How recent history affects perception: the normative approach and its heuristic approximation

Area of Science:

• Cognitive Neuroscience
• Computational Neuroscience

Background:

• Prior knowledge significantly influences perception.
• The Bayesian framework models how prior knowledge integrates with sensory data.

Purpose of the Study:

• Investigate how perceptual systems acquire and use prior distribution information over short timescales.
• Clarify the role of prior knowledge in perceptual decision-making.

Main Methods:

• A two-tone discrimination task was employed.
• Analysis of trial-by-trial responses and performance with atypical stimulus distributions.

Main Results:

• A 'contraction bias' was observed, consistent with Bayesian priors.
• Recent trials were overweighted, deviating from standard Bayesian models.
• Participants showed deviations from ideal Bayesian detection, suggesting heuristic approximations.

Conclusions:

• The brain utilizes Bayesian-like heuristics for perceptual decision-making.
• These heuristics show limitations in adapting to novel stimulus environments.
• A biologically plausible model involving decaying averages explains observed biases and deviations.