When is a cause the "same"? Incoherent generalization across contexts.

Area of Science:

• Cognitive Psychology
• Causal Inference
• Decision Making

Background:

• Cheng's power (p) model enables prediction of cause effectiveness across different contexts.
• Previous research indicated people detect changes in causal power but not contingency (∆p) when outcome base rates vary.
• Prior methodologies allowed scenario simplification, confounding key causal variables.

Purpose of the Study:

• To investigate if causal induction aligns with Cheng's principles in varied situations.
• To compare a simplified scenario procedure with one preventing simplification.
• To examine sensitivity to changes in P(O|C), p, and ∆p when confounding factors are controlled.

Main Methods:

• Four experiments (1a, 1b, 2a, 2b) were conducted.
• Compared a simplified causal scenario procedure with one that prevented simplification.
• Used contingencies that controlled for the probability of an outcome given a cause (P(O|C)).

Main Results:

• When simplification to a zero base rate was avoided, participants showed sensitivity to variations in P(O|C), p, and ∆p.
• This finding contrasts with previous conclusions that only changes in p are detected.
• Results challenge the notion that people exclusively use a single causal model for generalization.

Conclusions:

• Human causal induction is more nuanced than previously suggested, being sensitive to multiple factors beyond just causal power.
• Methodological choices significantly impact findings in causal inference research.
• Generalization of causal understanding across contexts may involve more than a single normative theory.