Contiguity and covariation in human causal inference.

Area of Science:

• Cognitive Science
• Psychology
• Causal Inference

Background:

• Causal induction theories typically assume covariation data is readily available.
• Experimental designs often simplify causal induction by using summary statistics or discrete trials.
• This overlooks the critical step of identifying event co-occurrences.

Purpose of the Study:

• To challenge the assumption that covariation data is always readily available in causal induction.
• To highlight the importance of event-parsing in real-world causal learning.
• To review existing and potential approaches to event-parsing within causal induction theories.

Main Methods:

• Review of existing literature on causal induction, associative learning, and causal power theories.
• Analysis of experimental designs in causal induction research.
• Theoretical argumentation regarding the role of event-parsing.

Main Results:

• Current experimental paradigms oversimplify the problem of causal induction.
• The identification of co-occurring cause-effect events is a fundamental aspect of the inductive process.
• Existing theories may need to incorporate event-parsing mechanisms.

Conclusions:

• Causal induction research needs to move beyond simplified assumptions about data availability.
• Event-parsing is a critical, yet often neglected, component of causal learning.
• Future research should explore how associative learning and causal power theories can address event-parsing.