My research interest is to understand how perception and action emerge from brain activity. My main approach is to record neural activity in the brains of monkeys performing various visual search and forced-choice discrimination tasks. Most of my research has focused on the frontal eye field (FEF), which is located in the prefrontal cortex, and participates in the transformation of visual information into commands to move the eyes. My recent research can be divided into two related projects, (1) Visual Search, and (2) Signal Detection. The goal of the visual search experiments is to understand how the brain chooses which of many possible visual objects will be used to guide eye movements. The signal detection experiments aim to understand the neural mechanisms of near-threshold perceptual judgments. My working hypothesis is that the FEF functions as a visual salience map that combines visual input and cognitive influences into a topographic map of visual conspicuity, or salience, at every location in the visual scene. The salience map is a prominent feature of many theoretical models of directed spatial attention and saccade target selection. I hypothesize that covert spatial attention and saccades are directed to the spatial location represented by the highest activity in the FEF. My coworkers and I have shown that during a visual search task for an oddball target among distractors, visually responsive FEF neurons specify the target before the saccade that directs gaze to the target. The results from our recent work with monkeys performing various visual search tasks strongly suggests that the target discrimination process in the frontal eye field indexes the outcome of visual processing, not saccade preparation. In Sato et al., 2001, we increased the monkeys' reaction times by making the target similar to the distractor or by occasionally changing the location of an easily discriminated target and requiring the monkey to withhold the saccade to the first location and make a single saccade to the new target location. Saccadic reaction time increased in both task conditions, but only the change in discrimination difficulty increased the amount of time it took for frontal eye field neurons to identify the saccade target. In Murthy et al., 2001, we analyzed those trials on which the target changed locations before the saccade. The activity of frontal eye field visual neurons very reliably identified the change in the target stimulus location regardless of whether or not the saccade was made to that stimulus. Recently we have investigated the effect of prior knowledge on the target discrimination process by manipulating the similarity of the distractors to an unchanging target during visual search. We found that the neural representation of the distractors that were more similar to the target was greater than the neural representation of the distractors that were less similar to the target, even when no target was present. This shows that the neural representation of objects in the frontal eye field is a combination of the physical properties of the image and a top-down knowledge of what is being looked for. An analysis of FEF visual activity on error trials showed that when the representation of the distractor exceeds that of the target, even by a small amount, the saccade is made to that distractor. These studies have extended our understanding about the frontal eye field far beyond its familiar role in controlling eye movements. Taken together, they validate computational models of selective attention by identifying a population of neurons that have all of the characteristics of a theoretical salience map. With this knowledge we can design experiments to investigate the flow of sensory information through the brain as it is transformed into perception and action.
Showing the most recent 10 out of 12 publications
