Causal Conditional Reasoning and Conditional Likelihood

We hypothesized that causal conditional reasoning reflects judgment of the conditional likelihood of causes and effects based on a probabilistic causal model of the scenario being judged. Although this proposal has much in common with Cummins’ (1995) theory based on the number of disabling conditions and alternative causes, it takes more variables into account and therefore makes some differing predictions. To test this idea we collected judgments of the causal parameters of the conditionals and used them to derive predictions from a model with zero free parameters. We compared these predictions to Cummins’ acceptability ratings and to analogous likelihood judgments that we also collected. The hypothesis was borne out for Affirming the Consequent and the analogous diagnostic likelihood judgments, where the model obtained close fits to both data sets. However, we found a surprising dissociation between Modus Ponens and judgments of predictive likelihood leading to a relatively poor fit to the Modus Ponens acceptability ratings. We propose an explanation for this in the discussion.