Investigating Local and Configural Shape Processing with Steady-State Visual Evoked Potentials

The perception of object shape underlies our ability to detect, recognize and manipulate objects. Both local shape (curvature) and non-local (configural) shape contribute, and recent work has used specialized stimuli and behavioural methods to dissociate these contributions. Here we used high-density EEG to explore the cortical mechanisms involved in both local and configural shape perception. Object shape silhouettes were presented during passive viewing in an SSVEP paradigm that allowed us to isolate differential brain processing between pairs of stimulus conditions. Stimuli included natural animal-shape silhouettes (upright or inverted), synthetic maximum-entropy shapes progressively matching local curvature statistics of natural shapes but lacking global (configural) regularities (Elder et al., 2018), and stimuli in which the top and bottom half have been flipped to disrupt configural shape, named Frankenstein stimuli (Baker & Elder, 2022). Our findings so far (n = 32) reveal differential activity in occipital and temporal cortices emerging 170–280 msec post-stimulus, influenced by both local curvature and global configural shape. We find clear effects of matching the local curvature statistics on brain processing in visual cortex, especially for the variance. However, even when controlling all the local statistics, responses to natural animal shapes are still quite distinct from the curvature-matched controls. Interestingly, the differential responses to natural animal shapes compared to curvature-matched controls is subject to an inversion effect, highlighting the potential influence of semantic and holistic processing on the measured responses. It is important to note, however, that inverted animals still produce measurable differential responses compared to curvature-matched controls, suggesting that some configural properties survive the inversion. Future work, including ongoing studies with the Frankenstein stimuli, will further explore what those properties are.