Replay and compositional computation

Medial temporal cortex supports object perception by integrating over visuospatial sequences

Perception unfolds across multiple timescales. For humans and other primates, many object-centric visual attributes can be inferred 'at a glance' (i.e., given <200 ms of visual information), an ability supported by ventral temporal cortex (VTC). Other perceptual inferences require more time; to determine a novel object's identity, we might need to represent its unique configuration of visual features, requiring multiple 'glances.' Here we evaluate whether medial temporal cortex (MTC), downstream from VTC, supports object perception by integrating over such visuospatial sequences. We first compare human visual inferences directly to electrophysiological recordings from macaque VTC. While human performance 'at a glance' is approximated by a linear readout of VTC, participants radically outperform VTC given longer viewing times (i.e., >200 ms). Next, we leverage a stimulus set that enables us to characterize MTC involvement in these temporally extended visual inferences. We find that human visual inferences 'at a glance' resemble the deficits observed in MTC-lesioned human participants. By measuring gaze behaviors during these temporally extended viewing periods, we find that participants sequentially sample task-relevant features via multiple saccades/fixations. These patterns of visuospatial attention are both reliable across participants and necessary for MTC-dependent visual inferences. These data reveal complementary neural systems that support visual object perception: VTC provides a rich set of visual features 'at a glance', while MTC is able to integrate over the sequential outputs of VTC to support object-level inferences.

The predictive nature of spontaneous brain activity across scales and species

Emerging research suggests the brain operates as a "prediction machine," continuously anticipating sensory, motor, and cognitive outcomes. Central to this capability is the brain's spontaneous activity-ongoing internal processes independent of external stimuli. Neuroimaging and computational studies support that this activity is integral to maintaining and refining mental models of our environment, body, and behaviors, akin to generative models in computation. During rest, spontaneous activity expands the variability of potential representations, enhancing the accuracy and adaptability of these models. When performing tasks, internal models direct brain regions to anticipate sensory and motor states, optimizing performance. This review synthesizes evidence from various species, from C. elegans to humans, highlighting three key aspects of spontaneous brain activity's role in prediction: the similarity between spontaneous and task-related activity, the encoding of behavioral and interoceptive priors, and the high metabolic cost of this activity, underscoring prediction as a fundamental function of brains across species.

A cognitive map for value-guided choice in the ventromedial prefrontal cortex

The prefrontal cortex (PFC) is crucial for economic decision-making. However, how PFC value representations facilitate flexible decisions remains unknown. We reframe economic decision-making as a navigation process through a cognitive map of choice values. We found rhesus macaques represented choices as navigation trajectories in a value space using a grid-like code. This occurred in ventromedial PFC (vmPFC) local field potential theta frequency across two datasets. vmPFC neurons deployed the same grid-like code and encoded chosen value. However, both signals depended on theta phase: occurring on theta troughs but on separate theta cycles. Finally, we found sharp-wave ripples-a key signature of planning and flexible behavior-in vmPFC. Thus, vmPFC utilizes cognitive map-based computations to organize and compare values, suggesting an alternative architecture for economic choice in PFC.

Human Sensorimotor Cortex Reactivates Recent Visuomotor Experience during Awake Rest

The re-emergence of task-related activation patterns during awake rest has been reported to play a role in memory consolidation and perceptual learning. This study aimed to test whether such reactivation occurs in the primary sensorimotor cortex following a visuomotor task. During functional magnetic resonance imaging (fMRI) scanning, 42 healthy participants (13 women and 29 men) learned visuomotor tracking, while a rotational perturbation was introduced between the cursor position and joystick angle. This visuomotor task block was interleaved with a control block, during which participants passively viewed a replay of their previously performed cursor movements. Half of the participants used their right hand, whereas the other half used their left hand to control the joystick. Resting-state scans were acquired before and after the visuomotor task sessions. A multivariate pattern classifier was trained to classify task and control blocks and was then tested on resting-state scans collected before and after the task session. Results revealed a significant increase in the number of volumes classified as "task" during post-task rest compared with pre-task rest, indicating re-emergence of task-related activity. Representational similarity analysis also showed a greater similarity to task-related patterns during the post-task rest period. Furthermore, this effect was specific to the left primary sensorimotor cortex contralateral to the hand used and significantly correlated with motor improvement following rest. Our findings reveal the reactivation of recent task-related experience in the primary sensorimotor cortex.

