S-070 | Quantitative analysis of autobiographical memory: validation of NLP-derived metrics from narrative data

Cognition, Behavior, and Memory
Author: Zoe Werner | Email: zoewerner27@gmail.com

Zoe Werner^1°, Corina Revora^2°, Luz Bavassi^2°3°,  Laura Kaczer^1°

^1° Laboratorio de Lenguaje y Cognición. Departamento de Fisiología, Biología Molecular y Celular, FCEyN, UBA
^2° Departamento de Física, FCEyN, UBA
^3° IFIBYNE, UBA, CONICET

The quantification of memory processes is an ever-evolving area in neuroscience. Autobiographical memories pose a particular challenge due to the high variability of individual experiences, which complicates systematic and quantitative analysis. These memories consist of an episodic component (related to the spatiotemporal context) and a semantic component (abstract details), both reflected in narratives. The primary objective of this project is to validate automated, NLP-derived measurements against the established gold standard of manual scoring (Levine et al., 2002).
For this, we have collected narratives from 60 participants and transcribed them. The scoring manual was initially adapted to Spanish and we quantified the number of internal and external details present in each narrative. We will then conduct a correlation analysis to compare these metrics with the ones previously obtained from NLP to identify which variable provides the most informative insight regarding the quantification of autobiographical memory. Through this approach, we aim to identify markers of memory plasticity in the structure and content of the narratives.