Abstract

L. Itti, Modeling Primate Visual Attention, In: Computational Neuroscience: A Comprehensive Approach, (J. Feng Ed.), pp. 635-655, Boca Raton:CRC Press, 2003. (Cited by 40)

Abstract: Selective visual attention is the mechanism by which we can rapidly direct our gaze towards objects of interest in our visual environment. From an evolutionary viewpoint, this rapid orienting capability is critical in allowing living systems to quickly become aware of possible preys, mates or predators in their cluttered visual world. It has become clear that attention guides where to look next based on both bottom-up (image-based) and top-down (task-dependent) cues. As such, attention implements an information processing bottleneck, only allowing a small part of the incoming sensory information to reach short-term memory and visual awareness. That is, instead of attempting to fully process the massive sensory input in parallel, nature has devised a serial strategy to achieve near real-time performance despite limited computational capacity: Attention allows us to break down the problem of scene understanding into rapid series of computationally less demanding, localized visual analysis problems. These orienting and scene analysis functions of attention are complemented by a feedback modulation of neural activity at the location and for the visual attributes of the desired or selected targets. This feedback is believed to be essential for binding the different visual attributes of an object, such as color and form, into a unitary percept. That is, attention not only serves to select a location of interest, but also enhances the cortical representation at that location. As such, focal visual attention is often compared to a rapidly shiftable spotlight, which scans our visual environment both overtly (with accompanying eye movements) or covertly (with the eyes fixed). Finally, attention is involved in triggering behavior, and consequently is intimately related to recognition, planning and motor control. Of course, not all of vision is attentional, as we can derive coarse understanding from presentations of visual scenes that are too brief for attention to explore the scene. Vision thus relies on sophisticated interactions between coarse, massively parallel, full-field pre-attentive analysis systems and the more detailed, circumscribed and sequential attentional analysis system. In what follows, we focus on several critical aspects of selective visual attention: First, the brain area involved in its control and deployment; second, the mechanisms by which attention is attracted in a bottom-up or image-based manner towards conspicuous or salient locations in our visual environment; third, the mechanisms by which attention modulates the early sensory representation of attended stimuli; fourth, the mechanisms for top-down or voluntary deployment of attention; and fifth, the interaction between attention, object recognition and scene understanding.

Themes: Computational Modeling, Model of Bottom-Up Saliency-Based Visual Attention, Model of Top-Down Attentional Modulation, Human Psychophysics, Review Articles and Chapters, Scene Understanding