Neuron

Internal states such as motivation and task engagement influence cognitive functions. Working memory, which maintains information over time, is an essential component of cognition and is modulated by motivation. Here, we show motivational states modulated attractor dynamics that supported working memory. Combining population recordings from mouse medial prefrontal cortex (mPFC) with data-constrained recurrent neural network (RNN) modeling, we found task engagement selectively modulated attractor dynamics within a memory-maintenance subspace, while stimulus-evoked responses remained intact. Reverse-engineering the RNNs revealed that task engagement reorganized the dynamical landscape by stabilizing memory-specific attractors. Specifically, task engagement modulated interactions between neurons to change the attractor dynamics. Finally, gradual changes in behavioral engagement were predicted by continuous modulation of attractor geometry in RNNs and mPFC. Together, these results suggest that internal state modulate working memory function by controlling the dynamical regime of mPFC circuits, providing a mechanistic link between internal state, neural dynamics, and cognitive function.

Learning to navigate a changing environment requires the ability to detect when a reward no longer matches a previous expectation. While the hippocampus is essential for adapting behavior strategies when reward expectations are violated and is known to integrate goal-related information into spatial maps, the precise dynamics by which these maps update during an unexpected reduction in reward magnitude is not well understood. Using longitudinal calcium imaging of neuronal activity in behaving mice, we found that hippocampal CA1 population activity encodes reward magnitude through elevated event rates at high value locations. Within this population, we discovered a specialized group of reward tethered place cells that bind spatial context to reward magnitude. These reward tethered place cells exhibit spatial fields across the environment while simultaneously exhibiting activity anchored to high-value reward locations. Upon reward reduction, CA1 population activity equalizes and these neurons undergo a selective and rapid remapping that precedes behavioral adjustment to the reward downshift. The broader spatial map remains intact, indicating that this change allows the animal to update the value of a goal while preserving a stable representation of its surroundings. This selective reorganization of hippocampal firing patterns could support adaptive decision making by updating internal models of the world when expectations are violated. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=148 SRC="FIGDIR/small/725501v1_ufig1.gif" ALT="Figure 1"> View larger version (66K): org.highwire.dtl.DTLVardef@1725be8org.highwire.dtl.DTLVardef@eff7c1org.highwire.dtl.DTLVardef@72d4b3org.highwire.dtl.DTLVardef@ea3e41_HPS_FORMAT_FIGEXP M_FIG O_FLOATNOGraphical abstract.C_FLOATNO Masala N., Donahue M., etal. C_FIG

Prefrontal-hippocampal communication, supported by both direct and indirect anatomical pathways, is essential for a range of cognitive functions, including spatial navigation. The midline thalamic nucleus reuniens (RE) has been proposed as a key hub along this indirect pathway, coordinating bidirectional interactions between the hippocampus (HC) and prefrontal cortex (PFC). Here, we investigated functional dynamics within the HC-RE-PFC network as adult male rats learned to navigate a complex maze. Initial behavioral analyses revealed three distinct learning phases: exploration, goal-oriented learning, and efficient navigation. Aligning neural data with subject-specific transitions between these phases uncovered distinct neural signatures associated with each learning phase. Notably, the transition from exploratory to goal-directed behavior was accompanied by the emergence of persistent HC-RE beta-band (15-25Hz) interactions, including elevated beta coherence, theta-beta phase-amplitude coupling, and HC-to-RE Granger causality. The interactions further scaled with navigational efficiency, showing increased HC-RE beta coherence in trials without errors. Together, these findings provide new evidence for dynamic HC-RE interactions during goal-directed navigation and support an emerging view that RE functions as a working memory buffer for route-related information, revealing a potential network mechanism underlying flexible spatial behavior. HighlightsO_LIBeta-band hippocampus-reuniens coupling emerges during goal-directed navigation. C_LIO_LIPAC between HC theta and RE beta increased during goal learning. C_LIO_LIHC-RE coupling strength increases with navigational efficiency. C_LIO_LIDirectional hippocampus-to-reuniens communication emerges during learning C_LIO_LIReuniens supports hippocampal-prefrontal integration for flexible behavior C_LI Significance StatementFlexible spatial navigation requires coordinated communication between the hippocampus (HC) and the prefrontal cortex (PFC), yet the thalamic mechanisms coordinating HC-PFC interaction remain poorly understood. This study reveals that the nucleus reuniens (RE) of the midline thalamus plays a critical role in mediating HC-PFC communication during spatial learning. By combining detailed behavioral analysis with simultaneous multi-site electrophysiological recording, we identify beta-band HC-RE coupling as a neural signature of the transition from exploration to goal-directed navigation. The strength and directionality of this interaction tracked improvements in navigational efficiency, revealing a thalamic mechanism for integrating hippocampal output into prefrontal circuits. These findings highlight the RE as a working memory buffer for route-related information and advance our understanding of how thalamic hubs contribute to the circuit-level dynamics underlying flexible behavior.

