American psychologist John B. Watson kick-started the behaviourist movement, arguing that psychologists should “never use the terms consciousness, mental states, mind, content, introspectively verifiable, imagery, and the like” (Watson, 1913, p. 166). Watson’s firebrand stance was fuelled in large part by the work of a Russian physiologist, Ivan Pavlov (1849–1936).

Pavlov was awarded the Nobel Prize in Physiology in 1904 for his work on the salivation of dogs. Pavlov studied the digestive processes of laboratory animals by surgically implanting test tubes into the cheeks of dogs to measure their salivary responses to different kinds of foods. Serendipitously, his explorations into spit and drool revealed the mechanics of one form of learning, which came to be called classical conditioning. Classical conditioning occurs when a neutral stimulus produces a response after being paired with a stimulus that naturally produces a response. In his classic experiments, Pavlov showed that dogs learned to salivate to neutral stimuli such as a bell or a tone after that stimulus had been associated with another stimulus that naturally evokes salivation, such as food.

Pavlov’s basic experimental setup involved cradling dogs in a harness to administer the foods and to measure the salivary response, as shown in FIGURE 7.1. He noticed that dogs that had previously been in the experiment began to produce a kind of “anticipatory” salivary response as soon as they were put in the harness, before any food was presented. Pavlov and his colleagues regarded these responses as annoyances at first because they interfered with collecting naturally occurring salivary secretions. In reality, the dogs were behaving in line with the four basic elements of classical conditioning:

Consider your own dog (or cat). Does your dog always know when her dinner is coming, preparing just short of pulling up a chair and tucking a napkin into her collar? It is as though she has one eye on the clock every day, waiting for the dinner hour. Alas, your dog is no clock-watching wonder hound. Instead, the presentation of food (the US) has become associated with a complex CS—your getting up, moving into the kitchen, opening the cabinet, working the can opener—such that the CS alone signals to your dog that food is on the way and therefore initiates the CR of her getting ready to eat.

When Pavlov’s findings first appeared in the scientific and popular literature (Pavlov, 1923a, 1923b), they produced a flurry of excitement because psychologists now had demonstrable evidence of how conditioning produced learned behaviours. This was the kind of behaviourist psychology John B. Watson was proposing: An organism experiences events or stimuli that are observable and measurable, and changes in that organism can be directly observed and measured. Dogs learned to salivate to the sound of a buzzer, and there was no need to resort to explanations about why it had happened, what the dog wanted, or how the animal thought about the situation. In other words, there was no need to consider the mind in this classical conditioning paradigm, and this appealed to Watson and the behaviourists. Pavlov also appreciated the significance of his discovery and embarked on a systematic investigation of the mechanisms of classical conditioning. Let us take a closer look at some of these principles. (As The Real World box shows, these principles help explain how drug overdoses occur.)

269

Remember when you first got your dog? Chances are she did not seem too smart, especially the way she stared at you vacantly as you went into the kitchen, not anticipating that food was on the way. That is because learning through classical conditioning requires some period of association between the CS and US. This period is called acquisition, the phase of classical conditioning when the CS and the US are presented together. During the initial phase of classical conditioning, typically there is a gradual increase in learning: It starts low, rises rapidly, and then slowly tapers off, as shown on the left side of FIGURE 7.3. Pavlov’s dogs gradually increased their amount of salivation over several trials of pairing a tone with the presentation of food, and similarly, your dog eventually learned to associate your kitchen preparations with the subsequent appearance of food. After learning has been established, the CS by itself will reliably elicit the CR.

After conditioning has been established, a phenomenon called second-order conditioning can be demonstrated: conditioning where a new, neutral CS is paired with a stimulus that became associated with the US in an earlier procedure. For example, in an early study Pavlov repeatedly paired a new CS, a black square, with the now reliable tone. After a number of training trials, his dogs produced a salivary response to the black square even though the square itself had never been directly associated with the food. Second-order conditioning helps explain why some people desire money to the point that they hoard it and value it even more than the objects it purchases. Money is initially used to purchase objects that produce gratifying outcomes, such as an expensive car. Although money is not directly associated with the thrill of a drive in a new sports car, it is directly associated with CSs that result in gratifying outcomes (like a sports car, a new smart phone, or a movie channel subscription). Such repeated exposure means that, eventually, money is desirable for its own sake.

