Logan: Basic Principles of Learning PRINCIPLE: EXPERIMENTAL EXTINCTION If the US fails to occur following the antecedent stimulus in classical conditioning, or if the reinforcer fails to occur following an operant/instrumental response, there will result a decrease in the probability of the recurrence of the learned response in the future. The Principle of Experimental Extinction presumes a prior history of reinforcement that resulted in an increased response probability; the principle states simply that this effect is reversible. Extinction should not be confused with forgetting, which refers to a decrease in response probability as a result of the passage of time. In contrast, extinction requires nonreinforced exposure to the situation, in which case the organism learns that the old contingencies no longer prevail and behavior changes appropriately. The resistance of a response to experimental extinction depends on a number of variables that pertained during acquisition but the most obvious and relevant variable is the number of acquisition trials given before the onset of extinction. As might reasonably be expected, increasing the amount of training increases the level of performance and correspondingly increases the number of responses that will occur during extinction. But as probably would not be expected, very extended training under conditions of continuous (100%) reinforcement actually reduces resistance to extinction. This means that there is no simple, consistent relationship between the level of performance attained during acquisition and the persistence of the response. Although experimental extinction could be interpreted as reflecting the "undoing" of whatever was done during acquisition, many theorists believe that some new process is involved. Specifically, the S-R notion is that an excitatory tendency was acquired during reinforcement, and an opposing inhibitory tendency is acquired during nonreinforcement. A comparable S-S notion is that a positive expectancy was acquired during reinforcement, a negative expectancy is acquired during nonreinforcement. In either case, these tendencies oppose each other such that performance during extinction represents the difference between them. An important finding in this connection is that re-acquisition of an extinguished conditioned response is rapid. This fact not only suggests that extinction does not eradicate the positive associative process, but whatever the effects of extinction may be, they can be counteracted rapidly. The decrease in resistance to extinction resulting from extended training is called the OVERTRAINING EXTINCTION EFFECT and remains something of a theoretical enigma. It clearly indicates that, although performance of a learned response may not be improving in any immediately observable sense, some further learning is still going on. One possibility is that the behavior becomes increasingly rigid, stereotyped, and habitual such that, although performance is smooth and efficient so long as reinforcement is being obtained, the experience of nonreinforcement is especially surprising, frustrating, and generally overwhelming of what had become highly routinized behavior. A major corollary of the Principle of Experimental Extinction is known as SPONTANEOUS RECOVERY. The phenomenon is that an extinguished response will spontaneously recover some of its lost strength simply as a result of the passage of time after extinction. This suggests that whatever process may be involved in extinction, it is not very durable. The Principle of Experimental Extinction is obviously of important adaptive significance. When old responses are no longer reinforced, and past contingencies no longer obtained, behavior should change appropriately. However, in attempting to employ this principle to eliminate undesirable responses it is first necessary to know what reinforcement was involved. For example, if a child's frequent complaints of feeling sick are actually maintained by the reinforcement of extra attention and affection bestowed upon a sick child, it is necessary for the parents to modify their own behavior first so as not to make sickness an attractive condition. In a similar vein, parents who consistently reward their children with praise for each and every little accomplishment may actually reduce a child's later persistence in trying to deal with problems in contexts where no one is there to maintain such a liberal reinforcement schedule. TERMS: Behavior modification, expectancy, extinction, overlearning extinction effect, inhibition, reinforcement (continuous), relearning, spontaneous recovery.