Learning & Memory

Reinforced conditioning allows individuals predicting future events with high confidence. However, many daily behaviours rely on unreinforced connections of neutral stimuli, called Incidental Associations (IAs), which enhance predictive capacity in unstable environments and are observed across species. IAs can be studied through sensory preconditioning paradigms, where two neutral stimuli (S1/S2) are presented together in a preconditioning phase, followed by classical conditioning of S1 with a potent reinforcer. As a result, subjects present a direct response to the S1 stimulus, but also display mediated responses to the S2 stimulus never explicitly reinforced, indicating IA formation during preconditioning. Our previous work demonstrated that type-1 cannabinoid receptors (CB1 receptors) in the hippocampus are essential for this IA formation. As dopamine signaling is also important for this process, we investigate the role of interactions between these dopamine-cannabinoid systems in IA memory formation in the hippocampus. Extending our previous work on odor-taste association, using light-sound association we showed that global CB1 receptor knock-out or specific hippocampal CB1 receptor deletion also blocked mediated responses to sound (S2) while direct response to light (S1) was unaltered. Focusing on dopamine, we then found hippocampal dopaminergic activity is enhanced during paired presentations of S1 and S2 and that blockade of dopamine D1 receptor during preconditioning S1-S2 associations abolished mediated response to S2. Interestingly, mice lacking CB1 receptors specifically in D1-receptor-expressing cells (D1-CB1-KO) failed to show mediated responses for either light-sound or odor-taste associations, identifying this CB1 receptor population as critical for IA formation. Enhanced activation of CB1 receptors, through either increase of endocannabinoids (using degradation enzyme inhibition) or exogenous stimulation by cannabis-derived Delta-9-tetrahydrocannabinol (THC), was able to promote the formation of IAs under insufficient conditions. The effect of endogenous CB1 activation, but not THC, was blocked in D1-CB1-KO mice indicating that IA-facilitation by endogenous and exogenous receptor activation rely on different mechanisms. Overall, these data uncover new mechanisms underlying unreinforced associative learning.

Animals detect and evaluate signs of danger and safety in their environment to ensure survival, yet the neural mechanisms that distinguish safety learning from other forms of conditioned inhibition, remain poorly understood. Here, we directly compared fear and safety learning in male rats. Fear conditioned rats showed high freezing to the tone and the conditioning context, whereas safety conditioned rats showed significant tone-specific reduction in freezing. This safety cue could also generalize to a novel, previously unassociated threat context leading to suppressed freezing when presented demonstrating that inhibitory actions of safety cues are not tied to its original training environment but can modify fear expression across settings. Fear and safety learning also produced unique patterns of neuronal activation and glutamatergic receptor expression in the medial prefrontal cortex (mPFC), basolateral amygdala (BLA), and central amygdala (CeA), as measured by c-Fos immunohistochemistry and Western blotting. Fear conditioning induced greater Fos expression in the BLA and CeA, as well as elevated amygdalar NMDA receptor (GluN1) levels, whereas safety learning increased amygdalar PSD-95 and AMPA receptor (GluA1) expression. Both safety and fear learning increased mPFC Fos expression without affecting glutamatergic receptors levels. Finally, safety conditioning was associated with lower tone-evoked freezing than fear conditioned rats across early extinction sessions and was accompanied by distinct patterns of prefrontal and amygdala activation across extinction. Together, these findings suggest that safety learning engages neural and behavioral mechanisms distinct from fear learning and extinction, while modifying amygdala-prefrontal circuits towards more rapid fear suppression.

