Recent Trends In Classical Conditioning Research Paper

This sample Recent Trends In Classical Conditioning Research Paper is published for educational and informational purposes only. If you need help writing your assignment, please use our research paper writing service and buy a paper on any topic at affordable price. Also check our tips on how to write a research paper, see the lists of psychology research paper topics, and browse research paper examples.

When most people hear the words “Pavlovian conditioning” or “Pavlov” they conjure a mental image of a dog salivating to the sound of a bell that has been previously paired with food. Although such an awareness of this basic learning process is surprising widespread, it is equally noteworthy that most people’s understanding of conditioning is restricted to this particular example. Pavlovian conditioning leads to a rich form of learning that allows animals to anticipate, and consequently prepare for, important events in their environment. Over the past few decades, our knowledge of the nature and mechanisms of Pavlovian conditioning has evolved dramatically from the layperson’s understanding of the conditioning process to that of a rich form of associative learning.

Laws Of Learning

A dominant early view of the conditioning process was that a handful of general principles, or laws of learning, were capable of explaining the conditioning process. Many of these laws of association were first suggested by philosophers, including Aristotle and David Hume (1739/1964). One such law is the law of contiguity, which states that an association will be formed between stimuli when these stimuli occur close together in time. Applied to Pavlovian learning, pairings of a conditioned stimulus (CS; e.g., a bell) with an unconditioned stimulus (US; e.g., food) generate the strongest conditioned responding to the CS when researchers present these stimuli close together in time (a procedure known as short-delay conditioning). However, equally important is the law of contingency, which states that the occurrence of the effect is dependent upon the occurrence of the cause—in other words, the US should only occur along with the CS. Any weakening of the CS-US contingency has a deleterious effect on conditioned responding. Take, for example, my dog learning to bark at the sound of my doorbell ringing. The first few times she heard the doorbell ring, there was no response. However, after several episodes of ringing being followed closely by my greeting people at the door, she began to bark whenever she heard a doorbell ring. In this particular example, that the doorbell was closely followed by people entering our home demonstrates temporal contiguity, whereas the consistency of the doorbell being followed by people entering our home (and entering our home only when the doorbell was rung) reflects the strong contingency between these events. Sometimes, however, my child watches a particular cartoon on television in which a doorbell rings with some frequency. At first, my dog would respond strongly (and loudly) to the sound of this doorbell, just as it did with ours. However, presentation of the doorbell in the absence of visitors weakens the CS-US contingency, in that the CS is sometimes experienced without the US. Likewise, some visitors merely walk into our home without ringing the bell, thereby leading to presentation of the US in the absence of the CS. Both of these examples reduce the CS-US contingency and consequently the likelihood of my dog barking at the sound of doorbells, because it is no longer as reliable a predictor of people entering the house as it had been. Researchers have long believed that these two factors, contiguity and contingency, are two of the major determinants of Pavlovian learning.

The Function Of Pavlovian Conditioning

Associations formed through Pavlovian conditioning can provide significant adaptive value to organisms in our environment. For example, learning to avoid foods that have previously made us sick (taste-aversion learning), identifying the sounds of an approaching predator, or developing a fear of potentially harmful animals (e.g., snakes or spiders) are the consequence of Pavlovian conditioning. Being able to form such associations provides animals with an adaptive edge in that avoidance of potentially dangerous stimuli will increase the likelihood that an organism will successfully reproduce and therefore pass on its genes to future generations. When one considers the value of Pavlovian conditioning from a functional perspective, it is clear that this form of learning extends far beyond learning to salivate to bells. Instead, Pavlovian learning allows animals to anticipate important events in their environment (e.g., food, water, sex, painful stimuli, and illness). Once an animal has learned to anticipate such events, this knowledge then allows the animal to prepare for these events though a variety of behaviors known as conditioned responses. As a result of this process, Pavlovian learning is responsible for the emotional responses that we have to many of the stimuli in our environment, whether it be life-or-death anticipatory responses to stimuli that signal impending danger, or the (hopefully) happy feeling that overcomes you when you hear the song that you first danced to with your significant other. (Notably, if your relationship with your significant other sours, so too will your response to the song, a process known as counterconditioning, in which changes in the value of a US result in corresponding changes in the value of the CS.)

Higher-Order Conditioning

Second-Order Conditioning

Interestingly, Pavlovian learning is not restricted to biologically significant events such as food and painful stimuli. Instead, the mechanisms underlying Pavlovian conditioning allow animals to associate nearly all events that occur close together in time. Take, for example, a group of phenomena known as higher-order conditioning. One example of higher-order conditioning is second-order conditioning (Rizley & Rescorla, 1972), in which a previously conditioned stimulus functions much like a US.

