Learning and memory in the human saccadic system

The human saccadic system maintains accurate registration between vision and action in the face of perturbations by adapting motor commands. Such plasticity may emerge through learning mechanisms that continually adapt the system to new sensorimotor transformations. We first tested the ability of the human saccadic system to adapt gaze shifts or saccades to minimize error between the visual target and the saccade endpoint of every saccade, even when such errors are not directionally consistent. We induced visual errors that were not directionally consistent by perturbing visual targets in random directions. Saccades adapted on a trial-by-trial basis despite that the perturbations were random, indicating that the saccadic system may employ a learning mechanism that is rapid, obligatory and unconscious.;We then tested the ability of the saccadic system to maintain accuracy in situations when visual error information is not immediately available and a behaviorally complex response is required. We modified a memory-guided saccade paradigm such that feedback was consistently displaced and delayed with respect to the saccade endpoint. In this situation, accuracy was achieved by making small corrective saccades to the first saccade estimate of the remembered location. Subjects, aware of systematic errors, developed strategies to compensate. This adaptive behavior suggests that the human saccadic system may utilize a learning mechanism dependent on awareness and working memory to strategically guide sensorimotor adaptation.;Finally, we used event-related functional magnetic resonance imaging to identify cortical areas that maintain spatial information to guide future behavior. We compared brain activity as subjects performed in two spatial working memory tasks designed to bias them towards maintaining spatial information via a retrospective code of visuospatial attention or a prospective code of saccade preparation. We found evidence of sustained activity in frontal and parietal regions that was undifferentiated by task type. Even oculomotor regions such as the frontal eye fields showed sustained activity in both tasks, despite that the saccade metrics were known in one condition and not the other. This finding indicates that higher order sensorimotor regions such as the frontal eye fields may maintain a prioritized attentional map of space, rather than the metrics for saccades.