A rational process model of reasoning causally with continuous variables.

Area of Science:

• Cognitive Science
• Psychology
• Causal Inference

Background:

• Humans are effective causal reasoners but prone to systematic errors.
• The mutation sampler model explains this by positing rational inference limited by cognitive resources.
• Previous tests focused on binary variables and generative causal relations.

Purpose of the Study:

• To empirically test the mutation sampler model with continuous variables.
• To investigate causal inference in common cause networks with mixed generative and inhibitory relations.
• To determine if human errors in causal inference extend to continuous variables.

Main Methods:

• Empirical testing of causal inference with continuous variables.
• Utilizing a common cause network structure.
• Employing a mixture of generative and inhibitory causal relations.
• Applying a novel version of the mutation sampler model for continuous variables.

Main Results:

• People commit similar qualitative errors in causal inference with continuous variables as with binary ones.
• A new version of the mutation sampler accurately explains these errors.
• The mutation sampler's sampling process quantitatively impacts all causal inferences, not just erroneous ones.

Conclusions:

• The mutation sampler model provides a robust explanation for human causal inference errors across variable types.
• Cognitive resource limitations significantly influence causal reasoning with continuous variables.
• The model's quantitative effects highlight the pervasive impact of sampling processes on inference.