Classical Conditioning: Extinction

EXTINCTION

Classical conditioning can have considerable adaptive value. Imagine a mouse that has several times seen the cat resting on a kitchen chair. It would serve the mouse well to learn about this asso-ciation between the cat and a particular location; that way, the mouse will likely feel afraid whenever it nears the kitchen. This fear, in turn, will probably lead the mouse to avoid that room—a habit that could save the mouse’s life!

At the same time, it would be unfortunate for the mouse if this association, once established, could never be undone. The cat might lose interest in that resting place or leave the household altogether. Either way, it would be useful for the mouse to lose its fearful response to the kitchen so it can return there to forage for food.

All is well for this mouse, though, because the effects of classical conditioning can be undone through a sequence of events similar to those that established the conditioning in the first place. Pavlov demonstrated that the CR will gradually disappear if the CS is pre-sented several times by itself—that is, without the US. For example, repeated pairings of light plus a blast of cold air will create a condi-tioned response, so that the animal will shiver (the CR) whenever the light (the CS) is presented. But if the light is then presented sev-eral times on its own, the shivering response will be extinguished.Extinction is the undoing of a previously learned response so that the response is no longer produced (Figure 7.8).

Let’s be clear, though, that extinction is not just the result of an animal forgetting what it learned earlier. Of course, animals (including humans) do eventually forget things they once learned, but that is not what’s going on in extinction. This point is evident, for exam-ple, in the speed of extinction. As Figure 7.8 shows, a response can be extinguished in just a half-dozen trials over a period of only a few minutes. In contrast, forgetting is far slower: To demonstrate this, we can condition an animal, then leave it alone for several weeks, and then test it by presenting the CS. In this cicumstance, we have arranged for no extinction trials, but we have provided an opportunity for forgetting. The result of this procedure is clear: Even after a substantial delay, the animal is likely to exhibit a full-blown conditioned response (B. Schwartz, Wasserman, & Robbins, 2005). It seems, then, that classically con-ditioned responses are forgotten only very slowly.

The difference between extinction and forgetting is also clear in another procedure. First we condition an animal by repeated pairings of CS and US; then we extinguish the learning by presenting the CS on its own. In a third step, we recondition the same animal—by presenting some more learning trials, just like those in the first step of the procedure. What happens? The reconditioning usually takes much less time than the initial conditioning did. The speed of relearning, in other words, is faster than the orig-inal rate of learning. Apparently, then, extinction doesn’t “erase” the original learning and return the animal to its original naive state. Instead, the animal still has some memory of the learning, and this gives it a head start in the reconditioning trials.

We can draw similar conclusions about extinction from the phenomenon of spontaneous recovery. This phenomenon is observed in animals that have beenthrough an extinction procedure and then left alone for a rest interval. After this rest period, the CS is again presented, and now the CS often elicits the CR—even though the CR was fully extinguished earlier (see Figure 7.8).

According to one view of this effect, the extinction trials lead the animal to recognize that a once informative stimulus is no longer informative. The bell initially signaled that food would be coming soon; but now, the animal learns, the bell signals nothing. However, the animal still remembers that the bell was once informative; so when a new experimental session begins, the animal checks to see whether the bell will again be informative in this new setting. Thus, the animal resumes responding to the bell, pro-ducing the result we call spontaneous recovery (Robbins, 1990).

Like all aspects of conditioning, spontaneous recovery can easily be observed out-side of the laboratory and in humans. For example, various anxiety disorders are often treated via exposure therapy—a process modeled after the extinction procedure. In this process, the person is repeatedly exposed to the specific stimulus or the particular situ-ation that has, for that person, been a source of anxiety—heights, say, or enclosed spaces, or the sight of a snake. During these exposures, the person is kept safe and comfortable—and so there’s no fearful US associated with the CS. As we’d expect, this sequence of events leads to extinction of the CR (the feelings of anxiety)—and with each exposure, the person feels less and less anxious.

When exposure therapy ends, however, people often relapse and again become anxious when exposed to the phobic stimulus. This relapse is not a sign that the therapy has failed. It’s simply an example of spontaneous recovery of a CR—a sign that more treatment is needed to eliminate the anxiety.