Abstract

Episodic memory often captures both the content of past events (e.g., people, places, and situations involved) and the specificity with which those events are remembered, ranging from summary-level to detailed recollection. Content-based theories, including the PMAT Framework, propose that cortical networks are organized around the types of information memories contain. In particular, the Anterior Temporal Network (ATN) supports entity representations, including characters, and the Posterior Medial Network (PMN) support contextual information, such as locations. Specificity-based theories, including the Trace Transformation Theory, propose that the level of recall specificity maps onto functional gradients within the hippocampus, with posterior regions (pHPC) supporting detailed recollection and anterior regions (aHPC) supporting gist-like recall. Although both dimensions are fundamental to episodic memory, they have rarely been investigated within the same experiment, and whether the organizational principles derived from these paradigms generalize to narratively complex, real-world events remains unclear. Here, I examined how content and specificity shape the neural organization of memory for naturalistic events. Participants underwent fMRI while encoding and recalling eight narrative events that systematically recombined two characters, two locations, and two themes. Retrieval specificity was manipulated by cueing participants to produce either gist summaries or detailed recollections of the events, both immediately and after a two-week delay. Additional exposure phases presented characters and locations in isolation across the experimental timeline to examine how individual feature representations contribute to event memories. Combining behavioral measures with univariate and multivariate fMRI analyses (i.e., model-based RSA, pattern reinstatement, and feature-to-event pattern similarity), I tested how content and specificity organize neural representations across cortico-hippocampal systems over time. Behaviorally, my specificity manipulation was effective and consistent across the delay. At the univariate level, cortical networks showed content-sensitive engagement consistent with the PMAT Framework. However, multivariate analyses revealed that representational structure during retrieval was organized around integrated event and thematic information rather than separable character and location dimensions. In the hippocampus, specificity modulated overall engagement but did not produce the predicted anterior-posterior hippocampal dissociation, which is inconsistent with the Trace Transformation Theory. Hippocampal contributions were instead most evident during active recall, where character cues selectively resembled recall patterns from events containing that character, which did not emerge for locations or in cortical networks. These findings motivate an account in which cortical networks maintain broad representations capturing content type and situational structure, while the hippocampus resolves the specific episode when retrieval demands it. This work demonstrates that organizational principles established in previous paradigms do not straightforwardly generalize to narratively complex events and highlights the importance of testing theoretical frameworks under conditions that approximate the complexities of real-world episodic memory.

Committee Chair

Zachariah Reagh

Committee Members

Alexandra Decker; Ian Dobbins; Jeffrey Zacks; Maureen Ritchey

Degree

Doctor of Philosophy (PhD)

Author's Department

Psychology

Author's School

Graduate School of Arts and Sciences

Document Type

Dissertation

Date of Award

4-21-2026

Language

English (en)

Author's ORCID

https://orcid.org/0000-0003-0121-7618

Download

Share

COinS