Procedurally, second-order conditioning occurs when a CS, which we will call CS1, is conditioned through pairings with the US (i.e., CS1—>US, a process often referred to as first-order conditioning) and then a second CS, CS2, is paired with CS1 (i.e., CS2—>CS1). Importantly, these latter CS2-CS1 pairings occur without the US present. As a consequence of this training, CS2 is capable of producing a conditioned response. Second-order conditioning is interesting for a couple of reasons. First, the phenomenon demonstrates that conditioning it not restricted to CS-US pairings, as CS2 is conditioned as a result of its pairings with another (previously conditioned) CS. As the name of this phenomenon implies, CS1 has a first-order (direct) association with the US, whereas CS has a second-order relation with the US (through CS1). Second-order conditioning is particularly interesting from an applied perspective, as it allows one to develop fears of stimuli that were never directly paired with a US. For example, imagine that you are afraid of snakes (a previously conditioned CS) and one day you take a hike and come across a snake on a particular trail. When you encounter the snake you become somewhat fearful and quickly retreat from the snake without anything bad happening to you (this is effectively a pairing of CS2, the trail, with CS1, the snake). The next time that you hike on this particular trail, you are likely to experience anxiety when you approach the location where you previously observed the snake. This process of higherorder conditioning allows conditioned fears to spread to stimuli as they become associated with a previously fearful stimulus, which is a potential reason for why my wife no longer takes out the garbage following an experience in which she found two mice (which she does not care for) in our garbage can!

Sensory Preconditioning

Another example of higher-order conditioning is sensory preconditioning (Brogden, 1939). Sensory preconditioning is similar to second-order conditioning, except that the order of learning is reversed, such that two neutral stimuli, CS2 and CS1, are paired together and then one of the stimuli, CS1, is made biologically important through pairings with the US. As a result of this training, CS2 becomes capable of producing a conditioned response. The primary difference between sensory preconditioning and second-order conditioning is the status of CS1 at the time of CS2-CS1 pairings. In sensory preconditioning, CS1 is neutral, whereas in second-order conditioning, CS1 is already conditioned. This procedural difference demonstrates that in the case of sensory preconditioning, learning can occur between two neutral stimuli. In other words, the presence of a US is not necessary for learning to occur. Instead, the US merely provides a reason for the animal to demonstrate what it has learned, a relation between CS2 and CS1 and between CS1 and the US. When most people think about the learning process, they commonly assume that we are most likely to learn about important stimuli and not seemingly trivial events in our environment. Sensory preconditioning demonstrates that Pavlovian learning allows us to associate seemingly unimportant events in our environment. Take, for example, a child who frequently goes to a park where many people walk their dogs. The presence of dogs at the park is likely to be associated. Should the child one day be attacked by a dog in a different place, the child is now likely to demonstrate a reluctance to go to the park that was previously associated with dogs. Examples of this nature point to the fact that what is trivial now might well be critically important at some time in the future. Were learning to be restricted to biologically important events, we would lack many of the potentially important predictors of events in our environment.

Timing Of Conditioned Responding

Pairings of a CS and a US are thought to result in the formation of an association between the two stimuli. This Pavlovian association is responsible for the ability of the CS to produce a conditioned response. Since Pavlov’s (1927) pioneering work, researchers have known that the conditioned responding can be controlled by a variety of factors, including the temporal (time) relation between the CS and the US. For instance, Pavlov found that following pairings of a long-duration CS with a US, conditioned responding was not constant throughout the CS. Instead, very little responding was observed during the early portions of the CS, but at the time at which the US was normally presented, conditioned responding increased. Pavlov termed this effect “inhibition of delay,” as he believed that the early portions of the CS signaled that the US was still some time away and consequently “inhibited” or prevented the conditioned response, whereas the latter portions of the CS were “excitatory” and triggered the conditioned response. Such effects demonstrate that the temporal relation is part of the knowledge gained through Pavlovian conditioning and that this information can control when an animal responds.

Researchers have demonstrated temporal control of conditioned responding in a number of different studies. For example, Rosas and Alonso (1997) presented rats with a long duration CS (a 150-s tone) followed by a brief, mild foot shock to condition a fear of the CS. Early in training, subjects demonstrated a fear of the tone that was constant throughout the entire CS. However, as training progressed, very little fear was observed during the early portions of the CS, whereas greater fear was observed toward the end of the CS, a result that replicates Pavlov’s inhibition of delay experiments. Interestingly, when researchers tested subjects after various delays between training and testing, the temporal control of conditioned responding was lost. That is, a few weeks after training, fear was strong throughout the CS. This finding suggests that the rats remembered the CS-US pairing, but gradually forgot the temporal relation. More generally, temporal information that influences when an animal responds appears to be independent from the more general association that influences whether an animal will respond to a CS.

Inhibitory Learning

Pavlov’s (1927) pioneering work on the study of associative learning focused not only on animals learning to anticipate the occurrence of events in their environment (what Pavlov called excitatory conditioning), but also on the ability to learn that some stimuli predict the absence of an event. Learning about the omission of a US was termed conditioned inhibition, as this form of learning led to the inhibition of a behavior that would have otherwise occurred in the presence of a CS. In a typical conditioned inhibition procedure, a CS is paired with the US except when another stimulus is present. This second stimulus becomes a conditioned inhibitor, as it is thought to signal the absence of the otherwise expected US.

