Abstract
Background. Electrophysiological studies have shown that people with schizophrenia are impaired at processing Kanizsa shapes (Foxe et al., 2005; Spencer et al., 2004), but the level at which this integration problem occurs is unclear. Does the problem owe to a poor representation of global shape, or can it instead be attributed to a more basic dysfunction in interpolating between locally relatable edges? Method. We addressed this question behaviorally by testing 18 clinically stable patients and 13 healthy controls on a "fat/thin" discrimination task (Ringach & Shapley, 1996). On each trial, a computer monitor briefly displayed four sectored circles ("pac-men"). In the "illusory" condition, the pac-men were oriented to form an illusory square and the task was to determine whether the square was "fat" or "thin". In the "fragmented" condition, the pac-men pointed downward, and the task was to indicate whether they were all rotated left or right. Half of the trials in each condition incorporated distractor lines, which are known to disrupt illusory contour formation. Performance was assessed with an entropy-based Bayesian adaptive staircase; threshold corresponded to the amount of pac-man rotation needed to achieve 80% accuracy. Results. Across all subjects, distractor lines raised thresholds more in the illusory condition than in the fragmented condition (p=.004). This interaction did not depend on participant group (p=.78), suggesting that patients and controls filled-in illusory contours comparably. Crucially, patients were worse than controls at discriminating illusory shapes (p<.004) and were the same as controls at discriminating fragmented configurations (p>.11). This interaction was significant (p=.04). Conclusion Illusory contour formation is intact in schizophrenia, but global shape integration is not. Patient impairments in representing objects from fragmented information may result from relatively high-level visual processing deficits.
Meeting abstract presented at VSS 2012