Newly acquired memories are initially fragile and are consolidated into long-term memory over time. Although consolidated memories were once thought to be resistant to interference, accumulating evidence shows that memory reactivation renders them transiently labile, making them susceptible to modification or interference before reconsolidation. Crucially, however, there is substantial variation in whether reactivated memories are disrupted by new information, remain protected from it, or even strengthened by it. The factors that guide this modification are largely unknown. To systematically investigate this, we examined motor memory interference using a classic A-B-A visuomotor rotation paradigm. Participants adapted to a 30-degree clockwise rotation (A) on Day1. On Day2, an interfering 30-degree counter-clockwise rotation (B) was introduced under varied conditions: directly without reactivation, after brief reactivation of A, after expression of A without feedback, or following a gradual transition from A to B. The final experiment used explicit contextual cues (a secondary follow-through target) to distinguish A and B trials. Contrary to the simple prediction that reactivation should increase vulnerability to interference, reactivating the original memory before introducing interference protected it, as evidenced by significant savings during relearning on Day3. In contrast, introducing interference directly, without reactivation, disrupted the original memory. This protection was consistent with a contextual-inference account: the large sensory prediction error experienced during the abrupt transition from A to B served as a latent contextual cue, signaling a new context and thereby shielding the original memory from being overwritten. Eliminating this prediction error through an immediate washout session with a similar error profile or through a gradual A-to-B transition abolished the protective effect and disrupted the original memory. Furthermore, when explicit contextual cues distinguished the two perturbations, memories were protected even in the absence of a salient prediction error. Our findings are consistent with a contextual inference account in which the fate of a consolidated memory, whether it is modified or protected, is shaped by the availability of explicit cues or latent signals such as sensory prediction error at the time when interference is introduced.

Self-directed learning allows learners to actively control their learning experience and has been shown to enhance memory compared with matched yoked learning. However, it remains unclear when and how active control modulates memory-related neural activity during learning and retrieval. We recorded electroencephalography (EEG) while participants encoded objects under active and yoked learning conditions and again during a delayed recognition test approximately 24 h later. We examined event-related potential (ERP) activity across earlier processing windows, including pre-stimulus slow potentials and early N2 activity, and later processing windows, including P300, late slow-wave, and post-stimulus slow-potential activity. We also examined encoding-retrieval similarity (ERS) between neural patterns during encoding and retrieval. Behaviorally, active control improved delayed recognition, with the advantage selectively expressed in detailed recognition. In the ERP analyses, earlier processing windows showed memory-related effects, with pre-stimulus slow-potential and N2 activity differentiating subsequently remembered from forgotten items, but were not modulated by active control. By contrast, active control modulated later memory-related ERP activity, with remembered-forgotten differences expressed during late stimulus-related and immediate post-stimulus processing only in the active condition. ERS showed a similar active-control modulation: memory-related encoding-retrieval pattern similarity was evident under active learning, but not under yoked learning, with this effect involving relatively late encoding and retrieval windows. Together, these findings support a constructive-processing account, suggesting that memory formation under active control depends more strongly on rich, detailed encoding representations that can be reinstated during retrieval.

Anxiety-related traits (ARTs) have been linked to altered fear learning, but previous studies have typically examined different experimental phases and response systems, limiting the comparability of findings and the accumulation of consistent evidence. Here, we comprehensively examined associations between ARTs and fear conditioning across acquisition, extinction and renewal and across subjective, physiological and neural response systems in a well-powered sample (N = 267) using a two-day differential conditioning paradigm. ARTs were operationalized as a composite of trait anxiety, neuroticism, and intolerance of uncertainty and conditioned responding was assessed using skin conductance responses, fear-potentiated startle, US expectancy ratings, fear ratings, and functional magnetic resonance imaging. Higher ARTs were consistently associated with elevated subjective fear and US expectancy to both threat and safety cues during extinction and renewal, without corresponding elevations in physiological responding. At the same time, ARTs were not associated with threat-safety discrimination in subjective or physiological measures across phases, while neural associations were limited to reduced dorsal anterior cingulate cortex discrimination during early renewal. These findings suggest that ARTs are characterized by a CS unspecific cognitive bias toward heightened threat expectancy and evaluation rather than altered associative fear learning, highlighting the importance of distinguishing conditioned discrimination from general levels of responding across response systems.

Planaria exhibit remarkable regenerative ability, including the capacity to regrow complete heads and brains after decapitation. Here, we re-investigated whether regenerated planaria can preserve learned avoidance behavior, a phenomenon that has been reported previously but has been difficult to study due to unreliable experimental protocols. Using a light-to-food associative conditioning paradigm, planaria were trained to override their normal photophobic preference and then decapitated. Following a two-week regeneration period, behavioral responses to the conditioned stimulus were re-evaluated. Results indicated that the majority of regenerated planaria retained the learned response, supporting a model in which behavioral patterns can regenerate as well as anatomical patterns. By establishing a consistent, low-cost, and effective protocol for studying memory persistence through regeneration, such work may help inform future research on memory loss, resilience, and recovery in neurodegenerative diseases.