Awake replay: off the clock but on the job

Hippocampal replay is widely thought to support two key cognitive functions: online decision-making and offline memory consolidation. In this review, we take a closer look at the hypothesized link between awake replay and online decision-making in rodents, and find only marginal evidence in support of this role. By contrast, the consolidation view is bolstered by new computational ideas and recent data, suggesting that (i) replay performs offline fictive learning for later goal-oriented behavior; and (ii) replay tags memories prior to sleep, prioritizing them for consolidation. Based on these recent advances, we favor an updated and refined role for awake replay - that is, supporting prioritized offline learning and tagging outside the hippocampus - rather than a direct, online role in guiding behavior.

Complementary roles for hippocampus and anterior cingulate in composing continuous choice

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. Examining hippocampus and anterior cingulate cortex (ACC) population dynamics during goal switches revealed distinct roles for the two regions in parameterizing continuous controller mixing and meta-control. Hippocampal ensemble dynamics encoded the controller blending dynamics, suggesting it implements a mixing of goal-specific control policies. In contrast, ACC ensemble activity exhibited value-dependent ramping activity before goal switches, linking it to a meta-control process that accumulates evidence for switching goals. Our results suggest that hippocampus and ACC play complementary roles corresponding to a generalizable mixture controller and meta-controller that dictates value dependent changes in controller mixing.

Neural mechanisms of relational learning and fast knowledge reassembly in plastic neural networks

Humans and animals have a striking ability to learn relationships between items in experience (such as stimuli, objects and events), enabling structured generalization and rapid assimilation of new information. A fundamental type of such relational learning is order learning, which enables transitive inference (if A > B and B > C, then A > C) and list linking (A > B > C and D > E > F rapidly 'reassembled' into A > B > C > D > E > F upon learning C > D). Despite longstanding study, a neurobiologically plausible mechanism for transitive inference and rapid reassembly of order knowledge has remained elusive. Here we report that neural networks endowed with neuromodulated synaptic plasticity (allowing for self-directed learning) and identified through artificial metalearning (learning-to-learn) are able to perform both transitive inference and list linking and, further, express behavioral patterns widely observed in humans and animals. Crucially, only networks that adopt an 'active' solution, in which items from past trials are reinstated in neural activity in recoded form, are capable of list linking. These results identify fully neural mechanisms for relational learning, and highlight a method for discovering such mechanisms.

Cognitive maps and schizophrenia

Structured internal representations ('cognitive maps') shape cognition, from imagining the future and counterfactual past, to transferring knowledge to new settings. Our understanding of how such representations are formed and maintained in biological and artificial neural networks has grown enormously. The cognitive mapping hypothesis of schizophrenia extends this enquiry to psychiatry, proposing that diverse symptoms - from delusions to conceptual disorganization - stem from abnormalities in how the brain forms structured representations. These abnormalities may arise from a confluence of neurophysiological perturbations (excitation-inhibition imbalance, resulting in attractor instability and impaired representational capacity) and/or environmental factors such as early life psychosocial stressors (which impinge on representation learning). This proposal thus links knowledge of neural circuit abnormalities, environmental risk factors, and symptoms.

Ensemble reactivations during brief rest drive fast learning of sequences

During motor learning, breaks in practice are known to facilitate behavioural optimizations. Although this process has traditionally been studied over long breaks that last hours to days1-6, recent studies in humans have demonstrated that rapid performance gains during early motor sequence learning are most pronounced after very brief breaks lasting seconds to minutes7-10. However, the precise causal neural mechanisms that facilitate performance gains after brief breaks remain poorly understood. Here we recorded neural ensemble activity in the motor cortex of macaques while they performed a visuomotor sequence learning task interspersed with brief breaks. We found that task-related neural cofiring patterns were reactivated during brief breaks. The rate and content of reactivations predicted the magnitude and pattern of subsequent performance gains. Of note, we found that performance gains and reactivations were positively correlated with cortical ripples (80-120 Hz oscillations) but anti-correlated with β bursts (13-30 Hz oscillations), which ultimately dominated breaks after the fast learning phase plateaued. We then applied 20 Hz epidural alternating current stimulation (ACS) to motor cortex, which reduced reactivation rates in a phase-specific and dose-dependent manner. Notably, 20 Hz ACS also eliminated performance gains. Overall, our results indicate that the reactivations of task ensembles during brief breaks are causal drivers of subsequent performance gains. β bursts compete with this process, possibly to support stable performance.

