CHAPTER 3 | PAVLOVIAN CONDITIONING [CHANCE] reflexes depend on experience, they vary considerably from individual to individual. Pavlov called these conditional PREVIEW reflexes because they “depend on very many conditions”. Pavlovian conditioning is not about teaching dogs to slobber at the sound of a bell, An unconditional reflex consists of an unconditional stimulus (US) nor is it about learning that one event predicts another. It is about one of the major and the behavior it evokes, the unconditional response (UR). ways that events change our behavior, including the behavior we call thoughts and Unconditional stimuli are typically events that are important to survival. feelings. Understanding it is central to understanding people, including ourselves. Pavlov’s next question was, How does a neutral stimulus (one that does not naturally evoke a reflex response) come to do so? How, for example, does a food dish become a CS for salivating? o Pavlov found that virtually any stimulus could become a 1. BEGINNINGS conditional stimulus if it regularly preceded an unconditional One of Pavlov’s goals was to understand how the body breaks food stimulus. down into chemicals that can be absorbed into the blood. This process Each pairing of CS and US is one trial, and the procedure (or starts with the salivary reflex: When food is taken into the mouth, it experience) is best known as Pavlovian or classical conditioning. triggers the flow of saliva. The saliva dilutes the food and starts o However, it also goes by a variety of other names, including breaking it down chemically. In a typical experiment on the salivary sign learning, signal learning, stimulus learning, S-S learning, reflex, Pavlov would bring a dog into the laboratory, put food into its respondent learning, and probably one or two others. mouth, and observe the result. There are two defining elements of Pavlovian conditioning. Pavlov was fascinated by the adaptability of the glands. o [1] First, the behavior elicited by the US is a reflex response, o So, the reflex action of the gland depended on the nature of such as salivating, eye blinking, sweating, or jumping in the stimulus. response to a loud noise. The cleverness of the glands did not end there, however. When a dog o [2] Second, the appearance of the two stimuli is independent had been fed several times, it began to salivate before anything was of behavior; the CS and the US are presented regardless of put into its mouth. In fact, it might start salivating as soon as it entered what the animal or person does. the laboratory. What fascinated Pavlov was that the dogs did not salivate when they were first brought into the laboratory, but only after they had been fed there repeatedly. 3. HIGHER-ORDER CONDITIONING o How could this be? How could experience alter the action of The basic Pavlovian procedure consists of presenting a neutral a gland? This question preoccupied Pavlov to the point of stimulus followed by an unconditional stimulus. Suppose, however, making him shift his attention to psychic reflexes. that a neutral stimulus precedes a well-established CS. The CS is not a US, yet thanks to conditioning it does elicit a CR. So, if you paired a neutral stimulus with the CS, would the neutral stimulus become another CS? 2. BASIC PROCEDURES o G. P. Frolov decided to find out. There are, Pavlov concluded, two distinct kinds of reflexes. First he trained a dog to salivate at the sound of a o [1] One kind is the largely inborn and usually permanent ticking metronome by following the ticking with food. reflex found in virtually all members of a species and that When the metronome was well established as a CS varies little from individual to individual. Pavlov called these for salivating, Frolov held up a black square and unconditional reflexes because they occur more or less then activated the metronome: unconditionally. o [2] The second type of reflex is not present at birth; it must be acquired through experience and is, compared to innate At first the dog salivated at the sound of the o This involves presenting the CS alone (i.e., without the US) metronome but not at the sight of the black square. periodically, perhaps on every fifth trial. If the dog salivates After several pairings of the two stimuli, however, the even when it gets no food, the salivation is clearly a dog began salivating when it saw the square. The conditional response to the tone. Sometimes, test trials are black square had become a CS for salivating even presented at random intervals. though it had never been followed by food: o Learning is thus represented as an increase in the frequency of the conditional response. One problem in attempting to measure Pavlovian learning is a phenomenon known as pseudoconditioning. Pseudoconditioning is the tendency of a neutral stimulus to elicit a CR after a US has elicited a reflex response. o A strong stimulus, such as a needle jab, can sensitize you to The procedure of pairing a neutral stimulus with a other stimuli so that you react to them more or less as you well-established CS is called higher-order would react to the strong stimulus. If a nurse jabs you with a conditioning. needle, you may then wince when he coughs, even if he did Higher-order conditioning greatly increases the importance of not cough before jabbing you. You wince, not because Pavlovian conditioning because it means that many more stimuli can conditioning has occurred, but because the needle jab has come to elicit conditional responses sensitized you to other stimuli. Higher-order conditioning may be even more important to humans If a stimulus has not been paired with a US, any effect it produces than it is to rats. Among people, words are particularly likely to become cannot be the result of conditioning. A problem arises, however, when conditional stimuli by being paired with conditional stimuli—including a stimulus has been paired with a strong US. other words. o Is the behavior that occurs a conditional response, or is it the result of the earlier exposure to a strong stimulus? We can determine the answer by presenting the CS and US to control 4. MEASURING PAVLOVIAN LEARNING group subjects in a random manner so that the stimuli In most studies of Pavlovian conditioning, the CS and US are sometimes appear alone and sometimes appear together. presented close together. Given that the US is by definition capable of o After this we can compare the performance of these control evoking the UR, how is it possible to tell when learning occurs? subjects with experimental subjects for which the CS and US o Suppose, for example, that you sound a tone and then, 2 always appear together. seconds after the tone stops, you put food into a dog’s mouth. o If subjects in the experimental group perform differently from How can you tell when the dog is salivating to the tone as subjects in the control group, the difference in behavior may well as to the food? be attributed to conditioning. o One answer is to note when salivation begins. In this case, you can measure the amount of learning in terms of the latency of the response—the interval between the onset of 5. VARIABLES AFFECTING PAVLOVIAN CONDITIONING the CS and the first appearance of saliva. As the number of The course of Pavlovian conditioning depends on several variables. CS–US pairings increases, the response latency diminishes; the dog may begin salivating even before the tone stops sounding In some conditioning studies, the interval between CS onset and the appearance of the US is so short that using response latency as a measure of learning is difficult. One way to test for conditioning in these situations is to use test trials (also called probe trials). 