Prior information differentially affects discrimination decisions and subjective confidence reports

Area of Science:

• Cognitive Psychology
• Computational Neuroscience
• Decision Science

Background:

• Bayesian models assume decisions and confidence equally integrate prior expectations and incoming information.
• This equal integration of prior information has not been empirically tested.

Purpose of the Study:

• To quantify how prior information influences both decisions and confidence.
• To test the assumption of equal weighting of priors in decision-making versus confidence judgments.

Main Methods:

• Three experiments using a dual-decision task to compare conditions matched for posterior information but differing in prior/likelihood informativeness.
• Utilized exogenous probabilistic cues as priors.
• Employed a Bayesian model to quantify prior weighting parameters.

Main Results:

• Priors were underweighted in discrimination decisions.
• Priors were less underweighted in confidence judgments compared to decisions.
• These patterns persisted with probabilistic cues and were not due to processing time differences.

Conclusions:

• Prior information is utilized more optimally in explicit confidence judgments than in discrimination decisions.
• Findings challenge the assumption of equal prior weighting in standard Bayesian models of decision-making and confidence.