What Is Transferred in Causal Generalization Across Contexts?

Area of Science:

• Cognitive Science
• Psychology
• Causal Inference

Background:

• Causal induction theories, including covariation and causal power accounts, offer different predictions for causal generalization.
• Understanding what information is transferred during causal generalization across contexts is crucial for cognitive models.

Purpose of the Study:

• To test predictions from covariation and causal power accounts of causal induction regarding information transfer in generalization.
• To investigate whether people rely more on observed covariation (ΔP) or causal power when generalizing causal knowledge.

Main Methods:

• Two experiments were conducted using hypothetical scenarios involving interventions.
• Participants evaluated the effect of interventions in between-group and within-group designs.
• Manipulations included ΔP, causal power, and their combination.

Main Results:

• Causal transfer was consistently determined by ΔP, irrespective of causal power.
• The observed transfer patterns supported the ΔP transfer account over other theoretical accounts.
• Causal transfers based on ΔP were found to be non-normative, violating the coherence criterion of the causal power framework.

Conclusions:

• Observed covariation (ΔP) is the primary driver of causal generalization, rather than underlying causal power.
• This suggests that people often employ a non-normative mental measure of causal strength in causal induction.
• Findings challenge normative models and highlight the prevalence of heuristic-based reasoning in causal judgments.