Neuron

Many cortical and subcortical regions contribute to complex social behavior; nevertheless, the network level architecture whereby the brain integrates this information to encode appetitive socioemotional behavior remains unknown. Here we measure electrical activity from eight brain regions as mice engage in a social preference assay. We then use machine learning to discover an explainable brain network that encodes the extent to which mice chose to engage another mouse. This socioemotional network is organized by theta oscillations leading from prelimbic cortex and amygdala that converge on ventral tegmental area, and network activity is synchronized with brain-wide cellular firing. The network generalizes, on a mouse-by-mouse basis, to encode socioemotional behaviors in healthy animals, but fails to encode an appetitive socioemotional state in a high confidence genetic mouse model of autism. Thus, our findings reveal the architecture whereby the brain integrates spatially distributed activity across timescales to encode an appetitive socioemotional brain state in health and disease.

Planning is critical for adaptive behaviour in a changing world, because it lets us anticipate the future and adjust our actions accordingly. While prefrontal cortex is crucial for this process, it remains unknown how planning is implemented in neural circuits. Prefrontal representations were recently discovered in simpler sequence memory tasks, where different populations of neurons represent different future time points. We demonstrate that combining such representations with the ubiquitous principle of neural attractor dynamics allows circuits to solve much richer problems including planning. This is achieved by embedding the environment structure directly in synaptic connections to implement an attractor network that infers desirable futures. The resulting spacetime attractor excels at planning in challenging tasks known to depend on prefrontal cortex. Recurrent neural networks trained by gradient descent on such tasks learn a solution that precisely recapitulates the spacetime attractor - in representation, in dynamics, and in connectivity. Analyses of networks trained across different environment structures reveal a generalisation mechanism that rapidly reconfigures the world model used for planning, without the need for synaptic plasticity. The spacetime attractor is a testable mechanistic theory of planning. If true, it would provide a path towards detailed mechanistic understanding of how prefrontal cortex structures adaptive behaviour.

The thalamus engages in attention by amplifying relevant signals and filtering distractors. Whether architectural features of thalamic circuitry offer a unique locus for attentional control is unknown. We developed a circuit model of excitatory thalamocortical and inhibitory reticular neurons, capturing key observations from task-engaged animals. We found that top-down inputs onto reticular neurons regulate thalamic gain effectively, compared to direct thalamocortical inputs. This mechanism enhances downstream readout, improving detection, discrimination, and cross-modal performance. The model revealed heterogeneous thalamic responses that enable decoding top-down versus bottom-up signals. Spiking activity from task-performing mice supported model predictions, with a similar coding geometry in auditory thalamus and readout strategy in auditory cortex. Dynamical systems analysis explained why reticular neurons are potent sites for control, and how lack of excitatory connectivity among thalamocortical neurons enables separation of top-down from bottom-up signals. Our work reveals mechanisms for attentional control and connects circuit architectures to computational functions.

Primates are deeply social, but the underlying neural basis has been elusive because brain activity is often recorded in artificially constrained conditions. Here, we recorded wireless neural activity simultaneously from two macaque monkeys interacting socially in a natural setting. Neural activity in each monkey encoded key social variables such as allogrooming state, partner identity and joint movements. Surprisingly, neurons in the high-level visual cortex of each monkey continuously tracked a social favor signal, i.e. net duration of allogrooming given versus received, providing a neural basis for grooming reciprocity. In both monkeys, while grooming their partner, neural activity was predicted better by the partners joints and his brain activity. Thus, the receiver of grooming drove the social interaction, not the groomer. Taken together, our findings elucidate the rich and dynamic neural basis of primate social interactions in an ecologically valid setting.

