How disconfirmatory evidence shapes confidence in decision-making.

Area of Science:

• Cognitive Neuroscience
• Decision Science
• Psychology

Background:

• Normative models suggest equal weighting of evidence for confidence.
• Recent findings indicate a positive-evidence bias (PEB), where decision-congruent information is overweight.
• Human confidence judgments may deviate from these models.

Purpose of the Study:

• To investigate how choice-confirming and choice-conflicting evidence influence confidence.
• To test whether the brain overweights decision-congruent or incongruent evidence in confidence formation.
• To explain the mechanisms behind confidence computation in decision-making.

Main Methods:

• Re-analysis of nine datasets (N=176) involving color judgment and confidence ratings.
• Fitting an extended log-posterior-ratio for confidence model.
• A preregistered experiment (N=32) manipulating category boundaries.

Main Results:

• Participants overweight choice-conflicting evidence in confidence judgments, contradicting the normative model and PEB.
• A robust averaging principle, similar to decision-making, explains these effects.
• Incongruent evidence is weighted more heavily due to proximity to the category boundary.
• Experimentally shifting the category boundary altered confidence computation.

Conclusions:

• Confidence is influenced by dis-confirmatory evidence.
• This effect arises from downstream consequences of decision-making mechanisms.
• Confidence computation is sensitive to the relationship between evidence and category boundaries.