Abstract
There is an ongoing debate on the effects of stimulus history on perception. Recent studies have shown that repeated perceptual decisions to similar stimuli lead to contextual effects, correlated both negatively with the past (negative aftereffects) and positively (serial dependence), sometimes at the same time (Chopin and Mamassian, Current Biology, 2012; Fritsche et al., Current Biology, 2017). However, less attention has been given to understanding how these effects evolve for stimuli further in the past. This task can be challenging because the effect is weak and difficult to dissociate with that of the very recent history. Here, we design a novel psychophysical paradigm to specifically target the influence of stimulus statistics at different points in the past. Observers are presented with oriented Gabor patches from a set number of orientations, randomly interleaved, and are asked to judge whether the orientation of each patch is clockwise or counter-clockwise from a reference orientation. For a specific orientation (targeted orientation), we manipulate the presentation of orientations at a specific point in the past, e.g. K trials in the past, while balancing the orientations shown in the immediate past. We are then able to measure the shift in the perception of the targeted orientation compared to a baseline measurement. By repeating the experiment for different values of K, we map the effect of the stimulus history independently of the immediate history. Our results suggest that stimulus regularities can have an influence on the current percept even when distant in time. Furthermore, due to the careful balancing and the lack of autocorrelations in the statistics of the stimulus, we can show that our findings represent a genuine effect of adaptation to the stimulus statistics and not an artefact of the stimulus pattern.
Meeting abstract presented at VSS 2018