5.1. HOW THE CS AND THE US ARE PAIRED Pavlovian conditioning involves the pairing of stimuli. The amount of learning that occurs depends to a large extent on how the stimuli are presented. There are four basic ways of pairing stimuli o [1] In trace conditioning, the CS begins and ends before the US appears. There is, then, a gap between the two stimuli. Trace conditioning gets its name from the assumption that the CS leaves some sort of neural trace. o [2] In delay conditioning, the CS and US overlap. That is, the US appears before the CS disappears Some researchers distinguish between short-delay and long-delay procedures. The difference refers to the length of time the CS is present before the US appears. In the short-delay procedure, the CS may be present from several milliseconds (a millisecond is 0.001 second) to a few seconds before the US appears. In the long-delay procedure, the CS may persist for several seconds or even minutes before the US appears. 5.2. CS-US CONTINGENCY Initially, short- and long-delay procedures produce A contingency is an if–then statement. One event, X, is contingent on similar results: A conditional response begins to another event, Y, to the extent that X occurs if and only if Y occurs. appear soon after the CS appears. But in the case Various experiments have suggested that the effectiveness of of long-delay conditioning, the CR latency (the Pavlovian procedures varies with the degree of contingency between interval between the onset of the CS and the CR) CS and US. gradually increases. o In one study, Robert Rescorla exposed rats to a tone followed o Both trace and delay procedures can produce conditional by a mild shock. responses, and most studies of Pavlovian conditioning o The results showed that the amount of learning depended on involve one of these two procedures. However, other how reliably the CS predicted shock. When the CS was procedures for pairing CS and US are possible. nearly always followed by the US, conditioning occurred. o [3] In simultaneous conditioning, the CS and US coincide When a shock was about as likely to occur in the absence of exactly. Both stimuli begin and end at the same time. The a CS as in its presence (the 40% group), little or no learning simultaneous appearance of CS and US is probably rare in took place. the natural environment, but something approximating it may o Rescorla concluded that contingency was essential to occur. Pavlovian learning, but later work raised doubts about this. Simultaneous conditioning is a weak procedure for o Nevertheless, we can say that, other things being equal, the establishing a conditional response. rate of Pavlovian conditioning will vary with the degree of o [4] In backward conditioning, the CS follows the US. CS–US contingency. It is very difficult to produce a CR with the backward o In the laboratory, it is a simple matter to ensure rapid learning procedure. Pavlov described some attempts made by creating a high degree of contingency between the CS at backward conditioning in his laboratory. and US. Pavlov was able to pair a ticking metronome with food so that if the metronome was ticking, the dog always got food, and if it was not ticking, the dog never got food. Outside the laboratory, however, life is more complicated. A stimulus will sometimes precede a particular US and other times will appear alone—or, to be more precise, it will appear with stimuli other than the US. 5.4. STIMULUS FEATURES Researchers have used all sorts of stimuli in conditioning experiments, but not all are highly effective. The physical characteristics of the 5.3. CS-US CONTIGUITY CS and US affect the pace of conditioning. Contiguity refers to the closeness in time or space between two It might seem that, given several neutral stimuli, one would serve as a events. In Pavlovian conditioning, contiguity usually refers to the CS as well as another. But although nearly any stimulus can become a interval between the CS and US. This interval is called the CS, some serve the purpose far more readily than others. interstimulus interval (ISI). o This is illustrated by experiments in which the CS consists of o In trace conditioning, the ISI is the interval between the two or more stimuli presented simultaneously. Such a termination of the CS and the onset of the US; in delay compound stimulus is paired with a US for one or more conditioning, where the two stimuli overlap, it refers to the trials, after which the experimenter tests for conditioning by interval between the onset of the CS and the onset of the US. presenting the compound stimulus and each component of In general, the shorter the ISI, the more quickly conditioning occurs. the CS alone. However, the simultaneous procedure, with no interval at all, is In one of the first studies of compound stimuli, one ineffective. The optimum interval depends on a number of variables. of Pavlov’s assistants simultaneously presented cold One important variable is the kind of response being learned. and tactile stimulation to a dog, followed by a few o For instance, in establishing a conditional eyeblink response, drops of mild acid in the mouth (a US for salivation). the ideal CS–US interval is usually less than one-half second. Then the experimenter tested the dog with the cold In fear conditioning, in which a stimulus is paired with a stimulus alone, the tactile stimulus alone, and the stimulus that arouses fear, the ISI may be much longer, compound stimulus. perhaps several minutes. The results revealed that although both the tactile o And it is possible to obtain very good results with long CS–US stimulus and the compound stimulus were effective intervals in studies of taste aversion. Taste aversion conditional stimuli, the cold stimulus alone was conditioning usually consists of pairing a distinctive taste utterly ineffective. with a substance that induces nausea. This phenomenon is known as overshadowing because, as Pavlov noted, “The effect of one [stimulus] was found very commonly to The optimum CS–US interval also varies according to the type of overshadow the effect of the others almost completely”. The conditioning procedure used, with short intervals generally being less overshadowed stimulus does not go entirely unnoticed; it simply does important in delay conditioning than in trace conditioning. However, not become an effective CS. even in trace conditioning, extremely short intervals may not work well, Perhaps the chief distinguishing characteristic of an effective CS is its as a study by Gregory Kimble demonstrates. intensity: Strong stimuli overshadow weak ones. It is difficult to generalize about the role of contiguity in Pavlovian The intensity of the US is also very important, with stronger stimuli conditioning. Although shorter intervals are generally preferable to producing better results, in general, than weaker ones. longer ones, the ideal interval varies in complex ways from situation to It is possible, however, for either a CS or a US to be too intense. In situation. eyelid conditioning, a bright light may make a better CS than a dim o Contiguity is complicated, but it cannot be ignored because it one, but if the light is very strong, it may be an unconditional stimulus affects the success of any given conditioning procedure one for blinking and will therefore interfere with learning. way or another. Some stimuli are inherently more likely to become conditional stimuli than others. The ability of a stimulus to become a CS also depends on the nature food or a danger such as a predator, helps us prepare for that event of the US. In general, conditioning proceeds best when both the CS and thereby survive. But there is little additional benefit from duplicate and the US affect internal receptors or when both affect external signals; in blocking, we are merely ignoring duplicate signals. receptors o Blocking might, however, work against us. When we eat foods that make us sick to our stomach, we tend to feel nauseated when we taste those foods again. We therefore 5.5. PRIOR EXPERIENCE WITH CS AND US avoid those foods, and that can help us