Causal explanation improves judgment under uncertainty, but rarely in a Bayesian way

Area of Science:

• Cognitive Psychology
• Decision Science
• Behavioral Economics

Background:

• Judgment under uncertainty often deviates from normative Bayesian principles.
• Understanding the causal origins of statistical information may influence decision-making.
• Previous research suggests potential benefits of causal explanations in improving reasoning.

Purpose of the Study:

• To investigate whether clarifying the causal origin of statistics enhances normative performance in Bayesian problems.
• To examine how causal explanations affect probability estimation and error magnitude.
• To explore the underlying reasoning strategies influenced by causal explanations.

Main Methods:

• Three experiments were conducted, including process-tracing methods in the third experiment.
• Participants solved Bayesian problems with varying statistical explanations.
• Error magnitudes and solution strategies were analyzed.

Main Results:

• Causal explanation did not increase the rate of normative solutions across experiments.
• A reduction in the magnitude of probability estimation errors was observed.
• The error reduction was most significant when statistics were presented in percentage formats.
• Process-tracing revealed that explanations influenced reasoning through increased attention and approximations of Bayes' rule, not solely normative Bayesian updating.

Conclusions:

• Clarifying the causal origin of statistics can alter mental representations of Bayesian problems.
• This alteration does not consistently lead to increased normative responding.
• Cognitive mechanisms, including non-Bayesian strategies, mediate the effects of causal explanations on probability estimation.