Who demonstrated the ways in which cognitive processes are involved in classical conditioning?

Abstract

Classical conditioning was first discovered by Ivan P. Pavlov in the early 1900s. It can be conceptualized as learning about event sequences that occur independently of one's actions in one's environment. Specifically, one learns that a preceding event (stimulus) becomes a signal for a subsequent event. This article touches on a variety of issues. It provides examples of classical conditioning with humans, describes different forms of classical conditioning, elaborates on one of many models of classical conditioning, presents a few of the key phenomena, and finally illustrates application of classical conditioning to some treatments of clinical conditions such as phobias and drug addiction.

V.B.1.b.iv. Fear Classical Conditioning

Classical conditioning has been increasingly used to study the learning of fear. This paradigm can be considered a hybrid of autonomic and somatic classical conditioning because fear causes numerous autonomic changes, which could be measured as the CR. However, in the rat, the most common subject for studies of this type, fear can also be measured with the somatic response of freezing. In the typical paradigm, a tone CS is paired with a shock US. The shock US is delivered to the rat through an electrified floor grid. With pairing of the CS and the US, a fear CR develops in response to the CS. In this case, the fear CR is freezing (the rat holds completely still).

VI. Summary

Classical conditioning theories have become considerably more complex since the early formulations by O. H. Mowrer and others. Modern conditioning models emphasize the role of cognitive factors such as memory processes and expectancies in the etiology and maintenance of conditioned responses. As theories of fear have developed, other pathways to fear acquisition have been added, although classical conditioning continues to be seen as important. Classical conditioning theories have led to a number of important treatments, with the most widely used being the exposure therapies for reducing fear. According to contemporary views, extinction of the CR can be regarded as a process of exposure to corrective information. Exposure involves having the person repeatedly exposure himself or herself to a feared stimulus until fear abates. Patients play an active role in choosing what they will be exposed to, and when the exposure will occur. Exposure therapies can successfully reduce conditioned fears and fears arising from other forms of learning.

Of the exposure therapies, graded in vivo exposure and flooding are among the most effective treatments of phobias, and play an important role in treating disorders in which fear plays a prominent role (e.g., social phobia, agoraphobia). For patients who are extremely phobic, the least demanding form of exposure (systematic de-sensitization) is typically the exposure intervention to be used first. Graded in vivo exposure is particularly important because it involves teaching patients skills for overcoming their fears. Patients can continue to apply these skills on their own, without the aid of a therapist. Exposure therapies can be combined with other psychological interventions, such as relaxation training and cognitive restructuring. For the average phobic patient, combination treatments tend to be no more effective than exposure alone. However, there are likely to be exceptions to this rule, and some patients may benefit most from a combination of psychotherapeutic procedures. Combining exposure with antianxiety drugs does not improve outcome, and may actually impair the effects of exposure. The benefits of exposure therapy tend to be long lasting, with no evidence of symptom substitution. Patients sometimes relapse, although their reemergent fears can usually be successfully treated with a further course of exposure therapy.

4 Classical-operant interactions

While the distinctions between classical and operant conditioning are clear and the two processes can be demonstrated to control behavior separately, for most conditioning preparations it is probably the case that both classical and operant processes are controlling behavior. For example, in operant discrimination training an animal may be trained to respond for food during a tone while during a light no food would be available. In this situation, the animal may learn to associate the tone with food in a manner analogous to classical conditioning. Even in the simplest operant conditioning situation where an animal's bar presses are followed by food with no other stimuli present, the animal may learn that approach and contact with the bar is associated with food in a classical conditioning manner. For classical conditioning, once a response is made it is necessarily followed by the reinforcer and a further strengthening of the response may be related to operant conditioning processes. For example, in the rabbit NMR preparation the eyeblink response may reduce the aversiveness of the shock or air-puff such that the operant processes of escape and avoidance come into play. These potential interactions between classical and operant conditioning processes have led to the development of a number of two-process learning theories (Anger, 1963;Rescorla and Solomon, 1969;Weiss, 1978). These theories postulate two types of associations that are formed during conditioning, a stimulus-reinforcer association analogous to classical conditioning and a response-reinforcer association analogous to operant conditioning.

While separating classical and operant conditioning processes theoretically is relatively easy, separating them in such a way that a drug effect can be studied independently for the two processes becomes much more difficult. One approach is to develop procedures where only classical or operant processes are operating. This is in fact much easier for classical conditioning than for operant conditioning (cf. Mackintosh, 1983). The use of the omission procedure (Sheffield, 1965) within a classical conditioning preparation can firmly establish that only classical conditioning processes are operating. With the omission procedure, any response during the CS negates the presentation of the US. Thus, the response is never followed by the reinforcer. For example, with the autoshaping procedure a key-light is paired with grain presentation for a pigeon. Eventually the pigeon will come to peck the key. If pecks at the key always cause the grain to be omitted, then only classical conditioning processes can maintain responding. In fact, interpretations based on operant conditioning processes would predict that key-pecking should be eliminated. With this procedure responding is reduced, but is still maintained at high levels indicating the classical conditioning processes can maintain substantial amounts of behavior (Williams and Williams, 1969).

To conclusively demonstrate that only operant conditioning is operating it is necessary to eliminate all stimuli associated with the reinforcer. This is difficult for many operant conditioning preparations, so alternative procedures are usually necessary. Many of these procedures involve restraint of the animals, however, some of the most convincing evidence that operant conditioning controls behavior independently of stimuli associated with a reinforcer comes from wheel-running experiments. Running in a wheel maintains the same stimulus conditions for the animal at all times and with these procedures animals can be trained to run to receive food or avoid shock (e.g. Bolles et al., 1966; Mackintosh and Dickinson 1979). Thus, procedures can be developed that conclusively involve only classical or operant conditioning processes.