Barnet and Miller (1996) investigated the content of inhibitory learning using rats as subjects. In their experiment, two types of conditioning trials were presented. In one type of trial, the excitatory CS was paired with a brief footshock in order to condition a fear of the CS. On other trials, the excitatory CS was presented along with the inhibitory CS and the shock US was omitted. Barnet and Miller expected that the excitatory CS would come to signal the presentation of the US at a particular moment in time (based upon studies like those described previously). Their question was whether the inhibitory CS would come to signal the omission of the US at a particular moment in time. It is worth noting that this is a particularly difficult question to study because it is relatively easy to examine the timing of a fear response, as the animal’s fear response can be measured throughout the presence of the CS, whereas the ability to measure an absence of fear is complicated by the fact that the inhibition of fear results in no explicit change in behavior. That is, when an inhibitory CS is presented by itself (and the animal presumably expects that the shock will not occur), the animal typically does not change its behavior in any noticeable way. Simply put, if you do not expect something bad to happen, you merely continue on with what you were doing. In order to deal with this measurement issue, Barnet and Miller used what is called a summation test for conditioned inhibition (Rescorla, 1969). In a summation test, the potential of the inhibitor to reduce an animal’s fear response to another CS is assessed. If the supposed inhibitor truly signals that the US will not occur, then it should be able to reduce fear triggered by some other fearful stimulus (i.e., it transfers its inhibitory properties to another stimulus). In their experiment, Barnet and Miller presented their subjects with two types of test trials. In the first, the inhibitor was presented along with a test stimulus that signaled the presentation of the US at the same time at which the subjects were thought to have developed an expectation for the absence of the US (based upon the training already described). In the second, subjects were presented with the inhibitor along with a test stimulus that signaled the presentation of the US at a different time than that at which the subjects were thought to have developed an expectation for the absence of the US. Their results revealed that fear was inhibited only when these two expectations (presentation of shock and omission of shock) occurred at the same time. These findings suggest that an inhibitor not only tells the animal that the US will not occur, but also signals when the US will not occur—thereby resulting in a stimulus that signals safety for a limited time period.

Extinction Of Conditioned Responding

Pairings of a CS and a US result in the acquisition of a conditioned response, a process that alters the way in which we respond to the CS. Likewise, presentations of the CS in the absence of the US results in a decrease in conditioned responding to the CS, a phenomenon termed extinction. Extinction of Pavlovian responding can be analyzed in three ways—operationally, empirically, and theoretically. Operationally, extinction simply involves the presentation of the CS without the US. Empirically, extinction results in a decrease in conditioned responding to the CS. In contrast to these two clear-cut means of understanding extinction, less agreement is reached at the theoretical level of analysis of why the conditioned response decreases when the CS is presented alone. Superficially, one might claim that the response deceases simply because the US is no longer presented, but this merely begs the question of why the response diminishes (or more precisely, what is the nature of learning that occurs during extinction?). Such a question has both great theoretical and implied implications given the pervasive use of extinction procedures to treat anxiety disorders.

Theories of Extinction

One simple theoretical explanation of extinction is that presentations of the CS in the absence of the US lead to a weakening of the CS-US association. This explanation essentially claims that the association is unlearned, effectively returning the CS to a neutral status (as if it had never been previously conditioned). However, several findings are problematic for this explanation. First, Pavlov (1927) found that following successful extinction of a CS, conditioned responding returns when a delay is imposed between extinction and subsequent testing, a phenomenon called spontaneous recovery (as the response recovers on its own following a delay). That the conditioned response is capable of returning demonstrates that the original excitatory association between the CS and US must still be intact, thereby requiring an alternative explanation for the loss of responding caused by the extinction procedure.

A second phenomenon that is often observed is that when a previously extinguished CS is reconditioned through additional pairings of the CS with the US, the conditioned response emerges faster than during initial acquisition training, a phenomenon termed facilitated reacquisition (Konorski & Szwejkowska, 1950, 1952). Again, this typically more rapid reacquisition of conditioned responding implies that the original CS-US association must still be intact. Finally, environmental changes can alter the effectiveness of extinction, a phenomenon termed the renewal effect.

Renewal of Conditioned Responding

Bouton and other researchers have spent many years studying the effect of environmental cues on extinction of conditioned responding. Bouton and Bolles (1979) provided rats with fear conditioning through pairings of a CS with a brief shock in a particular type of experimental chamber (or Skinner box). Following acquisition training, subjects received extinction of the CS in a different type of Skinner box. Typically, the experimental chambers are altered such that they appear, smell, and feel different through the addition of textured floors, patterned walls, odors, changes in illumination, and often differently shaped or sized chambers in order to alter as many sensory characteristics of the training environment (or context) as possible. Following extinction of the CS, subjects were returned to the original training environment in order to measure their fear of the CS. This procedure is known as an ABA renewal procedure, as acquisition training is provided in Context A, extinction in Context B, and testing in Context A. At test, subjects demonstrated a return (or renewal) of fear to the CS. This effect is somewhat different from that of spontaneous recovery, as the return of conditioned responding occurs as a consequence of a change in physical context rather than a delay imposed between extinction and testing. That the conditioned response returns when testing occurs outside of the place in which extinction was provided once again demonstrates that the original excitatory association is somehow preserved during extinction.

