Extinction 102
The first commentary in this series of three on extinction addressed some basic observations about extinction. As with many things in life, extinction is a little more complicated than it may appear to be at first blush. Extinction 102 considers some of these complexities and their implications for using extinction in managing behavior.
1. Extinguishing behavior takes time. Other things being constant, the reinforcement schedule maintaining the response prior to extinction governs both the speed and pattern of responding as response reduction proceeds.
The lesson: Don’t expect extinction to always proceed in the same way and at the same rate. If you want to optimize extinction, arrange a schedule of reinforcement just prior to extinction that yields the most rapid extinction (e.g., reinforce every response). 
2. Sometimes, but not always, the initial removal of reinforcement may increase responding before it reduces it. Research has yet to isolate the conditions under which extinction bursts occur. In any case, these bursts are temporary, although they might seem to go on forever when extinction is first implemented.
The lesson: Don’t be surprised when extinction bursts do or don’t happen.
3. There are circumstances under which eliminated behavior will recur. When an extinction period is interrupted, for example, transient responding is to be expected when the extinction period is reinstated. Such spontaneous recovery dissipates, however, across interruptions. Furthermore, if the stimulus conditions in effect during extinction are changed, either by introducing a new stimulus or by changing to one different from that in effect as the response was extinguished, the response is likely to recur, or, in the technical language of learning, renew.
The lesson: expect occasional reversals of the downward progress of the extinguishing response, and not to worry because the overall trend will be down. Furthermore, on the one hand if you want to maximize extinction, don’t change the stimulus conditions. But, on the other hand, if you want broad extinction under a variety of stimulus conditions, change the stimuli from time to time as extinction progresses.
4. When a target response is extinguished, other, different, responses may resurge. If these other responses are not reinforced they too will dissipate with time. If they are reinforced, as in a differential-reinforcement-of-alternative behavior, or DRA, procedure, the elimination of the target response will be hastened.
The lesson: Be on the watch for the recurrence of previously reinforced responding during extinction, because some of that recurring behavior may be less appropriate than the inappropriate behavior being extinguished.
5. One of the biggest challenges facing people trying to extinguish behavior is to consistently not reinforce the undesirable behavior. Extinguished behavior is easily reinstated when the target response is inadvertently reinforced, but it also can be equally easily reinstated, at least temporarily, by delivering the reinforcer independently of responding. The most successful extinction programs occur when the person doing the extinguishing “sticks to her guns” and doesn’t inadvertently reinforces the behavior being extinguished. Not always an easy thing to do, especially if you have a two-year-old having a full-blown tantrum in front of you, as happened to the hapless Mom whose daughter did just that not too long ago in front of the President of the United States!
The lesson: Be consistent!
In recent years, we have witnessed (dare I say it?) a resurgence of research related to extinction and its byproducts. But still much remains unknown. For example, even something seemingly as well establish and general as the extinction burst on further inspection remains controversial as a general phenomenon associated with the onset of extinction. Although we are beginning to map how spontaneous recovery, renewal, resurgence, and reinstatement contribute to the recurrence of extinguished operant behavior, many questions remain. Perhaps the most basic of these is whether all are manifestations of a similar behavioral process or whether different processes are involved in some of them.
Extinction can be a valuable procedure, either alone or in combination with other techniques, such as the concurrent reinforcement of an appropriate response (the method known as the differential reinforcement of alternative behavior, or DRA), in helping people eliminate unwanted, problematic behavior. It is most successfully used when its varied effects on both the response targeted for extinction and other behavior are understood.
