Competition between engrams influences fear memory formation and recall

Not the same as it ever was: A review of memory modification, updating, and distortion in humans and rodents

Memory is a reconstructive and continuous process that enables existing information to be modified in response to a changing environment. Being able to dynamically update outdated memories is critical to an organism's survival. Memory modifications have been extensively studied in both rodents and humans, and prior work has revealed many regional, cellular, neurotransmitter, and subcellular molecular mechanisms underlying this process. However, these diverse bodies of literature have not yet been fully integrated into a comprehensive cross-species review. Integrating the finding across rodent and human work is important for furthering our understanding of memory modifications and the underlying neural mechanisms that support memory modification in both species. Here, we discuss advances in our understanding of adaptive and maladaptive memory modifications in terms of both underlying mechanisms (regional, cellular, and molecular) and behavioral outcomes. By emphasizing findings from both humans and rodents, the two major model systems in which memory modifications have been studied, we are able to highlight converging mechanisms and point to open questions in the field. Specifically, we discuss the major findings from several memory paradigms including declarative, aversive and procedural memory designs and highlight paradigms and models that have been readily translated between rodent and human models. Ultimately, this review identifies key parallels underlying memory updating across species, paradigms, tasks, and models.

Prefrontal encoding of an internal model for emotional inference

A key function of brain systems mediating emotion is to learn to anticipate unpleasant experiences. Although organisms readily associate sensory stimuli with aversive outcomes, higher-order forms of emotional learning and memory require inference to extrapolate the circumstances surrounding directly experienced aversive events to other indirectly related sensory patterns that were not part of the original experience. This type of learning requires internal models of emotion, which flexibly track directly experienced and inferred aversive associations. Although the brain mechanisms of simple forms of aversive learning have been well studied in areas such as the amygdala1-4, whether and how the brain forms and represents internal models of emotionally relevant associations are not known5. Here we report that neurons in the rodent dorsomedial prefrontal cortex (dmPFC) encode a flexible internal model of emotion by linking sensory stimuli in the environment with aversive events, whether they were directly or indirectly associated with that experience. These representations form through a multi-step encoding mechanism involving recruitment and stabilization of dmPFC cells that support inference. Although dmPFC population activity encodes all salient associations, dmPFC neurons projecting to the amygdala specifically represent and are required to express inferred associations. Together, these findings reveal how internal models of emotion are encoded in the dmPFC to regulate subcortical systems for recall of inferred emotional memories.

A sensitive period for the development of episodic-like memory in mice

Episodic-like memory is a later-developing cognitive function supported by the hippocampus. In mice, the formation of extracellular perineuronal nets in subfield cornu ammonis (CA) 1 of the dorsal hippocampus controls the emergence of episodic-like memory during the fourth post-natal week. Whether the timing of episodic-like memory onset is hard-wired, or flexibly set by early-life experiences during a critical or sensitive period for hippocampal maturation, is unknown. Here, we show that the trajectories for episodic-like memory development vary for mice given different sets of experiences spanning the second and third post-natal weeks. Specifically, episodic-like memory precision developed later in mice that experienced early-life adversity, while it developed earlier in mice that experienced early-life enrichment. Moreover, we demonstrate that early-life experiences set the timing of episodic-like memory development by modulating the pace of perineuronal net formation in dorsal CA1, which is dependent on the brain-derived neurotrophic factor (BDNF)-tropomysin receptor kinase B (TrkB) signaling pathway. These results indicate that the hippocampus undergoes a sensitive period during which early-life experiences determine the timing for episodic-like memory development.

On the origin of memory neurons in the human hippocampus

The hippocampus is essential for episodic memory, yet its coding mechanism remains debated. In humans, two main theories have been proposed: one suggests that concept neurons represent specific elements of an episode, while another posits a conjunctive code, where index neurons code the entire episode. Here, we integrate new findings of index neurons in humans and other animals with the concept-specific memory framework, proposing that concept neurons evolve from index neurons through overlapping memories. This process is supported by engram literature, which posits that neurons are allocated to a memory trace based on excitability and that reactivation induces excitability. By integrating these insights, we connect two historically disparate fields of neuroscience: engram research and human single neuron episodic memory research.

Parallel processing of past and future memories through reactivation and synaptic plasticity mechanisms during sleep

Every day, we experience new episodes and store new memories. Although memories are stored in corresponding engram cells, how different sets of engram cells are selected for current and next episodes, and how they create their memories, remains unclear. Here we show that in male mice, hippocampal CA1 neurons show an organized synchronous activity in prelearning home cage sleep that correlates with the learning ensembles only in engram cells, termed preconfigured ensembles. Moreover, after learning, a subset of nonengram cells develops population activity, which is constructed during postlearning offline periods, and then emerges to represent engram cells for new learning. Our model suggests a potential role of synaptic depression and scaling in the reorganization of the activity of nonengram cells. Together, our findings indicate that during offline periods there are two parallel processes occurring: conserving of past memories through reactivation, and preparation for upcoming ones through offline synaptic plasticity mechanisms.

How Fear Memory is Updated: From Reconsolidation to Extinction?

Post-traumatic stress disorder (PTSD) is a psychiatric disorder caused by traumatic past experiences, rooted in the neurocircuits of fear memory formation. Memory processes include encoding, storing, and recalling to forgetting, suggesting the potential to erase fear memories through timely interventions. Conventional strategies such as medications or electroconvulsive therapy often fail to provide permanent relief and come with significant side-effects. This review explores how fear memory may be erased, particularly focusing on the mnemonic phases of reconsolidation and extinction. Reconsolidation strengthens memory, while extinction weakens it. Interfering with memory reconsolidation could diminish the fear response. Alternatively, the extinction of acquired memory could reduce the fear memory response. This review summarizes experimental animal models of PTSD, examines the nature and epidemiology of reconsolidation to extinction, and discusses current behavioral therapy aimed at transforming fear memories to treat PTSD. In sum, understanding how fear memory updates holds significant promise for PTSD treatment.

