Classical Conditioning: The Absence of Contingency

THE ABSENCE OF CONTINGENCY

The importance of contingency is also evident in another way—in an organism’s reac-tion when there’s no contingency at all. To explore this idea, let’s imagine two different procedures. In both, the animal hears 40 presentations of a tone and receives 20 elec-tric shocks. In the first procedure, these stimuli are presented randomly—and so there’s no contingency, no relationship, between hearing the tone and receiving a shock. In the second procedure, the stimuli are arranged so that half of the tones are followed by shock and half are not, and shocks are never presented without a tone preceding them. In this setting, hearing the tone signals a 50% chance that a shock is about to arrive; in the absence of the tone, the chance of shock is zero.

Animals react very differently to these two procedures. In the first, the tone conveys no information—shock is just as likely with a tone as without. Not surprisingly, the tone in this situation does not become a fearful stimulus. Indeed, after just a few pre-sentations of the tone—so that the novelty of this stimulus wears off—the animal’s behavior doesn’t change at all when the tone is sounded. In the second procedure, in contrast, the tone is informative, indicating that a shock quite likely is about to arrive. It’s no wonder that in this case the animal soon shows a fear response whenever the tone is presented.

These two procedures also differ in another way. The first procedure—with its ran-dom arrangement of tones and shocks—is far more aversive to animals. This is clear, for example, if we monitor the animals’ bodily state (e.g., their heart rates) during both procedures. The comparison tells us that the animals are more stressed in the noncon-tingent procedure. We can also, in effect, “ask” animals which of these procedures they find worse. We do this using a lab setup in which the animal can enter either of two chambers, one governed by the random procedure we’ve described and one governed by the contingent procedure. When given these options, animals reliably choose the second procedure.

What’s going on here? In the second procedure, there is a clear “danger signal” for the animal (the tone), and this signal reliably produces fear. But this setup also provides a clear “safety signal”—namely, the absence of the tone. When the tone isn’t sounded, the animal knows no shock is coming and it can relax. In the first (noncontingent) pro-cedure, in contrast, there’s never a danger signal, nothing to indicate when a shock is coming, and therefore no specific trigger for fear. But, in this setting, there’s also no indication when the animal is safe. As a result, in this procedure the animal is constantly afraid and constantly on guard.

Results like these highlight the importance of contingency—the relations among stimuli that allow us to anticipate upcoming events. When there is some contingency (i.e., one event allows predictions about another event), animals learn this; and it seems to be crucial for classical conditioning. But when there’s no contingency, ani-mals learn this, too—and learn that their environment is unpredictable. These points are crucial for, say, dogs in a conditioning experiment, but they’re no less important for humans. For example, think about why terrorist activities are so frightening: Terrorists hope to convey the message that they can strike you anytime, anywhere, so that you’re never safe. It’s this absence of contingency that makes terrorist threats so scary. Similarly—but on a more personal level—think about the distinction between fear and anxiety. According to some theorists, fear is a state triggered by a specific situation or object; anxiety, on the other hand, is chronic, has no object, and occurs in many situations. Some authors suggest that such unfocused anxiety is partly caused by unpredictability—that is, by an absence of safety signals (B. Schwartz et al., 2005; Seligman, 1975).