Task-relevance is causal in eye movement learning and adaptation

Abstract When a saccadic eye movement does not land accurately on its visual target, subsequent saccades to the same target are subject to a corrective adjustment, which has been called saccade adaptation. Saccade adaptation has emerged as a go-to model for sensorimotor learning. Because observers show limited awareness of image manipulations during saccades, adaptive changes in saccade amplitude have long been thought to rely on the passive processing of visual error signals. However, it turns out that task-relevance has a modulatory effect on adaptation and that it can even be a sufficient cause for adaptation. Indeed, adaptation can be driven by a shift in task-relevant information even in the absence of a bottom-up visual error. This task-driven adaptation shares similar characteristics to bottom-up adaptation, that is adaptation triggered by a displacement of the eye-movement target. The effect of task-relevance is consistent with an integrated view of the saccadic system, where bottom-up and top-down signals converge to define the saccade target and the orienting of attention. We point to possible neural substrates of top-down adaptation, which largely remains to be elucidated in contrast to the detailed experimental and modeling work linking the cerebellum to bottom-up adaptation.