Contour integration in static and dynamic scenes

A Grzymisch1, C Grimsen2, U A Ernst1

1Institute for Theoretical Physics, University of Bremen, Germany
2Human Neurobiology, University of Bremen, Germany

Contact: axel@neuro.uni-bremen.de

Contour integration is an integral part of visual information processing which requires observers to combine colinear and cocircular edge configurations into coherent percepts. Psychophysical experiments have shown that humans are efficient in these tasks, reaching considerable contour detection performances for presentation times as low as 20ms. These studies have mainly used static stimulus which are briefly flashed, or shown for an extended period. However, in nature we rarely encounter brief presentations of a visual scene, rather, we observe a scene for an extended period and develop a coherent picture which takes into account dynamic elements. It is unknown how contour integration is performed in dynamic situations, and how top-down cognitive processes, such as selective attention, interact with bottom-up feature integration. We investigate contour integration in dynamic stimulus configurations where slowly rotating Gabor elements generate dynamic contours at different times and locations on a screen. Preliminary results suggest that contours are better detected in flashed presentations of 200ms than in long presentations containing the same arrangements. Thus, contours 'pop-out' only when there is a sudden change in a visual scene, from a grey background to a field of Gabors, but require sustained attention to be detected in dynamic scenes. Acknowledgement: This work has been supported by the BMBF (Bernstein Award Udo Ernst, grant no. 01GQ1106)

Up Home