Integrating Aversive Memories in the Basolateral Amygdala

Decades of research have established a critical role of the basolateral complex of the amygdala (BLA) in the encoding and storage of aversive memories. Much of this work has utilized Pavlovian fear conditioning procedures in which animals experience a single aversive event. Although this effort has produced great insight into the neural mechanisms that support fear memories for an isolated aversive experience, much less is known about how amygdala circuits encode and integrate multiple emotional experiences. The emergence of methods to label and record neuronal ensembles over days allows a deeper understanding of how amygdala neurons encode and integrate distinct aversive episodes over time. Here, we review evidence that the BLA is an essential site for the persistent storage of long-term fear memory. As a long-term storage site for fear memory, a challenge for encoding multiple fear memories is the mechanisms by which BLA neurons allocate, integrate, and discriminate distinct experiences from one another. In this review, we discuss the historical evidence supporting the BLA as a critical site for long-term memory storage, as well as new evidence that stems from technological advances that allow researchers to simultaneously study the encoding and storage of multiple memory traces, including recent versus remote experiences. We explore the possibility that dysfunction in ensemble coding schemes contributes to the pathophysiology of posttraumatic stress disorder and argue that future studies should place increased emphasis on potential subregional differences in memory coding schemes in the amygdala to deepen our understanding of both normal and pathological emotional memory.

Role of brain-derived neurotrophic factor in dysfunction of short-term to long-term memory transformation after surgery and anaesthesia in older mice

BACKGROUND

Memory decline is one of the main manifestations in perioperative neurocognitive disorder. Short-term memory (STM) to long-term memory (LTM) transformation is one aspect of memory consolidation. Early-phase long-term potentiation (E-LTP) to late-phase long-term potentiation (L-LTP) is the molecular correlate of STM to LTM transformation. We examined whether the STM to LTM transformation was impaired after anaesthesia and surgery in older mice.

METHODS

Optogenetics and chemogenetics were used to confirm the role of Vglut1+ glutamatergic neurones in the STM to LTM transformation in older mice. Synaptosomes were isolated to analyse expression of brain-derived neurotrophic factor (BDNF). Golgi-Cox staining and hippocampal field potential recordings were also used to measure synaptic plasticity.

RESULTS

We found that the STM to LTM and E-LTP to L-LTP transformations were impaired after anaesthesia and surgery in older mice, and Vglut1+ excitatory neurone activity in the hippocampal CA1 region was reduced. BDNF expression decreased in the postsynaptic fraction, especially in Vglut1+ neurones, whereas cell-type specific overexpression of BDNF in Vglut1+ neurones reversed postoperative STM to LTM transformation dysfunction in older mice.

CONCLUSIONS

Reduced BDNF expression was involved in anaesthesia and surgery-induced impairment of the STM to LTM transition involving glutamatergic neurones in the hippocampal CA1 region of older mice. This provides a potential target that might be helpful for understanding and developing treatments for postoperative neurocognitive dysfunction.

Erasable hippocampal neural signatures predict memory discrimination

Memories involving the hippocampus can take several days to consolidate, challenging efforts to uncover the neuronal signatures underlying this process. Here, we use calcium imaging in freely moving mice to track the hippocampal dynamics underlying memory consolidation across a 10-day contextual fear conditioning task. We find two neural signatures that emerge following learning and predict memory performance: context-specific place field remapping and coordinated neural activity prior to memory recall (freezing). To test whether these signatures support memory consolidation, we pharmacologically induced amnesia in separate mice by administering anisomycin, a protein synthesis inhibitor, immediately following learning. We find that anisomycin paradoxically accelerates cell turnover. Anisomycin also arrests learning-related remapping and blocks coordinated activity predictive of memory-related freezing behavior, effects that are likewise absent in untreated mice that exhibit poor memory expression. We conclude that context-specific place field remapping and the development of coordinated ensemble activity underlie contextual memory consolidation.

Processing of pain and itch information by modality-specific neurons within the anterior cingulate cortex in mice

Pain and itch are aversive sensations with distinct qualities, processed in overlapping pathways and brain regions, including the anterior cingulate cortex (ACC), which is critical for their affective dimensions. However, the cellular mechanisms underlying their processing in the ACC remain unclear. Here, we identify modality-specific neuronal populations in layer II/III of the ACC in mice involved in pain and itch processing. Using a synapse labeling tool, we show that pain- and itch-related neurons selectively receive synaptic inputs from mediodorsal thalamic neurons activated by pain and itch stimuli, respectively. Chemogenetic inhibition of these neurons reduced pruriception or nociception without affecting the opposite modality. Conversely, activation of these neurons did not enhance stimulus-specific responses but commonly increased freezing-like behavior. These findings reveal that the processing of itch and pain information in the ACC involves activity-dependent and modality-specific neuronal populations, and that pain and itch are processed by functionally distinct ACC neuronal subsets.

Compartmentalized dendritic plasticity in the mouse retrosplenial cortex links contextual memories formed close in time

Events occurring close in time are often linked in memory, and recent studies suggest that such memories are encoded by overlapping neuronal ensembles. However, the role of dendritic plasticity mechanisms in linking memories is unknown. Here we show that memory linking is dependent not only on neuronal ensemble overlap in the mouse retrosplenial cortex, but also on branch-specific dendritic allocation mechanisms. The same dendritic segments are preferentially activated by two linked (but not independent) contextual memories, and spine clusters added after each of two linked (but not independent) contextual memories are allocated to the same dendritic segments. Importantly, we show that the reactivation of dendrites activated during the first context exploration is sufficient to link two contextual memories. Our results demonstrate a critical role for localized dendritic plasticity in memory integration and reveal rules governing how linked and independent memories are allocated to dendritic compartments.

A posterior insula to lateral amygdala pathway transmits US-offset information with a limited role in fear learning

During fear learning, associations between a sensory cue (conditioned stimulus, CS) and an aversive stimulus (unconditioned stimulus, US) are formed in specific brain circuits. The lateral amygdala (LA) is involved in CS-US integration; however, US pathways to the LA remain understudied. Here, we investigated whether the posterior insular cortex (pInsCx), a hub for aversive state signaling, transmits US information to the LA during fear learning. We find that the pInsCx makes a robust, glutamatergic projection specifically targeting the anterior LA. In vivo Ca2+ imaging reveals that neurons in the pInsCx and anterior LA display US-onset and US-offset responses; imaging combined with axon silencing shows that the pInsCx selectively transmits US-offset information to the anterior LA. Optogenetic silencing, however, does not show a role for US-driven activity in the anterior LA or its pInsCx afferents in fear memory formation. Thus, we describe a cortical projection that carries US-offset information to the amygdala with a limited role in fear learning.

