Examining the representational stabilization of lifetime period narratives in real time

Poster Session E - Monday, April 15, 2024, 2:30 – 4:30 pm EDT, Sheraton Hall ABC

Ziming Cheng^1,2 (zcheng@research.baycrest.org), Buddhika Bellana³, Samuel Fynes-Clinton¹, William Fisher³, Donna Rose Addis^1,2,4; ¹Baycrest Health Sciences, Toronto, Canada, ²University of Toronto, ³York University, Toronto, Canada, ⁴The University of Auckland

Lifetime periods represent prolonged periods of stable experience, both at the level of the individual and the collective. Retrospective studies are commonly used to study autobiographical and collective narratives for lifetime periods but cannot detect the development of narratives as that period is lived. To this end, we prospectively tracked the formation of lifetime period narratives of the COVID-19 pandemic. Participants wrote brief “chapters” about the pandemic for their autobiography and a history book, repeatedly in at least five of nine surveys conducted during 2020 and 2021. Universal Sentence Encoder, a machine-learning algorithm, quantified the similarity between pairs of narratives, enabling assessment of changes in the consistency of semantic meaning over time. First, similarity with the first narrative (May 2020) decreased over time likely reflecting the synthesis of diverse experiences; importantly, it was not simply an effect of increasing temporal distance as the reduction plateaued by early 2021. Second, the similarity of narratives generated at adjacent time-points increased over time, likely reflecting a stabilization as experiences became routine. Although for both analyses, similarities were higher for one’s collective versus autobiographical narratives, there were no time x type interactions suggesting these changes reflect a more fundamental stabilization process of lifetime period memories irrespective of whether the content is autobiographical or collective. More generally, this study offers a new application of natural language processing to life story research that could be used to advance our understanding of other aspects of memory and cognition.

Topic Area: LONG-TERM MEMORY: Other