Abstract
BACKGROUND: According to theories of figure-ground segmentation, observers ought not to perceive figural shape from 3D concave regions. While concave regions may have a global bounding contour, the figure-ground and part parsing cues provided by their minima of curvature always indicate that the bounded region should not form a coherent figure. Nonetheless, 3D concavities are perceived to have figural shapes, although they do not cue attention in an object-like way (Cate & Behrmann, 2010) and activate object-selective cortex less than convex regions (Haushofer et al., 2008), similar to 2D aperture holes (Albrecht et al., 2008; Vinberg & Grill-Spector, 2008). We propose that temporopolar cortex implicated in illusory depth reversals in the hollow-face illusion may be critical for perceiving figural shape from 3D concavities. METHODS: 19 healthy participants performed a fixation cue discrimination task during block design fmri. Three different blocks displayed truncated rectangular prisms oriented as concave with respect to the frontoparallel plane, which appeared at 3 different fixation disparities (including both crossed and uncrossed); three blocks displayed convex prisms. Four blocks displayed flat frontoparallel stereograms using the same fixation disparities as the concave and convex blocks, which served as controls for the effects of absolute disparity. Stereograms were projected into the scanner using a PROPixx DLP projector with a circular polarizing filter running at 120 Hz (60 Hz/eye). Activation in the superior temporal pole was analyzed with small volume correction, and otherwise activations were thresholded using family-wise error correction in SPM8. RESULTS: Contrasting concave versus convex 3D stereograms showed activation in the left hemisphere superior temporal pole (BA 38), while convex versus concave activated lateral occipital cortex bilaterally. CONCLUSION: The same process that produces illusory depth in the hollow-face illusion may enable perception of figural shape from concave regions.