A dual-process model of belief and evidence interactions in causal reasoning

In three experiments, we examined how reasoners' preexisting beliefs about causal relations constrained their evaluation of covariation-based empirical evidence. Reasoners were presented with causal candidates that were a priori rated to be either believable or unbelievable, as well as information regarding the degree to which the cause and the effect covaried. Several findings supported the conclusion that preexisting beliefs about causal relations reflect knowledge of both causal mechanisms and covariation relations, that these sources of knowledge are represented independently and contribute independently to causal judgments, and that the evaluation of new empirical evidence is influenced differently by mechanism-based and covariation-based beliefs. Finally, we observed that reasoners were relatively accurate in evaluating the degree to which their judgments were sensitive to empirical evidence but were less able to judge how much their judgments were influenced by their prior beliefs. We present a dual-process model that provides a descriptive account of the boundary conditions for belief and evidence interactions in causal reasoning.