survive. The effects of a conditioning procedure depend partly on the There is another way in which experience with a neutral stimulus can individual’s previous exposure to the stimuli that will serve as CS and affect later conditioning. US. o Suppose two neutral stimuli, such as a bell and a light, are Research shows that the appearance of a stimulus without the US repeatedly presented together but are not paired with a US. interferes with the ability of that stimulus to become a CS later. This Then one of these stimuli, perhaps the bell, is repeatedly phenomenon is called latent inhibition. paired with an unconditional stimulus so that it becomes a o It probably occurs at least partly because prior exposure CS. What effect will this procedure have on the capacity of undermines the contingency between the CS and US during the light to become a CS? training. o Wilfred Brogden, using dogs as subjects, paired a light and o If a CS sometimes appears alone during training, conditioning a bell for 2 seconds 20 times a day for ten days. Then, for proceeds more slowly, so it is hardly surprising that exposure some of the dogs, he repeatedly paired the bell with a mild to the CS alone before training interferes with conditioning. shock to one of the animal’s front legs to elicit a reflex Janice McPhee and colleagues (2001) added a 28-day movement. interval between the CS preexposure and the conditioning o Next, Brogden presented the light to see what would happen. session, yet the preexposure still interfered with conditioning. He found that this stimulus often elicited a CR even though it Latent inhibition suggests that novel stimuli (stimuli with which the had never been paired with the US. Brogden called this individual has had little or no experience) are more likely to become phenomenon sensory preconditioning. Dogs that had not conditional stimuli than are stimuli that have appeared many times in been exposed to the bell–light pairing did not respond to the the absence of the US. light in this way. In general, then, a stimulus will become a CS o But what if the novel stimulus is part of a compound stimulus more rapidly if it has been paired with another stimulus that that includes an effective CS? Suppose, for example, that a has since become a CS. researcher conducts an experiment on Pavlovian learning in rats, first by repeatedly pairing a tone and an electric shock, then by repeatedly pairing a compound stimulus consisting of 5.6. NUMBER OF CS-US PAIRINGS the tone and a novel stimulus (light) with the shock. What will Because conditioning requires the pairing of stimuli, it seems logical happen if the researcher now presents the light alone? that the more often the CS and US appear together, the more likely a o Leon Kamin performed this experiment and found that the conditional response is to occur. In general, nature accepts this logic. light did not become a CS. This phenomenon, called However, the relationship between the number of stimulus pairings blocking, resembles overshadowing in that one stimulus and the amount of learning is not linear: The first several pairings interferes with the ability of another to become a CS. In are more important than later ones. overshadowing, however, the effect is the result of From a survival standpoint, the curvilinear relationship between CS– differences between the stimuli in characteristics such as US pairings and learning makes excellent sense. If important events intensity; in blocking, the effect is due to prior experience reliably occur together, the sooner the individual adapts, the better. with one part of a compound stimulus. The number of CS–US pairings required to produce a CR varies Blocking would seem to be a useful phenomenon. Exposure to a greatly. In some cases a hundred or more pairings produce only a stimulus that signals an important event, such as the availability of weak CR; in others, a single pairing is effective. 5.7. INTERTRIAL INTERVAL 6. EXTINCTION OF CONDITIONAL RESPONSES The CS–US interval is important to learning. Another interval that If, however, the CS is repeatedly presented without the US, the affects the rate of conditioning is the gap between successive trials. conditional response will become weaker and weaker. The procedure (Recall that each pairing of the CS and US is one trial.) The intertrial of repeatedly presenting the CS alone is called extinction. interval can vary from about a second to several years. At first glance, extinction looks something like forgetting. However, In general, experiments comparing various intertrial intervals find that forgetting refers to a deterioration in performance following a period longer intervals are more effective than shorter ones. While the without practice. optimum interval between CS and US is often a second or less, the o Extinction is a very different procedure. Thus, we can view best intertrial interval may be 20 seconds or more. extinction as a form of conditioning in which the CS is paired with the absence of the US. Pavlov discovered that if, after a response seemed to have been 5.8. OTHER VARIABLES extinguished, he discontinued training for a time and then presented The variables discussed thus far are perhaps the most important, but the CS again, CR appears again. This reappearance of a CR after many others affect the course of Pavlovian conditioning. extinction is called spontaneous recovery. If Pavlov once again o Harry Braun and Richard Geiselhart compared eyelid presented the CS alone several times, the CR would rapidly conditioning in children, young adults, and senior citizens. extinguish. Learning was closely related to age; in fact, the procedure Multiple extinctions may eliminate spontaneous recovery, but this does was not effective in establishing a conditional eyeblink in the not mean the effects of conditioning have been entirely undone. This is oldest subjects. evident from Pavlov’s observation that an extinguished CR usually can o Temperament can also affect conditioning. Pavlov noticed be reestablished far more readily than it was established initially. that some dogs are highly excitable, whereas others are Like conditioning, extinction of conditioning is influenced by numerous much more sedate. He found that these differences, which variables. may be largely due to heredity, affect the rate of learning: the more excitable dogs learned faster. o Stress also affects conditioning. Janet Taylor found that 7. THEORIES OF CONDITIONING anxious students acquired conditional responses more There is no unified theory of Pavlovian conditioning, no set of quickly than those who were more relaxed. propositions that together account for all of the findings in the o Richard Servatius and colleagues also found that, in general, conditioning literature. However, a number of theories have been stress facilitated Pavlovian learning, but Michael Zorawski proposed, and each has its supporters. and his colleagues found that the effects of stress may vary with gender. The variables that affect conditioning interact in important ways. o For example, combining naturally significant stimuli (e.g., 7.1. STIMULUS SUBSTITUTION THEORY pictures of snakes) and intense USs is particularly likely to Pavlov’s attempt to understand conditioning focused on the nature of produce a strong CR quickly. It is easy to see how natural the conditional response. What exactly is the CR? His answer: The selection favors such propensities. CR is the UR. o Many other variables affect the course of conditioning. Once Remember that Pavlov was a physiologist, so his approach was to a conditional response is well established, however, it tends think about the physiological mechanisms involved in reflexes. He to be quite durable and may last for many years. In fact, assumed that in an innate reflex, the US stimulates nerve fibers, which undoing the effects of conditioning can be difficult, or