From birth, human infants engage in multi-modal social exchanges with caregivers that involve the coordination of gaze and touch to guide attention and support neurodevelopment. However, little is known about the association between these first interactive experiences and the functional organization of the developing brain during the first postnatal month, a window of remarkable brain growth in humans. We address this gap by combining microanalytic coding of caregiver- infant interactions with task-free functional connectivity (FC), measured using high-density diffuse optical tomography (HD-DOT) in infants homes during the first postnatal month. Task-free FC measures the intrinsic functional organization of the developing brain, shedding light on the early development of neural systems supporting perception, regulation, and social interactions. Infants were assessed up to three times (1 week, 2 weeks, 1 month), enabling characterization of both early FC and its rapid developmental change. Caregiver-infant interactions were associated with both concurrent organization and rapid longitudinal change in FC. Dyadic engagement in the context of face-to-face interaction was associated with the refinement of short-range connectivity and the integration of long-range connectivity particularly between social brain regions, while affectionate touch was associated with general increases in long-distance connectivity. These results demonstrate that caregiving experiences influence the development of the brains functional architecture in the first postnatal month, highlighting a critical window for shaping infant brain function. Significance StatementThe first postnatal weeks are a period of rapid brain development, when the brain may be especially sensitive to experience. A key early experience is social interaction between infants and their caregivers, yet it remains unknown if these interactions influence neonate brain function. By combining observations of caregiver-infant interactions at one month with repeated home-based brain imaging, we show that variations in dyadic engagement and affectionate touch influence how infant brain functional connectivity is organized and develops. These findings highlight the importance of supporting perinatal life and early interactions for infant brain development.

Tau is a microtubule-associated protein with diverse roles in the healthy brain but contributes to neurodegenerative disorders when dysregulated. Although tau ablation has shown protective effects in several disease models, how its absence confers this protection remain unclear. Here, we performed and analyzed single-nucleus RNA sequencing on cortices of aged tau knockout (Mapt-/-) mice in a wild type background as well as in a vascular amyloid model to evaluate the effect on disease context. Comparisons in a wild type setting revealed that tau ablation induced compensatory remodeling across multiple cell types. Excitatory neurons expanded into a distinct subtype with unique glutamatergic signaling, astrocytes adopted synaptoprotective states, oligodendrocytes upregulated genes supporting connectivity and plasticity, and microglia engaged structural remodeling programs. In contrast, in disease, tau removal not only restored functions disrupted by vascular amyloid pathology, but also generated new phenotypes. Excitatory neurons rewired receptor and postsynaptic signaling, astrocytes and oligodendrocytes recovered wild-type-like gene programs related to neurotransmitter cycling, synaptic support, and myelin integrity, and microglia reprogrammed toward sensing and mounting responses. Together, these findings demonstrate that tau ablation reshapes brain cellular programs in a context-dependent manner, exerting adaptive responses in the otherwise healthy brain while restoring homeostatic functions under vascular amyloid pathology. These results position tau as a key regulator of neuronal-glial network balance and highlight the importance of understanding how tau influences distinct cellular programs within specific disease environments.

Making decisions in complex, real-world environments is challenging. Biologically plausible strategies like reinforcement learning (RL) require attention toward reward-predictive stimuli to define task states, yet how attention and decision processes coordinate in the human brain remains unclear. We hypothesized this arises through interactions between orbitofrontal (OFC) value-based mechanisms and lateral prefrontal (LPFC) attention filtering. To test this, we combined behavioral modeling with local field potential (LFP) and single-unit recordings in 22 subjects performing a multidimensional RL task. Reward expectations were encoded in OFC and LPFC, as reflected in high-frequency LFP and OFC single-unit spiking, but modulated by attention only in LPFC. Theta LFPs encoded reward expectations and indexed attention-dependent LPFC-OFC coordination, with value-related coupling emerging pre-choice in high-attention subjects and post-choice in low-attention subjects. These findings show that prefrontal circuits dynamically coordinate to encode attention-weighted value signals, shaping state representations and providing a tractable solution to learning in complex environments.