Early-life stress induces persistent astrocyte dysfunction associated with fear generalisation

Early-life stress can have lifelong consequences, enhancing stress susceptibility and resulting in behavioural and cognitive deficits. While the effects of early-life stress on neuronal function have been well-described, we still know very little about the contribution of non-neuronal brain cells. Investigating the complex interactions between distinct brain cell types is critical to fully understand how cellular changes manifest as behavioural deficits following early-life stress. Here, using male and female mice we report that early-life stress induces anxiety-like behaviour and fear generalisation in an amygdala-dependent learning and memory task. These behavioural changes were associated with impaired synaptic plasticity, increased neural excitability, and astrocyte hypofunction. Genetic perturbation of amygdala astrocyte function by either reducing astrocyte calcium activity or reducing astrocyte network function was sufficient to replicate cellular, synaptic, and fear memory generalisation associated with early-life stress. Our data reveal a role of astrocytes in tuning emotionally salient memory and provide mechanistic links between early-life stress, astrocyte hypofunction, and behavioural deficits.

The influence of emotion on temporal context models

Temporal context models (TCMs) have been influential in understanding episodic memory and its neural underpinnings. Recently, TCMs have been extended to explain emotional memory effects, one of the most clinically important findings in the field of memory research. This review covers recent advances in hypotheses for the neural representation of spatiotemporal context through the lens of TCMs, including their ability to explain the influence of emotion on episodic and temporal memory. In recent years, simplifying assumptions of "classical" TCMs - with exponential trace decay and the mechanism by which temporal context is recovered - have become increasingly clear. The review also outlines how recent advances could be incorporated into a future TCM, beyond classical assumptions, to integrate emotional modulation.

Stress disrupts engram ensembles in lateral amygdala to generalize threat memory in mice

Stress induces aversive memory overgeneralization, a hallmark of many psychiatric disorders. Memories are encoded by a sparse ensemble of neurons active during an event (an engram ensemble). We examined the molecular and circuit processes mediating stress-induced threat memory overgeneralization in mice. Stress, acting via corticosterone, increased the density of engram ensembles supporting a threat memory in lateral amygdala, and this engram ensemble was reactivated by both specific and non-specific retrieval cues (generalized threat memory). Furthermore, we identified a critical role for endocannabinoids, acting retrogradely on parvalbumin-positive (PV+) lateral amygdala interneurons in the formation of a less-sparse engram and memory generalization induced by stress. Glucocorticoid receptor antagonists, endocannabinoid synthesis inhibitors, increasing PV+ neuronal activity, and knocking down cannabinoid receptors in lateral amygdala PV+ neurons restored threat memory specificity and a sparse engram in stressed mice. These findings offer insights into stress-induced memory alterations, providing potential therapeutic avenues for stress-related disorders.

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.

The Neurobehavioral State hypothesis

Since the early attempts to understand the brain made by Greek philosophers more than 2000 years ago, one of the main questions in neuroscience has been how the brain perceives all the stimuli in the environment and uses this information to implement a response. Recent hypotheses of the neural code rely on the existence of an ideal observer, whether on specific areas of the cerebral cortex or distributed network composed of cortical and subcortical elements. The Neurobehavioral State hypothesis stipulates that neurons are in a quasi-stable state due to the dynamic interaction of their molecular components. This increases their computational capabilities and electrophysiological behavior further than a binary active/inactive state. Together, neuronal populations across the brain learn to identify and associate internal and external stimuli with actions and emotions. Furthermore, such associations can be stored through the regulation of neuronal components as new quasi-stable states. Using this framework, behavior arises as the result of the dynamic interaction between internal and external stimuli together with previously established quasi-stable states that delineate the behavioral response. Finally, the Neurobehavioral State hypothesis is firmly grounded on present evidence of the complex dynamics within the brain, from the molecular to the network level, and avoids the need for a central observer by proposing the brain configures itself through experience-driven associations.

Linking new information to a short-lasting memory trace induces consolidation in the hippocampus

Novelty often influences the retention of nearby weak and transient memory traces, yet its precise role in shaping long-term memory storage remains elusive. Here, we demonstrate that a short-lasting memory is stabilized into a long-lasting one when new information is linked to the weak mnemonic trace in rats, resulting in the formation of long-term memories that are recalled together. An increased overlap between neuronal ensembles and de novo protein synthesis in the dorsal CA1 region of the hippocampus (dCA1) mediates this process. This intricate interconnectedness relies on both temporal and contextual relations between experiences, enhancing the adaptive value of memory consolidation. Finally, this phenomenon is negatively affected by aging, which is associated with reduced ensemble size after novelty exposure and diminished overlap between ensembles in aged dCA1. These findings provide valuable insights into the selectivity and malleability of memory consolidation and its decline during aging.

Erasable Hippocampal Neural Signatures Predict Memory Discrimination

Memories involving the hippocampus can take several days to consolidate, challenging efforts to uncover the neuronal signatures underlying this process. Using calcium imaging in freely moving mice, we tracked the hippocampal dynamics underlying memory formation across a ten-day contextual fear conditioning (CFC) task. Following learning, context-specific place field remapping correlated with memory performance. To causally test whether these hippocampal dynamics support memory consolidation, we induced amnesia in a group of mice by pharmacologically blocking protein synthesis immediately following learning. We found that halting protein synthesis following learning paradoxically accelerated cell turnover and also arrested learning-related remapping, paralleling the absence of remapping observed in untreated mice that exhibited poor memory expression. Finally, coordinated neural activity that emerged following learning was dependent on intact protein synthesis and predicted memory-related freezing behavior. We conclude that context-specific place field remapping and the development of coordinated ensemble activity require protein synthesis and underlie contextual fear memory consolidation.

