Abstract
Purpose: To assess the test-retest reliability in reporting depth-inverting illusions (DII) among healthy controls as a first step in testing differences with schizophrenia patients. Method: We assessed the illusion predominance (the percentage of time participants reported the illusory percept) for 12 subjects with bi-stable 3-D physical stimuli. Five stimuli were employed: two hollow faces (one realistically painted, PaintedHollowMask; the other unpainted, UnpaintedHollowMask); two reverse-perspective geometry scenes (one realistically painted, PaintedReverseScene; the other unpainted, UnpaintedReverseScene); and an unpainted convex face that should always be perceived as convex (CatchStimulus). Each stimulus was viewed twice for 2 minutes—once monocularly with subjects swaying left-to-right (to isolate motion parallax) and once binocularly with subjects on a chinrest (to isolate stereopsis). Subjects reported whether objects appeared convex or concave. Each subject repeated the same conditions in the same order, on average seven weeks later. Results: The illusion was strong for most stimuli. For first and second monocular/monocular and binocular/binocular viewings, illusion percent predominances were, respectively: PaintedHollowMask 90/62 and 59/42; UnpaintedHollowMask 65/63 and 34/38; PaintedReverseScene 92/90 and 74/54; UnpaintedReverseScene 75/70 and 38/37; CatchStimulus, as expected, 0/0 and 0/0. An intraclass correlation analysis was performed on the data to test for consistency. Overall, the IC coefficient (ICC) was acceptable. The average measures ICC’s for the monocular/binocular conditions were: PaintedHollowMask 0.609/0.698, UnpaintedHollowMask 0.029/0.464; PaintedReverseScene 0.933/0.789, UnpaintedReverseScene 0.748/0.131; ICC not valid for CatchStimulus, because subjects always reported veridical percept. Discussion: The results indicate strong illusions, suggesting significant top-down influences in visual processing for scenes with strong linear perspective cues and for faces. The low ICC values for UnpaintedHollowMask and UnpaintedReverseScene suggest that we should prefer the painted mask and the painted reverse-perspective stimuli for the schizophrenia experiments. High ICCs suggest that the painted DII stimuli produce reliable performances across time.
Meeting abstract presented at VSS 2013