Primate motor cortex (M1) contains distinct zones with dense projections to the spinal circuits of arm and hand muscles making them critical substrates for manual behavior. But how does neural activity encoding manual actions map in space and time onto these zones? We addressed this question by recording single unit activity (n=1,573) throughout M1 in two macaque monkeys performing a reach-to-grasp task. Recordings were made with linear electrode arrays that were registered to detailed motor maps obtained with intracortical microstimulation. When units were grouped by somatotopic location, the time-resolved profiles from the arm and hand zones closely resembled time-lagged versions of the corresponding muscle activity. Thus, activity in both M1 zones differentiated between task phases and target objects. Unlike the arm and hand muscles, however, neural activity did not differ significantly between M1 arm and hand zones. In contrast, examining the spatial organization of neural selectivity for task phase revealed clear functional distinctions: reach-selectivity was stronger in the arm zone than in the hand zone and manipulate-selectivity was stronger in the hand zone than in the arm zone. Similarly, task condition decoding from hand zone activity was more accurate than from arm zone activity. Our findings collectively show that the encoding of reach-to-grasp movements is spatially clustered within the M1 forelimb representation and that the clusters are selective for function. These distinctions are graded rather than categorical, but they reveal tighter coupling between the spatio-temporal organization of M1 single unit activity and underlying cortical structure than generally assumed.

Dopamine is essential for basal ganglia function. Striatal dopamine release can be triggered by dopamine cell firing, but also by coordinated cholinergic interneuron activity, which stimulates dopamine release via presynaptic nicotinic acetylcholine receptors on dopamine axons. While acetylcholine-dependent dopamine release is well-documented ex vivo and under artificial optogenetic stimulation in vivo, its role during natural behavior has remained unclear. One possible natural driver of acetylcholine-dependent dopamine release is thalamic input, which provides strong excitatory drive to cholinergic interneurons. To examine whether thalamic input provokes acetylcholine-dependent dopamine release during behavior, we performed simultaneous fiber photometry recordings of striatal dopamine (GRAB-rDA3m) and thalamic axon activity (gCaMP8m) in the dorsomedial (DMS) and dorsolateral striatum (DLS) of mice learning the accelerating rotarod, a striatal-dependent task that demands precise and effortful motor control. Recordings were obtained on- and off-task and across days of training to capture the full arc of learning. Dopamine transients in DMS, but not DLS, were frequently coupled to peaks in thalamic axon activity via an acetylcholine-dependent mechanism. The occurrence of these thalamic-evoked dopamine transients depended on learning, task engagement, and the recent history of striatal dopamine activity, but did not appear to signal motor errors. Together, these findings establish thalamic input as a physiological driver of acetylcholine-dependent dopamine release. Moreover, they reveal that striatal sensitivity to this local release mechanism is dynamically gated by dopaminergic history, providing a compelling framework for understanding how local and soma-triggered dopamine signals are coordinated to support learning.

Separating meaningful sensory stimuli from irrelevant ones requires learning sensorimotor associations, but how sensory-linked striatal circuits acquire and maintain these associations is unclear. We longitudinally imaged direct- and indirect-pathway (D1 and A2a) spiny projection neurons (SPNs) in the tail of the striatum (TS) as mice learned to push or pull a joystick in response to auditory cues in either a stimulus-response association (go/omit) task or a two-alternative forced choice (2AFC) task. Learning in both tasks increased the fraction and strength of task-modulated TS SPNs across the sound, action, and reward epochs, yet individual neuron selectivity often switched over days between behavioral epochs. In spite of individual neuron variability, population activity of direct and indirect pathways became aligned with characteristic behavioral features during learning: D1-SPNs dominated the action category, A2a-SPNs were biased toward the mixed category (multiple epochs), and both SPN types showed sound category specificity that depended on the sound-action association. Trial-wise modeling revealed a reweighting of behavioral predictors within the action window, with reward gaining and movement losing predictive weight. Learning the two-choice task led to a higher prevalence of association-preferring neurons and better behavioral decoding within the sound window than in the action/reward window, reflecting a task-dependent prioritization of sensory information. Association-preferring neurons also showed a stable local distance-similarity relationship, with nearby neurons more similar than distant neurons across learning. Together, our results support a population mechanism in TS during learning in which neurons from both direct and indirect pathways are recruited and take on distinct behavioral roles that vary with performance and task complexity.

