When an error is injected to saccade endpoints by displacing the target midflight during saccades, observers typically adjust their saccade amplitudes on later trials to reduce landing error. Since target displacements are much harder to see during a saccade than during fixation (termed a??saccadic suppression of displacementa??), it is often assumed that observers are unaware of the manipulation for typical displacement amplitudes. Different conceptions of saccade adaptation predict different effects of target visibility on learning rates. One states that when displacements are less likely to be seen the error is attributed to the motor system instead of the external world and learning should be faster. Another one gives no role to visual error attribution itself, but predicts that learning rates are a function of the uncertainty of the visual error and the uncertainty in the visuomotor mapping. In the latter case learning rates should increase with the visibility of the target. We tested the effect of target visibility on learning rates by measuring saccade adaptation towards targets of different contrasts (10 and 100%). The target was 12 degrees of visual angle (dva) from the fixation point and was displaced 1.5 dva backward during the adaptation phase. On the same session we measured psychometric functions for discriminating between forward and backward target displacements. Learning, indexed by the time-constant of exponential fits, was slower for observers who had higher perceptual thresholds, favoring the signal uncertainty account.
展开▼