Working memory is capacity-limited, but interactions with episodic memory may offset this constraint. We tested moment-by-moment contributions of episodic representations to working memory by combining the N-back and Mnemonic Similarity tasks. Thirty-one participants, undergoing eye-tracking, first encoded items in a one-back task, classifying them as "same" or "similar" to their predecessor. In a subsequent two-back task, mnemonic discrimination showed a graded, item-specific benefit of prior experience: performance was best for previously compared items, whereas recognition of identical repeats was unaffected. Successful discrimination of previously compared items was accompanied by greater pupil dilation, gradually emerging gaze patterns resembling those elicited by their similar pair-mate, and higher gaze-similarity between one-back and two-back target viewing. Diverging gaze patterns between pair-mates during one-back further predicted two-back discrimination. These findings challenge working memory's characterization as an isolated system, demonstrating how it recruits episodic computations - encoding distinct traces, predicting upcoming content, and reinstating it at retrieval.

Spatial navigation requires animals to integrate current environmental information with previously acquired spatial memories. The locus coeruleus provides neuromodulatory input to the hippocampus, but whether this pathway facilitates spatial learning in general or preferentially supports the updating of established representations remains unclear. Here, we selectively activated LC projections to the dorsal hippocampus while mice performed object-location recognition and an appetitive radial-maze task involving initial spatial learning followed by reversal. LC-hippocampal activation enhanced object-location memory and improved reversal learning, reducing total and working-memory errors, but did not affect initial spatial reference acquisition or retention. To characterize navigation beyond classical performance measures, we developed a graph-based analysis comparing each observed trajectory with paths generated from random, regular, small-world and heuristic goal-directed network models. Radial-maze trajectories contained a structured mixture of goal-directed-like and regular or serial-like patterns that evolved across learning. In addition, agreement with the goal-directed model was associated with fewer errors and greater proximity to the rewarded arm. Together, these findings indicate that LC inputs to the hippocampus preferentially facilitate spatial memory updating rather than uniformly enhancing spatial learning, and introduce a complementary framework for quantifying the organization of radial-maze trajectories.

The cerebellar nuclei (CN) has demonstrated its influence on cognitive behavior via the cerebello-cortico circuit, yet the role of CN critical period mechanisms and how they may influence cognitive behavior, such as parvalbumin (PV) expressing interneurons enwrapped by perineuronal nets (PNNs), is still unclear. Therefore, we investigated the role of the lateral CN (LCN) PV cell calcium activity while animals performed a visual discrimination touchscreen cognitive task. All animals received the PV cell calcium indicator GCaMP6f at postnatal day 21 (P21). We targeted the LCN critical period by manipulating neural activity in male mice using the inhibitory Designer Receptors Exclusively Activated by Designer Drugs (DREADDs) from postnatal day 21 to 35 or by injecting an Hapln1-AAV vector to selectively target LCN PNN development. After animals completed the visual discrimination task, cerebellar tissue was collected for viral recovery and antibody staining for PNN components, Hapln1 and aggrecan. Results revealed DREADD animals showed improved reversal learning, an increase in calcium response to learning-related activity and altered PNN expression (Hapln1 and aggrecan). Hapln1 treated animals displayed a decrease in final day acquisition performance, lower reversal performance compared to DREADD groups, a decrease in reversal calcium learning-related activity, and an increase in PNN expression (Hapln1). Together, these data provide further evidence of LCN mechanisms associated with learning as well as the importance of understanding region-specific critical periods of plasticity.

Decisions are often modeled as a sequential process in which evidence accumulates until it reaches a threshold, triggering a response. Studies in freely moving animals raise questions about how ongoing behavior, not just stimulus properties, shapes this process. We trained rats of both sexes on a visual detection task with three luminance levels, each associated with the same reward outcome. Rats controlled cue duration through sustained head entries into a center port, yielding a measurable index of active sampling. Females consistently sampled longer than males. Sampling durations were shorter on error than correct trials, and reaction times were longer on error trials. We used drift diffusion models to relate these behaviors to the decision process. Luminance selectively affected the rate of evidence accumulation, with drift rate increasing monotonically across low, mid, and high luminance levels. Active sampling time was associated with the decision threshold, with longer sampling predicting higher thresholds in both sexes. The relationship between sampling time and drift rate differed by sex. Females showed a negative association between sampling duration and drift rate that was absent in males. These findings suggest that cue properties and active sampling make separable contributions to the decision process. These findings suggest that cue properties and active sampling make separable contributions to decision making, with a negative association between sampling duration and drift rate evident in females but not males.