Sensory neuron spiking responses vary across repeated presentations of the same stimuli, but whether this trial-to-trial variability represents noise versus unidentified signals remains unresolved. Some of the variability can be attributed to correlations between neural activity and arousal, locomotion, and other overt movements. We hypothesized that correlations with global activity factors, i.e., patterns of neural activity observable in other brain regions, may explain additional variability in spike count responses of visual cortical neurons. To test this, we used Neuropixels 2.0 probes to record neural activity in mouse primary visual cortex (V1) while subjects passively viewed images. We recorded videos of behavior alongside neural activity from other brain regions, either spiking activity of neural populations in anterior cingulate area (ACA) or widefield calcium signals from across the dorsal cortex. We then used a model based on reduced rank regression to partition the explainable variability of visual cortical responses by source. Some of the trial-to-trial variability in V1 spike counts was attributable to locally shared patterns of activity uncorrelated with either behavior or global activity patterns. Locally shared activity patterns explained trial-to-trial variability that was in excess of Poisson spike generation. Of the parts of variability attributable to non-local sources, global cortical activity predicted significantly more V1 spike count variability than behavioral factors. Additionally, behavioral factors explained little variability uniquely and comprised a geometric subspace of the globally predictable V1 activity. Finally, optogenetically perturbing ACA directly impacted V1 activity, and ACA activity patterns predicted V1 spike count variability even on trials without overt behaviors. Our data indicate that globally shared factors from other cortical areas contribute substantially to shared spike count variability in V1, with only a minority of shared variability confined to local V1 circuits.

Naturalistic, goal directed behavior often requires continuous actions directed at dynamically changing goals. In this context, the closest analogue to choice is a strategic reweighting of multiple goal-specific control policies in response to shifting environmental pressures. To understand the algorithmic and neural bases of choice in continuous contexts, we examined behavior and brain activity in humans performing a continuous prey-pursuit task. Using a newly developed control-theoretic decomposition of behavior, we find pursuit strategies are well described by a meta-controller dictating a mixture of lower-level controllers, each linked to specific pursuit goals. We find that hippocampal neurons encode the policy blending variable in a value-invariant manner and monitor policy switches after they occur. ACC neurons encode policy switches in a value-dependent manner, with value related modulation detectable several hundred ms before the switch, alongside a ramping increase in mean firing rate toward the switch. Meanwhile, OFC activity is consistent with an encoding of the current value structure of the task, rather than policy switching. Together these results are consistent with a tripartite functional division in which hippocampus serves as a controller over behavior, ACC serves as a meta-controller, and OFC provides a value context signal. Overall, our results shed light onto the complex processes associated with choice during naturalistic continuous interactive behavior.

The decision criterion is foundational to theories of decision-making, yet little is known about its neural underpinnings. In memory-based decision making, the criterion sets the minimal memory strength for something to be familiar, but whether or how it is distinctly represented from memory strength is unknown. We recorded single neurons in the medial frontal cortex (MFC) and medial temporal lobe (MTL), both implicated in memory-based decisions, while participants made decisions under different decision criteria. We identified criterion-selective (CS) neurons in the MFC that tracked the criterion regardless of memory strength, and memory-selective (MS) neurons in both regions that tracked memory strength regardless of the criterion. CS neurons signaled the criterion before MS neurons signaled memory strength, and a race model incorporating both neuron types outperformed one using MS neurons alone. These findings reveal two independent cellular substrates, one for the decision criterion and one for memory strength, whose joint activity underlies memory-based decisions.

Natural vision is continuously shaped by an animal's own movements, which determine what enters the visual system and when visual input changes. Gaze shifts are known to initiate a temporally structured sequence of activity in primary visual cortex (V1), but how the visual content sampled by each movement contributes to this sequence has remained unclear. We recorded visual input, eye and head movements, and V1 activity in freely moving mice, and asked how the visual content sampled on each gaze shift shapes the response. The magnitude of visual change induced by each gaze shift scaled the amplitude of responses according to each neuron's characteristic response profile, while movement amplitude alone did not reproduce this modulation in darkness, supporting a role of visual input in driving the sequence. Activity following gaze shifts reflected each neuron's spatial receptive field structure, and visual filters estimated under head-fixed conditions predicted the relative timing of spike responses during gaze shifts of freely moving animals, demonstrating that gaze shift responses encode visual information. At the population level, decoded V1 activity shifted toward the scene sampled after each gaze shift. Thus, across single-neuron, spatial, temporal, and population measures, V1 activity tracked the content of each sample beyond movement timing alone, indicating that gaze shifts act as sampling events that result in V1 encoding visual information in temporally ordered responses.