270

After Pavlov and his colleagues had explored the process of acquisition extensively, they turned to the next logical question: What would happen if they continued to present the CS (tone) but stopped presenting the US (food)? As shown on the right side of the first panel in Figure 7.3, behaviour declines abruptly and continues to drop until eventually the dog ceases to salivate to the sound of the tone. This process is called extinction, the gradual elimination of a learned response that occurs when the CS is repeatedly presented without the US. The term was introduced because the conditioned response is “extinguished” and no longer observed.

271

Having established that he could produce learning through conditioning and then extinguish it, Pavlov wondered if this elimination of conditioned behaviour was permanent. Is a single session of extinction sufficient to knock out the CR completely, or is there some residual change in the dog’s behaviour so that the CR might reappear? To explore this question, Pavlov extinguished the classically conditioned salivation response and then allowed the dogs to have a short rest period. When they were brought back to the lab and presented with the CS again, they displayed spontaneous recovery, the tendency of a learned behaviour to recover from extinction after a rest period. This phenomenon is shown in the middle panel in Figure 7.3. Notice that this recovery takes place even though there have not been any additional presentations of the CS and US. Some spontaneous recovery of the conditioned response even takes place in what is essentially a second extinction session after another period of rest (see the right-hand panel in Figure 7.3). Clearly, extinction had not completely erased the learning that had been acquired. The ability of the CS to elicit the CR was weakened, but it was not eliminated.

If you get a new can opener, do you think your dog will be stumped, unable to anticipate the presentation of her food? Will a whole new round of conditioning need to be established with this modified CS?

Probably not. It would not be very adaptive for an organism if each little change in the CS–US pairing required an extensive regimen of new learning. Rather, the phenomenon of generalization tends to take place: The CR is elicited even though the stimulus is slightly different from the CS used during acquisition. This means that the conditioning generalizes to stimuli that are similar to the CS used during the original training. As you might expect, the more the new stimulus changes, the less conditioned responding is observed. If you replaced a manual can opener with an electric can opener, your dog would probably show a much weaker conditioned response (Pearce, 1987; Rescorla, 2006).

When an organism generalizes to a new stimulus, two things are happening. First, by responding to the new stimulus used during generalization testing, the organism demonstrates that it recognizes the similarity between the original CS and the new stimulus. Second, by displaying diminished responding to that new stimulus, it also tells us that it notices a difference between the two stimuli. In the second case, the organism shows discrimination, the capacity to distinguish between similar but distinct stimuli. Generalization and discrimination are two sides of the same coin. The more organisms show one, the less they show the other, and training can modify the balance between the two.

272

Before you conclude that classical conditioning is merely a sophisticated way to train your dog, let us revisit the larger principles of Pavlov’s work. Classical conditioning demonstrates that durable, substantial changes in behaviour can be achieved simply by setting up the proper conditions. It was this kind of simplicity that appealed to behaviourists. In fact, Watson and his followers thought that it was possible to develop general explanations of pretty much any behaviour of any organism based on classical conditioning principles. As a step in that direction, Watson embarked on a controversial study with his research assistant Rosalie Rayner (Watson & Rayner, 1920). To support his contention that even complex behaviours were the result of conditioning, Watson enlisted the assistance of 9-month-old “Little Albert.” Albert was a healthy, well-developed child, and, by Watson’s assessment, “stolid and unemotional” (Watson & Rayner, 1920, p. 1). Watson wanted to see if such a child could be classically conditioned to experience a strong emotional reaction, namely, fear.

Watson presented Little Albert with a variety of stimuli: a white rat, a dog, a rabbit, various masks, and a burning newspaper. Albert’s reactions in most cases were curiosity or indifference, and he showed no fear of any of the items. Watson also established that something could make him afraid. While Albert was watching Rayner, Watson unexpectedly struck a large steel bar with a hammer, producing a loud noise. Predictably, this caused Albert to cry, tremble, and be generally displeased.

