Abstract
Saccades are rapid eye movements that orient the central, high-acuity part of the retina toward objects of interest to enable their detailed visual analysis. The accuracy of saccadic eye movements is crucial for performance in visual tasks, and studies of saccadic adaptation (McLaughlin, 1967) have suggested that the oculomotor system is continually calibrating the link between motor commands and visual location to keep accuracy high. Probabilistic models of motor learning predict that the rate of adaptation should depend on the reliability by which oculomotor errors are measured; that is, as the uncertainty about the position of the target increases, the fraction of saccadic landing error that is used to calibrate the motor command should become progressively smaller, resulting in slower motor learning. While these models capture many aspects of motor learning in other types of movements, such as hand reaching (e.g. Burge, Ernst & Banks, 2008), one previous study has concluded that they fail to describe learning in saccadic eye movements (Souto, Gegenfurtner & Schütz, 2016). In that study, however, perceptual judgments in the high and low uncertainty conditions had similar precision, suggesting that the manipulation of uncertainty may not have been successful. Here we re-examined this conclusion using stimulus settings that have been recently shown to increase positional uncertainty (Lisi, Solomon & Morgan, 2019). Observers completed a backward saccadic adaptation task in a condition of high uncertainty (with stimulus adjusted to individual sensitivities in a pre-test session) and in a low uncertainty condition, run in different sessions on separate days. However, we found no difference in the saccadic adaptation rate between high and low uncertainty conditions. Overall, these results replicates previous work and suggest that saccadic adaptation, differently from the adaptation of other types of movements, is not well described by probabilistic models of motor learning.