Abstract
When observers make a saccade to a visual target, the endpoint is often non-foveal. How does the observer know that the non-foveal endpoint is due to oculomotor noise and not to real-object movement? The visual system needs to attribute discrepancies between fovea and target to an internal source (oculomotor noise) or an external source (real-world movement). Previous studies suggest that the visual system knows saccade amplitude (via, e.g., an extraretinal signal), and can thus predict non-foveal endpoints and discriminate between predicted and non-predicted discrepancies (which should then be attributed to the real world). However, previous studies have also shown that non-predicted discrepancies (such as those introduced by surreptitiously stepping the target back during saccade execution) can be attributed to an internal source rather than an external source provided the discrepancy is not too large. Systematic non-predicted internal discrepancies cause the saccadic system to adapt its amplitude to eliminate future errors. We examined how the characteristics of these non-predicted errors influence saccadic adaptation. We had 3 hypotheses: (1) If the saccadic system predicts large and variable discrepancies, then small non-predicted discrepancies are more likely to be attributed to an internal source and to cause adaptation. Therefore, saccades of large amplitude should adapt more than saccades of small amplitude; (2) If non-predicted discrepancies are inconsistent, then they should more likely be attributed to an external source. Therefore, the more consistent the target step across a session, the more adaptation there should be; (3) If the non-predicted discrepancies are too large, they should be attributed to the outside world. Thus, the smaller the average back-step, the less adaptation there should be. The results support these three hypotheses, suggesting that the saccadic system does indeed attribute non-predicted errors to an internal or external source based on the reliability of each source.
Meeting abstract presented at VSS 2012