Animals facing threat must integrate multiple streams of information -- about danger, environment, and internal state -- into a time-pressured escape decision. In mice, this computation is performed by glutamatergic neurons of the dorsal periaqueductal grey (dPAG), but how their convergent long-range inputs combine to drive flexible decisions is unknown. Here we find that the functional weight of each input is set predominantly by the temporal statistics of its presynaptic activity, rather than by pathway identity or synaptic placement. We first used multi-region single unit recordings during naturalistic behaviour and generalised linear models to estimate the functional connectivity from midbrain, hypothalamic, and cortical inputs onto dPAG neurons. We then combined synapse-resolution circuit tracing, two-photon dendritic stimulation with whole-cell somatic and dendritic recordings, and biophysical modelling to identify the mechanisms setting these weights. We found that dPAG neurons are electrotonically compact, generating broadly uniform somatic responses to inputs across the dendritic tree. As a result, presynaptic firing dynamics -- burstiness within neurons and population synchrony -- are the dominant determinants of input efficacy. This temporal-statistics framework accounts for the measured differences in functional connectivity across input regions and predicts that input weights should change dynamically whenever presynaptic temporal structure shifts -- which we confirm by showing rapid, context-dependent reweighting of cortical input during motivational conflict. We propose that the subcellular specialisations of dPAG neurons allow them to integrate signals from distributed sources into a single decision, with input weights that can be flexibly adjusted on behavioural timescales -- a principle that may extend to other brain hubs that compute survival decisions.

Frontal cortex plays critical roles in action control and motor skill learning. Within the layer 1 apical tuft dendrites of layer 5 (L5) neurons in frontal cortex, precise input patterns and back-propagating action potentials can trigger powerful regenerative events that may be essential for flexible computation and learning. However, it remains unclear whether tuft activity in frontal cortical L5 circuits encodes sensorimotor information that differs from the information conveyed by their outputs to downstream targets. Using longitudinal two-photon calcium imaging, we investigated sensorimotor encoding in the apical tuft dendrites and somata of L5 extratelencephalic (ET) neurons in the frontal cortex during learning of a discrete change to a cued dexterous action. During learning, movement errors either triggered corrective action or did not, allowing us to dissociate error signals from signals selective for corrective action. Somatic activity tracked both sensory cues and action, whereas tuft activity predominantly tracked sensory cues. Movement errors during learning revealed additional distinct tuft activity that was selectively associated with corrective actions. Furthermore, learning induced divergent changes in the response gain and net selectivity of tuft dendrites compared to somata. Our measurements uncover systematic differences between the tuft dendrites and somata in sensorimotor selectivity, sensitivity to corrective action, and functional plasticity, providing a foundation for investigating the contributions of dendritic computation to motor skill learning.

Working memory depends on the flexible representation of stimulus information in neural activity, which changes dynamically depending on task. Stimulus transformations are thought to be efficient in use of neural resources and optimal for task performance. However, these transformations are often opaque, and efficiency may conflict with optimal performance. Here we show that in a working memory task requiring selective recall of one of two stimuli based on a context cue, the prefrontal cortex of two male monkeys prioritized efficiency by overwriting information within a shared neural subspace rather than maintaining distinct subspaces for each stimulus. In neural activity and recurrent neural networks such efficiency incurs a cost, in that efficient representations are more prone to errors. Conversely, stimulation of the cholinergic forebrain which improves behavior altered this default mechanism by encoding distinct contexts in higher dimensions. These findings demonstrate a fundamental tradeoff between efficiency and effectiveness in flexibly updating working memory.

Primates pursue goals over extended periods and over multiple steps. The ability to track progress and its rate is critical for such pursuit, yet the neural mechanisms of this tracking are unclear. We recorded single-unit activity from macaque midcingulate cortex (MCC) and lateral prefrontal cortex (LPFC) whilst animals worked for immediate rewards and checked a gauge showing progress towards a larger reward. MCC expressed a temporally extended progress rate signal linked to the long timescale neurons in the region. LPFC expressed progress rate only when inferences on that property were required. In MCC, independent feedback-valence information was also preserved, enabling both readout of work success and rate-specific progress updating. An RNN trained under the same constraints showed that progress rate representations emerge spontaneously, and are critical for tracking goal progress. These results demonstrate a frontal mechanism for inferring and predicting goal progress by extracting progress rate information.

