Poster Session: Bayesian inference and adaptation in neural responses

• DOI: 10.1167/jov.23.11.60
• Publication Date: 2023-09-21T16:33:08Z
• Authors: Jake Manalansan; Kara Emery; Michael Rudd; Michael Webster
• Keywords: Stimulus (psychology)

Many of the perceptual changes induced by adaptation can be captured in a Bayesian framework, which renormalizes neural responses to weak or strong stimuli, consistent with correcting likelihood match prior (Emery and Webster VSS 2020). We explored what this model implies about estimation at level an individual neuron channel. The corresponds expects see, but causes it expects. If decoder does not "know" neuron's state (Series et al 2009), these adjustments effectively flatten priors for dimensions adapts to, preserving expectations only information that adapt such as how signals deviate from predicted value. Bayesian-like estimates deviations computed within because shape is instantiated contrast response function (CRF). For example, Gaussian corresponding CRF cumulative Gaussian. Steeper CRFs bias more toward mean. biases approximate do equal posterior. If, like adaptation, decoding account approximation, then interpreted stimulus value distorted relative Bayes estimate. This may underlie distortions evident some dimensions, nonlinearities perceived contrast.