NMDA Receptors Control Activity Hierarchy in Neural Network: Loss of Control in Hierarchy Leads to Learning Impairments, Dissociation, and Psychosis

While it is known that associative memory is preferentially encoded by memory-eligible "primed" neurons, in vivo neural activity hierarchy has not been quantified and little is known about how such a hierarchy is established. Leveraging in vivo calcium imaging of hippocampal neurons on freely behaving mice, we developed the first method to quantify real-time neural activity hierarchy in the CA1 region. Neurons at the top of activity hierarchy are identified as primed neurons. In cilia knockout mice that exhibit severe learning deficits, the percentage of primed neurons is drastically reduced. We developed a simplified neural network model that incorporates simulations of linear and non-linear weighted components, modeling the synaptic ionic conductance of AMPA and NMDA receptors, respectively. We found that moderate non-linear to linear conductance ratios naturally leads a small fraction of neurons to be primed in the simulated neural network. Removal of the non-linear component eliminates the existing activity hierarchy and reinstate it to the network stochastically primes a new pool of neurons. Blockade of NMDA receptors by ketamine not only decreases general neuronal activity causing learning impairments, but also disrupts neural activity hierarchy. Additionally, ketamine-induced super-synchronized slow oscillation during anesthesia can be simulated if the non-linear NMDAR component is removed to flatten activity hierarchy. Together, this study develops a unique method to measure neural activity hierarchy and identifies NMDA receptors as a key factor that controls the hierarchy. It presents the first evidence suggesting that hierarchy disruption by NMDAR blockade causes dissociation and psychosis.

A small population of stress-responsive neurons in the hypothalamus-habenula circuit mediates development of depression-like behavior in mice

Accumulating evidence has shown that various brain functions are associated with experience-activated neuronal ensembles. However, whether such neuronal ensembles are engaged in the pathogenesis of stress-induced depression remains elusive. Utilizing activity-dependent viral strategies in mice, we identified a small population of stress-responsive neurons, primarily located in the middle part of the lateral hypothalamus (mLH) and the medial part of the lateral habenula (LHbM). These neurons serve as "starter cells" to transmit stress-related information and mediate the development of depression-like behaviors during chronic stress. Starter cells in the mLH and LHbM form dominant connections, which are selectively potentiated by chronic stress. Silencing these connections during chronic stress prevents the development of depression-like behaviors, whereas activating these connections directly elicits depression-like behaviors without stress experience. Collectively, our findings dissect a core functional unit within the LH-LHb circuit that mediates the development of depression-like behaviors in mice.

Synaptic acetylcholine induces sharp wave ripples in the basolateral amygdala through nicotinic receptors

While the basolateral amygdala (BLA) is critical in the consolidation of emotional memories, mechanisms underlying memory consolidation in this region are not well understood. In the hippocampus, memory consolidation depends upon network signatures termed sharp wave ripples (SWR). These SWRs largely occur during states of awake rest or slow wave sleep and are inversely correlated with cholinergic tone. While high frequency cholinergic stimulation can inhibit SWRs through muscarinic acetylcholine receptors, it is unclear how nicotinic acetylcholine receptors or different cholinergic firing patterns may influence SWR generation. SWRs are also present in BLA in vivo. Interestingly, the BLA receives extremely dense cholinergic inputs, yet the relationship between acetylcholine (ACh) and BLA SWRs is unexplored. Here, using brain slice electrophysiology in male and female mice, we show that brief stimulation of ACh inputs to BLA reliably induces SWRs that resemble those that occur in the BLA in vivo. Repeated ACh-SWRs are induced with single pulse stimulation at low, but not higher frequencies. ACh-SWRs are driven by nicotinic receptors which recruit different classes of local interneurons and trigger glutamate release from external inputs. In total, our findings establish a previously undefined mechanism for SWR induction in the brain. They also challenge the previous notion of neuromodulators as purely modulatory agents gating these events but instead reveal these systems can directly instruct SWR induction with temporal precision. Further, these results intriguingly suggest a new role for the nicotinic system in emotional memory consolidation.

Beyond neurons and spikes: cognon, the hierarchical dynamical unit of thought

From the dynamical point of view, most cognitive phenomena are hierarchical, transient and sequential. Such cognitive spatio-temporal processes can be represented by a set of sequential metastable dynamical states together with their associated transitions: The state is quasi-stationary close to one metastable state before a rapid transition to another state. Hence, we postulate that metastable states are the central players in cognitive information processing. Based on the analogy of quasiparticles as elementary units in physics, we introduce here the quantum of cognitive information dynamics, which we term "cognon". A cognon, or dynamical unit of thought, is represented by a robust finite chain of metastable neural states. Cognons can be organized at multiple hierarchical levels and coordinate complex cognitive information representations. Since a cognon is an abstract conceptualization, we link this abstraction to brain sequential dynamics that can be measured using common modalities and argue that cognons and brain rhythms form binding spatiotemporal complexes to keep simultaneous dynamical information which relate the 'what', 'where' and 'when'.

A sensitive period for the development of episodic-like memory in mice

Episodic-like memory is a later-developing cognitive function supported by the hippocampus. In mice, the formation of extracellular perineuronal nets in subfield CA1 of the dorsal hippocampus controls the emergence of episodic-like memory during the fourth postnatal week (Ramsaran et al., 2023). Whether the timing of episodic-like memory onset is hard-wired, or flexibly set by early-life experiences during a critical or sensitive period for hippocampal maturation, is unknown. Here, we show that the trajectories for episodic-like memory development vary for mice given different sets of experiences spanning the second and third postnatal weeks. Specifically, episodic-like memory precision developed later in mice that experienced early-life adversity, while it developed earlier in mice that experienced early-life enrichment. Moreover, we demonstrate that early-life experiences set the timing of episodic-like memory development by modulating the pace of perineuronal net formation in dorsal CA1. These results indicate that the hippocampus undergoes a sensitive period during which early-life experiences determine the timing for episodic-like memory development.

Natural forgetting reversibly modulates engram expression

Memories are stored as ensembles of engram neurons and their successful recall involves the reactivation of these cellular networks. However, significant gaps remain in connecting these cell ensembles with the process of forgetting. Here, we utilized a mouse model of object memory and investigated the conditions in which a memory could be preserved, retrieved, or forgotten. Direct modulation of engram activity via optogenetic stimulation or inhibition either facilitated or prevented the recall of an object memory. In addition, through behavioral and pharmacological interventions, we successfully prevented or accelerated forgetting of an object memory. Finally, we showed that these results can be explained by a computational model in which engrams that are subjectively less relevant for adaptive behavior are more likely to be forgotten. Together, these findings suggest that forgetting may be an adaptive form of engram plasticity which allows engrams to switch from an accessible state to an inaccessible state.

