Shaya Samet¹, Yara Iskandar¹, Erez Freud¹, Peter J Kohler¹; ¹York University

Symmetry is a fundamental principle important for perceptual organization. Here we ask whether symmetry facilitates working memory in a task that is unrelated to object shape or identity. In addition, we investigate whether potential effects of symmetry on working memory are specific to symmetry in the image plane, or persist when symmetries are distorted due to perspective. Participants were presented with images of symmetrical and asymmetrical novel objects generated procedurally in the Blender 3D graphics software. The image-plane group (n = 60) was shown images where the symmetry axis of the object was orthogonal to viewing direction, leading to symmetry in the image plane. The distortion group (n = 60) was shown images where the objects were rotated relative to viewing direction, leading to perspective distortion. To replicate the online experiment under controlled viewing conditions, we conducted the experiment in the lab as well. In-person experiments recruited participants (N = 120) with normal or corrected-normal vision, with each experiment placing half in the image-plane group and half in the distortion group. The in-person experiments allowed us to control viewing distance and thus maintain a constant stimulus size in º/visual angle across participants. On each trial, 1-6 objects were shown at different locations on the screen, at different pseudo-randomly chosen orientations. After a retention interval (1000 ms), the participant was shown the same image and asked if the stimuli were presented in the same orientations as the original image. On half the trials this was true, and in the other half one of the images had changed orientation. In both online and in-person experiments, we found increased sensitivity for symmetrical objects, but only when symmetry was present in the image plane, not for perspective distorted symmetry. This suggests that image level symmetry facilitates representations of object orientation in working memory.

Acknowledgements: This work was supported by the Vision Science to Applications (VISTA) program funded by the Canada First Research Excellence Fund (CFREF, 2016–2023) and by a Discovery Grant from the Natural Sciences and Engineering Research Council of Canada awarded to PJK