Learning from aversive experiences often generalizes beyond the context in which they occurred. In rodents, a strong aversive event can induce retrospective memory linking (RLI), whereby fear generalizes to a previously neutral context encountered days earlier. Although prior work has shown that RLI is associated with increased co-activity of hippocampal CA1 neurons across neutral and aversive contexts, it remains unclear how broader representational changes support generalization without affecting the ability to discriminate between contexts. Here, we reanalyzed calcium imaging data from dorsal CA1 during RLI to examine how hippocampal representational geometry changes during fear generalization. Using robust, non-parametric measures of population similarity, we show that in mice exhibiting RLI, the representation of the neutral context not only changes over time but becomes more similar to the aversive context during recall. Beyond this similarity increase, we provide evidence for a higher-dimensional geometric transformation consistent with a shared "fear" operation that can be applied across contexts while preserving their identity. Crucially, these two representational signatures dissociate by behavioral state: similarity to the aversive context emerges during freezing, whereas a shared transformation is expressed during active exploration. Together, these findings demonstrate that hippocampal representations support retrospective fear generalization through state-dependent geometric transformations, highlighting representational geometry as a key computational mechanism to resolve the apparent tension between generalization and discriminability.

To comprehend language, the brain must navigate a high-dimensional semantic landscape while seamlessly contextualizing meaning. Inspired by recent advances in the mechanistic interpretability of large language models (LLMs), we hypothesized that the brain utilizes polysemanticity, a coding strategy wherein individual neurons represent multiple semantically unrelated features through high-dimensional superposition (Elhage et al., 2022; Olah et al., 2020). We recorded single-unit activity from the human hippocampus during podcast listening. We found that hippocampal neurons exhibit dense semantic codes characterized by multiple tuning peaks with an overdispersed, isotropic geometry. This geometry satisfies the theoretical requirements for interference minimization in superimposed codes. Furthermore, semantic responses are strongly modulated by lexical and speaker-identity context; nonetheless, the underlying population geometry remains stable. This coding strategy permits rapid contextualization without requiring specialized, context-specific neurons. Indeed, we show clear pattern separation of similar terms, along with pattern completion for held-out words. Together, these results demonstrate that the human brain leverages superposition to solve a universal computational problem: maximizing semantic capacity within a constrained representational space.

Humans solve problems adaptively by selecting strategies suited to the situation. For example, when missing the bus to an appointment, we may wait for the next bus, call a taxi, cancel, or reschedule depending on the circumstances. Yet the neural and computational principles that support such flexible problem solving remain poorly understood. To address this question, we designed a moderately complex decision task for monkeys that allows multiple plausible solution strategies. Animals learned the task rapidly, generalized to novel maze geometries, and their choices were inconsistent with any single fixed strategy. We then recorded large-scale neural activity from the frontal cortex and found that population dynamics varied systematically with maze geometry. Neural responses clustered into two distinct dynamical regimes with separable initial states, consistent with hierarchical and sequential strategies. A decoder trained on population activity revealed time-resolved decision dynamics that aligned with these regimes, and an unsupervised latent-space analysis provided convergent evidence that strategy use varied across trials. A behavioral model grounded in neurally inferred strategies accounted for choices better than fixed-strategy alternatives and captured trial-by-trial variability. Together, these results provide a neural and computational account of how the brain selects and implements distinct strategies during adaptive problem solving.

