Abstract
Microsaccade characteristics are associated with covert shifts of attention and response preparation, and therefore may carry fine directional information about upcoming errors. We tested whether microsaccades reflect angular errors in a localization task, after target appearance, but prior to manual response initiation. While maintaining central fixation, observers reported the locations of static noise patches within a Gaussian contrast envelope, at one of 12 angular locations at 7º eccentricity. Targets were presented for 50 ms, and following a 500 ms delay period, observers reported the target’s location by manually adjusting a cursor constrained at the same eccentricity as the targets. Microsaccade rate decreased near the time of stimulus onset and increased after approximately 200 ms. Microsaccades were analyzed from the delay period, 200-500 ms following stimulus offset, prior to initiation of observers’ manual response. On each trial, observers’ response errors were recorded as the angular difference between the physical location of the patch and the cursor. For targets that were within 90º of the microsaccade direction, errors in observers’ future responses were correlated with the angular difference between the microsaccade and the target location (r = 0.31; p < .01). Observers’ angular response errors were, on average, 1.14º clockwise (0.14 d.v.a) when microsaccades were clockwise relative to the target location, and 2.94º counterclockwise (0.36 d.v.a.) when microsaccades were counterclockwise relative to target location. Permutation tests indicated that these effects were trial-specific, and not an artifact of systematic angular biases in observers’ location reports across all trials. These results demonstrate that microsaccades generated between stimulus offset and non-ocular response initiation carry information about angular errors in another domain before they occur, and reveal errors related to peripheral action preparation on a more granular scale than previously shown.