Why extinction is disrupted with a change of context has been a hot topic of research over the past several decades. One potential explanation for the renewal effect is that when subjects are tested for renewal of conditioned responding in the context in which acquisition training was provided (Context A), that context “reminds” the animal of the fear conditioning that occurred in that setting, thereby producing a return of fear. However, this appears to be only a partial explanation for this effect. In another experiment, Bouton and Bolles (1979) found that when fear conditioning was provided in Context A followed by extinction in Context B, renewal of responding was also observed when testing was conducted in a new context, Context C (an ABC renewal design). This finding demonstrates that renewal is likely to be observed whenever testing with the CS occurs outside of the extinction context, suggesting that animals fail to remember the extinction experience once they leave the place in which extinction was provided. Notably, ABC renewal is typically somewhat weaker than ABA renewal, providing some evidence that a return to the fear-conditioning context does help remind the animal of the original fear. This reminder, in addition to a failure to fully generalize extinction, leads to a somewhat stronger return of fear.

At the theoretical level, researchers have offered several explanations of the renewal effect. One possibility is that inhibitory learning (which is thought to occur during extinction) does not generalize as well as excitatory learning (i.e., the original fear). Such an explanation is attractive from an evolutionary perspective as it might be safer for one to assume that something harmful will occur than to assume that the potentially harmful event will not occur. Despite the attractiveness of this theoretical perspective, recent evidence casts doubt upon this view. Nelson (2002) pointed out that in a typical renewal procedure, extinction, and consequently the formation of an inhibitory association, is always the second experience that an animal learns. Hence, inhibitory learning and order of training are confounded, such that it is difficult to separate whether it is inhibitory learning rather than second learned information that becomes context specific. To investigate this question, Nelson reversed the typical order of training so that rats received inhibitory training first and excitatory training second. When they tested subjects in a different context, the researchers found that the first learned association generalized and that the second learned association did not. Hence, second learned information (whether it be excitation or inhibition) tends to be more context speciic than first learned information.

Toward addressing why second learned associations become context specific, Bouton (1993, 1994) claimed that extinction of a CS results in multiple memories—one consisting of the original excitatory association and another consisting of an inhibitory CS-US association. Hence, the CS becomes ambiguous, as it now has two meanings. In other words, sometimes the CS is followed by the US and other times it is not. This effect is much like words in the English language that can have multiple meanings—for example, the word “fire” can mean very different things depending upon whether you are in a burning building or standing in front of a firing squad (Bouton, 1997). When we encounter ambiguous words, we use surrounding (context) words to clarify meaning. Likewise, animals use their physical surroundings to determine the current meaning of a CS following acquisition and extinction training. According to Bouton (1993, 1994), once a CS becomes ambiguous, the second learned information becomes context specific through a process called occasion setting. An occasion setter is a stimulus that helps differentiate an ambiguous stimulus by facilitating retrieval of a particular memory. For instance, in a typical occasion-setting procedure, a CS is followed by a US except when another stimulus (an occasion setter) is present. One of the special characteristics of occasion setting is that occasion setters do not typically impact responding directly. Instead, they clarify the meaning of an ambiguous stimulus by promoting retrieval of the appropriate experience. Bouton (1993, 1994) suggests that in order for the memory of extinction to be retrieved, the animal must be exposed to both the CS and the context in which extinction was carried out. In an extinction procedure, the extinction context acts as an occasion setter that facilitates recall of the memory of extinction. If the CS is presented outside of the extinction context, then the inhibitory association will not be retrieved, thereby resulting in a return of conditioned responding based upon the original memory of the CS being followed by the US. Based on this view, all second learned information about a stimulus is controlled in this manner.

Clinical Implications of the Renewal Effect

One of the reasons why the renewal effect has attracted substantial interest from researchers of animal learning as well as clinical psychologists is because of its considerable implications for the treatment of psychological disorders. Consider a person with a fear of snakes. Phobias are intense, irrational fears (irrational because the perceived threat is much greater than the actual threat) that cause marked impairment in the ability of an individual to function in his or her daily life. The behavioral treatment of choice is exposure therapy, which ultimately consists of exposure to the feared object or situation in the absence of any sort of traumatic outcome (i.e., extinction). In everyday life, psychological treatment is most commonly carried out in a therapist’s office. During therapy, the client with a fear of snakes is gradually exposed to snakes until the client no longer feels anxious. At the conclusion of therapy, it is hoped that once the client returns to normal, everyday life, he or she will no longer experience substantial fear in the presence of snakes. Such a treatment plan is similar to the ABA renewal studies already described, which suggests that although the treatment might work in the therapist’s office, there is a good chance that the client’s fear will return once he or she encounters a snake outside of the therapist’s office.

Mystkowski, Craske, and Echiverri (2002) have provided evidence that renewal of fear can occur following seemingly successful treatment of a fear of spiders. In their experiment, people with a fear of spiders showed a return of fear when they encountered a spider in a different place from where their treatment was provided. Findings such as these have resulted in research designed to study whether renewal of conditioned responding can be prevented. Given that extinction only holds up in the place where it was provided, one might consider providing treatment in the environment in which the client is most likely to encounter the fearful stimulus. This strategy is likely to prove more effective given that those environments should help retrieve the memory of therapy, thereby reducing the client’s fear.

