A new selective attention model is proposed in this paper, which integrates a top-down attention
mechanism into a bottom-up saliency map model to generate salient areas related with human
interest. Human selects the certain area from natural scene and decides whether the selected area is
preference or refusal. The fuzzy adaptive resonance theory (ART) network trains and memorizes the
characteristic of that area, also generates a refusal or a preference signal so that the sequence of test
areas is modified to be a desired scan path. The proposed model generates a plausible scan path
based on human interest by endowing weight values to feature maps in a course of constructing the
saliency map.