We recently reported that rats performed better at a task distance of 2 m than at 0 m in a T-maze delayed alternation paradigm using a movable home cage in the longer-delay condition (Takita & Ichitani, 2026). We simultaneously recorded local field potentials from the bilateral prefrontal cortex, intermediate hippocampus, and ventral hippocampus. Across task epochs, coherence and two cross-frequency measures (phase-locking value and modulation index [MI]) revealed differences between correct and error trials in prefrontal interactions with hippocampal subregions. Among these measures, only MI was affected by task distance during the pre-task delay epoch. MI was highest in 2-m error trials and lowest in correct trials. In 0-m error trials, MI transiently increased during arm entry to levels comparable to those in 2-m error trials before declining toward the levels observed in correct trials during the later post-task delay. These MI dynamics appeared to be consistent with distance-dependent differences in behavioral performance. In addition, normalized Correct-Error Indices calculated for each electrophysiological measure revealed differential contributions of prefrontal coupling with the intermediate and ventral hippocampus across task distances. These findings suggest the existence of distinct near and far working memory states underlying distance-dependent behavioral differences, with distinct yet complementary contributions of the intermediate and ventral hippocampus to prefrontal interactions.

Contextual fear memories become less specific as they age, modeling fear overgeneralization seen in post-traumatic stress disorder. Glucocorticoid receptor (GR) signaling in the dorsomedial prefrontal cortex (dmPFC) during the immediate post-learning period may govern both endocrine recovery from an aversive experience and the eventual specificity of the resulting memory, but this link remains untested. We infused vehicle or the GR antagonist mifepristone into the dmPFC of rats immediately after contextual fear conditioning, then measured corticosterone dynamics, fear expression at recent and remote time points, and c-Fos coactivation networks. Mifepristone accelerated corticosterone recovery without changing total hormone release, spared recent memory, and produced stronger, less context-specific freezing at the remote time point. This behavioral shift coincided with reorganization of the retrieval network from a salience-network-like to a default-mode-network-like configuration. These findings identify dmPFC glucocorticoid signaling as a mechanism constraining fear memory generalization as memories transition to a remote, cortically dependent state.

BACKGROUND: An individual's adaptation to threatening environments under uncertainty is reflected in stress responses. Predictability (the ability to anticipate events) and controllability (the ability to control outcomes) are central to how one adapts, yet their joint influence on aversive learning remains unclear. METHODS: Thirty healthy adults completed a probabilistic aversive learning task in which cue-outcome contingencies varied across levels of predictability and controllability, i.e. whether shock intensity depended on prediction accuracy. Prediction accuracy, reaction time, subjective stress ratings, and skin conductance responses were recorded throughout. Trial-wise learning dynamics were estimated using the Volatile Kalman Filter. RESULTS: Prediction accuracy reduced as environments became less predictable and negatively associated with higher learning rates across predictability levels, with the strongest relationship observed in highly predictable blocks. Skin conductance responses showed that moderately predictable environments elicited responses like those in highly predictable environments when accurate predictions reduced shock intensity, but resembled responses in unpredictable environments when shock intensity was uncontrollable. Model comparison revealed a double dissociation between subjective stress ratings and skin conductance responses. Subjective ratings were best explained by model-derived volatility when prediction accuracy determined shock intensity and by belief uncertainty when it was independent of prediction accuracy, whereas skin conductance responses showed the reverse pattern. Reaction times were best explained by belief uncertainty when predictions influenced shock intensity. Higher anxiety was associated with elevated learning rates in highly and moderately predictable blocks when predictions did not control shock intensity.

