Name: Liam Hart

Master’s Thesis Defense Meeting

Data: Friday, May 8th, 2026

Time: 12:00PM – 2:00PM

Location: Room 148 and virtual, link here: https://gatech.zoom.us/j/92396114398?pwd=oyyqwzRKjsfkPD7WfrK0UKtsdH6hEi.1

Thesis Chair/Advisor:
Hsiao-Wen Liao, Ph.D. (Georgia Tech)

Thesis Committee Members:

Sashank Varma, Ph.D. (Georgia Tech)

Audrey Leroux, Ph.D. (Georgia Tech)

Title: Gist-Based Compensation for Age-Related Episodic Decline in Emotional Autobiographical Memory: A Natural Language Analysis of Life’s Highs and Lows 

Abstract:

Age-related declines in episodic specificity during autobiographical recall are well-established. Recent work, however, suggests that older adults' preserved semantic knowledge may scaffold episodic retrieval. Sheldon et al. (2024) found that semantic content diversity — the breadth of semantic content within a memory narrative — predicted greater episodic, but not non-episodic, detail in memories collected via the Autobiographical Interview, a task that constrains recall toward maximizing episodic specificity. Whether this pattern generalizes to retrieval contexts that foreground identity and personal meaning remains untested. This preregistered study examined whether semantic content diversity predicted episodic detail in the Life Story Interview, in which memories are situated within a personal life story. A total of 4576 memory narratives spanning five memory types (high points, low points, turning points, critical events, and health events) from 932 older adults (ages 62–75; 56.3% female) were analyzed using LLM-based detail scoring and topic modeling–derived semantic content diversity. Multilevel models showed that semantic content diversity predicted greater episodic detail. This relationship held at both the memory and individual levels, with between-individual effects of greater magnitude. Neither association was moderated by memory type. While high point memories contained more episodic detail than low point memories, consistent with socioemotional aging accounts, the two memory types did not differ in semantic content diversity. Contrary to Sheldon et al. (2024), semantic content diversity was found to positively predict non-episodic detail. Taken together, the current study findings suggest that semantic knowledge supports detailed life story memories in older adults. The influence of retrieval context in shaping the scaffolding role of semantic processing in autobiographical memory is discussed.