Offline ensemble co-reactivation links memories across days

Memories are encoded in neural ensembles during learning1-6 and are stabilized by post-learning reactivation7-17. Integrating recent experiences into existing memories ensures that memories contain the most recently available information, but how the brain accomplishes this critical process remains unclear. Here we show that in mice, a strong aversive experience drives offline ensemble reactivation of not only the recent aversive memory but also a neutral memory formed 2 days before, linking fear of the recent aversive memory to the previous neutral memory. Fear specifically links retrospectively, but not prospectively, to neutral memories across days. Consistent with previous studies, we find that the recent aversive memory ensemble is reactivated during the offline period after learning. However, a strong aversive experience also increases co-reactivation of the aversive and neutral memory ensembles during the offline period. Ensemble co-reactivation occurs more during wake than during sleep. Finally, the expression of fear in the neutral context is associated with reactivation of the shared ensemble between the aversive and neutral memories. Collectively, these results demonstrate that offline ensemble co-reactivation is a neural mechanism by which memories are integrated across days.

A hierarchical active inference model of spatial alternation tasks and the hippocampal-prefrontal circuit

Cognitive problem-solving benefits from cognitive maps aiding navigation and planning. Physical space navigation involves hippocampal (HC) allocentric codes, while abstract task space engages medial prefrontal cortex (mPFC) task-specific codes. Previous studies show that challenging tasks, like spatial alternation, require integrating these two types of maps. The disruption of the HC-mPFC circuit impairs performance. We propose a hierarchical active inference model clarifying how this circuit solves spatial interaction tasks by bridging physical and task-space maps. Simulations demonstrate that the model's dual layers develop effective cognitive maps for physical and task space. The model solves spatial alternation tasks through reciprocal interactions between the two layers. Disrupting its communication impairs decision-making, which is consistent with empirical evidence. Additionally, the model adapts to switching between multiple alternation rules, providing a mechanistic explanation of how the HC-mPFC circuit supports spatial alternation tasks and the effects of disruption.

Episodic recombination and the role of time in mental travel

Mental time travel is often presented as a singular mechanism, but theoretical and empirical considerations suggest that it is composed of component processes. What are these components? Three hypotheses about the major components of mental time travel are commonly considered: (i) remembering and imagining might, respectively, rely on different processes, (ii) past- and future-directed forms of mental time travel might, respectively, rely on different processes, and (iii) the creation of episodic representations and the determination of their temporal orientation might, respectively, rely on different processes. Here, we flesh out the last of these proposals. First, we argue for 'representational continuism': the view that different forms of mental travel are continuous with regard to their core representational contents. Next, we propose an updated account of episodic recombination (the mechanism generating these episodic contents) and review evidence in its support. On this view, episodic recombination is a natural kind best viewed as a form of compositional computation. Finally, we argue that episodic recombination should be distinguished from mechanisms determining the temporal orientation of episodic representations. Thus, we suggest that mental travel is a singular capacity, while mental time travel has at least two major components: episodic representations and their temporal orientation. This article is part of the theme issue 'Elements of episodic memory: lessons from 40 years of research'.

A geometrical solution underlies general neural principle for serial ordering

A general mathematical description of how the brain sequentially encodes knowledge remains elusive. We propose a linear solution for serial learning tasks, based on the concept of mixed selectivity in high-dimensional neural state spaces. In our framework, neural representations of items in a sequence are projected along a "geometric" mental line learned through classical conditioning. The model successfully solves serial position tasks and explains behaviors observed in humans and animals during transitive inference tasks amidst noisy sensory input and stochastic neural activity. This approach extends to recurrent neural networks performing motor decision tasks, where the same geometric mental line correlates with motor plans and modulates network activity according to the symbolic distance between items. Serial ordering is thus predicted to emerge as a monotonic mapping between sensory input and behavioral output, highlighting a possible pivotal role for motor-related associative cortices in transitive inference tasks.

