Neural correlates of 3D versus 2D perception: An activation likelihood estimation meta-analysis

Despite hundreds of neuroimaging studies examining the neural correlates of 3D shape perception (as opposed to 2D), there is no consensus because of the diversity of stimuli and depth cues used. We addressed this problem through an activation likelihood estimation (ALE) coordinate-based meta-analysis, pooling together studies that examined the 3D vs 2D shape contrast across multiple depth cues used to render the 3D shapes. A systematic review was performed using Medline, PsychInfo and Embase databases and yielded 25 empirical studies after screening. Articles were split into depth cue types, yielding 11 for binocular disparity, 10 for motion, and 2 for shading and texture each. We performed three sets of ALE-based coordinate analysis—full-sample ALE analysis, sub-analyses testing individual depth cues separately, and a contrast analysis between disparity-defined 3D shapes and monocularly-defined 3D shapes. Results for the full-sample analysis showed that 3D shape perception is widespread throughout the high-level visual cortex regardless of depth cue. Disparity-defined 3D shapes seem to reliably engage higher-level dorsal stream areas, including bilateral intraparietal sulcus (IPS). Motion-defined 3D shape preferentially recruited ventral stream regions associated with object recognition processes. Monocularly-defined 3D shapes reliably recruited ventral stream areas, mainly the bilateral inferior lateral cortex and dorsal stream IPS for the right hemisphere. When contrasted with disparity, 2D vision reliably recruited the left lateral occipital cortex. The results suggest laterality in 3D versus 2D shape representations and that 3D shape representations occur in both ventral and dorsal pathways regardless of the depth cues that define them.