Prior probability and feature predictability interactively bias perceptual decisions.

Area of Science:

• Cognitive Neuroscience
• Decision Science
• Computational Psychiatry

Background:

• Expectations bias sensory perception, enhancing likely stimuli and suppressing alternatives.
• Feature-level predictions in expectation bias sensory processing and evidence accumulation.
• Prior knowledge precision and stimulus feature match influence perceptual biases.

Purpose of the Study:

• Investigate how expectation precision and stimulus match modulate pre- and post-stimulus bias mechanisms.
• Test models of choice bias in a probabilistic face/house discrimination task.
• Examine how feature-level uncertainty and stimulus feature match affect decision-making.

Main Methods:

• Utilized a probabilistic face/house discrimination task.
• Applied the drift-diffusion model to analyze choice bias mechanisms.
• Tested models of starting-point and drift-rate bias.

Main Results:

• Predictive cues biased both starting-point and evidence accumulation rate towards the probable category.
• Prior bias in starting-point depended on feature-level uncertainty of category expectations.
• Evidence accumulation gain increased for less ambiguous features, enhanced by valid expectations.

Conclusions:

• Predictive cues create bias in both initial bias and evidence accumulation rate.
• Prior bias is modulated by feature-level uncertainty, while dynamic bias depends on stimulus feature match.
• Findings bridge psychological decision-making models with predictive coding theories of perception.