Strange but true: Corroboration and base rate neglect.

Area of Science:

• Cognitive Psychology
• Decision Science
• Legal Reasoning

Background:

• Corroborative evidence is typically seen as a strong indicator of a hypothesis's validity.
• The interaction between the prior probability (base rate) of a hypothesis and corroboration remains underexplored.
• Understanding how base rates influence belief updating is crucial for accurate reasoning.

Purpose of the Study:

• To investigate the relationship between a hypothesis's base rate and the support provided by corroborating evidence.
• To examine how lay reasoners handle base rate information when updating beliefs based on testimony.
• To identify potential biases in belief updating, particularly concerning unlikely hypotheses.

Main Methods:

• Development of a Bayesian network model to illustrate the theoretical relationship between base rates and corroboration.
• Conduction of three experiments involving lay reasoners to assess their belief updating in diagnostic and intercausal reasoning tasks.
• Analysis of how participants account for the base rate of hypotheses when evaluating corroborative witness testimonies.

Main Results:

• A Bayesian network model demonstrated an inverse, nonlinear relationship: less likely hypotheses gain more support from each piece of corroborating testimony.
• Experimental findings revealed that participants consistently underestimated the impact of rare base rates on belief updating.
• Lay reasoners exhibited a failure to adequately adjust their beliefs, leading to substantial underestimation of hypothesis support.

Conclusions:

• This study presents a novel demonstration of an inverted base rate neglect, where rare base rates are disproportionately influential.
• The findings have significant implications for scenarios involving testimony where confirmation or disconfirmation is not uniform.
• Accurate belief updating requires a nuanced understanding of how prior probabilities modulate the impact of new evidence.