Cognitive biases as Bayesian probability weighting in context.

Area of Science:

• Cognitive Psychology
• Behavioral Economics
• Computational Modeling

Background:

• Human judgments under uncertainty often display systematic biases, including conservatism bias and base-rate neglect.
• Understanding the context-dependent nature of these biases is crucial for cognitive and economic theories.

Purpose of the Study:

• To investigate how task context influences systematic biases in human probability judgments.
• To explore the role of prior probabilities and evidence likelihoods in decision-making under uncertainty.
• To propose and validate a computational model explaining these context-dependent biases.

Main Methods:

• Forty-eight participants completed 12 scenarios requiring subjective probability judgments.
• Scenarios involved integrating prior probabilities with evidence likelihoods in varying contexts (small-world vs. large-world).
• Probabilistic information was presented both as probabilities and relative frequencies.

Main Results:

• Task context significantly mediated the weighting of priors and evidence.
• Small-world scenarios amplified conservatism bias by emphasizing priors.
• Large-world scenarios increased sensitivity to evidence, leading to base-rate neglect; relative frequencies did not mitigate biases.

Conclusions:

• The Adaptive Bayesian Cognition (ABC) model explains cognitive biases as adaptive strategies influenced by context, capacity constraints, and meta-learning.
• Findings bridge cognitive psychology, behavioral economics, and computational modeling for a unified understanding of decision-making under uncertainty.
• The study informs the design of decision support systems by highlighting the adaptive nature of cognitive biases.