Watson and Rayner then led Little Albert through the acquisition phase of classical conditioning. Albert was presented with a white rat. As soon as he reached out to touch it, the steel bar was struck. This pairing occurred again and again over several trials. Eventually, the sight of the rat alone caused Albert to recoil in terror, crying and clamouring to get away from it. In this situation, a US (the loud sound) was paired with a CS (the presence of the rat) such that the CS all by itself was sufficient to produce the CR (a fearful reaction). Little Albert also showed stimulus generalization. The sight of a white rabbit, a seal-fur coat, and a Santa Claus mask produced the same kinds of fear reactions in the infant.

What was Watson’s goal in all this? First, he wanted to show that a relatively complex reaction could be conditioned using Pavlovian techniques. Second, he wanted to show that emotional responses such as fear and anxiety could be produced by classical conditioning and therefore need not be the product of deeper unconscious processes or early life experiences as Freud and his followers had argued (see the Psychology: Evolution of a Science chapter). Instead, Watson proposed that fears could be learned, just like any other behaviour. Third, Watson wanted to confirm that conditioning could be applied to humans as well as to other animals. This study was controversial in its cavalier treatment of a young child, especially given that Watson and Rayner did not follow up with Albert or his mother during the ensuing years (Harris, 1979). Modern ethical guidelines that govern the treatment of research participants ensure that this kind of study could not be conducted today. At the time, however, it was consistent with a behaviourist view of psychology.

The kind of conditioned fear responses that were at work in Little Albert’s case were also important in the case of Jennifer, who, as described in the opening story, experienced fear and anxiety when hearing the previously innocent sound of an approaching helicopter as a result of her experiences in Iraq. If Jennifer has learned to associate fear and anxiety with previously neutral stimuli (so that they are now CSs) perhaps an effective treatment would be to help her learn to associate those CSs with safety and comfort? Indeed, a therapy that has proven effective in dealing with such trauma-induced fears is based directly on principles of classical conditioning: Individuals are repeatedly exposed to conditioned stimuli associated with their trauma in a safe setting in an attempt to extinguish the conditioned fear response (Bouton, 1988; Rothbaum & Schwartz, 2002). However, conditioned emotional responses include much more than just fear and anxiety responses. Advertisers, for example, understand that conditioned emotional responses can include various kinds of positive emotions that they would like potential customers to associate with their products, which may be why attractive women are commonly involved in advertisements for products geared toward young males, including beer and sports cars. Just as money becomes desirable through second-order conditioning (by association with other desirable things) so do advertisers hope that their products become desirable through association with other desirable things. The warm and fuzzy feeling that envelops you when watching playful kittens in a bathroom tissue commercial represents a type of conditioned emotional response.

273

As a form of learning, classical conditioning could be reliably produced: It had a simple set of principles and applications to real-life situations. In short, classical conditioning offered a good deal of utility for psychologists who sought to understand the mechanisms underlying learning, and it continues to do so today.

Since the time of Pavlov, though, classical conditioning has been subjected to deeper scrutiny in order to understand exactly how, when, and why it works. Let us examine three areas that give us a closer look at the mechanisms of classical conditioning: the cognitive, neural, and evolutionary elements.

As we have seen, Pavlov’s work was a behaviourist’s dream come true. In this view, conditioning is something that happens to a dog, a rat, or a person, apart from what the organism thinks about the conditioning situation. However, although the dogs salivated when their feeders approached (see Psychology: Evolution of a Science), they did not salivate when Pavlov did. Eventually someone was bound to ask an important question: Why not? After all, Pavlov also delivered the food to the dogs, so why did he not become a CS? Indeed, if Watson was present whenever the unpleasant US was sounded, why did Little Albert not come to fear him?

Somehow, Pavlov’s dogs were sensitive to the fact that Pavlov was not a reliable indicator of the arrival of food. Pavlov was linked with the arrival of food, but he was also linked with other activities that had nothing to do with food, including checking on the apparatus, bringing the dog from the kennel to the laboratory, and standing around and talking with his assistants.

