Another distinct line of work that we have spent a number of years developing focuses on the interactions between memory and perception/cognition that serve to organize our memories. One way of thinking about this question (and indeed the one that inspired our thinking on the topic) is to ask what constitutes an ‘episode’ in episodic memory? We have pursued this question using definitive assays of episodic memory, fMRI and MEG.
A key discovery from our lab is that changes in context, or event boundaries, embedded within a continuous sequence of to-be-encoded items mark changes in context such that within-event representations of stimuli are more tightly bound to each other in comparison to those that occur across (i.e., on either side of) an event boundary. This effect is highly reproducible and is evident using a variety of behavioral measures such as cued recall, serial recall, temporal order, subjective temporal distance and implicit measures of priming. We have used this paradigm (dubbed the Ezzyat-DuBrow-Davachi, or EDD paradigm) to show that regions in medial PFC and the MTL are sensitive to event structure and that stability in hippocampal neural patterns across time promotes sequence memory. Using MEG, we have also shown that event structure organizes low frequency oscillations. Further, theta-gamma coupling increases as items are encountered as part of the same event or context and drops off sharply at boundaries.
A parallel line of studies has used high-resolution fMRI scanning to address finer grain questions about how the hippocampus and related cortical regions shift between different states of encoding or between encoding and retrieval, One simple but powerful notion is that contextual novelty triggers encoding (or pattern separation) while old information will trigger retrieval (or pattern completion). Our research has shown that the hippocampus shows both match (old) and mismatch (novelty) signals but they were differentially localized to the anterior (for novelty mismatch signal) and posterior (match signal) hippocampus. Using high-resolution imaging, we have shown that hippocampal area CA1, more than any other hippocampal subregion, shows a parametric sensitivity to change in the environment and, thus, is in a good position to bias hippocampal processing towards incoming novel information for encoding or towards pattern completion to support retrieval. Our most recent report provided evidence that CA1 functional connectivity with CA3 is indeed significantly enhanced during episodic retrieval, compared to encoding, and predicts successful retrieval whereas CA1 connectivity with the ventral tegmental area (VTA) is related to encoding success. These results suggest a dynamic interplay between area CA1, in particular, and other brain regions (CA3, perirhinal cortex, VTA) in orchestrating system processing towards memory formation or retrieval. Related to this, we were the first to demonstrate that encoding and retrieval may indeed represent ‘states’ as these states can linger and influence subsequent memory decisions.
