Causality in thought.

Area of Science:

• Cognitive Science
• Psychology
• Artificial Intelligence

Background:

• Causal knowledge is fundamental to human cognition.
• The precise nature of causal representation and inference is debated.
• Existing models often focus on probabilistic dependencies.

Purpose of the Study:

• To explore whether human causal inference is solely based on probabilistic dependencies.
• To investigate richer forms of causal representation.
• To review research across reasoning, decision-making, judgment, and attribution.

Main Methods:

• Review of existing research in human reasoning and decision-making.
• Evaluation of causal Bayesian networks as a normative framework.
• Analysis of experimental work identifying key features of causal reasoning.

Main Results:

• Causal Bayesian networks provide a useful normative framework but are incomplete for explaining causal thinking.
• Three key hallmarks of causal reasoning identified: mechanism, narrative, and mental simulation.
• These hallmarks extend beyond probabilistic knowledge.

Conclusions:

• Human causal reasoning incorporates richer representations than mere probabilistic dependencies.
• Mechanism, narrative, and mental simulation are crucial components of causal thought.
• Mental simulations represent mechanisms over time and aggregate into narratives in multi-actor scenarios.