Abstract
It has been proposed that salient locations in a visual scene are represented in a spatial saliency map. For the calculation of such a saliency map different feature contrast maps for luminance, orientation, motion, color, etc. are combined. Brain regions that are candidates for encoding such a saliency map are the pulvinar, superior colliculus, the frontal eye fields and the posterior parietal cortex. But there is also evidence for a saliency map within the ventral visual pathway. In the experiment presented here we investigated the neural correlates of orientation pop-out in humans. Two types of stimuli were used: (a) Homogenous stimuli consisting of an array of bars all of which had the same orientation (randomly chosen to be 0°, 45°, 90° or 135°), and (b) pop-out stimuli consisting of homogenous arrays but with one bar in each quadrant orientated differently (0°, 30°, 60° or 90°) relative to the background. We used an event-related stimulation protocol and stimulated with pop-out stimuli for 4 sec after which the homogenous background stimuli were presented for 7, 10 or 13 sec. During stimulation attention was directed to the fixation using a task that demanded attending the center. This way no overt attention was directed to the salient positions at the two quadrants. BOLD activity in striate and extrastriate cortex showed a significant signal increase for salient positions in V1, V4 and intraparietal sulcus. Furthermore BOLD signals in V4 monotonically increased with the relative orientation of bars in the corresponding region of the visual field. However the encoding of relative orientation was not modulated by the presence of pop-out signals in other regions of the visual field. The results point to a retinotopically independent processing of pop-out signals in the absence of attention.