The mammalian basal ganglia (BG) orchestrate motor, cognitive, and affective functions, yet cell type-specific genetic access remains limited, especially beyond rodents. Key structures implicated in movement and psychiatric disorders, including pallidum, subthalamic nucleus, and dopaminergic midbrain, lack scalable tools for cross-species targeting. Here, we present a comprehensive enhancer-AAV library enabling selective labeling and manipulation of major BG neuronal populations: striatal projection neuron subtypes, pallidal and subthalamic neurons, and midbrain dopaminergic and GABAergic populations. Using an evolutionarily informed discovery pipeline, we identified enhancers targeting canonical, non-canonical, and disease-relevant cell types, with validation demonstrating robust cross-species conservation of specificity between mouse and macaque. Computational modeling revealed sequence features predictive of in vivo performance, including motif grammar, chromatin accessibility, and evolutionary conservation, and identified distinct regulatory architectures across glial, projection, and interneuron lineages. This work establishes a comprehensive cross-species viral toolkit for the BG, unlocking previously inaccessible cell types for circuit dissection.

A prevailing view in affective neuroscience holds that innate and learned behaviors are processed through distinct neuroanatomical pathways, one pre-wired, the other running on synaptic plasticity. However, here we show that processing innate and learned threats in the lateral amygdala deviates fundamentally from this view. We tracked the three core elements of circuit function (excitatory neurons, inhibitory neurons, and neuromodulators) in mice, as they were exposed to an innately aversive looming stimulus and as they learned a cued threat. Tracking the same neurons across sessions, revealed a subpopulation of excitatory neurons recruited by the innate threat that was preferentially potentiated following auditory threat learning. Furthermore, both forms of threat converged on the same modulatory mechanisms: the disinhibitory VIP/SST motif and norepinephrine release, but with a critical difference. While the innately aversive stimulus possessed privileged access to these pathways, the learned cue acquired access through synaptic plasticity. In this instance, learning about a new threat apparently recruits a circuit that protects animals from natural threats.

Decision-making involves dynamic evaluation of competing options. Recording 16,495 neurons across nine frontal and subcortical areas in macaques revealed brain-wide fluctuations between encoding the attributes of available options. These dynamic representations track deliberation and scale with decision difficulty. Instead of binding attribute information of different options together, our analyses show that these representations are associated with dynamic area- and attribute-specific changes in neural geometry.

Arousal is essential for survival, and maladaptive arousal processing leads to an inability to focus, anxiety-like behavior, and dysregulated affective states. Norepinephrine (NE) is known to regulate anxiety, arousal, and learning through locus coeruleus (LC) projections throughout the brain. Evidence for co-release of the NE precursor and neurotransmitter dopamine (DA) from LC neurons has been accumulating for years, yet definitive measures of DA release across regions, stimulus paradigms, and behaviors associated with the LC-NE system remain controversial. Here, we identified the physiological and behavioral properties that evoke DA release from LC axon terminals. Using concomitant approaches, we inhibited the LC and ventral tegmental area (VTA) to selectively isolate the contributions of LC-derived DA release. Together these findings establish the constraints by which LC neurons release DA in a modality-dependent manner.

Flexible behavior requires generalizable memory and learning. For example, we rapidly learn to commute in new cities by reusing our knowledge of Euclidean two-dimensional space and structures like roundabouts and subway systems without forgetting how to get to a favorite restaurant back home. Yet we lack a detailed understanding of how the brain uses existing knowledge to generalize while retaining the memory of specific past experiences. To address this gap, we combine behavioral measurements, neural recordings, and computational modeling in an abstract sequential image navigation task to study three forms of generalization: mnemonic generalization, from visual to mental navigation; transitive generalization, from trained to novel routes; and structural generalization, from familiar to new environments. In contrast to monkeys and humans, recurrent neural networks failed at all generalizations. We found that a structured entorhinal-hippocampal memory model, which provides a content-independent metric scaffold based on grid cells for storing experience, coupled to a policy recurrent network, succeeds at all three. The content-independent scaffold enables mnemonic and transitive generalization through path integration and facilitates structural generalization by allowing reuse of a previously learned action policy network. Moreover, the scaffold's high combinatorial capacity permits continual learning without catastrophic forgetting. We recorded neural activity from the entorhinal cortex and posterior parietal cortex of two monkeys performing the task and found two distinct computations across the neural population. Modularizing an entorhinal and parietal action policy network to separately track distance and initiate actions captured the distinct population dynamics and improved model performance. Finally, we added a reinforcement learning module to the network that enabled it to learn an appropriate scale factor to align the grid periodicity with the environmental temporal structure. Our findings reveal that an architecture which factorizes invariant metric representations from rapid sensory associations and a transferable policy learns, generalizes, and remembers like the brain.