Group Coordination Catalyzes Individual and Cultural Intelligence

A large program of research has aimed to ground large-scale cultural phenomena in processes taking place within individual minds. For example, investigating whether individual agents equipped with the right social learning strategies can enable cumulative cultural evolution given long enough time horizons. However, this approach often omits the critical group-level processes that mediate between individual agents and multi-generational societies. Here, we argue that interacting groups are a necessary and explanatory level of analysis, linking individual and collective intelligence through two characteristic feedback loops. In the first loop, more sophisticated individual-level social learning mechanisms based on Theory of Mind facilitate group-level complementarity, allowing distributed knowledge to be compositionally recombined in groups; these group-level innovations, in turn, ease the cognitive load on individuals. In the second loop, societal-level processes of cumulative culture provide groups with new cognitive technologies, including shared language and conceptual abstractions, which set in motion new group-level processes to further coordinate, recombine, and innovate. Taken together, these cycles establish group-level interaction as a dual engine of intelligence, catalyzing both individual cognition and cumulative culture.

Shared structure facilitates working memory of multiple sequences

Daily experiences often involve the processing of multiple sequences, yet storing them challenges the limited capacity of working memory (WM). To achieve efficient memory storage, relational structures shared by sequences would be leveraged to reorganize and compress information. Here, participants memorized a sequence of items with different colors and spatial locations and later reproduced the full color and location sequences one after another. Crucially, we manipulated the consistency between location and color sequence trajectories. First, sequences with consistent trajectories demonstrate improved memory performance and a trajectory correlation between reproduced color and location sequences. Second, sequences with consistent trajectories show neural reactivation of common trajectories, and display spontaneous replay of color sequences when recalling locations. Finally, neural reactivation correlates with WM behavior. Our findings suggest that a shared common structure is leveraged for the storage of multiple sequences through compressed encoding and neural replay, together facilitating efficient information organization in WM.

A mathematical theory of relational generalization in transitive inference

Humans and animals routinely infer relations between different items or events and generalize these relations to novel combinations of items. This allows them to respond appropriately to radically novel circumstances and is fundamental to advanced cognition. However, how learning systems (including the brain) can implement the necessary inductive biases has been unclear. We investigated transitive inference (TI), a classic relational task paradigm in which subjects must learn a relation ([Formula: see text] and [Formula: see text]) and generalize it to new combinations of items ([Formula: see text]). Through mathematical analysis, we found that a broad range of biologically relevant learning models (e.g. gradient flow or ridge regression) perform TI successfully and recapitulate signature behavioral patterns long observed in living subjects. First, we found that models with item-wise additive representations automatically encode transitive relations. Second, for more general representations, a single scalar "conjunctivity factor" determines model behavior on TI and, further, the principle of norm minimization (a standard statistical inductive bias) enables models with fixed, partly conjunctive representations to generalize transitively. Finally, neural networks in the "rich regime," which enables representation learning and improves generalization on many tasks, unexpectedly show poor generalization and anomalous behavior on TI. We find that such networks implement a form of norm minimization (over hidden weights) that yields a local encoding mechanism lacking transitivity. Our findings show how minimal statistical learning principles give rise to a classical relational inductive bias (transitivity), explain empirically observed behaviors, and establish a formal approach to understanding the neural basis of relational abstraction.

A sensory-motor theory of the neocortex

Recent neurophysiological and neuroanatomical studies suggest a close interaction between sensory and motor processes across the neocortex. Here, I propose that the neocortex implements active predictive coding (APC): each cortical area estimates both latent sensory states and actions (including potentially abstract actions internal to the cortex), and the cortex as a whole predicts the consequences of actions at multiple hierarchical levels. Feedback from higher areas modulates the dynamics of state and action networks in lower areas. I show how the same APC architecture can explain (1) how we recognize an object and its parts using eye movements, (2) why perception seems stable despite eye movements, (3) how we learn compositional representations, for example, part-whole hierarchies, (4) how complex actions can be planned using simpler actions, and (5) how we form episodic memories of sensory-motor experiences and learn abstract concepts such as a family tree. I postulate a mapping of the APC model to the laminar architecture of the cortex and suggest possible roles for cortico-cortical and cortico-subcortical pathways.

Dynamic diversity is the answer to proxy failure

We argue that a diverse and dynamic pool of agents mitigates proxy failure. Proxy modularity plays a key role in the ongoing production of diversity. We review examples from a range of scales.