even in turn stimulate other nerve fibers that evoke the UR. These neural impossible. links are what would now be called hard wired; that is, part of the innate structure of the organism. What happens during conditioning? Pavlov proposed that conditioning One interpretation of this result is that the rats involves the formation of a new neurological connection between the innately respond to painful stimuli by jumping, but CS neurons and the US neurons. By repeatedly pairing a bell with respond to events that precede pain by freezing. In food, Pavlov said, a new neural link is formed from the CS neurons to the wild, rats do not often get electrical shocks, but the US neurons so that the bell comes to stimulate the US area of the they do get attacked by animals—snakes, cats, brain, which then triggers the UR. dogs, foxes, and people, among others. o A US and a well-established CS both stimulate the area of If a rat is quietly snacking on some grain in a barn the brain that evokes the UR. Thus, the CR and the UR are and suddenly a house cat sinks its teeth into the the same; the difference is that in one case the neural link to rat’s neck, the rat might break free and survive if it the US area is innate, and in the other it is acquired. suddenly leaps into the air. However, if the rat leaps According to Pavlov, then, conditioning does not involve the into the air when it merely sees a cat or hears it acquisition of any new behavior, but rather the tendency to respond in meow, this could lead to a fatal attack. By freezing, old ways to new stimuli. The CS merely substitutes for the US in the rat improves the odds that it will go unnoticed, evoking the reflex response, hence the name stimulus substitution and prepares it to jump or run if the cat attacks. theory. Higher-order conditioning might also be accounted for by stimulus substitution theory: A CS2 presumably stimulates the CS2 region of 7.3. COMPENSATORY RESPONSE THEORY the brain, which stimulates the CS1 region, which stimulates the US Shepard Siegel offers a variation of the preparatory theory called region, which evokes the CR. compensatory response theory. Siegel argues that the CR prepares But not all the facts from conditioning research support the animal for the US by compensating for its effects. stimulus substitution. o The unconditional response to morphine, for instance, o A critical problem is that there is evidence that the CR and includes decreased sensitivity to pain, but the CR to stimuli UR are not the same. As a rule, the conditional response is associated with morphine is increased sensitivity to pain. weaker than, occurs less reliably than, and appears more o In this case, the person prepares for the drug by suppressing slowly than the UR. These differences are not too the body’s response to it. This means that when people troublesome; however The CS may simply be said to provide repeatedly take a drug in a particular setting, aspects of the weaker stimulation of the US area of the brain; hence, the CR setting may become CS for reduced responses to the drug. is weaker than the UR. Siegel’s theory may also account for certain cases of sudden death But there are often qualitative differences between conditional and following drug use. unconditional responses. o Such deaths are commonly attributed to accidental overdose, A more serious problem for Pavlov’s theory is the finding that the CR is but sometimes they occur following a dose that, given the sometimes the opposite of the UR. person’s history of drug use, should not have been fatal. Anecdotal evidence suggests that the deaths are sometimes due to the absence of stimuli normally present when people 7.2. PREPARATORY RESPONSE THEORY take the drug. Gregory Kimble proposed that the UR is an innate response designed to deal with a US, but the CR is a response designed to prepare for the US. This is known as preparatory response theory. 7.4. RESCORLA-WAGNER MODEL o A study by M. S. Fanselow illustrates. If a rat receives an The Rescorla-Wagner model argues that there is a limit to the amount electric shock, it responds by jumping. of conditioning that can occur in the pairing of two stimuli. One o Fanselow preceded the shock with a tone. The rats soon determinant of this limit is the nature of the US. responded to the tone, but not by jumping; instead they “froze.” o Pairing a bell with a bit of juicy steak, for example, is more amount of learning that can occur and multiply that number likely to produce strong conditioned salivation than pairing a by a constant representing the conditionability of the stimuli bell with a piece of dry bread, and dry bread is likely to work paired. better than a piece of cardboard. The [1] first thing you are likely to notice is that increasing the potency Nor are the characteristics of the US the only variables affecting of the stimuli used results in a huge increase in the amount of learning learning. Certain stimuli become CS in a few pairings, while other on the first trial. The [2] second thing you might notice is that more stimuli can be paired with a US hundreds of times without much effect. learning occurs in three trials than occurred in five trials with a weaker There is a limit to how much a dog can salivate, and this means that US. there is a limit to the amount of conditioning that can occur, even with o Blocking occurs because by the time CS1 is combined the the most effective CS and US. Further, each time a CS and US are stimuli with CS2, nearly all the learning that can occur has paired and some learning occurs, the individual gets closer to reaching already occurred. In a sense, the first CS “uses up” what the the maximum amount of conditioning possible. US has to offer, and there just is not much conditioning The rate at which conditioning proceeds is not, however, uniform. The available. first pairing of CS and US usually produces more learning than the Blocking occurs, then, because there’s very little learning left to occur. second pairing, and the second produces more learning than the third, In a sense, the US has “taught” nearly everything it can. and so on. o Rescorla and Wagner realized that these ideas—that there is a limit to the amount of conditioning that can occur, that the 7.5. OTHER CS THEORIES limit depends on characteristics of the CS and US, and that The Nicholas Mackintosh (1974) offers one. His theory suggests that each successive pairing produces less learning—can be learning depends on which events in the environment we attend stated mathematically in a formula: to, and that depends largely on how well that event predicts the US. o Thus, if a tone is followed by a shock 90% of the time, it is likely to be noticed and so will become a CS; if it is followed by shock 10% of the time, it is far less likely to attract attention and therefore unlikely to become a CS. o Unfortunately, sometimes an event reliably predicts the US yet does not become a CS as readily as one that is less predictive. John Pearce and Geoffrey Hall also assume that attention to the CS The formula looks complicated, but it merely expresses in is of critical importance in conditioning. According to their theory, mathematical form the ideas just covered. organisms pay attention to novel events, not familiar ones. So, if the The amount of learning on any given trial is indicated as a change appearance of a US is surprising, we will pay attention to the events (∆, delta, the symbol for change) in the strength of the association that precede it. (V, for value) on that trial (n). This number is determined by a o Thus, if a CS occurs alone on one trial, and then the US constant (c, a number falling between 0 and 1) based on the follows it on the next, this is a surprise and draws attention to characteristics of the CS and US being paired; the total amount of the CS, so learning occurs. This theory also has its critics, learning that