement is consistent with muscle-like biomechanical constraints on motor cortical activity, which must be accounted for to improve coordinated BCI control.

Emerging evidence suggests that hippocampus contributes to visual short-term memory (VSTM). However, the role of hippocampal ripple activity--brief high-frequency oscillations associated with memory replay--in supporting VSTM of naturalistic objects remains largely unknown. Here, using intracranial EEG recordings from human participants performing a delayed match-to-sample task, we found that hippocampal ripple rates progressively ramped up during the maintenance period and supported successful VSTM. More critically, hippocampal ripples were temporally coupled with the ripples in the lateral temporal lobe (LTL), and these coupled ripples were associated with the memory reactivation in the LTL. These findings provide direct evidence that hippocampal-neocortical interaction via coupled ripples supports VSTM, extending the hippocampal ripples role to short-term mnemonic processes.

The dorsal anterior midline thalamus (aMT) consists of several closely packed nuclei that are important for motivated and emotional behavior. Previous work on aMT has focused on cells and synapses in the paraventricular thalamus (PVT), and little is known about the adjacent paratenial thalamus (PT). Here we examine neural circuits involving PT using a combination of molecular profiling, anatomical tracing, electrophysiology, and optogenetics. We first find that Protein Kinase C-delta (PKCd) selectively labels thalamocortical (TC) cells concentrated in PT but largely absent from neighboring PVT. We show that TC cells in PT project to the infralimbic region (IL) of the medial prefrontal cortex (mPFC), where they contact and drive L2/3 pyramidal cells. In return, we find that IL mPFC primarily projects to PT over nearby PVT, making connections onto reciprocally connected TC cells. However, these cortical inputs are even stronger onto thalamostriatal (TS) and thalamoamygdala (TA) cells, allowing the mPFC to broadcast to the subcortex. Together, our findings help to parcellate aMT, highlight PT as a distinct thalamic nucleus, establish reciprocal connectivity between PT and IL mPFC, and show cortico-thalamic throughput to the subcortex.

Human language depends on highly specialized left-hemispheric brain networks. Damage to these networks causes severe language impairments (aphasia), one of the most common, debilitating and costly consequences of left-hemispheric brain injury, especially stroke. The limited recovery of aphasia despite intensive rehabilitation efforts emphasizes the need to understand the basis of residual language abilities at the single-neuron level, which has remained unexplored so far. Here, we report large-scale microelectrode recordings with single-unit resolution over a period of ten months from the right-hemispheric prefrontal and parietal association cortex of an individual with stroke-induced chronic non-fluent aphasia. Single neurons exhibited regionally specific responses during comprehension, retrieval and articulation of words, the core operations of language. Distinct subpopulations encoded linguistic information in a task-specific manner, despite correlated firing patterns across tasks. Both single-neuron activity and temporally coordinated population dynamics were predicted by semantic and phonological embeddings derived from large language models (LLMs), revealing a regional dissociation in which semantic features preferentially accounted for prefrontal activity and phonological features for parietal activity. Our findings suggest that right-hemispheric circuits, homotopic to the left language network, can support language processing through structured, functionally organised activity at the level of single neurons. This study opens an avenue for developing mechanistically specific neurorehabilitation and neurorestorative strategies for aphasia, such as brain-computer interfaces (BCIs), that leverage right-hemispheric language resources.

Accumulating evidence suggests that cardiac information influences neuronal activity and fundamental brain functions across species. Yet, how the brain represents cardiovascular signals and how this may impact the processing of emotion states remains unclear. Here, we show that intracellular dynamics and single unit activity in the posterior insular cortex (pInsCtx) are precisely regulated by cardiac signals, with individual neurons tuned to heartbeats in a frequency-dependent manner. Furthermore, heartbeat tuning in the pInsCtx occurred preferentially during the first phase of the cardiac cycle, at systole. Heartbeat tuning increased during both appetitive and aversive emotion states, a process not simply explained by increases in heart rate. Manipulation of sympathetic cardiac arousal using beta-adrenergic blockade disrupted neuronal encoding of emotion states in the pInsCtx and blunted behavioral and bodily emotion expression. These findings reveal precise sensory coding of cardiovascular signals in the pInsCtx, which supports its role in encoding emotion states.