The medial prefrontal cortex leaves the hippocampus when it prepares for the future

Our memories help us plan for the future. In some cases, we use memories to repeat the choices that led to preferable outcomes in the past. The success of these memory-guided decisions depends on close interactions between the hippocampus and medial prefrontal cortex. In other cases, we need to use our memories to deduce hidden connections between the present and past situations to decide the best choice of action based on the expected outcome. Our recent study investigated neural underpinnings of such inferential decisions by monitoring neural activity in the medial prefrontal cortex and hippocampus in rats. We identified several neural activity patterns indicating awake memory trace reactivation and restructuring of functional connectivity among multiple neurons. We also found that these patterns occurred concurrently with the ongoing hippocampal activity when rats recalled past events but not when they planned new adaptive actions. Here, we discussed how these computational properties might contribute to success in inferential decision-making and propose a working model on how the medial prefrontal cortex changes its interaction with the hippocampus depending on whether it reflects on the past or looks into the future.

Emergent neural dynamics and geometry for generalization in a transitive inference task

Relational cognition-the ability to infer relationships that generalize to novel combinations of objects-is fundamental to human and animal intelligence. Despite this importance, it remains unclear how relational cognition is implemented in the brain due in part to a lack of hypotheses and predictions at the levels of collective neural activity and behavior. Here we discovered, analyzed, and experimentally tested neural networks (NNs) that perform transitive inference (TI), a classic relational task (if A > B and B > C, then A > C). We found NNs that (i) generalized perfectly, despite lacking overt transitive structure prior to training, (ii) generalized when the task required working memory (WM), a capacity thought to be essential to inference in the brain, (iii) emergently expressed behaviors long observed in living subjects, in addition to a novel order-dependent behavior, and (iv) expressed different task solutions yielding alternative behavioral and neural predictions. Further, in a large-scale experiment, we found that human subjects performing WM-based TI showed behavior inconsistent with a class of NNs that characteristically expressed an intuitive task solution. These findings provide neural insights into a classical relational ability, with wider implications for how the brain realizes relational cognition.

Prioritizing replay when future goals are unknown

Although hippocampal place cells replay nonlocal trajectories, the computational function of these events remains controversial. One hypothesis, formalized in a prominent reinforcement learning account, holds that replay plans routes to current goals. However, recent puzzling data appear to contradict this perspective by showing that replayed destinations lag current goals. These results may support an alternative hypothesis that replay updates route information to build a "cognitive map." Yet no similar theory exists to formalize this view, and it is unclear how such a map is represented or what role replay plays in computing it. We address these gaps by introducing a theory of replay that learns a map of routes to candidate goals, before reward is available or when its location may change. Our work extends the planning account to capture a general map-building function for replay, reconciling it with data, and revealing an unexpected relationship between the seemingly distinct hypotheses.

Hippocampal sequences span experience relative to rewards

Hippocampal place cells fire in sequences that span spatial environments and non-spatial modalities, suggesting that hippocampal activity can anchor to the most behaviorally salient aspects of experience. As reward is a highly salient event, we hypothesized that sequences of hippocampal activity can anchor to rewards. To test this, we performed two-photon imaging of hippocampal CA1 neurons as mice navigated virtual environments with changing hidden reward locations. When the reward moved, the firing fields of a subpopulation of cells moved to the same relative position with respect to reward, constructing a sequence of reward-relative cells that spanned the entire task structure. The density of these reward-relative sequences increased with task experience as additional neurons were recruited to the reward-relative population. Conversely, a largely separate subpopulation maintained a spatially-based place code. These findings thus reveal separate hippocampal ensembles can flexibly encode multiple behaviorally salient reference frames, reflecting the structure of the experience.

