Donna Dean¹, Abe Leite, Gregory Zelinsky; ¹Stony Brook University

How is a person’s ability to understand and remember non-social scenes influenced by their psychological traits? In the initial encoding session, 44 participants viewed 150 non-social scenes (from a 300-image subset of the Visual Genome dataset) for a controlled number of fixations (1-10) and were prompted to describe what they saw in as much detail as they could. In the retrieval session (2-21 days later), participants viewed a partially overlapping set of 150 scenes for 250ms each, indicating which scenes they remembered with a button-press following each trial. Finally, they completed a battery of psychometric questionnaires including IPIP (personality), AQ (autistic traits), PHQ-9 (depression), and GAD-7 (anxiety). We performed a forward stepwise regression analysis to assess which traits significantly predicted performance. During encoding, participants who scored high on the AQ attention-to-detail subscale (p < .005) and IPIP agreeableness (p < .05) mentioned more nouns in their descriptions, indicating finer-grained scene understanding. During the retrieval session, participants who had stronger depressive tendencies (r = 0.51, p < .005) and IPIP openness-to-experience (p < .05) had better memory performance (hits - false alarms) during the scene recall task. The openness-to-experience result was expected since prior research shows that openness can lead to improved long-term memory performance in older adults (Stephan et al., 2020). The depression finding is more surprising. It is known that individuals who meet the DSM-5 criteria for Major Depressive Disorder show decreased memory for positive events, intensified memory for negative events, and impaired memory recollection overall (Dillon & Pizzagalli, 2018); does this mean that the neutral non-social scenes in our dataset are processed differently than the positive and negative stimuli used in prior work? In future work, we will correlate these measures with the eye-tracking data in the encoding session to seek a mechanistic explanation for these intriguing results.

Acknowledgements: Stony Brook University