Neural mechanisms of top-down selection during visual search

Abstract

Abstract-The brain's representation of visual information depends greatly on the behavioral relevance of the viewed stimuli. While in some instances behavioral significance is derived from conspicuity, in many situations significance depends on top-down factors such as the viewer's goals and knowledge.Studies combining neural recordings and behavioral observations have begun to elucidate how the brain selects visual stimuli based on top-down information. While many visual areas of the brain that are selective for visual attributes participate in the selection process, the outcome of the selection process across these areas appears to be represented in structures like the frontal eye field, a key stage in the transformation of visual selection into a command to move the eyes. Evidence shows that the frontal eye field exhibits all the characteristics of a visual salience map in which the behavioral significance of stimuli derived from bottom-up and top-down influences is represented. The patterns of neural modulation in structures like the frontal eye field can be used to design more efficient machine-vision algorithms for target selection.Keywords-Visual selection, attention, eye movements, saccade, cognition, memory, salience map, frontal eye field INTRODUCTIONDespite our subjective feeling that we "see" everything within our field of vision, not every part of a visual scene is processed to the same degree [1], [2]. Instead, we attend to objects of interest while ignoring irrelevant ones. Influences on visual selection can be broadly divided into bottom-up and top-down categories. While bottom-up influences are derived from the intrinsic conspicuity of features in a visual scene (e.g., a bright stimulus), top-down influences are derived from the goals and knowledge of the viewer. A large body of literature shows that eye movements reflect cognitive processes during various viewing contexts, which include visual search [3], natural scene perception [4], and reading [5].Neural mechanisms underlying visual selection have been investigated extensively. Correlates of visual selection appear to be reflected in nearly all visual or visualassociation areas (for reviews, see [6], [7]), and perhaps even as early as V1 (e.g., [8]). Except for those at the earliest stages of the visual system, most of these areas project to the frontal eye field (FEF) [9], which appears to be one of the highest points of convergence of the dorsal and ventral visual information streams in the brain [10]. In turn, FEF projects strongly to brainstem oculomotor structures [11], and thus is ideally positioned to transform the outcome of visual processing into a command to move the eyes. Neural correlates of bottom-up selection in FEF have been reviewed [12]. This paper focuses on studies demonstrating selection in FEF based on top-down factors, completing the argument that this area contains a visual salience map in which behavioral significance, regardless of its source, is represented. NEURAL MECHANISMS OF TOP-DOWN SELECTION DURING VISUAL SEA...

Priming in Macaque Frontal Cortex during Popout Visual Search: Feature-Based Facilitation and Location-Based Inhibition of Return

“…We have previously shown that F EF exhibits the characteristics of a visual salience map in which the behavioral relevance of stimuli is represented (for review, see Bichot, 2001a;). This conclusion was reached based on neuronal modulation in F EF during both bottom-up and top-down (Bichot, 2001b) visual search tasks, and is consistent with the findings of brain imaging and lesion studies of covert attention and overt saccade production in humans (Nobre et al, 1997;Corbetta, 1998;Mesulam, 1999;Donner et al, 2000). Here, we investigated the neuronal mechanisms underlying changes of performance during popout visual search caused by the repetition of stimulus features and target position.…”

Edge Detection Based on Attention Mechanism of Vision Perception

“…It is well known that the selective visual attention is driven both by a signal-based (Bottom-up) [7] and by a task-based mechanism (Top-Down) [8] . The design of computational models simulating the human selective visual attention is a difficult issue.…”

“…Stimulus-driven attention appears to act as a rapid, bottom-up, task-independent mechanism, while goal-driven selective-attention appears to act in a slower top-down volition-controlled manner. Several computational models of selective attention have been proposed [7] [8] .…”