Extinction is the process of eliminating (extinguishing) a conditioned response (CR) by repeatedly presenting the CS while withholding reinforcement. It presents interesting paradoxes. One, which both Pavlov (1928) and Hull (1943) discussed at length, is that the learning that takes place during extinction is caused by what Dickinson (1989) has called a “no-US,” an event that ought to have occurred but did not. The problem is that a no-US has no physical attributes-no intensity, no qualia, no locus in space, and, perhaps most important from a theoretical standpoint, no locus in time. In a materialist and neurobiologically oriented endeavor, there is no room for nonphysical causes. Contemporary models of conditioning, such as the Rescorla-Wagner (1972) model deal with this problem, but only through the artifice of dividing the continuous flow of time into discrete trials. The notion of a trial is itself deeply problematic, for reasons we elaborate in a later chapter.