Year: 2004
Type: article
Abstract: A necessary condition for correctly assessing causality is the absence of confounding causes.This set of experiments assesses whether people are sensitive to confounding when they infer causation.Two stories were constructed, one in which two candidate causes and an outcome perfectly covaried (confounded), and one in which one candidate cause occurred independently of a second candidate cause that perfectly covaried with an outcome (unconfounded).If people reason by forming contrast groups that hold alternative causes constant, then in the confounded case, subjects should say that it is impossible to determine causality when two candidate causes are confounded.In the unconfounded case, subjects should be able to say that the candidate is causal.If people are not sensitive to confounding, then they should attribute causality to both candidates in the confounded case, and the results for the target candidate should be the same as those in the unconfounded case.This experiment was conducted with children and adults.Children saw one of the two conditions, while adults saw both conditions.Both children and adults make a distinction between confounded and unconfounded causes when making attributions of causality.Our results show that children are able to state the indeterminacy of confounded causes at an age much earlier than previously documented.
Authors: E. Christina Schofield, Patricia W. Cheng
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
Award: G9000326947
OpenAlex ID: W2585063475