Free viewing of emotional pictures activates motivational circuits as a function of arousal and valence, yet, because no real outcomes are delivered, anticipatory wanting and consummatory liking cannot be dissociated. Gustatory neuroimaging paradigms deliver actual rewards but use impoverished, affectively neutral cues that do not engage natural selective attention. We bridged these paradigms by presenting emotional images alongside food photographs that either predicted juice delivery (Food+) or did not (Food-), all within a single fMRI session. On each Food+ trial, participants indicated in real time whether they wanted the juice, enabling a within-subject dissociation of anticipatory from consummatory signals. Nucleus accumbens showed a large and selective response to Food+ cues that exceeded activation to both pleasant (erotica) and unpleasant (mutilation) high-arousal images, establishing that mesolimbic engagement tracked outcome prediction rather than emotional arousal, affective valence, or visual content. A temporal dissociation further revealed that nucleus accumbens carried the dominant anticipatory signal during the cue period, while ventromedial prefrontal cortex carried the dominant outcome-period signal at juice delivery, a pattern consistent with the wanting and liking distinction. Representational similarity analysis confirmed that outcome prediction, rather than emotional arousal, affective valence, or visual category, was the dominant organizing principle of the multivariate neural response across the full region-of-interest network. Together, these findings show that whether a visual stimulus engages reward circuitry depends less on what it depicts than on what it predicts, and provide a framework for studying individual differences in appetitive motivation and cue-induced eating. Significance statementWhether a food image engages brain reward circuitry depends not on what it depicts, but on what it predicts. We scanned participants while they viewed food images that either preceded a real opportunity to receive juice or did not, alongside erotic, threatening, and neutral scenes. Nucleus accumbens, a core reward region, responded selectively to food images predicting juice, with a response that exceeded even the response to erotic images. Ventromedial-prefrontal cortex, by contrast, tracked actual juice receipt, dissociating anticipation from consumption. Across a ten-region network, learned reward prediction, rather than emotional arousal, valence, or visual category, organized the neural response. These findings establish a human neuroimaging paradigm for studying how cue-driven motivation goes awry in obesity, addiction, and compulsive eating.

Working memory, the transient maintenance and manipulation of information, is fundamental to human cognition. While working memory is typically thought to rely on frontoparietal cortex, recent neuroimaging evidence suggests the involvement of the cerebellum in a host of cognitive functions, including memory. It is currently unknown whether cerebellar processing is necessary for the persistent maintenance of visual input or if cerebellar signatures of working memory are simply a downstream reflection of cortical activity. Using a combination of functional magnetic resonance imaging (fMRI) and transcranial magnetic stimulation (TMS) in humans, we show that cerebellar perturbation broadly degrades cortical spatial tuning and impairs spatial working memory recall. This impairment matches that observed following perturbation of canonical frontoparietal working memory areas. These findings establish a causal role for the cerebellum in the persistent maintenance of cognitive representations, necessitating a revision of prevailing accounts of human working memory.

The entorhinal cortex (EC) supports a coordinate system for spatial memories, organised in a hierarchy along the EC dorso-ventral axis. Recent theories suggest that a similar coordinate system could scaffold non-spatial memories. Here we show that an abstract hierarchical coordinate system supports arbitrary sequence memories in the human medial temporal lobe (MTL). In single-unit recordings from MTL, we find abstract, coordinate-like coding in a simple sequential memory task. In fMRI we find that abstract coordinate representations are arranged hierarchically along the entorhinal cortex, mirroring the anatomical gradient of grid cells in the rodent EC but now for non-spatial sequences. We replicate this finding in an independent cohort of participants. These data suggest that memories are scaffolded on a hierarchical coordinate system aligned to preserved anatomy across domains and species.

Social dominance hierarchies are a common form of social organization across animal species. We have previously shown that both male and female CD-1 mice form highly linear dominance hierarchies. In mice and other vertebrates, the medial preoptic area (mPOA) is a key hypothalamic sub-region regulating aggressive and defensive behaviors that support hierarchical social structures, but the transcriptional mechanisms in mPOA neurons underlying dominance behaviors and the influence of sex on these neuron populations in the context of social dominance hierarchies remain largely unresolved. Using single-nucleus RNA sequencing (snRNA-seq) to profile mPOA neurons from dominant and subordinate mice, we identified highly consistent social status-dependent changes in the transcriptomes of neuronal nuclei expressing neuropeptide transcripts. Oxytocin expression was remarkably widespread across mPOA neurons, and we found it to be the primary driver of group differences in neuropeptide co-expression networks. Overall, dominant males and females exhibited markedly decreased expression of oxytocin and vasopressin and had a lower proportion of neurons co-expressing multiple neuropeptide transcripts compared to subordinate individuals. In contrast, subordinates displayed widespread reorganization of the transcriptomic neuropeptidome and strikingly enhanced coupling of neuropeptide expression in mPOA neurons. Despite the strong pattern of concordant gene expression in dominant and subordinate individuals, the number of genes that were differentially expressed by status was substantially reduced in males compared to females. In sum, these results demonstrate that maintenance of social status dynamically reconfigures hypothalamic transcriptomic neuropeptidome in a sex-dependent manner and establishes how social status is encoded at a single-cell resolution.