Another potential strategy for reducing the renewal effect is treatment in multiple contexts. Gunther, Denniston, and Miller (1998) provided fear conditioning to rats in one context, followed by extinction in a second context. When subjects were tested with the extinguished CS in a new context (an ABC renewal design), the typical return of fear was observed. However, if extinction was divided over three different contexts (B, C, and D), less renewal of fear was observed when testing was conducted in a new context. Findings such as these suggest that providing treatment in multiple settings might help reduce relapse in phobic individuals.

So far, I have discussed context only as the physical environment of an animal. However, many other cues can function as contexts. For instance, physiological states triggered by drugs or emotions can have effects similar to those described above. Anecdotally, when you are feeling down, you are more likely to recall a variety of negative experiences from your life, rather than your successes and other “happy thoughts.” Emotional states can act as retrieval cues for other memories. Researchers have studied effects of this nature in the laboratory using drugs to induce different physiological states. Bouton, Kenney, and Rosengard (1990) provided rats with fear conditioning and then extinguished the animals when they were under the influence of an anti anxiety drug (Valium). When subjects were tested for fear once they were no longer under the influence of the drug, the researchers observed a return of fear. This finding is particularly troublesome given that clients are often prescribed medications to reduce anxiety while they undergo treatment of their fear. Once the client is anxiety free and medication is terminated, a return of fear is possible, given the change in physiological state. Studies using human participants provide support for this possibility. Mystkowski, Mineka, Vernon, and Zinbarg (2003) provided extinction treatment to college students while they were under the influence of either caffeine or a placebo in order to reduce their fear of spiders. When the researchers tested the students in a different physiological state than that in which treatment was provided, a greater return of fear was observed. Such effects have potentially profound implications for the treatment of conditioned fear and will hopefully lead to more promising outcomes for clients as the implications of these studies become better appreciated.

Cue Competition

The Blocking Effect

For many years, researchers of Pavlovian learning assumed that conditioning was a relatively automatic process, provided that the CS and US occurred in a consistent manner close together in time (i.e., with a strong contingency and contiguity, respectively). However, in the late 1960s this view was seriously challenged by a group of phenomena known as cue competition. Cue competition refers to a family of effects in which the ability of a stimulus to produce a conditioned response following conditioning is impaired by the presence of other conditioned stimuli. That is, CSs (cues) appear to compete with one another for the role of predicting a US. One of the most thoroughly studied examples of cue competition is that of blocking (Kamin, 1969). Kamin provided rats with fear conditioning in which a noise was followed by shock. Once the noise was well conditioned, subjects then received additional training in which the noise and a new stimulus (a light) were again paired with shock multiple times. For ease of presentation (so that you don’t have to keep track of which stimulus is which), the noise will be referred to as CS A and the light will be referred to as CS X (the target of testing). When subjects were subsequently tested with the CS X, no fear was observed. Kamin claimed that the prior conditioning of the tone “blocked” learning about the CS X. This result was particularly surprising, given the numerous pairings of CS X (and CS A) with shock, and seriously questioned whether contiguity alone is sufficient for learning.

Toward explaining the blocking effect, Kamin (1969) claimed that Pavlovian learning will only occur when an animal is surprised by the events in its environment. Specifically, Kamin stated that if an animal is surprised by a US, then this will stimulate the learning process, essentially causing the animal to search for causes of the surprising event. Applied to blocking, when the animal first receives CS A followed by shock, the presentation of shock would be unexpected (shocking, if you will), thereby activating the learning process and resulting in conditioning of CS A as a predictor of shock. Following sufficient training, the shock should no longer be surprising, as it is fully predicted by CS A. Now consider the next part of the blocking study, in which both the CS A and CS X are paired with shock. When CSs A and X are presented, the animal should expect the shock, thereby making it less surprising once it is received. This absence of surprise should greatly limit learning about the added stimulus (CS X). In essence, CS X provides no new information to the animal (it is redundant) and is blocked from the learning process.

The Rescorla-Wagner Model

Kamin’s (1969) analysis of blocking truly revolutionized theories of Pavlovian learning for decades to come. Rescorla and Wagner (1972) developed a mathematical model of learning based in large part on Kamin’s idea of surprise. In their model, surprise is viewed as a discrepancy between what an animal actually receives and what the animal expects to receive—the greater the difference, the greater the learning that will occur in a given conditioning trial. Essentially, learning reflects the difference between the size (or magnitude) of the US received and the magnitude of the US expected (based upon the CSs present on a given conditioning trial; i.e., US received minus US expected). Applied to blocking, the first stimulus, CS A, gains associative strength when the unexpected shock occurs because the US received is much greater than the US that is expected. As conditioning progresses, surprise (the difference between the US received and the US that is now predicted by CS A) is reduced as CS A becomes a better predictor of shock. When CS X is added to the conditioning process in the second part of the blocking procedure, surprise is reduced by CS A, thereby limiting learning about CS X. Hence, according to the Rescorla-Wagner model, the absence of responding to the CS X during testing reflects a learning deficit that cannot be reversed unless CS X is directly paired with the US in the absence of the previously conditioning CS A (as the shock would now be surprising when CS A is no longer present to predict it).

