Voxel-wise binocular energy models for binocular disparity decoding

Abstract

Binocular disparity is a powerful cue for depth perception in three-dimensional (3D) space. Some neurophysiological studies proposed the binocular phase-shift and position-shift energy models to predict the responses of individual disparity-tuned neurons in cats and macaques. By far, it is unclear how to use binocular energy models to characterize the voxels’ responses in human visual cortex. In this study, we introduced the binocular energy models to the functional magnetic resonance imaging study and constructed the position-shift receptive-field model (Position-RFM) and the phase-shift receptive-field model (Phase-RFM) to predict voxel responses to disparity and to identify novel disparity levels from voxel responses. The results revealed that Phase-RFM outperformed Position-RFM in fitting the voxel responses for all the visual regions. Moreover, the novel disparity levels can be better identified from voxel’s responses in visual regions by Phase-RFM than Position-RFM. The findings may suggest that Phase-RFM can better encode the responses of disparity-tuned neuron populations than Position-RFM for human visual regions.