AO_SCPLOWBSTRACTC_SCPLOWAuditory-motor learning is critical in mastering the production of complex sounds, such as speaking and playing music. It is anchored upon internal models of interactions between actions and their sensory consequences, which are fine-tuned by minimizing the errors between the predicted and received sound. Here, we applied the concept of surprisal to a piano-playing task to probe the neural dynamics of sensorimotor learning. Specifically, during play, the key-pitch map was changed unpredictably among three map configurations: normal, inverted, and shifted-inverted. At the change boundaries, a signature of violated motor-to-auditory predictions was found in the auditory evoked responses at N100 which could not be attributed to either purely auditory surprisals or motor execution errors. This surprisal is modulated by short-term context, with greater surprise following longer periods of no map change, indicating that the brain continuously tracks short-term map contexts and rapidly adapts to them. In contrast, 30 minutes of extended goal-directed training on a single map modulated P50 amplitude only for that map, which can be explained by a slow, persistent modulation of motor predictions from the auditory signals. Hence, while auditory predictions from motor actions are rapidly and implicitly learned within short-term contexts, the complementary process of adjusting motor inferences from auditory inputs requires targeted training sustained over time. Our approach of studying auditory-motor surprisal in time-varying sequences reveals that auditory-motor learning is fast, context-sensitive, and shaped by both short- and long-term experience. Significance statementUnderstanding how the brain links motor actions with their sensory consequences is key to explaining how complex skills are acquired and how they adapt to changing environments. Prior work has shown that short-term sensory feedback supports rapid adaptation. Yet, the neural mechanisms underpinning the evolution of internal sensorimotor associations across different stages of learning remain to be elucidated. We address this challenge by extending the concept of surprisal, traditionally used in studies of perception, to the sensorimotor domain. Results show that surprisal responses are modulated by both short-term sensory feedback and longer-term training, suggestive of two distinct neural mechanisms underlying sensorimotor learning. These findings advance our understanding of the neural dynamics of sensorimotor learning and inform development of technologies that interface with sensorimotor systems, such as virtual reality and brain-machine interfaces.

Adaptive behavior depends on the ability to rapidly evaluate options and select those that promise the greatest benefit. Such decisions rely on neural representations of value distributed across multiple brain regions, including the orbitofrontal (OFC) and ventromedial prefrontal cortex (vmPFC), yet the neural code underlying these value representations remains unresolved. The dominant account proposes that OFC/vmPFC neurons encode value through a linear rate code, resulting in a single point estimate at the population level. However, this framework is difficult to reconcile with the heterogeneous tuning observed in individual OFC/vmPFC neurons, which can exhibit both positive and negative correlations with subjective value. Here, we test the alternative hypothesis--derived from theories of neural coding in perceptual systems-- that the OFC/vmPFC implements a probabilistic population code based on non-linear tuning functions. Such tuning allows population activity to represent not only subjective value but also the uncertainty surrounding it, in the form of a flexible posterior probability distribution. Using a population receptive field framework, we fitted non-linear value-tuning functions to functional magnetic resonance imaging data acquired during a value judgment task. Bayesian inversion of this encoding model enabled robust out-of-sample decoding of subjective value across several brain regions, including the OFC/vmPFC. Importantly, value uncertainty estimated from decoded medial OFC/vmPFC posteriors predicted within-subject preference instability, choice stochasticity, and confidence in option values, demonstrating its behavioral relevance and suggesting that participants had conscious access to this information. Complementary single-unit recordings from a subset of monkey OFC neurons similarly revealed nonlinear value-selective tuning. Together, these findings establish a probabilistic, non-linear population code for value in the OFC/vmPFC. This provides a neural foundation for the probabilistic code through which value, and uncertainty about value, can guide choice.

Motor control depends on the continuous integration of motor and sensory signals to maintain accurate estimates of body state, yet neural evidence for this integration remains elusive. Here, we investigated the interaction of motor corollary discharge and proprioceptive feedback signals in area 2 of monkey somatosensory cortex during voluntary and externally-perturbed reaching tasks. Though single neurons had mixed responses to corollary discharge and sensory feedback, we disentangled these signals at the population level to discover they occupy approximately orthogonal subspaces. Integrating information across these subspaces enabled accurate body state estimation prior to feedback arrival during voluntary movements. Moreover, the orthogonal population geometry of corollary discharge and sensory feedback signals enabled the decoding of external perturbations. Together, these results identified orthogonality as a population-level coding strategy for flexible integration of motor and sensory signals in somatosensory cortex.