Robert Rescorla and Allan Wagner (1972) were the first to theorize that classical conditioning occurs when an animal has learned to set up an expectation. The sound of a bell, because of its systematic pairing with food, served to set up this cognitive state for the laboratory dogs; Pavlov, because of the lack of any reliable link with food, did not. In fact, in situations such as this, many responses are actually being conditioned. When the bell sounds, the dogs also wag their tails, make begging sounds, and look toward the food source (Jenkins et al., 1978). In short, what is really happening is something like the situation shown in FIGURE 7.4.

274

The Rescorla–Wagner model introduced a cognitive component that accounted for a variety of classical conditioning phenomena that were difficult to understand from a simple behaviourist point of view. For example, the model predicted that conditioning would be easier when the CS was an unfamiliar event than when it was familiar. The reason is that familiar events, being familiar, already have expectations associated with them, making new conditioning difficult. In short, classical conditioning might appear to be a primitive process, but it is actually quite sophisticated and incorporates a significant cognitive element.

One issue that arises from this cognitive view of classical conditioning concerns the role of consciousness. In the Rescorla–Wagner model, the cognitive elements are not necessarily conscious. Rather, they likely reflect the operation of nonconscious associative mechanisms that do more than just record co-occurrences of events: They link those co-occurrences to prior experiences, generating an expectation. Studies of classical conditioning in humans indicate that conditioning can occur without conscious awareness of the relationship between the CS and US. The exception is when special conditioning procedures are used in which a brief time interval is inserted between the CS and the US. Under these circumstances, consciousness of the relationship between the CS and US appears to be necessary for conditioning to occur (Clark, Manns, & Squire, 2002; Clark & Squire, 1998). However, even the tiny sea slug Aplysia exhibits some evidence for this kind of conditioning, and scientists do not believe that Aplysia is capable of consciousness. After all, conditioning in Aplysia involves just a few neurons, as opposed to millions of neurons and complex brain systems in humans (Bekinschtein et al., 2011). This observation reminds us that even though there are many cross-species similarities in classical conditioning, there are also differences.

Pavlov saw his research as providing insights into how the brain works. After all, he was trained in medicine, not psychology, and was a bit surprised when psychologists became excited by his findings. Recent research has clarified some of what Pavlov hoped to understand about conditioning and the brain.

A series of pioneering experiments conducted across several decades by Richard Thompson and his colleagues focused on classical conditioning of eyeblink responses in the rabbit. In the most basic type of eyeblink conditioning, the CS (a tone) is immediately followed by the US (a puff of air), which elicits a reflexive eyeblink response. After many CS–US pairings, the eyeblink response occurs in response to the CS alone. Thompson and colleagues showed convincingly that the cerebellum is critical for the occurrence of eyeblink conditioning (Thompson, 2005). Studies of people with lesions to the cerebellum supported these findings by demonstrating impaired eyeblink conditioning (Daum et al., 1993). Rounding out the picture, more recent neuroimaging findings in healthy young adults show activation in the cerebellum during the eyeblink conditioning (Cheng et al., 2008). As you learned in the Neuroscience and Behaviour chapter, the cerebellum is part of the hindbrain and plays an important role in motor skills and learning.

In addition to eyeblink conditioning, fear conditioning has been extensively studied. Also in the Neuroscience and Behaviour chapter, you saw that the amygdala plays an important role in the experience of emotion, including fear and anxiety. So, it should come as no surprise that the amygdala, particularly an area known as the central nucleus, is also critical for emotional conditioning.

Consider a rat who is conditioned to a series of CS–US pairings where the CS is a tone and the US is a mild electric shock. When rats experience sudden painful stimuli in nature, they show a defensive reaction, known as freezing, where they crouch down and sit motionless. In addition, their autonomic nervous systems go to work: Heart rate and blood pressure increase, and various hormones associated with stress are released. When fear conditioning takes place, these two components—one behavioural and one physiological—occur, except that now they are elicited by the CS.