Memory engram stability and flexibility

Many studies have shown that memories are encoded in sparse neural ensembles distributed across the brain. During the post-encoding period, often during sleep, many of the cells that were active during encoding are reactivated, supporting consolidation of this memory. During memory recall, many of the same cells that were active during encoding and reactivated during consolidation are reactivated during recall. These ensembles of cells have been referred to as the memory engram cells, stably representing a specific memory. However, recent studies question the rigidity of the "stable memory engram." Here we review the past literature of how episodic-like memories are encoded, consolidated, and recalled. We also highlight more recent studies (as well as some older literature) that suggest that these stable memories and their representations are much more dynamic and flexible than previously thought. We highlight some of these processes, including memory updating, reconsolidation, forgetting, schema learning, memory-linking, and representational drift.

Neuronal enhancers fine-tune adaptive circuit plasticity

Neuronal activity-regulated gene expression plays a crucial role in sculpting neural circuits that underpin adaptive brain function. Transcriptional enhancers are now recognized as key components of gene regulation that orchestrate spatiotemporally precise patterns of gene transcription. We propose that the dynamics of enhancer activation uniquely position these genomic elements to finely tune activity-dependent cellular plasticity. Enhancer specificity and modularity can be exploited to gain selective genetic access to specific cell states, and the precise modulation of target gene expression within restricted cellular contexts enabled by targeted enhancer manipulation allows for fine-grained evaluation of gene function. Mounting evidence also suggests that enduring stimulus-induced changes in enhancer states can modify target gene activation upon restimulation, thereby contributing to a form of cell-wide metaplasticity. We advocate for focused exploration of activity-dependent enhancer function to gain new insight into the mechanisms underlying brain plasticity and cognitive dysfunction.

Sex and the facilitation of cued fear by prior contextual fear conditioning in rats

Previous studies have shown that the formation of new memories can be influenced by prior experience. This includes work using Pavlovian fear conditioning in rodents that has shown that an initial fear conditioning experience can become associated with and facilitate the acquisition of new fear memories, especially when they occur close together in time. However, most of the prior studies used only males as subjects, resulting in questions about the generalizability of the findings from this work. Here we tested whether prior contextual fear conditioning would facilitate later learning of cued fear conditioning in both male and female rats, and if there were differences based on the interval between the two conditioning episodes. Our results showed that levels of cued fear were not influenced by prior contextual fear conditioning or by the interval between training; however, females showed lower levels of cued fear. Freezing behavior in the initial training context differed by sex, with females showing lower levels of contextual fear, and by the type of initial training, with rats given delayed shock showing higher levels of fear than rats given immediate shock during contextual fear conditioning. These results indicate that contextual fear conditioning does not prime subsequent cued fear conditioning and that female rats express lower levels of cued and contextual fear conditioning than males.

Reactivation-dependent transfer of fear memory between contexts requires M1 muscarinic receptor stimulation in dorsal hippocampus of male rats

Memory updating is essential for integrating new information into existing representations. However, this process could become maladaptive in conditions like post-traumatic stress disorder (PTSD), when fear memories generalize to neutral contexts. Previously, we have shown that contextual fear memory malleability in rats requires activation of M1 muscarinic acetylcholine receptors in the dorsal hippocampus. Here, we investigated the involvement of this mechanism in the transfer of contextual fear memories to other contexts using a novel fear memory updating paradigm. Following brief reexposure to a previously fear conditioned context, male rats (n = 8-10/group) were placed into a neutral context to evaluate the transfer of fear memory. We also infused the selective M1 receptor antagonist pirenzepine into the dorsal hippocampus before memory reactivation to try to block this effect. Results support the hypothesis that fear memory can be updated with novel contextual information, but only if rats are reexposed to the originally trained context relatively recently before the neutral context; evidence for transfer was not seen if the fear memory reactivation was omitted or if it occurred 6 h before neutral context exposure. The transferred fear persisted for 4 weeks, and the effect was blocked by M1 antagonism. These findings strongly suggest that fear transfer requires reactivation and destabilization of the original fear memory. The novel preclinical model introduced here, and its implication of muscarinic receptors in this process, could therefore inform therapeutic strategies for PTSD and similar conditions.

Higher-order interactions between hippocampal CA1 neurons are disrupted in amnestic mice

Across systems, higher-order interactions between components govern emergent dynamics. Here we tested whether contextual threat memory retrieval in mice relies on higher-order interactions between dorsal CA1 hippocampal neurons requiring learning-induced dendritic spine plasticity. We compared population-level Ca2+ transients as wild-type mice (with intact learning-induced spine plasticity and memory) and amnestic mice (TgCRND8 mice with high levels of amyloid-β and deficits in learning-induced spine plasticity and memory) were tested for memory. Using machine-learning classifiers with different capacities to use input data with complex interactions, our findings indicate complex neuronal interactions in the memory representation of wild-type, but not amnestic, mice. Moreover, a peptide that partially restored learning-induced spine plasticity also restored the statistical complexity of the memory representation and memory behavior in Tg mice. These findings provide a previously missing bridge between levels of analysis in memory research, linking receptors, spines, higher-order neuronal dynamics and behavior.

Synaptic rearrangement of NMDA receptors controls memory engram formation and malleability in the cortex

Initially hippocampal dependent, memory representations rely on a broadly distributed cortical network as they mature over time. How these cortical engrams acquire stability during systems-level memory consolidation without compromising their dynamic nature remains unclear. We identified a highly responsive "consolidation switch" in the synaptic composition of N-methyl-d-aspartate receptors (NMDARs), which dictates the progressive embedding and persistence of enduring memories in the rat cortex. Cortical GluN2B subunit-containing NMDARs were preferentially recruited upon encoding of associative olfactory memory to support neuronal allocation of memory engrams. As consolidation proceeds, a learning-induced redistribution of GluN2B subunit-containing NMDARs outward synapses increased synaptic GluN2A subunit contribution and enabled stabilization of remote memories. In contrast, synaptic reincorporation of GluN2B subunits occurred during subsequent forgetting. By manipulating the surface distribution of GluN2A and GluN2B subunit-containing NMDARs at cortical synapses, we uncovered that the rearrangement of GluN2B-containing NMDARs constitutes an essential tuning mechanism that determines the fate of cortical memory engrams and controls their malleability.