can occur (lambda, λ, a number equal to or greater however than 0) with the US being used; and the amount of learning that has occurred as of the previous trial. o The formula means that if you want to predict the amount of learning on a given trial, you subtract the amount of learning that has occurred as of the previous trial from the maximum MODULE 2: WEEK 4 [FEB 19-23] comes before the Pavlovian conditioning trials. CS CHAPTER 4 | CLASSICAL CONDITIONING: MECHANISMS [DOMJAN] preexposure makes the CS highly familiar and of no particular significance because at this point the CS is CHAPTER PREVIEW presented alone. Chapter 4 continues the discussion of classical conditioning, focusing on the o [2] After the preexposure phase, the CS is paired with a US mechanisms and outcomes of this type of learning. The discussion is organized using conventional classical conditioning procedures. The around three key issues. First, I will describe features of stimuli that determine common result is that participants are slower to acquire their effectiveness as conditioned and unconditioned stimuli. Then, I will responding because of the CS preexposure. discuss factors that determine the types of responses that come to be made to o Thus, CS preexposure disrupts or retards learning. conditioned stimuli and how conditioning alters how organisms respond to Latent inhibition is like habituation. Both phenomena serve to limit the unconditioned stimulus. In the third and final section of the chapter, I will processing and attention to stimuli that are presented without a discuss the mechanisms of learning involved in the development of US and are therefore inconsequential. conditioned responding. The primary theoretical explanation of latent inhibition is that CS preexposure reduces attention to the CS. 1.2.2. THE US-PREEXPOSURE EFFECT Experiments on the importance of US novelty are similar in design to 1. WHAT MAKES EFFECTIVE CONDITIONED AND UNCONDITIONED STIMULI? CS-preexposure experiments. The question was first posed by Pavlov but continues to attract the The US-preexposure effect has been observed not only in appetitive attention of contemporary researchers. conditioning but also in a variety of other situations, including fear 1.1. INITIAL RESPONSES TO THE STIMULI conditioning, taste aversion learning, and drug conditioning. 1.3. CS AND US INTENSITY AND SALIENCE According to their definitions, the CS does not elicit the conditioned Another important stimulus variable for classical conditioning is the response initially but comes to do so because of becoming associated with the US. By contrast, the US is effective in eliciting the target intensity of the conditioned and unconditioned stimuli. response from the outset (unconditionally) without any special training. o More vigorous conditioned responding occurs when more Pavlov’s definitions were stated in terms of the elicitation of the intense conditioned and unconditioned stimuli are used. response to be conditioned. Because of this, identifying potential CSs Stimulus intensity is one factor that contributes to what is more and USs requires comparing the responses elicited by each stimulus generally called stimulus salience. The term salience is not well before conditioning. Such a comparison makes the identification of defined, but it roughly corresponds to significance or noticeability. CSs and USs relative o Theories of learning typically assume that learning will occur 1.2. NOVELTY OF CONDITIONED AND UNCONDITIONED STIMULI more rapidly with more salient stimuli. One can make a The behavioral impact of a stimulus depends on its novelty. Highly stimulus more salient or significant by making it more intense familiar stimuli elicit less vigorous reactions than do novel stimuli. and, hence, more attention-getting. Novelty is also important in classical conditioning. If either the o One can also make a stimulus more salient by making it more conditioned or the unconditioned stimulus is highly familiar, learning relevant to the biological needs of the organism. occurs more slowly than if the CS and US are novel. Another way to increase the salience of a CS is to make it more like 1.2.1. THE LATENT-INHIBITION OR CS-PREEXPOSURE EFFECT the kinds of stimuli an animal is likely to encounter in its natural Numerous studies have shown that if a stimulus is highly familiar, it will environment. not be as effective as a CS than if it were novel. This phenomenon is called the latent-inhibition effect or CS-preexposure effect. Experiments on the latent-inhibition effect involve two phases. 1.4. CS-US RELEVANCE, OR BELONGINGNESS o [1] Participants are first given repeated presentations of the Another variable that governs the rate of classical conditioning is the CS by itself. This is called the preexposure phase because it extent to which the CS is relevant to or belongs with the US Garcia and Koelling’s experiment demonstrates the principle of CS– As the term “higher order” implies, conditioning may be considered to US relevance, or belongingness. operate at different levels. If after becoming conditioned, CS2 were o Learning depended on the relevance of the CS to the US. used to condition yet another stimulus, CS3, that would be third-order Taste became readily associated with illness, and conditioning. audiovisual cues became readily associated with The procedure for second-order conditioning is similar to the standard peripheral pain. procedure for inhibitory conditioning. Rapid learning occurred only if the CS was o In both cases, one conditioned stimulus is paired with the US, combined with the appropriate US and a second CS is paired with the first one without the The CS–US relevance effect obtained by Garcia and Koelling was not unconditioned stimulus readily accepted at first. However, numerous subsequent studies have Why does such a procedure produce conditioned inhibition in some confirmed the original findings cases and excitatory second-order conditioning under other o The selective-association effect occurs even in rats one circumstances? day after birth. This observation indicates that extensive o One important factor appears to be the number of non-US or experience with tastes and sickness (or audiovisual cues and nonreinforced trials. With relatively few noUS trials, peripheral pain) is not necessary for the stimulus-relevance secondorder excitatory conditioning occurs. With extensive effect. training, conditioned inhibition develops o Rather, the phenomenon appears to reflect a genetic The existence of second-order conditioning is of considerable predisposition for the selective learning of certain significance because it greatly increases the range of situations in combinations of conditioned and unconditioned stimuli. which classical conditioning can take place. With second-order 1.5. LEARNING WITHOUT AN UNCONDITIONED STIMULUS conditioning, classical conditioning can occur without a primary US. As it turns out, Pavlovian conditioning can also take place in situations The only requirement is the availability of a previously conditioned where you do not encounter a US. There are two different forms of stimulus classical conditioning without a US. One is higher-order conditioning and the other is sensory preconditioning. 