Sofia Varon¹, Carol Coricelli¹, Eva Deligiannis¹, Karsten Babin¹, Kevin Stubbs¹, Laurie M. Wilcox², Jody C. Culham¹; ¹Western University, London, ON, Canada, ²York University, Toronto, ON, Canada

Information about 3D layout is crucial for many tasks in real-world scenes, including reachable spaces, where binocular disparity may be particularly effective for estimating the distances of objects to guide adept reaching and grasping actions. While extensive neuroimaging research has investigated 3D vision using simple stimuli (e.g., stereograms) and scene processing using static 2D photographs, little is known about how neural scene processing utilizes realistic 3D information. We used 3-Tesla functional magnetic resonance imaging to investigate how scene processing is affected by the addition of binocular disparity and motion parallax in 24 participants with normal binocular vision. We rendered five virtual scenes of reachable spaces (e.g., desk, kitchen table) with viewing geometry and binocular disparities consistent with the real world. These scenes were presented under five viewing conditions: 2D Static (same static image presented to each eye); 3D Static (binocularly disparate images presented to each eye); 2D Translation (video of the scene shifting without relative movement between objects); 2D Parallax (video of the scene with relative movement between objects consistent with motion parallax); 3D Parallax (video of the scene with both parallax and binocular disparity). Compared to 2D Static images, 3D Static images showed enhanced activation in the right parahippocampal place area, occipitoparietal regions (V3A, V3B, IPS0, with partial overlap with the occipital place area), MT+/LO, and the supramarginal gyrus. Compared to 2D Translation, 2D Parallax led only to a modest increase in activation in MT+. Compared to 2D Parallax, 3D Parallax led only to a modest increase in activation in the left occipital place area. While we found little impact of depth from motion, the information provided by binocular vision about the distances of objects from oneself and each other may play a more important role in scene processing than previously appreciated.

Acknowledgements: Canadian Institutes of Health Research (PJT 190159), Canada First Research Excellence Fund "BrainsCAN" grant