Considering Psychological Mechanisms Can Change the Interpretation of Bayesian Models

Area of Science:

• Cognitive Science
• Psychology
• Computational Neuroscience

Background:

• Bayesian models of cognition offer a powerful framework for understanding human reasoning.
• A key challenge lies in bridging the gap between abstract Bayesian assumptions and underlying psychological mechanisms.
• Understanding the cognitive processes involved in inductive inference is crucial.

Purpose of the Study:

• To investigate the psychological mechanisms of hypothesis generation and evaluation in inductive inference.
• To examine whether hypothesis generation and evaluation are separable cognitive processes.
• To test the predictions of a model that distinguishes between generating and evaluating hypotheses.

Main Methods:

• Developed a computational model separating hypothesis generation and evaluation.
• Conducted three experiments manipulating the ease of hypothesis generation and hypothesis plausibility.
• Analyzed behavioral data to assess the distinct effects of generation ease and plausibility.

Main Results:

• Demonstrated that hypothesis generation and evaluation are separable cognitive processes.
• Showed that manipulating the ease of hypothesis generation has distinct effects compared to manipulating hypothesis plausibility.
• Found that people's behavior is not fully explained by Bayesian models assuming fixed prior probabilities.

Conclusions:

• The ease of hypothesis generation and a hypothesis's plausibility are distinct factors in inductive reasoning.
• Psychological mechanisms, such as hypothesis generation, must be considered to fully understand Bayesian inference in cognition.
• Results challenge standard Bayesian models by highlighting the importance of separable generative and evaluative processes.