The US-Preexposure Effect

Cue competition can also occur between a CS and the context in which it is trained. One example of this is the US-preexposure effect (Randich & LoLordo, 1979), in which an animal receives exposure to the US prior to Pavlovian conditioning training. In a typical US preexposure procedure, an animal is placed into a particular type of experimental chamber (context) and is presented with the US (e.g., shock) multiple times. Following preexposure to the US, conditioning trials are provided in which a CS (X) is paired with a US (e.g., CS X—’shock). At test, subjects that received preexposure to the US are less fearful of CS X than are control subjects that did not experience preexposure to the US (i.e., that merely received X— shock pairings).

One potential explanation of the USpreexposure effect is that the animal develops a fear of the conditioning context as a result of its being associated with the shock. The Rescorla-Wagner model (1972) describes this as a consequence of the shock being unexpected (i.e., surprising), thereby allowing the context to become associated with the shock. Once the context becomes a good predictor of shock, surprise is reduced. Now consider the conditioning trials in which CS X is paired with shock. When the animal is placed into the context, the prior conditioning should generate an expectation of shock, such that when CS X is later paired with shock, surprise is reduced and learning about CS X is diminished. (Note the similarity between this explanation and the one provided for the blocking effect— US preexposure is essentially blocking by the context.)

A theoretically related effect to that of US preexposure is the effect of trial spacing on conditioning. Massed training in which CS-US pairings occur close together in time tends to produce less effective conditioning than does spacing the same number of conditioning trials over time (Gibbon, Baldock, Locurto, Gold, & Terrace, 1977). Essentially, spreading learning over time, as opposed to “cramming,” leads to more effective learning. Again, the context is thought to play a role in this effect. From the perspective of the Rescorla-Wagner model (1972), both the CS and the context become good predictors of the US, given the reduced time between conditioning trials. In this instance, the context functions like a CS, thereby allowing learning about both the context and the CS, as they are both potential predictors of the surprising US. However, the context and the CS must share their prediction of the US, leaving the CS with less of an association than if the trials had been spread out in time. With spaced training, the context is present for some time between conditioning trials, which should result in an expectation of the US that is not presented. The absence of the US when it is otherwise expected, based upon the context, should result in extinction of the context so that the animal no longer expects the US except when the CS is present.

Comparator Theories

The Rescorla-Wagner model (1972) has been one of the most successful theories of Pavlovian learning. The reason for its success is not necessarily its accuracy, but rather its ability to inspire research and new theories of conditioning. One family of theories that followed the Rescorla-Wagner model is comparator theories (e.g., Gibbon & Balsam, 1981; Miller & Matzel, 1988), which view conditioning as a comparison of the predictive ability of the CS to that of other stimuli. When the CS is a relatively better predictor of the US than are other stimuli, conditioned responding will be observed. Scalar Expectancy Theory (Gibbon & Balsam, 1981) views this comparison as taking place between the CS and the context in which training is provided. According to this theory, two time intervals are compared—the time between the CS and the US (known as the trial time, T) and the time between conditioning trials (known as the cycle time, C). When the ratio of C/T exceeds some specified value, conditioned responding emerges. Applied to the effect of trial spacing, spacing conditioning trials adds more time to the trial time and thereby inflates the value of the numerator of this comparison (above the threshold for responding). Conversely, massed conditioning trials reduces the cycle time and consequently the ratio, thereby impairing conditioned responding. Notably, learning of these time intervals is not impaired by massed training. Instead, conditioned responding simply fails to emerge because the context is essentially too good of a predictor of the US.

The account of the trial spacing effect provided by Scalar Expectancy Theory (Gibbon & Balsam, 1981) differs from that provided by the Rescorla-Wagner (1972) model. Whereas the Rescorla-Wagner model views all response deficits as a consequence of learning impairments in which the CS does not readily condition due to the presence of other stimuli, Scalar Expectancy Theory attributes the lack of responding to a performance deficit resulting from the C/T comparison. Central to this theoretical difference is whether these response deficits are due to performance variables rather than to deficits in learning (as posited by the Rescorla-Wagner model). Efforts to test these competing views have provided US preexposure training prior to conditioning (Matzel, Brown, & Miller, 1987). Once the typical US preexposure effect was observed, further training was provided in which the context was extinguished by exposing the animals to the context without any CS or US presentations. This training should increase the time between trials (the cycle time) and thereby increase the C/T ratio. Once this training was provided, the researchers observed greater conditioned responding to the CS, as would be anticipated by Scalar Expectancy Theory.

