ABSTRACT
A central proposal of the probabilistic paradigm in the psychology of reasoning is that most reasoning, both in everyday life and in science, is from premises that are not certain but are only held with varying degrees of belief. If we retain the classical logical definition of deduction as preservation of truth, or of certainty, then it follows that most real-world reasoning is not deductive. However, the question of what role deduction plays in reasoning arises anew when central deductive concepts are generalized to cover uncertain degrees of belief. Specifically, binary consistency can be generalized to coherence, and binary validity to probabilistic validity (p-validity for short). This chapter introduces the concepts of coherence and p-validity and outlines the complementary relation between them. It then surveys empirical evidence for them, and describes implications of this probabilistic generalization of deduction for dual-process theories of reasoning. The chapter concludes with a discussion of limitations in the scope of probabilistic deduction as a criterion for argument quality, and in the study of dynamic reasoning.
