Is saccade adaptation driven by reinforcement?

City College Of New York, New York NY

Investigators

Linked publications & trials

Abstract

DESCRIPTION (provided by applicant): When one chooses to look at something, one makes a saccadic eye movement to it. Because saccades are so brief, they cannot be guided by visual feedback, but must rely on feedback after each saccade to check the accuracy of that movement and, if needed, to adjust subsequent saccades; this is saccade adaptation. Although the specific mechanisms underlying saccadic adaptation are largely unknown, it is generally inferred that saccades are recalibrated depending on the retinal error after each saccade, i.e., the distance of the fovea from the target of the saccade. In the laboratory, saccade adaptation is induced by consistently and surreptitiously moving the target during each saccade (when vision is poor), with the result that the oculomotor system compensates as though its saccades had been in error. We propose to explore whether saccade adaptation is best described in engineering terms as a servo mechanism that reduces an error signal, or in behaviorist terms as an example of motor learning driven by reinforcement. We propose to compare two forms of saccade adaptation: (a) that produced conventionally by moving the target during the saccade, and (b) that produced by rewarding subjects for making saccades of a particular magnitude without the target being present after the saccade. Specifically, we propose to see how similar the two forms of adaptation are by comparing the temporal parameters known to affect the efficacy of conventional saccade adaptation (for example, delaying the feedback after each saccade). Furthermore, we propose a test that delineates a crucial difference between a system guided by error and one guided by reinforcement: is adaptation more effective if the subject only receives feedback on trials that land directly on the target versus trials that do not (and therefore present a visual error)? The conventional servo theory would predict that only the latter type of trials would be effective; the reinforcement theory would predict that the former one would be effective. We also propose to look for natural visual features that might act as natural reinforcers for saccades (image contrast, sharpness or duration). Finally, because saccade characteristics are modulated by basal ganglia activity, and reinforcement-guided adaptation would almost certainly involve the basal ganglia, and because patients with Parkinson's Disease have damage in this area and show deficits in other forms of motor learning, we propose to study saccade adaptation in Parkinson's patients, in collaboration with Felice Ghilardi of the CUNY Medical School.