Stephanie Kristensen¹, Julie Golomb¹; ¹Department of Psychology, The Ohio State University

Accurate perception of positions in depth is essential for human behavior. During natural vision, binocular disparity (the slight image difference between left and right eyes) is an important cue to perceive depth. Neural selectivity for binocular disparity is distributed across the visual cortex in regions as early as V1, though some prior fMRI studies suggest a transition from 2D to 3D processing along the visual hierarchy with more specialized processing of 3D depth perception in later visual areas. Our goal is to describe how neural populations across human cortex represent perceived depth and whether there exists a systematic organization of depth-related neuronal tuning, similar to 2D retinotopic maps. We reanalyzed data from an fMRI population receptive field (pRF) modeling experiment (Golomb, 2018) during which participants (N=8) passively viewed a random dot motion stereogram while wearing red/green anaglyph glasses. The stimulus depicts a full-field plane of dots located at varying positions-in-depth relative to fixation. Within a session, the plane moved forward or backward through 13 depth positions ranging from –18 arcmin (perceptually near, in front of fixation) to 18 arcmin (perceptually far, behind fixation). We adapted standard pRF modeling with a 1D Gaussian to identify the preferred depth and tuning width (range of preferred disparities) for each voxel. To improve model fits, we also incorporated a parameter estimating each voxel’s relative sensitivity to depth vs depth-invariant visual stimulation. The results revealed that voxels in early visual cortex exhibit primarily depth-invariant visual preferences, while tuning to position-in-depth increases in later visual areas. In particular, a cluster along the transverse occipital sulcus (overlapping with V3a/b and IPS0) shows a high degree of voxel-specific depth selectivity, with spatially separated voxels preferring negative or positive depths. Overall, this study builds on recent studies indicating that nuanced position-in-depth representations may reside in later visual areas.

Acknowledgements: NIH R01-EY025648 (JG)