Abstract
This study explores the effects of high-level image interpretation on responses in early visual areas. In a recent fMRI study Hsieh et al. (2010) found that the pattern of activation in early visual cortex to an identified two-tone "Mooney" image of an object was more similar to the response to its full grayscale image than to the same two-tone image when it was not identified. We report results of an analogous electrophysiological experiment: similar stimuli and experimental design were used, while evoked responses (ERPs) and spectral power responses (EEGs) were recorded. 5 images of natural objects were presented to 14 subjects in consecutive blocks: two-tone (no recognition), full-grayscale (recognition), two-tone (recognition) for each image. Each presentation consisted of 1 second image interval followed by 1 second blank screen, repeated 40 times. Central fixation was maintained. We found that, unlike fMRI, electrophysiological responses to two-tone and grayscale images were dissimilar, even after two-tone images were recognized. ERPs were almost identical between the two "Mooney" image blocks (recognition - no recognition) except for small (but significant) differences. In contrast, the grayscale block results were very different from the "Mooney" blocks results. In particular, the grayscale image ERPs were characterized by strong occipito-lateral activations 300 - 400 msec after the stimulus onset - those were missing in "Mooney" blocks. Comparing EEGs across the 3 blocks we found that image identification was accompanied by significant power increase in all frequency bands. The identification of two-tone images caused larger power increase compared to identification of grayscale images, and the time-course of EEGs was generally different between these images. We conclude that although image identification affected processing in early visual areas, the effects were quite distinct for bottom-up (grayscale) and top-down (two-tone) identifications.
Meeting abstract presented at VSS 2012