Integrating information about mechanism and covariation in casual reasoning

Causal reasoning is an important and complex process, in which individuals have multiple sources of information available to inform their judgments. An enduring tension exists between what cues to causality people use and focus on in acquiring causal knowledge and making causal judgments. Much research on causal reasoning has focused on how people use information about covariation in this process. More recently, research has shown that people seek and use information about causal mechanisms to inform their causal inferences. Consequently, an important theoretical question is how people combine knowledge about causal mechanisms, that is, how a candidate cause works to bring about or produce a given effect, with information about covariation, which is the empirical relation between these two variables.Very little research has investigated how these two sources of information are integrated in determining people's causal judgments. Two general models of how these sources of input are combined currently exist: covariation and mechanism precedence models. Both these models account for people's causal judgments on the basis of the primacy of one source of information over the other.The research presented in this thesis investigates several variables that are hypothesized to be key in the integration of covariation and mechanism information. It is hypothesized that the scope of the covariation information available, as well as the strength of the covariation present between possible cause and effect are important dimensions of covariation input. It is also hypothesized that the nature and structure of mechanism information available to the reasoner is an important variable influencing the integration, specifically the detailedness of the representation of mechanism information. In four experiments the effects of these different variables on judgments of causality were assessed in combination. Overall it was found that how information about covariation strength is used depends on the detailedness of mechanism information and the scope over which covariation information is defined. The results indicate that one source of information does not have primacy over the other. Thus, an interactive model of how these sources of input are integrated is proposed