Animals must generalize from limited experience, yet behavioral experiments in the laboratory setting rarely assess whether or how rapidly they generalize. This contrasts with machine learning systems, where generalization is considered a fundamental test of learning, and emphasizes performance evaluation with new in-distribution or out-of-distribution examples. Here, we used an olfactory categorization task to investigate rules of generalization versus memorization in mice. We trained mice to discriminate between two target odorants mixed with a variable number (0-13) of background odors. There are 32766 possible mixture stimuli to be classified, yet mice learn to generalize from as few as 8 unique mixtures. This generalization is not due to limited memory capacity: mice successfully learned to group the same set of mixtures when category labels were randomly shuffled. Analysis of individual variability revealed features in learning dynamics during training that predict performance in the generalization phase. A linear supervised learning algorithm could describe the generalization from few exemplars well, whereas nonlinear classifiers were necessary to explain memorization. Our experiments suggest that mice have an inductive bias towards generalization, consistent with a preference for simple rules, and will memorize only when forced to do so.

Reinforcement learning theory formulates distinct decision-making strategies, including reactive model-free and deliberative model-based strategies. This study investigates how mice adjust their reinforcement learning strategies while learning decision-making in dynamic environments. Unlike previous studies that focused on behaviors after extensive training periods, we analyzed changes in learning strategies in the course of training of a two-step decision-making task with probabilistic state transition and fluctuating reward probabilities. Our statistical behavioral analysis showed that the stay-probability following common and rare transitions diverged with training, a signature of strategies that utilize knowledge of task structure. We fit various reinforcement learning strategies to behavioral data and found that structure-informed strategies became increasingly dominant in their behaviors during training. Whereas previous studies emphasized transition from goal-directed to habitual strategies after extensive training, which were often associated with model-based and model-free strategies, respectively, our results newly demonstrate a shift from model-free to structure-informed strategies in early training in mice. Author summaryReinforcement learning theory allows us to examine how we make decisions and what approaches we use to optimize rewards. Most previous research, however, has examined animal behavior only after extensive training. Here we analyzed how mice adjust their reinforcement learning strategies as they are trained in a two-step decision-making task. Initially, mice relied on reactive model-free strategies, but as training progressed, their behavior began to incorporate knowledge of task structure. While previous studies suggested transition from model-based to model-free strategies with extensive training, our study revealed the opposite in the early stage of training.

Stress-related experiences alter glutamatergic signaling and neuronal excitability, but the mechanisms that couple experience to dentate granule cell function remain incompletely understood. Here, we examined how protein kinase N1a (PKN1a), a protein kinase C-like serine/threonine kinase, and repeated swim exposure regulate mouse hippocampal dentate granule cell excitability, with a focus on the neuronal glutamate transporter excitatory amino acid transporter 3 (EAAT3) and group I metabotropic glutamate receptors (mGluRs). Five days of repeated swim exposure increased spike firing in mature dentate granule cells from wild-type mice. PKN1a knockout produced a similar increase, and repeated swim did not further enhance firing in knockout mice. The enhanced firing observed after repeated swim exposure and in PKN1a knockout mice was reduced by co-application of an mGluR1 antagonist (LY367385) and an mGluR5 antagonist (MPEP). Inhibition of glutamate transporters with DL-TBOA increased granule cell firing in control wild-type mice but did not further increase firing in repeated-swim wild-type or PKN1a knockout mice, suggesting occlusion of transporter-dependent regulation of excitability. Repeated swim exposure and PKN1a knockout also reduced total and surface expression of EAAT3 in the hippocampus, whereas expression of the glial glutamate transporter EAAT2 was not significantly altered. Finally, PKN1a knockout and repeated swim exposure reduced anxiety-related behavior in the elevated plus maze test. Thus, PKN1a-dependent regulation of EAAT3 may restrain group I mGluR-dependent excitability in dentate granule cells, whereas repeated swim exposure and PKN1a knockout shift this system toward a lower-EAAT3, higher-excitability state accompanied by reduced anxiety-related behavior.

