Alexander James Barnett

Recent frameworks of naturalistic memory encoding emphasize coordinated interactions between the hippocampus and the default mode network (DMN), especially at event boundaries that segment continuous experience. Notably absent from these accounts, however, is the thalamus—now increasingly recognized as an integrative hub with wide-ranging relevance for memory. Here, we expand existing hippocampal-DMN models of event cognition by explicitly incorporating thalamic circuitry and dynamics, with a focus on functional connectivity (FC). We reanalyzed a previously acquired fMRI dataset in which participants viewed two animated movies and subsequently recalled them either immediately or after a 2-day delay. Combining naturalistic event-based analytic approaches with contemporary Bayesian multilevel modeling, we quantified how FC among the thalamic subnuclei (anterior, mediodorsal, and lateral geniculate), hippocampus, and DMN relates to subsequent event memory. Our analyses revealed robust thalamic connectivity signatures predictive of durable memory encoding. Hippocampal–anterior thalamic FC at event boundaries was associated with better recall at 2-day delay. Conversely, thalamo–DMN FC at those boundaries generally corresponded to poorer delayed recall performance. Weaker but convergent patterns also emerged across the full encoding period: hippocampal FC with the anterior thalamus and lateral geniculate nucleus covaried positively with the richness of retrieved event details, whereas mediodorsal thalamus–DMN FC selectively predicted reduced detail following delayed retrieval. These findings identify thalamo-hippocampal and thalamo-cortical dynamics as candidate contributors to the formation and longer-term stability of episodic memories. More broadly, they position the thalamus as an active participant in shaping how complex real-world experiences are transformed into lasting mnemonic representations.

The human hippocampus is a unique cortical structure central to brain function, plasticity, and disease. Unravelling its complex organization requires the integration of multiscale data, linking molecular features to mesoscale anatomy and macroscale functional patterns. Gene expression is a fundamental microscale phenotype, and its profiling can provide a reference description of how molecular features are distributed across the brain. Capitalizing on recent imaging-transcriptomic analyses, we introduce HippoGenes, a repository of fine-grained gene expression patterns across human hippocampal subregions. We leveraged spatial statistical models and hippocampal surface mapping to reconstruct dense transcriptomic maps from sparse post-mortem tissue samples of the Allen Human Brain Atlas, generating continuous expression estimates for thousands of genes aligned to a common surface-based coordinate system. We illustrate the utility of HippoGenes to (i) map medial-lateral and anterior-posterior transcriptomic gradients that align with subfield and tripartite subdivisions of the hippocampal formation, (ii) examine associations between gene expression and canonical microstructural and functional features of the hippocampus, and (iii) perform a molecular decoding of subregional alterations in neurological patients with hippocampal pathology. HippoGenes provides a framework for exploring the molecular organization of the hippocampus, opening avenues for multiscale integration in health and disease, and is openly available on https://hippogenes.readthedocs.io/.

The human hippocampus exhibits distinct genetic, cellular, and connectivity profiles along its anterior–posterior long-axis, resulting in a differential sensitivity to visuospatial information. Long-axis development, therefore, may contribute to developmental increases in visuospatial memory capacity. To test this, we developed and applied a precision functional mapping technique to identify functional systems in the hippocampus of single subjects using BOLD-fMRI (N = 471, aged 5–21). We discovered considerable developmental remodeling of the hippocampal long-axis, particularly for the posterior functional system. With age, surface-area (mm²) of the posterior system decreased by 39.4%, while BOLD activity became increasingly independent and showed a sharper topographic boundary with the rest of the hippocampus. Notably, the posterior system showed strong preferential connectivity with medial-parietal regions which correlated with both age and age-adjusted memory scores. These results indicate the hippocampal long-axis becomes topographically and functionally specialized with development, potentially contributing to developmental increases in memory capacity.

People can experience the same event yet form distinct memories shaped by individual interpretations. Prior research shows that multivariate activity patterns in the Default Mode Network (DMN) are correlated across individuals during shared experiences, suggesting a role in representing high-level event features. However, it remains unclear whether these shared neural patterns reflect similarity in subsequent memory content. Here, we examined whether memory similarity correlates with intersubject spatial patterns in the DMN using a pre-existing dataset. Twenty-four individuals watched and recounted two cartoon movies during fMRI scanning. Using topic modeling, we transformed verbal recall into vectors of latent topics to quantify memory similarity across participants. We found that greater similarity in recalled content was associated with stronger shared activation patterns at encoding and retrieval, particularly in the posterior medial, medial prefrontal and anterior temporal cortices. These findings highlight the utility of natural language processing tools in linking memory representations to brain activity and underscore the DMN’s role in encoding, interpreting, and recalling complex event features.

Fiber-tracts exhibit distinct projection patterns along the anterior–posterior axis of the hippocampus, promoting a specialization in function. This specialization becomes increasingly pronounced throughout child development, with important implications for neurocognitive outcomes. Developmental changes in fiber-tract properties, including intra-axonal cross-sectional area and myelin content, may contribute to this anterior-posterior functional specialization. To test this, we developed a diffusion-MRI tractography pipeline to identify hippocampal fiber-tracts in single subjects, then examined whether age-related differences in total intra-axonal cross-sectional area and myelin content (T1w/T2w) could predict functional specialization in a large cross-sectional sample ( N =539, aged 5–21 years). With age, we found that the cross-sectional area of short-range medial-temporal tracts, which primarily target the anterior/body of the hippocampus, exhibited rapid growth. Concomitantly, the cross-sectional area of long-range occipito-parietal tracts, which primarily target the posterior hippocampus, exhibited a modest pruning. Increases in myelin content were relatively homogenous across fiber-tracts. In support of our hypothesis, we found that the cross-sectional area of fiber-tracts significantly predicts the surface-area of an fMRI-defined posterior system; a sensitive marker of functional specialization in the hippocampus. Tracts targeting early visual cortex (V2, V3, V4) showed the strongest association, with statistical modeling indicating a mediating effect of early-visual tract development on the relation between age and functional specialization. These findings provide evidence consistent with a mechanism whereby anatomical neurodevelopment contributes to functional specialization in the human hippocampus.

