Arijit De¹ (), Shao Feng Liu¹, Yong Zhong Liang¹, Moaz Shoura¹, Adrian Nestor¹; ¹University of Toronto

Human cognition and behavior are intricately shaped by self-perception, self-concept and self-esteem. While prior work has begun to uncover visual traits of self-representations, detailed depictions of mental self-images and their evaluation with respect to systematic biases are still largely missing from the field. To address this, our study aims to uncover self-images from perception and memory, as well as to assess their sensitivity to specific perceptual abilities and personality traits. To this end, female Caucasian adults (N=30) evaluated the visual similarity between pairs of female face stimuli, including images of their own faces, as well as between mental images of themselves, as recalled from memory, and other face stimuli. Perception and memory-based self-images were then derived through behavior-based image reconstruction as applied to similarity data and assessed with respect to their visual content. Our investigation revealed significant levels of reconstruction accuracy relative to actual face images of the participants. It further revealed systematic correspondence between perception and memory-based representations of the self. Further, it demonstrated the impact of specific factors (e.g., as captured by self-attractiveness ratings) on the content of self-images. Thus, our findings shed new light on self-representations, on their visual content and on the factors that impact their veracity.

Acknowledgements: Natural Sciences and Engineering Research Council of Canada