275

The central nucleus of the amygdala plays a role in producing both of these outcomes through two distinct connections with other parts of the brain. If connections linking the amygdala to the midbrain are disrupted, the rat does not exhibit the behavioural freezing response. If the connections between the amygdala and the hypothalamus are severed, the autonomic responses associated with fear cease (LeDoux et al., 1988). Hence, the action of the amygdala is an essential element in fear conditioning, and its links with other areas of the brain are responsible for producing specific features of conditioning. The amygdala is involved in fear conditioning in people as well as rats and other animals (Olsson & Phelps, 2007; Phelps & LeDoux, 2005).

In addition to this cognitive component, evolutionary mechanisms also play an important role in classical conditioning. As you learned in the Psychology: Evolution of a Science chapter, evolution and natural selection go hand in hand with adaptiveness: Behaviours that are adaptive allow an organism to survive and thrive in its environment. In the case of classical conditioning, psychologists began to appreciate how this type of learning could have adaptive value. Much research exploring this adaptiveness has focused on conditioned food aversions.

Consider this example: A psychology professor was once on a job interview in Southern California, and his hosts took him to lunch at a Middle Eastern restaurant. Suffering from a case of bad hummus, he was up all night long, and developed a lifelong aversion to hummus.

On the face of it, this looks like a case of classical conditioning. The hummus was the CS, a bacterium or some other source of toxicity was the US, and the resulting nausea was the UR. The UR (the nausea) became linked to the once-neutral CS (the hummus) and became a CR (an aversion to hummus). But there are several peculiar aspects to this case. The psychologist’s hosts also ate the hummus, yet none of them reported feeling ill. It is not clear, then, what the US was; it could not have been anything that was actually in the food. Furthermore, the time between the hummus and the distress was several hours; classical conditioning of a CR typically requires that the CS occurs at the same time as the UR (or only very shortly before it). Most baffling, this aversion was cemented with a single acquisition trial. Usually it takes several pairings of a CS and UR to establish learning.

These peculiarities are not so peculiar from an evolutionary perspective. Any species that forages or consumes a variety of foods needs to develop a mechanism by which it can learn to avoid any food that once made it ill. To have adaptive value, this mechanism should have several properties:

John Garcia and his colleagues illustrated the adaptiveness of classical conditioning in a series of studies with rats (Garcia & Koelling, 1966). They used a variety of CSs (visual, auditory, tactile, taste, and smell) and several different USs (injection of a toxic substance, radiation) that caused nausea and vomiting hours later. The researchers found weak or no conditioning when the CS was a visual, auditory, or tactile stimulus, but a strong food aversion developed with stimuli that have a distinct taste and smell.

Indeed, this feature of conditioned taste aversion has been used to deal with an unanticipated side effect of radiation and chemotherapy. Cancer patients who experience nausea from their treatments often develop aversions to foods they ate before the therapy. Broberg and Bernstein (1987) reasoned that a simple technique should minimize the negative consequences of this effect. They gave their patients an unusual food (coconut- or root-beer–flavoured candy) at the end of the last meal before undergoing treatment. Sure enough, the conditioned food aversions that the patients developed were overwhelmingly for one of the unusual flavours and not for any of the other foods in the meal. Other than any root beer or coconut fans among the sample, patients were spared developing aversions to common foods that they were more likely to eat.

Studies such as these suggest that evolution has provided each species with a kind of biological preparedness, a propensity for learning particular kinds of associations over others, so that some behaviours are relatively easy to condition in some species but not others. For example, the taste and smell stimuli that produce food aversions in rats do not work with most species of birds. Birds depend primarily on visual cues for finding food and are relatively insensitive to taste and smell. However, as you might guess, it is relatively easy to produce a food aversion in birds using an unfamiliar visual stimulus as the CS, such as a brightly coloured food (Wilcoxon, Dragoin, & Kral, 1971). Indeed, most researchers agree that conditioning works best with stimuli that are biologically relevant to the organism (Domjan, 2005).