The Comparator Hypothesis

The comparator hypothesis (Miller & Matzel, 1988) is related to scalar expectancy theory (Gibbon & Balsam, 1981) in numerous ways, except, primarily, for the nature of the comparison. The comparator hypothesis claims that learning occurs whenever two or more stimuli occur close together in time (and space). Note that this view differs from the laws of learning discussed earlier that claimed that contiguity alone was not sufficient for learning (instead, factors such as surprise were also critical). However, according to the comparator hypothesis, performance is based upon relative associative strengths. What does this mean? Take, for example, a typical conditioning procedure in which a CS is paired with the US. According to the comparator hypothesis, three associations will be formed: (a) between the CS and the US, (b) between the CS and the context (the place in which the CS is trained), and (c) between the context and the US. At test, the CS is presented by itself, and the animal compares the associative strength of the CS to other stimuli that were present during training with the CS (the so-called comparator stimuli). The comparison works in the following manner: When the CS is presented at test, it activates memories (representations) of all of the stimuli that were present during its training, including the context. When the CS activates a memory of the context, the context can now (potentially) activate a memory of other stimuli that were present in that context (e.g., the US). Thus, the CS can directly activate a memory of the US, but it can also indirectly activate a memory of the US through the comparator stimulus (the context).

The comparator hypothesis claims that animals compare the strength of the US representation activated directly by the CS (association 1) to the strength of the US representation that is indirectly activated by the CS (through the comparator stimulus). This latter association will only be activated when the association between the CS and the comparator stimulus and between the comparator stimulus and the US are both strong. If either association is relatively weak, then the indirectly activated memory of the US will not be activated. Hence, in order to see a strong conditioned response to a CS, it must have a relatively strong association to the US (compared to other possible predictors of the US). Essentially, the CS needs to be a better predictor of the US than any other stimulus. This can potentially be accomplished in a couple of ways. The first way is if the CS has no comparator stimulus, but this isn’t practical in that training must take place somewhere (i.e., in some type of context). The second way is if the training stimulus’s comparator stimulus is only weakly associated with the US. Applied to a typical acquisition procedure—a CS is paired with the US. This would allow the CS to become associated with the US as well as with the context (because they are present at the same time— i.e., with good contiguity). At the same time, the context would become associated with the US (again, due to good contiguity). However, if there is a relatively long period of time between conditioning trials, the animals will be exposed to the context without the US. This would result in a weakening (extinction) of the context-US association and therefore a relatively ineffective comparator stimulus.

Now consider an instance in which conditioning is conducted with close spacing between trials. Yin, Barnet, and Miller (1994) provided rats with fear conditioning training in which a CS was trained in either a massed or spaced manner (i.e., trials were presented either close together or spread out in time). The comparator hypothesis would predict that the CS-US association would be the same in both conditions, as the CS and US are presented together. Similarly, the association between the CS and the context would also be equivalent across both groups because the CS and the context are paired the same number of times. What should differ between groups is the strength of the context-US association, in which this association would be expected to be stronger in the massed group than in the spaced group because the context-US association would extinguish more in the spaced group (due to the greater time between trials). As a result, the memory of the US activated through the comparator stimulus (the context) should be greater in Group Massed than in Group Spaced, thereby leading to less responding in Group Massed. This is what Yin et al. found—strong conditioned responding in Group Spaced and weak conditioned responding in Group Massed (i.e., the trial spacing effect). What they did next was even more interesting. They then extinguished the context for both groups by placing the animals in the context without giving them the US. Note that this should weaken the context-US association for the massed group, thereby leading to a less effective comparator stimulus. At test, they found that context extinction resulted in strong conditioned responding in both groups. That is, the trial spacing effect went away. Interestingly, this suggests that the trial spacing effect doesn’t influence “learning”; instead, it influences performance.

The comparator hypothesis can also explain the blocking effect that was originally observed by Kamin (1969). Recall that in a typical blocking procedure, a CS (A) is paired with the US in the first stage of training and then CS A (the blocking stimulus) and a new stimulus, CS X (the blocked stimulus), are paired with the US in the second stage of training. At test, conditioned responding to the blocked stimulus (X) is weak relative to a control condition lacking the critical first stage of training. The comparator hypothesis explains this effect as being due to a performance deficit resulting from the comparator process, not to a failure to learn an association between CS X and the US. That is, animals learn an X-US association, but they also learn associations between X and A and between A and the US. These strong X-A and A-US associations result in decreased performance to CS X (again, CS A is too good of a predictor of the US). The comparator hypothesis makes the prediction that weakening the AUS association through extinction of CS A should result in a weakened memory of the US activated through the comparator stimulus (A), and consequently an increase in conditioned responding to CS X. Blaisdell, Gunther, and Miller (1999) confirmed this prediction when they provided rats with blocking training followed by extinction of CS A and observed an increase in responding to CS X. These findings suggest that animals learned about the blocked stimulus (X) but failed to respond to it due to its strong associations to the blocking stimulus (A), which in turn is associated with the US.

Summary

Over the past several decades, researchers have learned much about the nature and function of Pavlovian learning. These developments have their roots in the classic work of Pavlov (1927) and have extended his work to the current state of knowledge, which points to the rich nature of Pavlovian learning. Pavlovian conditioning is a robust form of learning that provides knowledge to animals central to their survival. Through the mechanisms of Pavlovian learning, a wide variety of knowledge about our world can be acquired and used to understand not only what will occur in our environment, but also when these events will occur. Future research will continue to analyze the critical determinants of learning and performance of Pavlovian associations.

