The classical view of reasoning holds that people solve problems by applying formal inferential rules such as the syntactic rules of deductive logic. Evidence based on previous work by Nisbett, Cheng, and Holyoak shows that people typically do not reason using such rules; instead, they use what they term pragmatic reasoning schemas, which are clusters of abstract rules organized according to goals and conditions of applicability. The pragmatic schemas studied so far include regulation schemas, such as permissions and obligations, and a qualitative version of the law of large numbers. Teaching such abstract pragmatic schemas improved reasoning, both immediately following training and after a delay of a week or two, whereas teaching formal rules did not improve reasoning, even immediately following training (or, in other studies, improved reasoning immediately, but the improvement decayed rapidly). One plausible explanation is that pragmatic rules are organized according to the conditions under which they apply, and therefore can be applied and practiced once they are learned. In contrast, formal rules do not specify conditions of applicability and therefore are difficult to apply. The first set of experiments will extend the pragmatic schema approach to the evaluation of evidence on cause-and-effect relations. The first experiment will test whether people use pragmatic rather than formal rules in evaluating causal relations. It will also investigate the reasons for common fallacies about causal relations. The second experiment will measure the effectiveness of current graduate training on causal reasoning in psychology and chemistry. Psychology students are exposed to a wider range of causal relations than chemistry students, and therefore should be more accurate in classifying events according to the type of causal relation involved. The third experiment will test whether causal reasoning can be taught effectively by abstract means, as predicted by the pragmatic hypothesis. The second set of experiments will extend the study of the duration of training effects of formal and pragmatic rules. Pragmatic training will include rules on obligations and causal relations. The third set of experiments will examine whether it is the naturalness or the pragmatic nature of the rules that lead to effective training. The last study will evaluate a new course on reasoning based on pragmatic rules. This research will lead to the development of instructions on reasoning more effective than training in formal logic. The new instructions, being theoretically based, will be more systematic as well as more general than the case-based methodnow prevalent in law, medical, and graduate schools. The instructions will cover reasoning on social regulations, explanation of variability based on the law of large numbers, and the evaluation of evidence in accepting or rejecting scientific hypotheses.