Choosing optimal causal backgrounds for causal discovery.

Area of Science:

• Cognitive Psychology
• Causal Inference

Background:

• Understanding how humans discover causal relationships is crucial for cognitive science.
• Inferential approaches suggest optimal context selection for efficient causal discovery.

Purpose of the Study:

• To investigate human strategies in identifying generative and preventive causal relationships.
• To test whether participants select informative contexts aligned with inferential theories of causal discovery.

Main Methods:

• Two experiments involved participants testing unknown target causes in various contexts with differing base rates of effects.
• Participants chose contexts to evaluate generative and preventive causes, with probabilistic and deterministic relationships employed.

Main Results:

• Participants generally preferred testing generative causes in low base rate contexts and preventive causes in high base rate contexts, aligning with inferential predictions.
• A significant minority of participants continued to select less informative contexts even after the causal power was evident.

Conclusions:

• While inferential strategies guide initial causal discovery, persistent use of suboptimal contexts suggests other factors, like the matching law from operant conditioning, may also play a role.
• Human causal discovery involves both inferential reasoning and potentially habit-based or reinforcement-driven behaviors.