References:

Barnet, R. C., & Miller, R. R. (1996). Temporal encoding as a determinant of inhibitory control. Learning and Motivation, 27, 73-91.
Blaisdell, A. P., Gunther, L. M., & Miller, R. R. (1999). Recovery from blocking through deflation of the block stimulus. Animal Learning & Behavior, 27, 63-76.
Bouton, M. E. (1993). Context, time, and memory retrieval in the interference paradigms of Pavlovian learning. Psychological Bulletin, 114, 80-99.
Bouton, M. E. (1994). Context, ambiguity, and classical conditioning. Current Directions in Psychological Science, 3, 49-53.
Bouton, M. E. (1997). Signals for whether versus when an event will occur. In M. E. Bouton & M. S. Fanselow (Eds.), Learning, motivation, and cognition: The functional behaviorism of Robert C. Bolles (pp. 385-409). Washington, DC: American Psychological Association.
Bouton, M. E., & Bolles, R. C. (1979). Contextual control of the extinction of conditioned fear. Learning and Motivation, 10, 445-466.
Bouton, M. E., Kenney, F. A., & Rosengard, C. (1990). State-dependent fear extinction with two benzodiazepine tranquilizers. Behavioral Neuroscience, 104, 44-55.
Brogden, W. J. (1939). Sensory preconditioning. Journal of Experimental Psychology, 25, 323-332.
Gibbon, J., Baldock, M. D., Locurto, C. M., Gold, L., & Terrace, H. S. (1977). Trial and intertrial duration in autoshaping. Journal of Experimental Psychology: Animal Behavior Processes, 3, 264-284.
Gibbon, J., & Balsam, P. (1981). Spreading association in time. In C. M. Locurto, H. S. Terrace, & J. Gibbon (Eds.), Autoshaping and conditioning theory (pp. 219-253). New York: Academic Press.
Gunther, L. M., Denniston, J. C., & Miller, R. R. (1998).
Conducting exposure treatment in multiple contexts can prevent relapse. Behaviour Research and Therapy, 36, 75-91.
Hume, D. (1964). Treatise of human nature (L. A. Selby-Bigge, Ed.). London: Oxford University Press. (Original work published 1739)
Kamin, L. J. (1969). Predictability, surprise, attention, and conditioning. In B. A. Campbell & M. R. Church (Eds.), Punishment and aversive behavior (pp. 279-296). New York: Appleton-Century-Crofts.
Konorski, J., & Szwejkowska, G. (1950). Chronic extinction and restoration of conditioned reflexes. I. Extinction against the excitatory background. Acta Biologiae Experimentalis, 15, 155-170.
Konorski, J., & Szwejkowska, G. (1952). Chronic extinction and restoration of conditioned reflexes. III. Defensive motor reflexes. Acta Biologiae Experimentalis, 16, 91-94.
Matzel, L. D., Brown, A. M., & Miller, R. R. (1987). Associative effects of US preexposure: Modulation of conditioned responding by an excitatory training context. Journal of Experimental Psychology: Animal Behavior Processes, 13,65-72.
Miller, R. R., & Matzel, L. D. (1988). The comparator hypothesis: A response rule for the expression of associations. In G. H. Bower (Ed.), The psychology of learning and motivation (Vol. 22, pp. 51-92). San Diego, CA: Academic Press.
Mystkowski, J. L., Craske, M. G., & Echiverri, A. M. (2002). Treatment context and return of fear in spider phobia. Behavior Therapy, 33, 399-416.
Mystkowski, J. L., Mineka, S., Vernon, L. L., & Zinbarg, R. E. (2003). Changes in caffeine states enhance return of fear in spider phobia. Journal of Consulting and Clinical Psychology, 71, 243-250.
Nelson, J. B. (2002). Context specificity of excitation and inhibition in ambiguous stimuli. Learning and Motivation, 33, 284-310.
Pavlov, I. P. (1927). Conditioned reflexes. London: Oxford University Press.
Randich, A., & LoLordo, V. M. (1979). Preconditioning exposure to the unconditioned stimulus affects the acquisition of a conditioned emotional response. Learning and Motivation, 10, 245-277.
Rescorla, R. A. (1969). Pavlovian conditioned inhibition. Psychological Bulletin, 72, 77-94.
Rescorla, R. A., & Wagner, A. R. (1972). A theory of Pavlovian conditioning: Variations in the effectiveness of reinforcement and nonreinforcement. In A. H. Black & W. F. Prokasy (Eds.), Classical conditioning. II: Current research and theory (pp. 64-99). New York: Appleton-Century-Crofts.
Rizley, R. C., & Rescorla, R. A. (1972). Associations in second-order conditioning and sensory preconditioning. Journal of Comparative and Physiological Psychology, 81, 1-11.
Rosas, J. M., & Alonso, G. (1997). Forgetting of the CS duration in rats: The role of retention interval and training level. Learning and Motivation, 28, 404-423.
Yin, H., Barnet, R. C., & Miller, R. R. (1994). Trial spacing and trial distribution effects in Pavlovian conditioning: Contributions of a comparator mechanism. Journal of Experimental Psychology: Animal Behavior Processes, 20, 123-134.