Detection of Memory Engrams in Mammalian Neuronal Circuits

It has long been assumed that activity patterns persist in neuronal circuits after they are first experienced, as part of the process of information processing and storage by the brain. However, these "reverberations" of current activity have not been directly observed on a single-neuron level in a mammalian system. Here we demonstrate that specific induced activity patterns are retained in mature cultured hippocampal neuronal networks. Neurons within the network are induced to fire at a single frequency or in a more complex pattern containing two distinct frequencies. After the stimulation was stopped, the subsequent neuronal activity of hundreds of neurons in the network was monitored. In the case of single-frequency stimulation, it was observed that many of the neurons continue to fire at the same frequency that they were stimulated to fire at. Using a recurrent neural network trained to detect specific, more complex patterns, we found that the multiple-frequency stimulation patterns were also retained within the neuronal network. Moreover, it appears that the component frequencies of the more complex patterns are stored in different populations of neurons and neuron subtypes.

Comparing behaviours induced by natural memory retrieval and optogenetic reactivation of an engram ensemble in mice

Memories are thought to be stored within sparse collections of neurons known as engram ensembles. Neurons active during a training episode are allocated to an engram ensemble ('engram neurons'). Memory retrieval is initiated by external sensory or internal cues present at the time of training reactivating engram neurons. Interestingly, optogenetic reactivation of engram ensemble neurons alone in the absence of external sensory cues is sufficient to induce behaviour consistent with memory retrieval in mice. However, there may exist differences between the behaviours induced by natural retrieval cues or artificial engram reactivation. Here, we compared two defensive behaviours (freezing and the syllable structure of ultrasonic vocalizations, USVs) induced by sensory cues present at training (natural memory retrieval) and optogenetic engram ensemble reactivation (artificial memory retrieval) in a threat conditioning paradigm in the same mice. During natural memory recall, we observed a strong positive correlation between freezing levels and distinct USV syllable features (characterized by an unsupervised algorithm, MUPET (Mouse Ultrasonic Profile ExTraction)). Moreover, we observed strikingly similar behavioural profiles in terms of freezing and USV characteristics between natural memory recall and artificial memory recall in the absence of sensory retrieval cues. Although our analysis focused on two behavioural measures of threat memory (freezing and USV characteristics), these results underscore the similarities between threat memory recall triggered naturally and through optogenetic reactivation of engram ensembles. This article is part of a discussion meeting issue 'Long-term potentiation: 50 years on'.

Chromatin plasticity predetermines neuronal eligibility for memory trace formation

Memories are encoded by sparse populations of neurons but how such sparsity arises remains largely unknown. We found that a neuron's eligibility to be recruited into the memory trace depends on its epigenetic state prior to encoding. Principal neurons in the mouse lateral amygdala display intrinsic chromatin plasticity, which when experimentally elevated favors neuronal allocation into the encoding ensemble. Such chromatin plasticity occurred at genomic regions underlying synaptic plasticity and was accompanied by increased neuronal excitability in single neurons in real time. Lastly, optogenetic silencing of the epigenetically altered neurons prevented memory expression, revealing a cell-autonomous relationship between chromatin plasticity and memory trace formation. These results identify the epigenetic state of a neuron as a key factor enabling information encoding.

Experience-dependent information routing through the basolateral amygdala shapes behavioral outcomes

It is well established that the basolateral amygdala (BLA) is an emotional processing hub that governs a diverse repertoire of behaviors. Selective engagement of a heterogeneous cell population in the BLA is thought to contribute to this flexibility in behavioral outcomes. However, whether this process is impacted by previous experiences that influence emotional processing remains unclear. Here we demonstrate that previous positive (enriched environment [EE]) or negative (chronic unpredictable stress [CUS]) experiences differentially influence the activity of populations of BLA principal neurons projecting to either the nucleus accumbens core or bed nucleus of the stria terminalis. Chemogenetic manipulation of these projection-specific neurons can mimic or occlude the effects of CUS and EE on behavioral outcomes to bidirectionally control avoidance behaviors and stress-induced helplessness. These data demonstrate that previous experiences influence the responsiveness of projection-specific BLA principal neurons, biasing information routing through the BLA, to drive divergent behavioral outcomes.

Kdm4a is an activity downregulated barrier to generate engrams for memory separation

Memory engrams are a subset of learning activated neurons critical for memory recall, consolidation, extinction and separation. While the transcriptional profile of engrams after learning suggests profound neural changes underlying plasticity and memory formation, little is known about how memory engrams are selected and allocated. As epigenetic factors suppress memory formation, we developed a CRISPR screening in the hippocampus to search for factors controlling engram formation. We identified histone lysine-specific demethylase 4a (Kdm4a) as a negative regulator for engram formation. Kdm4a is downregulated after neural activation and controls the volume of mossy fiber boutons. Mechanistically, Kdm4a anchors to the exonic region of Trpm7 gene loci, causing the stalling of nascent RNAs and allowing burst transcription of Trpm7 upon the dismissal of Kdm4a. Furthermore, the YTH domain containing protein 2 (Ythdc2) recruits Kdm4a to the Trpm7 gene and stabilizes nascent RNAs. Reducing the expression of Kdm4a in the hippocampus via genetic manipulation or artificial neural activation facilitated the ability of pattern separation in rodents. Our work indicates that Kdm4a is a negative regulator of engram formation and suggests a priming state to generate a separate memory.

Systemic pharmacological suppression of neural activity reverses learning impairment in a mouse model of Fragile X syndrome

The enhancement of associative synaptic plasticity often results in impaired rather than enhanced learning. Previously, we proposed that such learning impairments can result from saturation of the plasticity mechanism (Nguyen-Vu et al., 2017), or, more generally, from a history-dependent change in the threshold for plasticity. This hypothesis was based on experimental results from mice lacking two class I major histocompatibility molecules, MHCI H2-Kb and H2-Db (MHCI KbDb-/-), which have enhanced associative long-term depression at the parallel fiber-Purkinje cell synapses in the cerebellum (PF-Purkinje cell LTD). Here, we extend this work by testing predictions of the threshold metaplasticity hypothesis in a second mouse line with enhanced PF-Purkinje cell LTD, the Fmr1 knockout mouse model of Fragile X syndrome (FXS). Mice lacking Fmr1 gene expression in cerebellar Purkinje cells (L7-Fmr1 KO) were selectively impaired on two oculomotor learning tasks in which PF-Purkinje cell LTD has been implicated, with no impairment on LTD-independent oculomotor learning tasks. Consistent with the threshold metaplasticity hypothesis, behavioral pre-training designed to reverse LTD at the PF-Purkinje cell synapses eliminated the oculomotor learning deficit in the L7-Fmr1 KO mice, as previously reported in MHCI KbDb-/-mice. In addition, diazepam treatment to suppress neural activity and thereby limit the induction of associative LTD during the pre-training period also eliminated the learning deficits in L7-Fmr1 KO mice. These results support the hypothesis that cerebellar LTD-dependent learning is governed by an experience-dependent sliding threshold for plasticity. An increased threshold for LTD in response to elevated neural activity would tend to oppose firing rate stability, but could serve to stabilize synaptic weights and recently acquired memories. The metaplasticity perspective could inform the development of new clinical approaches for addressing learning impairments in autism and other disorders of the nervous system.

