A consensus seems to be emerging: What animals learn in instrumental situations is the relationship between their behavior and its programmed consequences. Thus, if a biologically significant event, S*, such as food or shock, is programmed to follow some response, R, such as pressing a bar, we can attribute changes in the animal’s behavior specifically to a learned R-S* association: The animal learns that bar pressing is followed by food or that bar pressing is followed by shock. This view of instrumental learning has been advocated by such diverse theoretical writers as Irwin (1971), Mackintosh (1974), and Seligman and Johnston (1973). It is an attractive view. It is compatible with a cognitive orientation, and it seems to be consistent with a variety of experimental findings.