Evaluations of Causal and Conditional Hypotheses

Abstract

This study examined evaluations of causal and conditional hyptheses on the basis of individual exemplars. Subjects were presented scenes describing something that is malfunctioning and is taken to an expert. The expert makes a causal diagnosis and a prediction concerning an outcome following treatment. Four types of outcome evidence were provided: (1) treatment is provided and the problem is eliminated; (2) treatment is not provided and the problem is eliminated; (3) treatment is provided and the problem is not eliminated; and (4) treatment is not provided and the problem is not eliminated. Subjects were required to judge whether individual exemplars prove the diagnosis, disprove the diagnosis, or fail to test the diagnosis. A second form of problem presented the same scenes but required evaluation of the conditional prediction. A third form of problem did not present the causal scenes; the subject evaluated the conditional predication and no context. Three content domains were used: medical, mechanical, and arbitrary. There were three general results: (1) diagnoses were judged as less open to testing than predictive conditional hypotheses; (2) medical problems were judged differently from mechanical problems; and (3) problems lacking meaningful content or contexts led to more primitive response strategies.