The electrophysiological correlate of contour integration is modulated by task demands.

Psychophysical studies demonstrated under which task conditions contour integration based on orientation cues succeeds or fails. We investigated how the electrophysiological correlate of contour integration is modulated by changes in task demands. In two experiments, event-related potentials (ERPs) were recorded. We presented open or nearly closed contours defined by Gabor elements embedded in a background of randomly oriented distracters. In experiment 1, randomly oriented Gabors preceded the test stimulus, while in experiment 2, the screen was blank before appearance of the test stimulus. Correct performance was above 96% for aligned and slightly misaligned contours but did not significantly differ from chance level for random orientations of the contour elements. Detectable contours elicited a negative shift over posterior recording sites. This effect started about 150 ms after stimulus onset but was delayed for more difficult contour integration. Presenting the stimulus after a random display elicited a frontal selection positivity (FSP). In both experiments, contours elicited a P3, which was larger for aligned compared to misaligned contours. Timing, duration and localisation of the negative enhancement were similar to both the texture segmentation VEP (tsVEP) and the selection negativity (SN). Our results indicate that contours are processed similar to textures and that decreasing saliency of detectable contours is compensated by increasing processing time which possibly results from increasing allocation of visual selective attention.