1.5.2. SENSORY PRECONDITIONING Associations can also be learned between two stimuli, each of which 1.5.1. HIGHER-ORDER CONDITIONING elicits only a mild orienting response before conditioning. Irrational fears are often learned through higher-order conditioning. o Consider, for example, two flavors (i.e., vanilla and cinnamon) Higher-order conditioning occurs in two phases. that you often encounter together in pastries without ill o [1] During the first phase, a cue (call it CS1) is paired with a effects. Because of these pairings, the vanilla and cinnamon US often enough to condition a strong response to CS1. flavors may become associated with one another. o Once CS1 elicited the conditioned response, pairing CS1 with What would happen if you then acquired an aversion a new stimulus CS2 was able to condition CS2 to also elicit to cinnamon through food poisoning or illness? the conditioned response. The conditioning of CS2 occurred Chances are your acquired aversion to cinnamon in the absence of the US. would lead you to also reject things with the taste of vanilla because of the prior association of vanilla with cinnamon. o This is an example of sensory preconditioning. As with higher-order conditioning, sensory preconditioning involves a two-stage process. o [1] The cinnamon and vanilla flavors become associated with one another in the first phase when there is no illness or US. Let’s call these stimuli CS1 and CS2. The association between CS1 and CS2 that is established during the sensory preconditioning phase is usually not evident in any behavioral Another variable that is a major factor in what kinds of responses responses because neither CS has been paired with a US become conditioned is the interval between the conditioned stimulus yet, and, therefore, there is no reason to respond. and the unconditioned stimulus. o [2] During the second phase, the cinnamon flavor (CS1) is Generally, conditioning with a short CS–US interval activates paired with illness (US), and a conditioned aversion (CR) responses that are appropriate for immediately dealing with the US. In develops to CS1. Once this first-order conditioning has been contrast, conditioning with a long CS–US interval activates responses completed, the participants are tested with CS2 and now that prepare the organism for the US over a longer time horizon. show an aversion to CS2 for the first time. The response to 2.4. CONDITIONED RESPONDING AND BEHAVIOR SYSTEMS CS2 is noteworthy because CS2 was never directly paired The most successful framework for addressing these issues so far has with a US. been behavior systems theory. o Different systems of behavior have evolved to enable animals to accomplish various critical tasks such as procuring and 2. WHAT DETERMINES THE NATURE OF THE CONDITIONED RESPONSE eating food, defending their territory, avoiding predation, Classical conditioning is usually identified by the development of a producing and raising offspring, and so on. new response to the conditioned stimulus. A large variety of responses Behavior systems theory assumes that the presentation of a US in a can become conditioned. Pavlovian conditioning procedure activates the behavior system Why does one set of responses become conditioned in one situation relevant to that US but other responses develop to the CS in other circumstances? o Classical conditioning procedures involve superimposing a 2.1. THE US AS A DETERMINING FACTOR FOR THE CR CS–US relationship on the behavioral system activated by The most obvious factor that determines the nature of the conditioned the US. As a CS becomes associated with the US, it response is the unconditioned stimulus that is used. becomes integrated into the behavioral system and elicits Interestingly, even small variations in the nature of the US can produce components of that system. changes in the nature of the CR. Behavior systems theory readily explains why the US is an important The fact that the form of the conditioned response is determined by determinant of the CR. The theory is also consistent with the fact that the US encouraged Pavlov to propose the stimulus substitution the nature of the CR depends on the type of CS that is employed. model. The most innovative prediction of behavior systems theory is that the o According to this model, the association of a CS with a US form of the CR will also depend on the CS–US interval that is used. turns the conditioned stimulus into a surrogate US. The The CS–US interval is assumed to determine where the CS becomes conditioned stimulus comes to function much like the US did incorporated into the sequence of responses that makes up the previously. Thus, the CS is assumed to activate neural behavior system. circuits previously activated only by the US and elicit 2.5. S-R VERSUS S-S LEARNING responses similar to those elicited by the US. Conditioned behavior was viewed as a response elicited directly by the The stimulus substitution model correctly emphasizes that the nature CS. According to this idea, conditioning establishes a new stimulus– of the CR depends a great deal on the US that is used in a response, or the S–R. conditioning procedure. However, in many situations the CR does not An important alternative view is that subjects learn a new stimulus– resemble the UR. stimulus (S–S) connection between the CS and the US. According to 2.2. THE CS AS A DETERMINING FACTOR FOR THE CR this interpretation, participants respond to the CS not because it elicits Another important factor that determines the form of the CR is the a CR directly but because the CS activates a representation or nature of the conditioned stimulus. memory of the US. Conditioned responding is assumed to reflect the The stimulus-substitution model predicts that CS–US pairings will status of the activated US representation. generate responses to the CS that are similar to responses elicited by How might we decide between S–R learning and S–S learning the food US. mechanisms? A popular research method that has been used to 2.3. THE CS-US INTERVAL AS A DETERMINING FACTOR FOR THE CR decide between these alternatives involves the technique of US However, learning to respond to a CS is useful to an organism only if devaluation. the CR helps it cope with the US. o [1] The basic strategy of a US devaluation experiment is To be of biological benefit, Pavlovian conditioning should enable the illustrated in Figure 4.11. organism to interact with the US more effectively. Two groups of mildly food-deprived rats received Two different experimental designs that are used to demonstrate conditioning in which a tone was repeatedly paired conditioned modification of the UR. with pellets of food. This initial phase of the o [1] In the common testing design, two groups are experiment was assumed to establish an association compared. During training, one group receives a conditioned between the tone CS and the food US, as well as to stimulus (A) paired with the US while the other group gets get the rats to form a representation of the food that stimulus A and the US unpaired. Following these contrasting was used. Conditioned responding was evident in histories, both groups receive stimulus A followed by the US increased activity elicited by the tone. during a test trial. However, instead of focusing on how the o [2] In the next phase, the experimental group received a organism responds to A, investigators measure responding treatment designed to make the US less valuable to them. during the US. This US devaluation was accomplished by giving the rats o [2] In the common training design, all of the participants sufficient free food to completely satisfy their hunger. receive a procedure in which stimulus A is paired with the US Presumably satiation reduced the value of food and thus and stimulus B is presented unpaired. Following this common devalued the US representation. The deprivation state of the training, responding to the US is evaluated following control group was not changed in Phase 2, and, therefore, presentations of stimuli A and B the US representation remained intact for those rats (Figure Research has shown that Pavlovian conditioning modifies responding 4.11). Both groups then received a series of test trials with the to the US in a wide range of situations. tone CS. During these tests, the experimental group showed o This phenomenon has come to be called conditioned significantly less conditioned responding than the control diminution of the UR. group. 