Abstract
Background: Perception is flexible, and influenced by prior experience. However, the mechanisms underlying the effects of experience on subsequent perception are yet unclear. Here we investigated the influence of behavioral relevance of the prior experience on biasing subsequent perception, and examined the extent to which these influences are mediated by high-level decisions vs. low-level sensory or motor involvement. Method: A series of visual stimuli (uniformly filled circles) were presented on a screen. Participants judged whether each "target" circle was to the left or right of the screen's center (2AFC via button press). Each target circle was preceded by several "prior" circles with randomly distributed locations, which were either (on average) centered or biased to the left/right of the screen center. To dissociate behavioral relevance and motor responses, different task conditions were applied: i) prior circles' task was the same as for the target circles, ii) motor responses were inhibited for prior circles, iii) prior circle judgments were reported via different buttons, or prior circle color (and not location) was judged, such that (iv) motor responses alone were biased, (v) or stimulus locations were biased, but not explicitly judged (behaviorally irrelevant). Subjective perceived center was assessed for each condition. Results: Subjective perceived center shifted after relevant judgments, but not when motor responses alone were biased, or when stimulus locations were biased without being explicitly judged. Importantly, perceptual bias (for relevant judgments) ensued even when motor responses were dissociated. Conclusion: Behavioral relevance has a key role in biasing future perception by prior experience; low-level sensory bias is not always enough to influence subsequent perception. However, if that same information is behaviorally relevant, it would lead to a strong perceptual bias. This effect is not mediated by motor repetition, and occurs without external feedback. Relationship to attention, top-down and bottom-up processing is discussed.
Meeting abstract presented at VSS 2017