Open research questions in Memory and Neural Mechanisms

What the literature leaves open

• This form of metaplasticity, originally formalized in the Bienenstock, Cooper, and Munro (BCM) plasticity rule, generates competitive interactions between synapses and is thought to support sparse information encoding during memory formation.

  A novel, complex-spike burst-dependent form of BCM-like metaplasticity regulates the induction of behavioral timescale synaptic plasticity · 2026 · DOI

• Importantly, although a non-Hebbian, burst-dependent form of synaptic plasticity known as behavioral timescale plasticity (BTSP) is essential for hippocampal memory encoding, little is known about the role of metaplasticity in BTSP.

  A novel, complex-spike burst-dependent form of BCM-like metaplasticity regulates the induction of behavioral timescale synaptic plasticity · 2026 · DOI

• The orbitofrontal cortex (OFC) is implicated in representing outcome expectancies, but its role in memory-updating, beyond value-based learning, remains unclear.

  Chemogenetic Inhibition of the Ventrolateral Orbitofrontal Cortex Disrupts Prediction Error and Salience-Guided Memory-Updating at the Hippocampal Engram Level · 2026 · DOI

• Although proximal CA3 is functionally aligned with the DG through dense reciprocal connectivity, how DG output contributes to memory discrimination within proximal CA3 remains elusive.

  Dentate gyrus facilitates rate, but not global, remapping along the CA3 transverse axis · 2026 · DOI

• Despite this, it remains unclear exactly how ripples shift the activity of CA1 neuronal populations to accommodate the consolidation of specific memories and how sublayer differences manifest.

  A cortical-hippocampal communication undergoes rebalancing after new learning · 2026 · DOI

• Although learning new skills is especially prominent throughout childhood, remarkably few studies have examined the neural underpinnings of motor sequence learning (MSL) and memory consolidation in children.

  Neural Correlates of Motor Sequence Learning and Enhanced Offline Consolidation in 7–11‐Year‐Old Children · 2026 · DOI

• This latter finding should be further investigated because, although the functional requirement of corticostriatal neurons in goal- directed learning is well established, the long-range and local intracortical inputs that are engaged during such learning are not yet clear. Just as IT-specific transgenic mice have allowed us to establish specificity of the effects reported here, there remain many open questions and, by combining these mice with in-vivo imaging tools, the subtle dynamics of IT neuron activity in the different cortical layers and various stages of action-outcome learning may be further revealed.

  Intratelencephalic neurons in the medial prefrontal cortex mediate the acquisition of goal-directed actions. · 2026 · DOI

• However, the roles of specific prefrontal projections to the amygdala in contextual threat discrimination are not yet fully understood, particularly regarding more complex learning tasks and when disentangling the functionally distinct prelimbic (PL) subunit of the mPFC.

  The prefrontal cortex outputs to the amygdala facilitate threat-discrimination learning · 2026 · DOI

• The hippocampus and cortices of the medial temporal lobe (MTL) are increasingly implicated in working memory, particularly for tasks requiring complex, high-resolution relational binding, but their precise contributions--particularly during delay maintenance--remain debated.

  Phase-Specific Hippocampal and Cortical Medial Temporal Lobe Involvement in Allocentric Working Memory · 2026 · DOI

• While converging evidence suggests that persistent activity in the hippocampus and other areas of the brain con-tributes to WM storage, this link has not been examined by manipulating neural activity experimentally.

  Disrupting selectivity persistent activity with electrical stimulation impairs human working memory · 2026 · DOI

• Prior studies utilizing the PMA task have primarily investigated avoidance, freezing, and food-seeking behaviors, but few studies have thoroughly assessed darting behavior, a more recently identified measure of fear that has been largely explored in conditional fear paradigms.

  Rats use darting as a strategy to navigate between reward and safety during platform-mediated active avoidance under different social contexts. · 2026 · DOI

• Although entorhinal cortex lesions or inactivation impair latent inhibition 7-10, the specific contribution of the lateral entorhinal cortex (LEC) remains unclear.

  Dissecting the Role of the Lateral Entorhinal Cortex in Memory Interference · 2026 · DOI

• Neuronal activity in hippocampal-amygdala circuits has been shown extensively to underlie the conditioning and recall of contextual fear memories, yet the direct contribution of astrocytes to integrating stimulus and contextual information in this task remains unknown.