Behavioral flexibility is crucial to animal survival in dynamic environments, and a failure to update actions in response to recent outcomes is a hallmark of many neuropsychiatric disorders. However, the cellular and circuit mechanisms in the brain that support behavioral flexibility remain poorly understood. Forms of short-term plasticity such as synaptic facilitation have been theorized to transiently maintain information in neural circuits, and flexibly modulate how circuits process information depending on recent activity. Despite theoretical support, there is no direct experimental evidence linking synaptic facilitation to flexible decision-making. Recently, the presynaptic calcium sensor Synaptotagmin-7 (Syt7) has been shown to be required for synaptic facilitation at many synapses in the mammalian brain. Here, we assess operant learning in male and female Syt7 KO mice to determine how facilitation contributes to learning both stable and dynamic reward contingencies. We find that Syt7 KO mice acquired stable contingencies similarly to wild-type controls. However, KO mice were impaired in learning dynamic contingencies, showed more perseverative responding, and were delayed in applying a new task rule to all trial types following reversal. Behavioral modeling revealed a reduced influence of recent trial history on decisions in KO mice compared to wild-type controls. The behavioral deficits could not be explained by differences in motivation or memory. These results suggest that synaptic facilitation supports adaptive decision-making and that disruptions of short-term plasticity impair animals ability to use recent outcomes to update behavior.

Encoding-related pupil responses predict later memory performance, but the neural mechanisms linking these autonomic dynamics to memory formation remain unclear. This study examined whether pupil responses during encoding track activity in the brains memory network and whether they reflect functional interactions between memory-related regions and neural systems involved in pupil control. Participants performed an incidental encoding task involving object stimuli while undergoing simultaneous fMRI and pupillometry; recognition memory was subsequently assessed outside the scanner. Greater pupil constriction during encoding predicted both the strength and quality of later memory. These pupil dynamics correlated with activity in memory-related brain regions, notably the hippocampus and the parahippocampal cortex. Connectivity analyses indicated that encoding-related pupil responses were supported by functional interactions between the hippocampus and the midbrain Edinger-Westphal nucleus, the striatum, and the orbitofrontal cortex. The findings suggest that interactions between memory-related regions and parasympathetic pupil-control systems may modulate encoding efficiency. Together, the results identify encoding-related pupil constriction as a non-invasive marker of memory-network engagement and suggest a hippocampal-midbrain pathway through which autonomic pupil dynamics are coupled with successful memory formation.

People often must decide how long to continue waiting for rewards that will arrive at an uncertain time in the future. We propose that these persistence decisions involve weighing the benefits of continued waiting against opportunity costs of waiting, a balance that may shift over time. This framework suggests that persistence decisions share neural mechanisms with foraging decisions, which require ongoing comparisons between a current resource and possible alternatives. Dopamine and serotonin have been proposed to play opposing roles in foraging, with dopamine promoting exploration and serotonin promoting exploitation. Here we investigated their roles in persistence. In a within-subjects, double-blind, placebo-controlled study in young adults (n = 42), we examined the effects of increasing dopamine with L-dopa and increasing serotonin with escitalopram. We predicted that L-dopa would decrease persistence and escitalopram would increase it. Participants also completed patch-foraging, time perception, risk tolerance, and temporal discounting tasks to explore potential mechanisms of drug effects on persistence. Escitalopram increased persistence, after adjusting for the effects of anxiety and condition order, such that participants waited longer for rewards after taking the serotonergic drug. L-dopa did not influence persistence. In exploratory analyses controlling for age, however, L-dopa reduced persistence and increased exploration in foraging.

Long-term memory is widely thought to depend on activity-dependent synaptic plasticity and the structural remodeling that accompanies it, yet direct causal evidence that this remodeling is required for memory storage has been lacking. A case in point is the formation of dendritic spines: new spines appear following learning, but whether they constitute a physical substrate of memory remains unresolved. To address this, we developed the New Spine Elimination Tool (NSET), a chemical-genetic strategy, that ablates dendritic spines formed within a defined window of synaptic plasticity. NSET combines inducible expression of a degradable form of the actin-binding protein Drebrin with ligand-triggered proteasomal degradation. Because mainly nascent spines incorporate the degradable proteins into their cytoskeleton, ligand application eliminates them while sparing pre-existing ones. In hippocampal slice cultures, NSET eliminated recently formed spines without altering overall spine density or affecting pre-existing spines. Applied in vivo in the mouse basolateral amygdala, selective removal of learning-induced spines disrupted auditory fear memory, whereas consolidated memories and the capacity for new learning remained intact. These findings provide direct causal evidence that newly formed dendritic spines are required for long-term memory storage.