Transitive reasoning distorts induction in causal chains.

Area of Science:

• Cognitive psychology
• Causal inference
• Probabilistic reasoning

Background:

• Causal chains (A→B→C) are often assumed to be transitive.
• Transitivity in probabilistic causal relations requires the Markov condition to hold.
• Violations of the Markov condition can lead to intransitive probabilistic relationships.

Purpose of the Study:

• To investigate how people make probabilistic judgments about indirect relationships (A→C) in causal chains violating the Markov condition.
• To test the hypothesis that people default to transitive inferences based on the Markov condition, even with counterevidence.
• To examine the interplay between mental causal chain models and observed data in probabilistic judgments.

Main Methods:

• Two experiments were conducted involving probabilistic causal judgments.
• Participants were presented with data for A→B and B→C, while A and C were shown to be statistically independent.
• This setup created a violation of the Markov condition, demonstrating intransitivity.

Main Results:

• Transitive reasoning, mediated by event B, significantly influenced and distorted judgments about the indirect relationship between A and C.
• Participants' judgments were affected by an interaction between transitive, causal-model-based inferences and the presented data.
• Observed data indicating statistical independence between A and C did not prevent participants from inferring a transitive relationship.

Conclusions:

• People tend to construct mental causal chains that adhere to the Markov condition, facilitating transitive reasoning.
• This tendency persists even when empirical data contradicts the warranted transitive inferences.
• Cognitive biases in causal reasoning can lead individuals to prioritize intuitive transitive models over statistical evidence.