Discounting and Conditionalization Dissociable Cognitive Processes in Human Causal Inference

When people are asked to judge the strengths of two potential causes of an effect, they often demonstrate discounting—devaluing the strength of a target cause when it is judged in the presence of a strong (relative to a weak) alternative cause. Devaluing the target cause sometimes results from conditionalization—holding alternative causes constant while evaluating the target cause. Yet discounting not attributable to conditionalization also occurs. We sought to dissociate conditionalization and discounting (beyond that accounted for by conditionalization) by having subjects perform either a spatial or a verbal working memory task while learning a causal relation. Conditionalization was disrupted by the verbal task but not the spatial task; however, discounting was disrupted by the spatial task but not the verbal task. Conditionalization and discounting are therefore cognitively dissociable processes in human causal inference. On a daily basis, humans make causal inferences based on the contingency between events. For instance, to determine whether your new shampoo gives you healthy hair, you compare the state of your hair after using the new shampoo with its state when you were buying the discount brand. Unfortunately, the world rarely provides isolated contingency information; instead, people must make causal inferences given contingency information about multiple potential causes. Is the new shampoo giving you that great shine, or is the new water softener responsible? When determining the cause of an event given information about multiple competing causes, humans both control for alternative causes and discount moderately effective causes in the presence of a highly causal alternative (Goedert & Spellman, 2005). In this article, we demonstrate that these aspects of causal inference rely on dissociable cognitive processes.