Dominant activities of fear engram cells in the dorsal dentate gyrus underlie fear generalization in mice

Over-generalized fear is a maladaptive response to harmless stimuli or situations characteristic of posttraumatic stress disorder (PTSD) and other anxiety disorders. The dorsal dentate gyrus (dDG) contains engram cells that play a crucial role in accurate memory retrieval. However, the coordination mechanism of neuronal subpopulations within the dDG network during fear generalization is not well understood. Here, with the Tet-off system combined with immunostaining and two-photon calcium imaging, we report that dDG fear engram cells labeled in the conditioned context constitutes a significantly higher proportion of dDG neurons activated in a similar context where mice show generalized fear. The activation of these dDG fear engram cells encoding the conditioned context is both sufficient and necessary for inducing fear generalization in the similar context. Activities of mossy cells in the ventral dentate gyrus (vMCs) are significantly suppressed in mice showing fear generalization in a similar context, and activating the vMCs-dDG pathway suppresses generalized but not conditioned fear. Finally, modifying fear memory engrams in the dDG with "safety" signals effectively rescues fear generalization. These findings reveal that the competitive advantage of dDG engram cells underlies fear generalization, which can be rescued by activating the vMCs-dDG pathway or modifying fear memory engrams, and provide novel insights into the dDG network as the neuronal basis of fear generalization.

Fear learning induces synaptic potentiation between engram neurons in the rat lateral amygdala

The lateral amygdala (LA) encodes fear memories by potentiating sensory inputs associated with threats and, in the process, recruits 10-30% of its neurons per fear memory engram. However, how the local network within the LA processes this information and whether it also plays a role in storing it are still largely unknown. Here, using ex vivo 12-patch-clamp and in vivo 32-electrode electrophysiological recordings in the LA of fear-conditioned rats, in combination with activity-dependent fluorescent and optogenetic tagging and recall, we identified a sparsely connected network between principal LA neurons that is organized in clusters. Fear conditioning specifically causes potentiation of synaptic connections between learning-recruited neurons. These findings of synaptic plasticity in an autoassociative excitatory network of the LA may suggest a basic principle through which a small number of pyramidal neurons could encode a large number of memories.

Engram mechanisms of memory linking and identity

Memories are thought to be stored in neuronal ensembles referred to as engrams. Studies have suggested that when two memories occur in quick succession, a proportion of their engrams overlap and the memories become linked (in a process known as prospective linking) while maintaining their individual identities. In this Review, we summarize the key principles of memory linking through engram overlap, as revealed by experimental and modelling studies. We describe evidence of the involvement of synaptic memory substrates, spine clustering and non-linear neuronal capacities in prospective linking, and suggest a dynamic somato-synaptic model, in which memories are shared between neurons yet remain separable through distinct dendritic and synaptic allocation patterns. We also bring into focus retrospective linking, in which memories become associated after encoding via offline reactivation, and discuss key temporal and mechanistic differences between prospective and retrospective linking, as well as the potential differences in their cognitive outcomes.

Hippocampus-to-amygdala pathway drives the separation of remote memories of related events

The mammalian brain can store and retrieve memories of related events as distinct memories and remember common features of those experiences. How it computes this function remains elusive. Here, we show in rats that recent memories of two closely timed auditory fear events share overlapping neuronal ensembles in the basolateral amygdala (BLA) and are functionally linked. However, remote memories have reduced neuronal overlap and are functionally independent. The activity of parvalbumin (PV)-expressing neurons in the BLA plays a crucial role in forming separate remote memories. Chemogenetic blockade of PV preserves individual remote memories but prevents their segregation, resulting in reciprocal associations. The hippocampus drives this process through specific excitatory connections with BLA GABAergic interneurons. These findings provide insights into the neuronal mechanisms that minimize the overlap between distinct remote memories and enable the retrieval of related memories separately.

Sex and the facilitation of cued fear by prior contextual fear conditioning in rats

Previous studies have shown that the formation of new memories can be influenced by prior experience. This includes work using pavlovian fear conditioning in rodents that have shown that an initial fear conditioning experience can become associated with and facilitate the acquisition of new fear memories, especially when they occur close together in time. However, most of the prior studies used only males as subjects resulting in questions about the generalizability of the findings from this work. Here we tested whether prior contextual fear conditioning would facilitate later learning of cued fear conditioning in both male and female rats, and if there were differences based on the interval between the two conditioning episodes. Our results showed that levels of cued fear were not influenced by prior contextual fear conditioning or by the interval between training, however, females showed lower levels of cued fear. Freezing behavior in the initial training context differed by sex, with females showing lower levels of contextual fear, and by the type of initial training, with rats given delayed shock showing higher levels of fear than rats given immediate shock during contextual fear conditioning. These results indicate that contextual fear conditioning does not prime subsequent cued fear conditioning and that female rats express lower levels of cued and contextual fear conditioning than males.

Intrinsic Neural Excitability Biases Allocation and Overlap of Memory Engrams

Memories are thought to be stored in neural ensembles known as engrams that are specifically reactivated during memory recall. Recent studies have found that memory engrams of two events that happened close in time tend to overlap in the hippocampus and the amygdala, and these overlaps have been shown to support memory linking. It has been hypothesized that engram overlaps arise from the mechanisms that regulate memory allocation itself, involving neural excitability, but the exact process remains unclear. Indeed, most theoretical studies focus on synaptic plasticity and little is known about the role of intrinsic plasticity, which could be mediated by neural excitability and serve as a complementary mechanism for forming memory engrams. Here, we developed a rate-based recurrent neural network that includes both synaptic plasticity and neural excitability. We obtained structural and functional overlap of memory engrams for contexts that are presented close in time, consistent with experimental and computational studies. We then investigated the role of excitability in memory allocation at the network level and unveiled competitive mechanisms driven by inhibition. This work suggests mechanisms underlying the role of intrinsic excitability in memory allocation and linking, and yields predictions regarding the formation and the overlap of memory engrams.

