A Causal-Power Theory of Focal Sets

This chapter reviews the probabilistic contrast model and its notion of focal sets, as well as the power PC explanation of why people use certain specific types of focal sets to evaluate causal relations. The power analysis generates a variety of testable predictions about people's causal judgments. It describes and discusses some experimental tests that differentiate the predictions of the power PC theory from those of the major competing account, the Rescorla-Wagner model. The latter model, originally proposed as a model of classical conditioning in animals, has more recently been applied to human causal induction. The central difference between the two approaches involves the distinction between the computational and algorithmic levels of cognitive analysis. It discusses that the issues of what function is being computed by an information process and why it is computed logically precede the issue of how a given function is computed. The Rescorla-Wagner model is founded on an algorithm for discrepancy reduction on a trial-by-trial basis. The chapter describes describe some recent findings from our laboratory that test discriminating predictions of the power PC theory and the Rescorla-Wagner model, using causal analogues of two well-known paradigms that have been used in studies of animal conditioning: overexpectation and extinction of conditioned inhibition. It also discusses other phenomena that discriminate between these two theories.