2.6.1. CONDITIONED ANALGESIA Conditioned diminution of the UR is a prominent phenomenon in aversive conditioning and in conditioning experiments where pharmacological agents serve as unconditioned stimuli. In both these cases, a conditioned stimulus elicits physiological processes that serve to counteract the effects of the US. o An aversive stimulus activates the defensive behavior system… Interestingly, exposure to an aversive stimulus or physical injury results in the release of endogenous opiates that counteract the pain induced by the injury. 2.6.2. CONDITIONED DRUG TOLERANCE Tolerance to a drug is said to develop when repeated administrations of the drug have progressively less effect. Tolerance develops with nearly all psychoactive drugs. There is now substantial evidence that drug tolerance can result from Pavlovian conditioning. Pavlovian conditioning is involved because 2.6. PAVLOVIAN CONDITIONING AS A MODIFICATION OF RESPONSES TO each administration of a drug constitutes a conditioning trial in which THE UNCONDITIONED STIMULUS cues that accompany administration of the drug are paired with the Standard definitions of Pavlovian conditioning emphasize that the CS pharmacological effects of the drug. Thus, drug administration cues comes to elicit a new response as a result of being paired with a US. constitute the CS and the pharmacological effects are the US. 2.6.3. CONDITIONED REPRODUCTION AND FERTILITY experimental group than in the control group. (For a more Pavlovian conditioning also results in changes in responding to the US detailed discussion of controls for blocking, see Taylor et al., in appetitive conditioning situations. These effects have been 2008.) examined most extensively in studies of sexual conditioning. The blocking effect was initially investigated in fear conditioning using the conditioned suppression technique with rats. Subsequently, the phenomenon has been demonstrated in various other conditioning 3. HOW DO CONDITIONED AND UNCONDITIONED STIMULI BECOME preparations with both human participants and laboratory animals. ASSOCIATED? Since the time of Aristotle, temporal contiguity has been considered What are the mechanisms of learning, or the underlying processes the primary means by which stimuli become associated. The blocking that are activated by conditioning procedures to produce learning? effect is a landmark phenomenon in classical conditioning because it 3.1. THE BLOCKING EFFECT calls into question the assumption that temporal contiguity is sufficient To get an intuitive sense of the blocking effect, consider the following for learning. The blocking effect clearly shows that pairings of a CS scenario. Each Sunday afternoon, you visit your grandmother who with a US are not enough for conditioned responding to develop always serves bread pudding that slightly disagrees with you. Not Why does the presence of the previously conditioned stimulus A block wanting to upset her, you politely eat the pudding during each visit the acquisition of responding to the added cue B? and, consequently, acquire an aversion to bread pudding. One of the o Kamin, who discovered the blocking effect, explained the phenomenon by visits falls on a holiday, and to make the occasion a bit more festive, proposing that a US has to be surprising to be effective in producing your grandmother makes a special sauce to serve with the bread learning. If the US is signaled by a previously conditioned stimulus (A), it pudding. You politely eat the bread pudding with the sauce, and as will not be surprising. usual you get a bit sick to your stomach. Will you now develop an aversion to the sauce? Probably not. Knowing that bread pudding disagrees with you, you probably will attribute your illness to the 3.2. THE RESCORLA-WAGNER MODEL proven culprit and not learn to dislike the newly added sauce. The idea that the effectiveness of a US is determined by how Two conditioned stimuli are employed (example: the taste of the bread surprising it is forms the basis of a formal mathematical model of pudding and the taste of the special sauce). conditioning proposed by Robert Rescorla and Allan Wagner o [1] In Phase 1, the experimental group receives repeated With the use of this model, investigators have extended the pairings of one of the stimuli (A) with the US. This phase of implications of the concept of US surprise to a wide variety of training is continued until a strong CR develops to stimulus A. conditioning phenomena. In the next phase of the experiment, stimulus B is According to the Rescorla–Wagner model, an unexpectedly large US presented together with stimulus A and paired with is the basis for excitatory conditioning and an unexpectedly small US the US. After several such conditioning trials, (or the absence of the US) is the basis for inhibitory conditioning stimulus B is presented alone in a test trial to see if it 3.2.1. APPLICATION OF THE RESCORLA-WAGNER EQUATION TO THE also elicits the CR. Interestingly, very little BLOCKING EFFECT responding occurs to stimulus B even though B was The Rescorla–Wagner model clearly predicts the blocking effect. In repeatedly paired with the US during Phase 2. applying the model, it is important to keep in mind that expectations of o In Phase 2 of the blocking design, the control group receives the US are based on all of the cues available during the conditioning the same kind of conditioning trials (A+B paired with the US) trial. as the experimental group. However, for the control group, 3.2.2. LOSS OF ASSOCIATIVE VALUE DESPITE PAIRINGS WITH THE stimulus A is not conditioned prior to these compound- US stimulus trials. Rather, during Phase 1, the control group One unusual prediction is that the associative value of a CS can receives presentations of stimulus A and the US in an decrease despite continued pairings with the US. unpaired fashion. In many replications of this design, stimulus B invariably produces less conditioned responding in the o How might this happen? Stimuli are predicted to lose expected US, not much learning is necessary and associative value if they are presented together on a the CS commands less attention on the next trial. conditioning trial after having been trained separately. o More recent attention theories assume that there are 3.2.3. CONDITIONED INHIBITION several different forms of attention relevant to learning Excitatory conditioning involves the acquisition of positive associative and conditioned behavior. value and ceases once the organism predicts the US perfectly on each For example, Hogarth, Dickinson, and Duka reinforced trial. suggested that there are three types of attention. o On nonreinforced trials, the CS+ and CS– are presented The first category, “looking for action,” is together. Once the CS+ has acquired some degree of the attention a stimulus commands after it conditioned excitation (because of its presentation on has become a good predictor of the US and reinforced trials), the organism will expect the US whenever can generate a CR with minimal cognitive the CS+ occurs, including on nonreinforced trials. However, effort. Looking for action is similar to the the US does not happen on nonreinforced trials. Therefore, attentional mechanism of Mackintosh this is a case of overexpectation. (1975) and reflects the behavioral control 3.2.4. EXTINCTION OF EXCITATION AND INHIBITION by well-trained cues. In an extinction procedure, the CS is presented repeatedly without the The second category, called “looking for US. learning,” is the type of attention that is Repeated nonreinforced presentations of the CS+ will result in a involved in processing cues that are not yet progressive reduction of the associative value of the CS+ until VCS+ good predictors of the US and therefore reaches zero have much to be learned about. Thus, 3.2.5. PROBLEMS WITH THE RECORLA-WAGNER MODEL looking for learning is similar to the Pearce One of the difficulties with the model is that its