Normative accounts of illusory correlations

Area of Science:

• Cognitive Psychology
• Decision Making
• Behavioral Economics

Background:

• Humans often exhibit biases when inferring joint occurrences of variables.
• Illusory correlations, perceiving non-existent or incorrect associations, are a key example of these biases.
• Previous explanations include selective processing, pseudocontingency heuristics, and a normative account using Laplace's Rule of Succession.

Purpose of the Study:

• To critically evaluate the empirical and theoretical limitations of the normative account of illusory correlations based on Laplace's Rule of Succession.
• To propose and validate an alternative normative account for illusory correlations.
• To demonstrate that the proposed model can capture observed patterns in published studies.

Main Methods:

• Critique of existing normative models for illusory correlations.
• Development of a novel normative account based on Bayesian reasoning and marginal frequencies.
• Empirical validation against a corpus of published studies on illusory correlations.

Main Results:

• The study identifies significant limitations in the normative account using Laplace's Rule of Succession.
• The proposed Bayesian model, relying on marginal frequencies, successfully accounts for the qualitative patterns observed in illusory correlation studies.
• The new model provides a more robust normative explanation for these cognitive biases.

Conclusions:

• The normative account of illusory correlations based on Laplace's Rule of Succession is insufficient and should be dismissed.
• A Bayesian framework utilizing marginal frequencies offers a compelling normative explanation for illusory correlations.
• This research reframes illusory correlations not as errors but as expected outcomes of rational Bayesian inference under specific conditions.