Causal Induction and Category Formation
Dissertation, University of California, Los Angeles (
1993)
Copy
BIBTEX
Abstract
An interaction between the processes of causal induction and categorization is hypothesized. On one hand, by assigning greater weight to causally-relevant features with respect to an effect, causal induction shapes the categories formed. On the other hand, based on the principle of coherence, categorical knowledge helps to determine whether or not a regularity is causal. Coherence is operationalized in terms of consistency across regularities at different levels of abstraction. Coherent regularities are assumed to be judged more causal than incoherent ones. ;The first part of the hypothesis provides a plausible answer to a problem with the similarity-based approach to categorization, namely, how to determine what counts as a feature and its relative importance in a similarity comparison. The second part of the hypothesis provides a criterion for distinguishing genuine causes from accidental regularities, which is a problem for an inductive approach to causal inference as proposed by early versions of the covariational view. ;Four experiments were conducted as the empirical basis of support for the hypothesis. The first experiment was designed to test the role of coherence in causal judgments involving a single candidate cause, whereas the second experiment tested its role in cases involving two competitive candidates. The third experiment showed that subjects formed categories defined by causally-relevant features with respect to a given effect when they actively searched for the cause of the effect. In contrast, in a non-causal context, people were less likely to derive such categories. The last experiment showed that people not only induce categories with respect to the effects of concern, but also define members' goodness as examples of the categories according to how potently each member produces the effect