Do We "do"?

Area of Science:

• Cognitive Science
• Philosophy of Science
• Causal Inference

Background:

• Causal and counterfactual reasoning are fundamental cognitive processes.
• Distinguishing between observational and interventional reasoning is crucial for accurate causal modeling.
• The do-operator framework provides a formal method for representing interventions.

Purpose of the Study:

• To investigate how individuals reason about causal relationships when interventions are involved.
• To test the psychological implications of the do-operator in counterfactual reasoning.
• To explore the phenomenon of 'undoing' in causal judgments.

Main Methods:

• Six empirical studies were conducted to examine human reasoning.
• Participants were presented with causal (A causes B) and conditional (if A then B) arguments.
• Reasoning patterns were analyzed under conditions of observation versus intervention.

Main Results:

• Evidence supports the 'undoing' effect for causal arguments where interventions render variables independent of their causes.
• The 'undoing' effect was not consistently observed for parallel conditional arguments.
• Intervened-on variables were treated as nondiagnostic, unlike observed variables which were diagnostic.

Conclusions:

• Human counterfactual reasoning aligns with the do-operator framework when dealing with direct causal claims.
• The distinction between intervention and observation is psychologically significant.
• Existing theories lacking this distinction fail to explain the observed reasoning patterns.