Drift of neural ensembles driven by slow fluctuations of intrinsic excitability

Representational drift refers to the dynamic nature of neural representations in the brain despite the behavior being seemingly stable. Although drift has been observed in many different brain regions, the mechanisms underlying it are not known. Since intrinsic neural excitability is suggested to play a key role in regulating memory allocation, fluctuations of excitability could bias the reactivation of previously stored memory ensembles and therefore act as a motor for drift. Here, we propose a rate-based plastic recurrent neural network with slow fluctuations of intrinsic excitability. We first show that subsequent reactivations of a neural ensemble can lead to drift of this ensemble. The model predicts that drift is induced by co-activation of previously active neurons along with neurons with high excitability which leads to remodeling of the recurrent weights. Consistent with previous experimental works, the drifting ensemble is informative about its temporal history. Crucially, we show that the gradual nature of the drift is necessary for decoding temporal information from the activity of the ensemble. Finally, we show that the memory is preserved and can be decoded by an output neuron having plastic synapses with the main region.

Excitability mediates allocation of pre-configured ensembles to a hippocampal engram supporting contextual conditioned threat in mice

Little is understood about how engrams, sparse groups of neurons that store memories, are formed endogenously. Here, we combined calcium imaging, activity tagging, and optogenetics to examine the role of neuronal excitability and pre-existing functional connectivity on the allocation of mouse cornu ammonis area 1 (CA1) hippocampal neurons to an engram ensemble supporting a contextual threat memory. Engram neurons (high activity during recall or TRAP2-tagged during training) were more active than non-engram neurons 3 h (but not 24 h to 5 days) before training. Consistent with this, optogenetically inhibiting scFLARE2-tagged neurons active in homecage 3 h, but not 24 h, before conditioning disrupted memory retrieval, indicating that neurons with higher pre-training excitability were allocated to the engram. We also observed stable pre-configured functionally connected sub-ensembles of neurons whose activity cycled over days. Sub-ensembles that were more active before training were allocated to the engram, and their functional connectivity increased at training. Therefore, both neuronal excitability and pre-configured functional connectivity mediate allocation to an engram ensemble.

Integrating and fragmenting memories under stress and alcohol

Stress can powerfully influence the way we form memories, particularly the extent to which they are integrated or situated within an underlying spatiotemporal and broader knowledge architecture. These different representations in turn have significant consequences for the way we use these memories to guide later behavior. Puzzlingly, although stress has historically been argued to promote fragmentation, leading to disjoint memory representations, more recent work suggests that stress can also facilitate memory binding and integration. Understanding the circumstances under which stress fosters integration will be key to resolving this discrepancy and unpacking the mechanisms by which stress can shape later behavior. Here, we examine memory integration at multiple levels: linking together the content of an individual experience, threading associations between related but distinct events, and binding an experience into a pre-existing schema or sense of causal structure. We discuss neural and cognitive mechanisms underlying each form of integration as well as findings regarding how stress, aversive learning, and negative affect can modulate each. In this analysis, we uncover that stress can indeed promote each level of integration. We also show how memory integration may apply to understanding effects of alcohol, highlighting extant clinical and preclinical findings and opportunities for further investigation. Finally, we consider the implications of integration and fragmentation for later memory-guided behavior, and the importance of understanding which type of memory representation is potentiated in order to design appropriate interventions.

Distinct engrams control fear and extinction memory

Memories are stored in engram cells, which are necessary and sufficient for memory recall. Recalling a memory might undergo reconsolidation or extinction. It has been suggested that the original memory engram is reactivated during reconsolidation so that memory can be updated. Conversely, during extinction training, a new memory is formed that suppresses the original engram. Nonetheless, it is unknown whether extinction creates a new engram or modifies the original fear engram. In this study, we utilized the Daun02 procedure, which uses c-Fos-lacZ rats to induce apoptosis of strongly activated neurons and examine whether a new memory trace emerges as a result of a short or long reactivation, or if these processes rely on modifications within the original engram located in the basolateral amygdala (BLA) and infralimbic (IL) cortex. By eliminating neurons activated during consolidation and reactivation, we observed significant impacts on fear memory, highlighting the importance of the BLA engram in these processes. Although we were unable to show any impact when removing the neurons activated after the test of a previously extinguished memory in the BLA, disrupting the IL extinction engram reactivated the aversive memory that was suppressed by the extinction memory. Thus, we demonstrated that the IL cortex plays a crucial role in the network involved in extinction, and disrupting this specific node alone is sufficient to impair extinction behavior. Additionally, our findings indicate that extinction memories rely on the formation of a new memory, supporting the theory that extinction memories rely on the formation of a new memory, whereas the reconsolidation process reactivates the same original memory trace.

Memory allocation at the neuronal and synaptic levels

Memory allocation, which determines where memories are stored in specific neurons or synapses, has consistently been demonstrated to occur via specific mechanisms. Neuronal allocation studies have focused on the activated population of neurons and have shown that increased excitability via cAMP response element-binding protein (CREB) induces a bias toward memoryencoding neurons. Synaptic allocation suggests that synaptic tagging enables memory to be mediated through different synaptic strengthening mechanisms, even within a single neuron. In this review, we summarize the fundamental concepts of memory allocation at the neuronal and synaptic levels and discuss their potential interrelationships. [BMB Reports 2024; 57(4): 176-181].

Microglia: Activity-dependent regulators of neural circuits

It has been more than a century since Pío del Río-Hortega first characterized microglia in histological stains of brain tissue. Since then, significant advances have been made in understanding the role of these resident central nervous system (CNS) macrophages. In particular, it is now known that microglia can sense neural activity and modulate neuronal circuits accordingly. We review the mechanisms by which microglia detect changes in neural activity to then modulate synapse numbers in the developing and mature CNS. This includes responses to both spontaneous and experience-driven neural activity. We further discuss activity-dependent mechanisms by which microglia regulate synaptic function and neural circuit excitability. Together, our discussion provides a comprehensive review of the activity-dependent functions of microglia within neural circuits in the healthy CNS, and highlights exciting new open questions related to understanding more fully microglia as key components and regulators of neural circuits.