Stereotypes as Bayesian prediction of social groups.

Area of Science:

• Social psychology
• Cognitive science
• Bayesian modeling

Background:

• Stereotypes are generalizations used for probabilistic predictions about individuals.
• The cognitive basis of stereotypes, particularly their alignment with probabilistic reasoning, remains an area of investigation.

Purpose of the Study:

• To investigate whether social stereotypes can be understood as conditional probabilities following Bayesian principles.
• To test if human judgments of social categories align with Bayes' rule predictions.

Main Methods:

• Replication and extension of previous research on stereotypes and Bayesian reasoning.
• Conducted three studies with a total of 2,652 participants.
• Analyzed participants' direct posterior predictions for social categories and traits.

Main Results:

• People's judgments of social categories demonstrated Bayesian alignment, matching predictions from Bayes' rule.
• Traits with high diagnostic ratios effectively distinguished between stereotypic and non-stereotypic attributes across social categories.
• Explored the influence of cognitive ability, political orientation, and motivated cognition on stereotyping.

Conclusions:

• Social stereotypes appear to function as conditional probabilities consistent with Bayesian inference.
• The diagnostic ratio of traits is a key factor in distinguishing stereotypic from non-stereotypic representations.
• Further research should examine cognitive and motivational factors influencing these Bayesian stereotype representations.