Ineffective cues for contextual saccade adaptation.

Abstract

Contextual saccadic adaptation is investigated through a variant of the double-step paradigm, where two directions of intra-saccadic steps are signaled by two cues. This enables the simultaneous induction of two distinct saccadic adaptations. Surprisingly, contextual adaptation is effective only with motor-related cues, whereas visual cues such as target color and shape do not elicit significant adaptation. We tested nine different contextual cues to signal intra-saccadic steps in a contextual double-step paradigm: visual stimulus duration, lateralization of a sound, various statistical regularities across trials, symbolic cues, starting location of the target, as well as the amplitude of the first step or the target color and shape. Robust systematic contextual learning was found under the amplitude and the starting location experiments, while no learning occurred with any other cues. This lack of contextual learning further confirms that the prediction of the intra-saccadic steps depends on the nature of the context. In two additional experiments replicating those using target color and shape, as well as symbolic cues, participants were periodically prompted to explicitly report the contextual cue they had just experienced. Again, no systematic contextual adaptation was observed despite participants achieving reporting the contextual cue accurately. This dissociation between perceptual reports and motor tasks involving the same visual information aligns with previous results on the constraints for contextual learning. The saccadic system, evolutionarily specialized for spatial targeting, exhibits selective learning that prioritizes localization cues, effectively ignoring non-motor cues in its learning processes, even when such cues are explicitly perceived.