Social saliency

Saliency historically refers to the bottom-up visual properties of an object that automatically drive attention. It is an ordinal property that depends on the relative saliency of one object with respect to others in the scene. Simple examples are a red spot on a green background, a horizontal bar among vertical bars, or a sudden onset of motion. Researchers have introduced the idea of a saliency map, an abstract and featureless map of the ‘winners’ of attention competition, to model the dynamics of visual attention. The standard saliency map involves channels like color, orientation, size, shape, movement or unique onset. But how do complex stimuli, especially stimuli with social meaning such as faces, pop out and attract attention? Suppose you are attending a big party: your attention might be captured by someone in a fancy dress, someone looking at you, someone who is attractive, familiar, or distinctive in some way. This happens essentially automatically, and encompasses a huge number of different stimuli that are all competing for your attention. What determines which is the most salient, and how can we best measure this?.