analysis of the and Hall (1980) attentional mechanism I extinction of conditioned inhibition is not correct. described earlier. Another difficulty is that the Rescorla–Wagner model views extinction The third category, “looking for liking,” as the reverse of acquisition, or the return of the associative value of a refers to the attention that stimuli command CS to zero. because of their emotional value (how o Rather extinction appears to involve the learning of a new much they are liked or disliked). relationship between the CS and the US (namely, that the US In addition to specifying different categories of no longer follows the CS). attention, investigators are starting to identify the neural circuits responsible for these differences An important feature of attention theories is that they assume that the outcome of a given trial alters the degree of attention commanded by 3.3. ATTENTIONAL MODELS OF CONDITIONING the CS on future trials Attentional theories differ in their assumptions about what The assumption that the outcome of a given trial influences what is determines how much attention a CS commands on a learned on the next trial has made attention models unique in conditioning trial. explaining a number of interesting. However, attention models cannot Early theories postulated a single attentional mechanism. explain one-trial blocking. o For example, Pearce and Hall proposed that the amount of attention an animal devotes to a CS is determined by how surprising the US was on the preceding trial. Animals have a lot to learn if the US was surprising, 3.4. TIMING AND INFORMATION THEORY MODELS and that increases attention to the CS on the next Neither the Rescorla–Wagner model nor attentional models were trial. In contrast, if a CS was followed by an designed to explain the effects of time in conditioning. However, time is obviously a critical factor. One important temporal variable is the CS– In contrast, with a low I/T ratio the US waiting time US interval. during the intertrial interval is similar to the US The generally accepted view now is that in a Pavlovian procedure, waiting time during the CS. In this case, the CS participants learn not only that a CS is paired with a US, but when that provides little new information about the next US, US will occur. Based on their findings, Williams and colleagues went and not much conditioned responding develops. even further to claim that learning when the US occurs trumps learning whether it occurs. The idea that participants learn about the point in time when the US 3.5. THE COMPARATOR HYPOTHESIS occurs is called temporal coding. The temporal coding hypothesis The relative-waiting-time hypothesis and related theories were states that participants learn when the US occurs in relation to a CS developed to explain the effects of temporal factors in excitatory and use this information in blocking, second-order conditioning, and conditioning. other paradigms in which what is learned in one phase of training o One of their important contributions was to emphasize that influences what is learned in a subsequent phase conditioned responding depends not only on what happens Another important temporal variable is the interval between during the CS but also on what happens in other aspects of successive trials. Generally, more conditioned responding is the experimental situation. observed with longer intertrial intervals. In addition, the intertrial The idea that both these factors influence learned performance is also interval and the CS duration (or CS–US interval) act in combination to central to the comparator hypothesis and its successors developed determine responding. by Ralph Miller and his collaborators. Conditioning trials consisted of an (white noise) presented just before The comparator hypothesis was motivated by an interesting set of delivery of food into a cup. The conditioned response that developed findings known as revaluation effects. to the CS was nosing of the food cup (goal tracking). Each group was o As we discussed, the Rescorla–Wagner model interprets the conditioned with one of two CS durations, either 10 seconds or 20 blocking effect as a failure of learning to CSB. The presence seconds, and one of six intertrial intervals (ranging from 15 seconds to of CSA blocks the conditioning of CSB. 960 seconds). (Intertrial intervals were measured from one US The comparator hypothesis takes a different approach. It assumes that delivery to the next.) Each procedure could be characterized in terms what is blocked is responding to CSB, not learning about CSB. If that of the ratio of the intertrial interval (I) and the CS duration, which is true, then responding to CSB should become evident if the block is Holland called the trial duration (T). removed somehow. Why is conditioned responding determined by the I/T ratio? o A number of studies have shown that such extinction of CSA o A ratio suggests a comparison, in this case between following the blocking procedure unmasks conditioned events during the intertrial interval (I) and the responding to CSB. This is called a revaluation effect conditioning trial (T). because it involves changing the conditioned value of a o What is being compared has been expressed in various ways stimulus (CSA) that was present during the training of the over the years. target stimulus CSB. The unmasking of responding to CSB o According to the relative waiting-time hypothesis, the shows that blocking did not prevent the conditioning of CSB comparison is between how long one has to wait for the US but disrupted performance of the response to CSB. during the CS versus how long one has to wait for the US Inspired by revaluation effects, the comparator hypothesis is a during the intertrial interval (the interval from one US theory of performance rather than learning. It assumes that presentation to the next). conditioned responding depends not only on associations between a When the US waiting time during the CS is much target CS and the US but also on associations that may be learned shorter than during the intertrial interval, the I/T ratio between the US and other stimuli that were present when the target is high. Under these circumstances, the CS is highly CS was being conditioned. informative about the next occurrence of the US and o These other stimuli are called the comparator cues and high levels of responding occur. include the experimental context and other discrete CSs. In the blocking experiment, the target stimulus is CSB and the primary comparator cue is the previously trained CSA that is present during the conditioning of CSB. Another key assumption of the comparator hypothesis is that it only allows for the formation of excitatory associations with the US. Whether conditioned responding reflects excitation or inhibition is assumed to be determined by the relative strengths of excitation conditioned to the target CS as compared to the excitatory value of the comparator stimuli that were present with the target CS during training. Unlike the relative-waiting-time hypothesis, the comparator hypothesis emphasizes associations rather than time. In its simpler form, the theory assumes that organisms learn three associations during the course of conditioning. o A comparison of the direct and indirect activations determines the degree of excitatory or inhibitory responding that occurs An important corollary to the theory is that the comparison that determines responding is made when participants are tested for their conditioned behavior. Because of this assumption, the comparator hypothesis makes the unusual prediction that extinction of comparator–US associations following training of a target CS will enhance responding to that target CS. o It is through this mechanism that the comparator hypothesis is able to predict that the blocking effect can be reversed by extinguishing the blocking stimulus (CSA).