  Hippocampal astrocytes contribute to encoding context-specific aversive stimuli to regulate fear-related behavior · 2026 · DOI

• inhibitory interneurons, Several limitations of the present study should be considered. First, the identity of PAG neurons receiving mPFC SST input remains unresolved. Although our data demonstrate functional inhibitory connectivity, we were unable to determine whether these projections excitatory target the precise circuit premotor neurons, or both. As a result, mechanism, whether direct inhibition or disinhibition, remains to be established. While it is possible that mPFC SST projections engage local disinhibitory circuits similar to those described for central amygdala projections, our data do not allow us to directly test this hypothesis and it should therefore be considered speculative. This interpretation is consistent with recent findings showing that activation of prelimbic SST neurons can recruit neuronal activity in the ventrolateral PAG (Cummings et al., 2022), although the extent to which these SST populations overlap with the projections studied here remains unclear. is necessary Second, while optogenetic inhibition revealed that the mPFC SST→PAG pathway for maintaining freezing behavior, complementary activation experiments did not alter freezing, this pathway may not be sufficient to drive fear expression. These findings suggest that mPFC SST projections act as modulatory components within a broader defensive network rather than as primary drivers of behavior. indicating that reflect the PAG, synaptic sparse Third, although anatomical tracing revealed widespread functional SST axonal projections throughout recordings indicated that only a subset (∼10%) of neurons exhibited discrepancy connectivity despite broad may specific axonal PAG during recordings, or state-dependent recruitment of synaptic inputs trough neuromodulators fully captured in ex vivo conditions. responses. This functional arborization, subpopulations preferential not of sampled uniformly targeting apparent is not that Finally, viral expression spanned multiple mPFC subdivisions, including prelimbic, infralimbic, and anterior cingulate cortices, and optogenetic manipulation was performed at the level of axon terminals in the PAG. Consequently, the relative contribution of individual mPFC subregions could not be determined. Given the known functional differences between prelimbic and infralimbic circuits in fear expression and suppression, future studies using region-specific targeting strategies will be required to disentangle their respective roles. Addressing these limitations will require projection-specific and cell-type-specific approaches to resolve the circuit architecture and functional organization of mPFC SST projections to the PAG.

  Prefrontal long-range somatostatin inhibitory projections modulate fear expression · 2026 · DOI

• In this study, we showed that use of LLMs for text-based scene segmentation can accurately segment naturalistic narrative stimuli. This approach eliminates subjective bias of human annotators and reduces LLM-specific biases by combining the output of multiple LLMs. The ensemble based LLM segmentation of scenes’s timings was used to construct FC features from DMN and auditory networks for scene level recall classification. We also applied low- rank decomposition on raw FC and found that the low rank DMN network, when combined with a RF classifier yielded a modest but statistically reliable improvement in accuracy over raw FC. At the same time, few limitations undermine these find- ings. Overall discriminability between high and low recall scenes remained moderate, and a class imbalance likely constrained performance despite our use of appropriate met- rics. The analysis focused on four stories of narrative fMRI datasets, two well-defined networks, and relatively simple classifiers, and the generalizability of the approach to other modalities (i.e video stimulus) and networks is needed. Future work could extend this framework to additional brain networks and modalities, explore richer multi-model behavioral labels, and investigate more advanced modeling approaches —like foundation models to better capture how shared and idiosyncratic neural patterns support memory for naturalistic events. In addition, the low-rank decomposition algorithm could be further optimized, for example by incor- porating non-linear methods, to capture a broader range of connectivity patterns related to recall.

  Classification of decomposed neural data in memory networks and LLM-based stimuli processing · 2026 · DOI

• Decoding was formulated as a series of binary classification tasks (e.g., encoding vs. retrieval, letter D vs. letter J) rather than as multiclass simultaneous decoding of all behavioral states. Development of multiclass decoding models that jointly classify all behavioral states and comparison with the pairwise binary approach would assess whether hierarchical or holistic patterns emerge.

  Planar, spiral, and concentric traveling waves distinguish behavioral states in human memory · 2026 · DOI

• The data rebalancing procedure equalized the number of epochs per behavioral state prior to five-fold cross-validation, but the class imbalance ratios before rebalancing and potential bias introduced by the rebalancing strategy are not reported. Analysis of how rebalancing affects minority behavioral states (if any existed) and sensitivity analysis across different rebalancing methods would clarify robustness.

  Planar, spiral, and concentric traveling waves distinguish behavioral states in human memory · 2026 · DOI

• The paper reports that SNR (signal-to-noise ratio) differences across task conditions did not drive decoding performance (all p's > 0.05 for pairwise correlations), yet SNR was not systematically manipulated or controlled as a covariate. Explicit quantification of SNR across behavioral states and task periods, with controlled SNR matching across conditions, would definitively establish wave pattern independence from signal quality variations.

  Planar, spiral, and concentric traveling waves distinguish behavioral states in human memory · 2026 · DOI

• The four-layer neural network architecture (input layer, two hidden layers with 32 and 16 neurons, sigmoid output) was kept constant across all spatial and verbal memory datasets, but hyperparameter optimization details are not specified. Systematic comparison of alternative architectures (recurrent neural networks, convolutional approaches to spatial electrode structure, attention mechanisms) for temporal traveling wave pattern classification is absent.