The hippocampus is critical for recollecting and imagining experiences. This is believed to involve voluntarily drawing from hippocampal memory representations of people, events, and places, including the hippocampus map-like representations of familiar environments. However, whether the representations in such "cognitive maps" can be volitionally and selectively accessed is unknown. We developed a brain-machine interface to test if rats could control their hippocampal activity in a flexible, goal-directed, model-based manner. We show that rats can efficiently navigate or direct objects to arbitrary goal locations within a virtual reality arena solely by activating and sustaining appropriate hippocampal representations of remote places. This should provide insight into the mechanisms underlying episodic memory recall, mental simulation/planning, and imagination, and open up possibilities for high-level neural prosthetics utilizing hippocampal representations.

Flexible behavior requires using past experiences to configure cortical computations to suit current task demands. A central question in neuroscience is how memory representations control such reconfigurations. Although hippocampal (HPC) engrams can drive learned behaviors, prior studies have been largely limited to a fixed stimulus-response computation. Thus, whether engrams can drive retrieval of a set of stimulus-response mappings, rather than a specific response itself, remains unresolved. Moreover, how engrams affect cortical task representations and dynamics so as to produce engram-consistent behavior remains unstudied. Here, we address both questions by combining tagging and reactivation of HPC engrams with simultaneous large-scale recordings in medial prefrontal cortex (mPFC) during a context-dependent task-switching paradigm in mice. We report that HPC engram reactivation caused mice to apply engram-consistent decision-rules rather than a specific motor output. Simultaneous mPFC recordings revealed that reactivating the HPC engram for a given context reinstated the representation of that context in mPFC within hundreds of milliseconds, indicating that it was mediated by rapid network effects, leading to choice behavior consistent with the reactivated HPC engram. Other aspects of endogenous dynamics in mPFC were left remarkably intact. Together, our findings provide direct causal evidence that HPC engrams can configure task-relevant population states of downstream cortical circuits in real time, establishing a neural mechanism by which memory traces control flexible behavior.

Substantia nigra pars compacta (SNc) dopaminergic neurons integrate synaptic inputs through dendrites that project laterally within the SNc and ventrally into the substantia nigra pars reticulata (SNr), which then determines dopaminergic output. We use spatially-specific optogenetics to map excitatory subthalamic nucleus (STN) and pedunculopontine nucleus (PPN) as well as inhibitory SNr inputs that provide feedforward inhibition onto SNc neurons. Using multiple GABAergic mouse lines, we find that SNr inputs selectively inhibit SNc somas and proximal dendrites without preference for SNr versus SNc dendrites. Using glutamatergic mouse lines, we find that PPN inputs selectively excite somas and proximal dendrites, while STN inputs selectively excite SNr dendrites, mirroring dendrite-selective inhibition from striosomes. These findings reveal a high level of spatial organization in SNc synaptic inputs and point to separate functional roles for STN versus PPN control of dopaminergic activity. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=147 SRC="FIGDIR/small/742307v1_ufig1.gif" ALT="Figure 1"> View larger version (25K): org.highwire.dtl.DTLVardef@ac8c5borg.highwire.dtl.DTLVardef@32cb89org.highwire.dtl.DTLVardef@d08f79org.highwire.dtl.DTLVardef@7b546d_HPS_FORMAT_FIGEXP M_FIG C_FIG

Psilocybin profoundly alters visual perception, yet the neuronal mechanisms underlying these effects remain unclear. Here we combined large-scale Neuropixels recordings with cell-type specific optogenetics in head-fixed mice performing a visual change-detection task. Psilocybin severely impaired task performance without overt motor deficits. In cortex, the drug modestly suppressed activity of layer 5 neurons while preserving representations of image identity. By contrast, psilocybin imposed a 4-Hz oscillation on visually evoked activity that preferentially affected neurons encoding image change rather than image identity. Under psilocybin, expected image repetitions aberrantly recruited change-encoding ensembles and shifted cortical population dynamics towards trajectories normally evoked by genuine stimulus changes. These effects were strongest in somatostatin-expressing (SST) interneurons in visual cortex. The strength of this modulation depended on image structure and was greatest for images with clear, continuous contours, which preferentially recruited change-encoding ensembles. These findings demonstrate that psilocybin drives internally generated cortical surprise signals, providing a circuit mechanism for altered perception in the acute psychedelic state.

