Components of visual bias: a multiplicative hypothesis.

Attentional selection can be viewed as having two aspects: selection with respect to particular objects and selection with respect to particular categories. Both aspects are mathematically modeled in the theory of visual attention (TVA). In this paper, we expand the rate equation of the TVA and propose that the visual bias toward seeing an object x as a member of category i is a product of three factors: the expectancy (prior probability) of being presented with members of category i, the subjective importance (utility) of seeing objects in category i as members of that category, and the general level of alertness. Together, the three factors also determine the level of arousal in the visual system. The hypothesized multiplicative interaction between the three components of visual bias seems consistent with the function of an ideal observer and also paves the way for a Bayesian interpretation of the TVA.