When people remember everyday experiences, they tend to focus on a subset of moments rather than everything that happened. Recent work shows that people reliably identify key moments within narratives, and that these moments disproportionately organize event representations in the brain. However, what these key moments represent remains an important unanswered question. Here we test the hypothesis that key moments function as semantic anchors in memory for extended events. Using spoken recall from three naturalistic datasets, we derived latent semantic representations of each narrative using topic modeling. We projected time-resolved movie annotations into this semantic space and systematically deleted narrative segments of the annotation to quantify the contribution of individual segments to the overall semantic structure. Removing segments associated with key moments produced significantly greater degradation of semantic structure than removing other segments, including event boundaries. These findings reveal a mechanism by which narratives are compressed into semantically meaningful representations.

Understanding how individuals perceive and recall information in their natural environments is critical to understanding potential failures in perception (e.g., sensory loss) and memory (e.g., dementia). Event segmentation, the process of identifying distinct events within dynamic environments, is central to how we perceive, encode, and recall experiences. This cognitive process not only influences moment-to-moment comprehension but also shapes event specific memory. Despite the importance of event segmentation and event memory, current research methodologies rely heavily on human judgements for assessing segmentation patterns and recall ability, which are subjective and time-consuming. A few approaches have been introduced to automate event segmentation and recall scoring, but validity with human responses and ease of implementation require further advancements. To address these concerns, we leverage Large Language Models (LLMs) to automate event segmentation and assess recall of written narratives, employing chat completion and text-embedding models, respectively. We validated these models against human annotations and determined that LLMs can accurately identify event boundaries, and that human event segmentation is more consistent with LLMs than among humans themselves. Using this framework, we advanced an automated approach for recall assessments which revealed semantic similarity between segmented narrative events and participant recall can estimate recall performance. Our findings demonstrate that LLMs can effectively simulate human segmentation patterns and provide recall evaluations that are a scalable alternative to manual scoring. This research opens avenues for studying the intersection between perception, memory, and cognitive impairment using methodologies driven by artificial intelligence.

Declarative memory deficits represent considerable challenges to adequate functioning and wellbeing in temporal lobe epilepsy (TLE) and frontal lobe epilepsy (FLE), two of the most common pharmaco-resistant epilepsies. TLE and FLE are impacted differently however, with TLE affecting primarily episodic, and mildly semantic memory, while FLE presenting with overall lesser declarative impairments and a greater involvement of language processes. Although functional magnetic resonance imaging (fMRI) studies are overall compatible with differential disruptions of medial temporal and fronto-limbic networks in both syndromes, direct comparisons of brain activity and connectivity remain scarce. The current study investigated how alterations in intrinsic functional brain organization, as mapped with resting-state connectivity (rsFC), shapes altered brain network activations in both episodic and semantic memory states. To this end, we acquired task- and rs-fMRI data in 28 TLE patients and 17 FLE patients and 87 age- and sex-matched healthy controls (HCs). We used activity flow mapping (AFM), a generative machine learning technique that derives plausible task activation patterns via individualized rsFC as well as normative task activation data. Previous studies in HC as well as patient populations have shown that this technique has been effective in identifying mechanisms contributing to atypical functional organization across different context. Overall, AFM reliably predicted task activations across HC, TLE, and FLE, but prediction accuracy was consistently reduced in both patient groups, indicating impaired propagation of task-relevant signals. These reductions co-occurred with abnormal episodic and semantic task activation patterns, atypical rsFC, and behavioral profiles marked by preserved semantic but impaired episodic memory performance in both patient groups. Importantly, although neither task-evoked abnormalities nor rsFC disruptions alone were associated with clinical variables, lower AFM accuracy in patients robustly tracked poorer episodic and semantic memory performance and longer disease duration. Prediction accuracy was driven more by intrinsic functional than structural network features, suggesting that altered network communication, rather than gross anatomy, constrains AFM in pharmaco-resistant epilepsy. Our work revealed syndrome-specific yet convergent disruptions in paralimbic and heteromodal association systems, linking intrinsic dysconnectivity to a shared episodic vulnerability and semantic resilience across TLE and FLE. By modelling how intrinsic connectivity shapes task-evoked responses, AFM provides a mechanistic account of imbalanced memory-state activations and isolates network-flow pathways as targets for intervention and rehabilitation.

Temporal lobe epilepsy (TLE) is the most common pharmacoresistant epilepsy in adults, yet few patients receive curative surgery due to diagnostic and prognostic uncertainty. In a multicenter cohort, we analyzed multimodal MRI and clinical data from 282 TLE patients, 298 healthy controls, and 45 disease controls. Patient-specific deviations from typical lifespan trajectories of intrinsic brain function were mapped using normative modeling. Regional functional alterations were heterogeneous but overlapped most in the mesiotemporal cortex. Connectome-based simulations revealed abnormality spread followed structural network architecture, highlighting the hippocampus as well as paralimbic and medial default-mode regions as epicenters. Multimodal integration implicated superficial white-matter microstructural alterations as a key contributor. Supervised models achieved AUCs of 0.77 for distinguishing TLE from disease controls, 0.74 for lateralizing seizure focus, and 0.64 for predicting postsurgical seizure freedom; greater contralateral temporal deviations predicted poorer outcomes. These findings support individualized functional biomarkers for precision presurgical care in focal epilepsy.