  Planar, spiral, and concentric traveling waves distinguish behavioral states in human memory · 2026 · DOI

• The decoding analysis was performed separately for spatial episodic memory (Treasure Hunt task, N=9 patients) and verbal working memory (Sternberg task, N=15 patients) without direct comparison of whether the same traveling wave patterns and their behavioral associations generalize across memory task types. Cross-task validation of traveling wave-behavior relationships would strengthen claims about domain-general mechanisms.

  Planar, spiral, and concentric traveling waves distinguish behavioral states in human memory · 2026 · DOI

• Higher-order modes (modes beyond 6) in the CICA decomposition showed progressively less contribution to test decoding accuracy and were characterized as complex wave shapes lacking specific structure. Systematic characterization of what constitutes meaningful versus noise-like wave structure in modes 6+ and determination of an optimal mode truncation threshold based on behavioral relevance rather than empirical accuracy plateauing is needed.

  Planar, spiral, and concentric traveling waves distinguish behavioral states in human memory · 2026 · DOI

• The decoding analysis trained separate neural network classifiers for each oscillation cluster and individual subject rather than developing a unified cross-subject decoder for traveling wave patterns. A transfer learning or meta-learning approach that can generalize traveling wave-behavior relationships across subjects and clusters would address the current limitation of cluster-specific and subject-specific model training.

  Planar, spiral, and concentric traveling waves distinguish behavioral states in human memory · 2026 · DOI

• Mnemonic technique effectiveness in older adults encountering practical problems during application (Brooks et al., 1993) has not been systematically linked to metacognitive monitoring abilities; Bjork (1994) emphasized metamemory considerations in training, but individual differences in metamemory accuracy during Method of Loci encoding in aging populations remain uncharacterized.

  The Method of Loci: A Survey Review · 2026 · DOI

• The neurobiological mechanisms linking prefrontal cortex activation changes (documented in Belleville et al., 2023) to improved memory performance with the Method of Loci have not been compared across age groups; memory plasticity varies across the lifespan (Brehmer et al., 2007), but whether encoding strategy effects on prefrontal and hippocampal networks differ between younger and older adults requires investigation.

  The Method of Loci: A Survey Review · 2026 · DOI

• Virtual reality implementation of the Method of Loci has been tested only in healthy older adults with memory complaints (Boller et al., 2021); systematic evaluation is needed in ADHD populations and schizophrenia patients who show documented impairments in sustained attention and strategic memory encoding, to determine whether VR-enhanced spatial encoding alleviates their specific mnemonic deficits.

  The Method of Loci: A Survey Review · 2026 · DOI

Most-cited papers in Memory and Neural Mechanisms

• The evolution of foresight: What is mental time travel, and is it unique to humans? · Behavioral and Brain Sciences · 2007 · 1,452 citations
• Norepinephrine ignites local hotspots of neuronal excitation: How arousal amplifies selectivity in perception and memory · Behavioral and Brain Sciences · 2015 · 531 citations
• Active Forgetting: Adaptation of Memory by Prefrontal Control · Annual Review of Psychology · 2020 · 233 citations
• The Temporal Context Model in Spatial Navigation and Relational Learning: Toward a Common Explanation of Medial Temporal Lobe Function Across Domains. · Psychological Review · 2005 · 232 citations
• Distributed Representations in Memory: Insights from Functional Brain Imaging · Annual Review of Psychology · 2011 · 231 citations
• Context modulation of memory for fear extinction in humans · Psychophysiology · 2005 · 230 citations
• Systemic or intrahippocampal delivery of histone deacetylase inhibitors facilitates fear extinction. · Behavioral Neuroscience · 2007 · 202 citations
• Acute stress and episodic memory retrieval: neurobiological mechanisms and behavioral consequences · Annals of the New York Academy of Sciences · 2016 · 189 citations
• LONG-TERM POTENTIATION AND LEARNING · Annual Review of Psychology · 1996 · 163 citations
• Optimizing exposure therapy with an inhibitory retrieval approach and the OptEx Nexus · Behaviour Research and Therapy · 2022 · 162 citations

Most recent work

• Hippocampal mechanisms resolve competition in memory and perception · bioRxiv · 2026
• A hierarchical coordinate system for sequence memory in human entorhinal cortex. · bioRxiv · 2026
• Conserved neural population geometry supports behavioral generalization · bioRxiv · 2026
• Anterior hippocampal integration tracks the developmental emergence of flexible navigation through changing environments · bioRxiv · 2026
• Disrupting selectivity persistent activity with electrical stimulation impairs human working memory · bioRxiv · 2026
• Classification of decomposed neural data in memory networks and LLM-based stimuli processing · Brain Imaging and Behavior · 2026
• Putting the Amygdala and Hippocampus in PTSD to Rest · American Journal of Psychiatry · 2026
• Stress disrupts hippocampal integration of overlapping events and memory inference in humans · Science Advances · 2026
• Navigational strategy dictates hippocampal representation of space in an everyday memory task · bioRxiv · 2026
• The hippocampus as a small-world cognitive map · bioRxiv · 2026

Related topics in Neuroscience