Event integration and temporal pattern separation: how hierarchical knowledge emerges in hippocampal subfields through learning
Bein, O.; Davachi, L.
Show abstract
Everyday life is composed of events organized by changes in contexts, with each event containing an unfolding sequence of occurrences. A major challenge facing our memory systems is how to integrate sequential occurrences within events while also maintaining their details and avoiding over-integration across different contexts. Here we asked if and how distinct hippocampal subfields come to hierarchically, and in parallel, represent both event context and subevent occurrences with learning. Participants viewed sequential events defined as sequences of objects superimposed on shared color frames while undergoing high-resolution fMRI. Importantly, these events were repeated to induce learning. Event segmentation, as indexed by increased reaction times at event boundaries, was observed in all repetitions. Temporal memory decisions were quicker for items from the same event compared to across different events, indicating that events shaped memory behavior. With learning, hippocampal CA3 multivoxel activation patterns clustered to reflect the event context, with more clustering correlated with behavioral facilitation during event transitions. In contrast, in the dentate gyrus temporally proximal items that belonged to the same event became associated with more differentiated neural patterns. A computational model explained these results by dynamic inhibition in the dentate gyrus. Additional similarity measures support the notion that CA3 clustered representations reflect shared voxel populations, while dentate gyrus distinct item representations reflect different voxel populations. These findings suggest an interplay between temporal differentiation in dentate gyrus and attractor dynamics in CA3. They advance our understanding of how knowledge is structured through integration and separation across time and context. SignificanceA major challenge of our memory system is to integrate experiences occurring in the same context to generalize context-appropriate knowledge, while also maintaining distinct representations of these same occurrences to avoid confusion. Here, we uncover a novel mechanism for hierarchical learning in the human hippocampus that might help to resolve this tension. In the CA3 subregion of the hippocampus, the neural representations of items presented sequentially in the same context, but not in different contexts, became more overlapping with learning. In contrast, adjacent items, appearing close in time and in the same context, became increasingly more differentiated in the dentate gyrus. Thus, multiple representations in different hippocampal subregions encoded in parallel might enable simultaneous generalization and specificity in memory.
Matching journals
The top 5 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- Progression from feature-specific brain activity to hippocampal binding during episodic encoding 98%
- An enduring role for hippocampal pattern completion in addition to an emergent non-hippocampal contribution to holistic episodic retrieval after a 24-hour delay 98%
- Signed Reward Prediction Errors in the Ventral Striatum Drive Episodic Memory 98%
Similar papers in this journal
- The locus of recognition memory signals in human cortex depends on the complexity of the memory representations 97%
- Prestimulus Activity in the Cingulo-Opercular Network Predicts Memory for Naturalistic Episodic Experience 97%
- Subcortical nuclei of the human ascending arousal system encode anticipated reward but do not predict subsequent memory 97%
Similar papers in this journal
Similar papers in this journal
- Neural similarity between overlapping events at learning differentially affects reinstatement across the cortex 97%
- Frontoparietal action-oriented codes support novel instruction implementation 97%
- Temporal contiguity training does not affect size-tolerant representations in object-selective cortex 96%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.