INPC 2003: Cheng Abstract

A single causal factor often contributes towards producing an effect, yet is insufficient to produce it on its own. From a set of observations, how should one determine whether multiple factors interact to produce or prevent an effect, rather than influence it independently? This paper will present a theory (Novick and Cheng, 2003) of how people intuitively evaluate interactive causal influence. In its assessment procedure, the theory explicitly represents possible causal relations, the desired unknowns, by variables representing their probability of influence (Cartwright, 1989). The theory posits that people interpret observations in terms of such unobservable causal relations. It explains psychological phenomena regarding the discovery of conjunctive causal relations (i.e., relations involving interactive causal influence) that are inexplicable by a "purely" observational approach, one that does not explicitly represent causal relations. An implication of the explicit representation of causal relations for null-hypothesis testing will also be discussed.