BAPS 2021 - Bayesian modeling of hysteresis and adaptation in visual perception

Abstract

Several studies of serial dependence in visual perception presented evidence for the existence of hysteresis and adaptation effects (e.g., Bosch, Fritsche, Ehinger, & de Lange, 2020). Hysteresis is the attractive effect of a previous percept on the current percept and is assumed to help stabilize the perceptual system. In contrast, adaptation is the repulsive effect of a previous stimulus on the current percept and might be involved in emphasizing relevant and characteristic changes in the stimuli (Snyder et al., 2015). To improve the understanding of hysteresis and adaptation and to look for latent individual differences in the two effects, we attempted to model them. We used the data from the experiment by Schwiedrzik and colleagues (2014), in which they observed hysteresis and adaptation in the perceived orientation of dot lattice stimuli. Additionally, we adapted the code from an efficient coding model (Wei & Stocker, 2015), which takes noisy mapping between stimulus and sensory space into account when computing the likelihood in Bayesian inference of orientation perception. To investigate whether the efficient coding model can predict the hysteresis and adaptation effects in orientation perception of dot lattices, we adjusted it to the experimental design by Schwiedrzik et al. We performed hierarchical Bayesian fitting and model simulations. We will present the results from the model fitting and simulations. This study will provide further insight into the individual differences and the mechanism behind hysteresis and adaptation and hopefully inspire future models of the two effects in various contexts and experimental designs.

Date
May 28, 2021
Location
online