Abstract
Given the attentional bottleneck, the visual system must select a limited aspect of inputs for detailed processing.
Much of the selection is by bottom up mechanisms to direct gaze to the selected or most salient visual
locations for detailed processing. While saliency has been investigated extensively in behavioral studies,
its physiological basis remains controversial. I will present the theoretical proposal that the primary visual
cortex (V1) creates a saliency map of the visual space, such that the receptive field location of the most
responsive V1 neuron to a scene is most likely selected for attentional processing. This proposal generates
non-trivial predictions, and their experimental tests will also be presented. More details are available at www.gatsby.ucl.ac.uk/ zhaoping
192
