Abstract
To what extent are real and illusory contours represented in a similar manner by the visual system? Much evidence indicates that extrastriate area V2 responds well to illusory contours, but the role of V1 in illusory contour perception is still debated. Using decoding methods that we recently developed (Kamitani & Tong, Nat Neurosci, 2005), we measured orientation-selective responses in the human visual cortex to real and illusory contours. Subjects performed a demanding letter discrimination task at central fixation while real or illusory gratings were presented in the periphery. Illusory gratings consisted of abutting segments of real tilted lines (45 or 135 deg) that formed higher-order vertical or horizontal patterns. Orientation discrimination performance of each visual area was assessed by using a linear classifier to categorize their activity patterns according to the viewed orientation. We found that individual visual areas (V1–V4) could discriminate the orientation of real gratings with high accuracy (∼90% correct, chance level 50%). In comparison, orientation-selective responses to illusory contours were much weaker but still reliable, as indicated by moderate classification accuracy (65–70% correct). Interestingly, training on illusory orientations led to accurate classification of real orientations (∼90% correct) for all visual areas. This indicates that the pattern of orientation-selective responses evoked by real and illusory gratings are very similar but simply weaker for illusory stimuli. Finally, we tested whether the real orientation component contained within illusory stimuli could be decoded. Whereas V1 and V2 could accurately distinguish real orientations embedded in illusory contours, higher areas performed poorly, suggesting that the presence of the illusory contours disrupted their sensitivity to real contours. Taken together, these findings indicate that V1 and higher areas are sensitive to the orientation of illusory contours, but that early areas (V1–V2) maintain greater sensitivity to the real components that define the illusory stimuli.
Research support: NEI R01-EY14202 to FT and Nissan Science Foundation Grant to YK.