Primates can richly parse sensory inputs to infer latent information, and adjust their behavior accordingly. It has been hypothesized that such flexible inferences are aided by simulations of internal models of the external world. However, evidence supporting this hypothesis has been based on behavioral models that do not emulate neural computations. Here, we test this hypothesis by directly comparing the behavior of humans and monkeys in a ball interception task to that of recurrent neural network (RNN) models with or without the capacity to "simulate" the underlying latent variables. Humans and monkeys had strikingly similar behavioral patterns suggesting common underlying neural computations. Comparison between primates and a large class of RNNs revealed that only RNNs that were optimized to simulate the position of the ball were able to accurately capture key features of the behavior such as systematic biases in the inference process. These results are consistent with the hypothesis that primates use mental simulation to make flexible inferences. Moreover, our work highlights a general strategy for using model neural systems to test computational hypotheses of higher brain function.

Behavior is naturally organized into categorically distinct states with corresponding patterns of neural activity; how does the brain control those states? We propose that states are regulated by specific neural processes that implement meta-control that can blend simpler control processes. To test this hypothesis, we recorded from neurons in the dorsal anterior cingulate cortex (dACC) and dorsal premotor cortex (PMd) while macaques performed a continuous pursuit task with two moving prey that followed evasive strategies. We used a novel control theoretic approach to infer subjects moment-to-moment latent control variables, which in turn dictated their blend of distinct identifiable control processes. We identified low-dimensional subspaces in neuronal responses that reflected the current strategy, the value of the pursued target, and the relative value of the two targets. The top two principal components of activity tracked changes of mind in abstract and change-type-specific formats, respectively. These results indicate that control of behavioral state reflects the interaction of brain processes found in dorsal prefrontal regions that implement a mixture over low-level control policies.

Multiple brain regions need to coordinate activity to support cognitive flexibility and behavioral adaptation. Neural activity in both the hippocampus (HPC) and prefrontal cortex (PFC) is known to represent spatial context and is sensitive to reward and rule alterations. Midbrain dopamine (DA) activity is key in reward seeking behavior and learning. There is abundant evidence that midbrain DA modulates HPC and PFC activity. However, it remains underexplored how these networks engage dynamically and coordinate temporally when animals must adjust their behavior according to changing reward contingencies. In particular, is there any relationship between DA reward prediction change during rule switching, and rule representation changes in PFC and CA1? We addressed these questions using simultaneous recording of neuronal population activity from the hippocampal area CA1, PFC and ventral tegmental area (VTA) in male TH-Cre rats performing two spatial working memory tasks with frequent rule switches in blocks of trials. CA1 and PFC ensembles showed rule-specific activity both during maze running and at reward locations, with PFC rule coding more consistent across animals compared to CA1. Optogenetically tagged VTA DA neuron firing activity responded to and predicted reward outcome. We found that the correct prediction in DA emerged gradually over trials after rule-switching in coordination with transitions in PFC and CA1 ensemble representations of the current rule after a rule switch, followed by behavioral adaptation to the correct rule sequence. Therefore, our study demonstrates a crucial temporal coordination between the rule representation in PFC/CA1, the dopamine reward signal and behavioral strategy. Significance StatementThis study examines neural activity in mammalian brain networks that support the ability to respond flexibly to changing contexts. We use a rule-switching spatial task to examine whether the key reward-responsive and predictive dopamine (DA) activity changes in coordination with changes in rule representations in key cognitive regions, the prefrontal cortex (PFC) and hippocampus. We first established distinct rule representations in PFC and hippocampus, and predictive coding of reward outcomes by DA neuronal activity. We show that the rule-specific DA reward prediction after a rule switch develops in temporal coordination with changes in rule representations in PFC, eventually leading to behavioral changes. These results thus provide an integrated understanding of reward prediction, cognitive representations of rules and behavioral adaptation.