The effect of sequence stability on serial dependence

Perception and decision-making are strongly influenced by previously presented information. Research on serial dependence has revealed significant attractive biases toward previous stimuli, suggesting that integrating information from both past and present stimuli is beneficial when representing information about the environment. Strong serial dependence is mainly found when consecutive stimuli are similar providing evidence that serial dependence supports continuity. Small changes in the stimulus are not perceived as meaningful stimulus changes and should, hence, be integrated. In the current study we examined the role of continuity further and presented sequences of Gabors that were similar in their orientation, but their spatial frequency was either constant or random. This created stable and unstable environments. We consistently found serial dependence. Interestingly, serial dependence was stronger when the spatial frequency was random compared to constant. In stable contexts, predictive coding mechanisms might sharpen neural representations of the expected input, improve discrimination, and reduce the need to integrate past information. Additionally, a form of optimal integration was observed when the uncertainty of the final Gabor was low. Using similar orientations across longer sequences seemed to induce a form of optimal integration of past and present information, potentially creating the continuous perception of a single object when the current trial's uncertainty was low.