Abstract
Purpose: Previous studies have used a variety of psychophysical techniques to estimate the temporal properties of visual completion (1–3). Here, we use response classification (4) to directly measure the time course of visual completion by tracing the changes that take place in observers' templates over time. Methods: “Fat” and “thin” Kanisza figures were produced by rotating the corners of Kanizsa squares by ± 1.75 degrees. Individual trials consisted of the presentation of a fat or thin Kanizsa square defined by either Illusory or Occluded contours in 43 unique frames of high contrast Gaussian white noise over a 500 ms period. Signal contrast was varied across trials to maintain 71% correct performance. For each observer in each condition, the noise movies presented across trials were classified, combined and smoothed with a space-time convolution kernel to form a spatiotemporal classification movie. Results: The resulting classification movie for two observers in both the Illusory and Occluded conditions after 30,000 to 50,000 trials showed that they were using only the regions near the corners at the beginning of the stimulus presentation. Later, by about 150–200 ms, the observers were also using the regions that fell between the corners of the square. Conclusions: These results are consistent with the idea that the visual system requires approximately 150–200 ms to fully construct a completed representation of the Kanizsa square. To test the possibility that observers simply use the regions between the corners at a later point in time during the stimulus presentation, we are currently measuring spatiotemporal classification images in an additional “Real” condition where these regions are defined by true physical luminance contours.
1. SekulerPalmerJ Exp Psychol Gen 1992, 121: 95–111; 2. RingachShapleyVis Res 1996, 36:3037–3050; 3. MurraySekulerBennettPsychol Bull & Rev 2001, 8(4):713–720; 4. AhumadaLovellJASA 1971, 49:1751–1756.