Abstract
Purpose. When a target is surreptitiously displaced forward or backward during a saccade the saccadic amplitude progressively changes. This saccadic adaptation is often viewed as being driven by the mismatch between a predicted and an actual post-saccadic visual position error. However, we recently demonstrated that saccadic adaptation may be induced by reinforcement learning in the absence of a visual position error (Madelain et al, J.Neurophysiology, 2011-106) suggesting that normal saccade adaptation might involve general learning mechanisms rather than only specialized motor calibration mechanisms. We now ask whether changes in saccade amplitude may be induced by the ability to perform a visual discrimination task. We designed a gaze contingent paradigm in which the difficulty of a post-saccadic visual discrimination task depends on the saccadic amplitude. Methods. Subjects were instructed to make a saccade to a peripheral mask stimulus. Immediately after completion of the saccade, a target briefly (20ms) replaced the mask at the saccade goal location. Subjects were required to perform a four-alternative forced-choice discrimination about target identity. In a first condition the luminance of the discrimination stimulus was either high (50% of the trials, discrimination performance well above chance) or low (50% of the trials, discrimination performance close to chance). In a second condition, the discrimination stimulus luminance depended on the saccade gain: if the gain of the current saccade was lower than the median gain, luminance was high; if the saccadic gain was higher than the median gain, luminance was low. Importantly, the saccade target position was unchanged across trials. Results. Saccadic amplitude was reduced by 7.06% on average. In some subjects the amount of gain reduction was higher than 15%. Conclusions. Importantly, these changes in amplitude increased the post-saccadic position error. These results demonstrate that saccadic gain may be affected by the ability to perform a discrimination task.
Meeting abstract presented at VSS 2012