A cognitive map for value-guided choice in ventromedial prefrontal cortex

The prefrontal cortex is crucial for economic decision-making and representing the value of options. However, how such representations facilitate flexible decisions remains unknown. We reframe economic decision-making in prefrontal cortex in line with representations of structure within the medial temporal lobe because such cognitive map representations are known to facilitate flexible behaviour. Specifically, we framed choice between different options as a navigation process in value space. Here we show that choices in a 2D value space defined by reward magnitude and probability were represented with a grid-like code, analogous to that found in spatial navigation. The grid-like code was present in ventromedial prefrontal cortex (vmPFC) local field potential theta frequency and the result replicated in an independent dataset. Neurons in vmPFC similarly contained a grid-like code, in addition to encoding the linear value of the chosen option. Importantly, both signals were modulated by theta frequency - occurring at theta troughs but on separate theta cycles. Furthermore, we found sharp-wave ripples - a key neural signature of planning and flexible behaviour - in vmPFC, which were modulated by accuracy and reward. These results demonstrate that multiple cognitive map-like computations are deployed in vmPFC during economic decision-making, suggesting a new framework for the implementation of choice in prefrontal cortex.

Consistency and Variation in Reasoning About Physical Assembly

The ability to reason about how things were made is a pervasive aspect of how humans make sense of physical objects. Such reasoning is useful for a range of everyday tasks, from assembling a piece of furniture to making a sandwich and knitting a sweater. What enables people to reason in this way even about novel objects, and how do people draw upon prior experience with an object to continually refine their understanding of how to create it? To explore these questions, we developed a virtual task environment to investigate how people come up with step-by-step procedures for recreating block towers whose composition was not readily apparent, and analyzed how the procedures they used to build them changed across repeated attempts. Specifically, participants (N = 105) viewed 2D silhouettes of eight unique block towers in a virtual environment simulating rigid-body physics, and aimed to reconstruct each one in less than 60 s. We found that people built each tower more accurately and quickly across repeated attempts, and that this improvement reflected both group-level convergence upon a tiny fraction of all possible viable procedures, as well as error-dependent updating across successive attempts by the same individual. Taken together, our study presents a scalable approach to measuring consistency and variation in how people infer solutions to physical assembly problems.

Generative replay underlies compositional inference in the hippocampal-prefrontal circuit

Human reasoning depends on reusing pieces of information by putting them together in new ways. However, very little is known about how compositional computation is implemented in the brain. Here, we ask participants to solve a series of problems that each require constructing a whole from a set of elements. With fMRI, we find that representations of novel constructed objects in the frontal cortex and hippocampus are relational and compositional. With MEG, we find that replay assembles elements into compounds, with each replay sequence constituting a hypothesis about a possible configuration of elements. The content of sequences evolves as participants solve each puzzle, progressing from predictable to uncertain elements and gradually converging on the correct configuration. Together, these results suggest a computational bridge between apparently distinct functions of hippocampal-prefrontal circuitry and a role for generative replay in compositional inference and hypothesis testing.

Gamma oscillations and episodic memory

Enhanced gamma oscillatory activity (30-80 Hz) accompanies the successful formation and retrieval of episodic memories. While this co-occurrence is well documented, the mechanistic contributions of gamma oscillatory activity to episodic memory remain unclear. Here, we review how gamma oscillatory activity may facilitate spike timing-dependent plasticity, neural communication, and sequence encoding/retrieval, thereby ensuring the successful formation and/or retrieval of an episodic memory. Based on the evidence reviewed, we propose that multiple, distinct forms of gamma oscillation can be found within the canonical gamma band, each of which has a complementary role in the neural processes listed above. Further exploration of these theories using causal manipulations may be key to elucidating the relevance of gamma oscillatory activity to episodic memory.

Aversive experience drives offline ensemble reactivation to link memories across days

Memories are encoded in neural ensembles during learning and stabilized by post-learning reactivation. Integrating recent experiences into existing memories ensures that memories contain the most recently available information, but how the brain accomplishes this critical process remains unknown. Here we show that in mice, a strong aversive experience drives the offline ensemble reactivation of not only the recent aversive memory but also a neutral memory formed two days prior, linking the fear from the recent aversive memory to the previous neutral memory. We find that fear specifically links retrospectively, but not prospectively, to neutral memories across days. Consistent with prior studies, we find reactivation of the recent aversive memory ensemble during the offline period following learning. However, a strong aversive experience also increases co-reactivation of the aversive and neutral memory ensembles during the offline period. Finally, the expression of fear in the neutral context is associated with reactivation of the shared ensemble between the aversive and neutral memories. Taken together, these results demonstrate that strong aversive experience can drive retrospective memory-linking through the offline co-reactivation of recent memory ensembles with memory ensembles formed days prior, providing a neural mechanism by which memories can be integrated across days.