Nathaniel Goldstein¹ (), Laurie M. Wilcox¹, Erez Freud¹; ¹York University, Toronto, Canada

Humans rely on visual depth cues in our day-to-day lives, however the extent to which specific sources of depth information are involved in reach-to-grasp actions is not clear. There is evidence that binocular depth cues (vergence, stereopsis) improve grasping efficiency; however, this conclusion is typically based on comparisons of discrete timepoints from grasping sequences (e.g., the maximum grip aperture). While valuable, these measurements overlook potentially rich information from the full reach-to-grasp trajectory. To better understand the role of depth cues in grasping, we compare the spatiotemporal structure of grip aperture trajectories under two viewing modalities: binocular and monocular, with and without motion parallax. In two experiments (n=30 each) participants grasped 3D-printed discs and moved them to a nearer or further location. In Experiment 1 motion parallax was available via natural head movements. In Experiment 2 the head was stabilized on a chinrest, thereby limiting parallax. Consistent with previous studies, in both experiments we found that kinematic attributes like movement time and wrist velocity were modulated by viewing modality. To compare the trajectories across conditions, we used Dynamic Time Warping (DTW) to compute dissimilarities between all grip aperture trajectories and built participant-level representational dissimilarity matrices (RDM). We then conducted a representational similarity analysis (RSA) to compare empirical RDMs to a theoretical model encoding only viewing modality. Our results showed a significant correlation between participant-level dissimilarity matrices and the viewing modality model, confirming that stereopsis shapes grasping dynamics. Also, binocular trajectories were more consistent than monocular trajectories. In both cases the availability of motion parallax did not impact these result patterns. Our work confirms that under these natural grasp conditions, motion parallax does not compensate for the lack of stereopsis. Further, our analysis provides new evidence that stereoscopic information modulates not just specific timepoints, but the overall structure of grasping aperture sequences.

Acknowledgements: Natural Sciences and Engineering Research Council of Canada (NSERC); CFREF Vision: Science to Applications (VISTA); CFREF Connected Minds (CM)