Characterizing Human Disparity Tuning Properties Using Population Receptive Field Mapping.

Abstract

Our visual percept of small differences in depth is largely informed by binocular stereopsis, the ability to decode depth from the horizontal offset between the retinal images in each eye. While multiple cortical areas are associated with stereoscopic processing, it is unclear how tuning to specific binocular disparities is organized across the human visual cortex. We used 3 T functional magnetic resonance imaging to generate population receptive fields (pRFs) in response to modulation of binocular disparity to characterize the neural tuning to disparity. We also used psychophysics to measure stereoacuity thresholds compared with backgrounds at different depths (pedestal disparity). Ten human participants (seven females) observed correlated or anticorrelated random-dot stereograms with disparity ranging from -0.3 to 0.3°, and responses were modeled as one-dimensional tuning curves along the depth dimension. First, we demonstrate that lateral and dorsal visual areas show the greatest proportion of vertices selective for binocular disparity. Second, with binocularly correlated stimuli, we show a polynomial relationship between preferred disparity and tuning curve width, with sharply tuned disparity responses at near-zero disparities, and broader disparity tuning profiles at near or far disparities. This relationship held across visual areas and was not present for anticorrelated stimuli. Finally, the individual thresholds for psychophysical stereoacuity at the three different pedestal disparities were broadly related to pRF tuning width in area V1, suggesting a possible limit for fine stereopsis at the earliest level of cortical processing. Together, these findings point to heterogeneity of disparity processing across human visual areas, comparable with nonhuman primates.