Confidence reports in decision-making with multiple alternatives violate the Bayesian confidence hypothesis

Area of Science:

• Cognitive Science
• Neuroscience
• Psychology

Background:

• Decision confidence is crucial for evaluating choices and guiding behavior.
• Previous studies on confidence primarily used two-alternative tasks.
• Existing theories link confidence to the posterior probability of the chosen option.

Purpose of the Study:

• To investigate an alternative theory of decision confidence.
• To test if confidence depends on the difference between the best and second-best options.
• To examine confidence in a three-alternative visual categorization task.

Main Methods:

• Participants performed a three-alternative visual categorization task.
• Confidence reports were collected.
• Computational models were used to analyze confidence data.

Main Results:

• Confidence was best explained by the difference in posterior probabilities between the best and second-best options.
• Confidence was not solely determined by the posterior probability of the chosen option.
• Overall uncertainty (entropy) did not adequately explain confidence reports.

Conclusions:

• The study challenges the prevailing theory of decision confidence.
• Confidence may reflect the subjective probability of having made the optimal decision.
• Decision confidence is influenced by the relative difference between competing options.