Rather than attempting to completely eliminate either classical or operant contingencies to study drug effects, a more productive line of research would appear to be the use of procedures which allows for the independent measurement of both operant and classical responding simultaneously. For example,Spealman et al. (1978) trained pigeons on a multiple schedule of reinforcement where key-peck responses were reinforced under operant contingencies on one key whose key color remained constant, while a second key was used as the discriminative stimulus to signal schedule components. Responding was not required to the discriminative stimulus key, however. On a mult VI-Extinction schedule, responding was maintained at a high rate on both response keys. The responses on the discriminative stimulus key can be thought of as a reflection of the strength of the stimulus-reinforcer contingencies (i.e. classical conditioning), while responses on the second key can be thought of as a reflection of the strength of the response-reinforcer contingencies (i.e. operant conditioning). On this schedule, Spealman et al. found that pentobarbital increased responding to the discriminative stimulus key while amphetamine did not. Neither drug increased responding on the constant key. Thus, the increases in responding produced by pentobarbital may be related to its effects on the stimulus-reinforcer relation.

2 Basic Conditioning Procedure

The procedure of classical conditioning consists of the repeated presentation of two stimuli in temporal contiguity. First, a neutral stimulus (NS) is presented—that is, a stimulus that does not elicit regular responses or responses similar to the unconditioned response (UR). Immediately after that, the US is presented. Because of this pairing, the NS will become a CS and, therefore, will be capable of provoking a conditioned response (CR) similar to the UR that, initially, only the US could elicit (Fig. 1).

On the initial trials, only the US will elicit the salivation response. However, as the conditioning trials continue, the dog will begin to salivate as soon as the CS is presented. In salivary conditioning, the CR and the UR are both salivation. However, in many other conditioning situations, the CR is very different from the UR. According to Pavlov, the animals learn the connection between stimulus and response (CS–UR). Currently, it is understood that animals learn the connection between stimuli.

B. The Rabbit Model System of Nictitating Membrane/Eyeblink Classical Conditioning

Much of the general literature on classical conditioning is based on data collected with the human eyeblink conditioning paradigm and in the rabbit nictitating membrane or “third eyelid” paradigm first introduced by Isadore Gormezano (Gormezano et al., 1962; Schneiderman et al., 1962). Evidence has converged from a number of sources to suggest that the cerebellum ipsilateral to the conditioned eye is essential for eyeblink classical conditioning in rabbits and humans.

The most extensive body of literature linking the cerebellum and eyeblink classical conditioning comes from research with animals. This research is eloquently described by its primary instigator and motivator, Richard F. Thompson et al., this volume, and will be mentioned only briefly here. A variety of techniques including electrophysiological recording of multiple and single units, electrolytic and chemical lesions, physical and chemical reversible lesions, neural stimulation, genetic mutations, and pharmacological manipulation have been used to demonstrate that the dorsolateral interpositus nucleus ipsilateral to the conditioned eye is the essential site for acquisition and retention (Berthier and Moore, 1986, 1990; Chen et al., 1996; Clark et al., 1992; Gould and Steinmetz, 1994; Krupa et al., 1993; Lavond et al., 1985; Lincoln et al., 1982; McCormick et al., 1981; McCormick and Thompson, 1984a,b; Steinmetz et al., 1992; Thompson, 1986, 1990; Yeo et al., 1985). Involvement of the cerebellar cortex has also been demonstrated during normal acquisition, although it may not be essential (Chen et al., 1996; Lavond and Steinmetz, 1989).

Introduction

Classical conditioning1 and operant conditioning2,3 are two important forms of associative learning that allow animals, including humans, to survive in a changing environment. Despite the individual analysis of the neuronal processes underlying classical and operant conditioning,4–12 it has been difficult to test experimentally whether, at some fundamental level, they are mechanistically distinct or similar.13–15 This difficulty is largely due to the lack of a suitable model system that exhibits both operant and classical conditioning of the same behavior and, at the same time, is tractable at the cellular and molecular levels.16–19

Feeding behavior in the marine mollusk Aplysia californica20 is a useful model system to overcome the previously mentioned limitations and conduct a comparative analysis of the mechanisms of classical and operant conditioning.21 Feeding behavior is controlled by a well-characterized neural circuit22–24 and, importantly, is modulated by both forms of associative learning.25–28

This chapter reviews studies that analyzed and compared the mechanisms underlying classical and operant conditioning. These analyses examine the neural substrates of learning at the levels of changes in neural circuit activity, the properties of individual neurons, and molecular processes that mediate the cellular modifications. Intriguing similarities and differences between these two forms of learning are being elucidated. For example, both forms of learning lead to increased biting in vivo and increased fictive ingestion in the neural circuit in vitro. A reinforcement pathway, which uses dopamine (DA) as the reinforcement transmitter, is common to both classical and operant conditioning. Finally, a cellular locus of plasticity that is common to both forms of learning, neuron B51, is altered in opposite ways by the two conditioning paradigms. The excitability of B51 is increased by operant conditioning, whereas it is decreased by classical conditioning.

Classical conditioning occurs when neutral stimuli become associated with a psychologically significant event. The main result is that the ‘neutral’ stimuli come to evoke responses or emotions that can contribute to many clinical disorders. Recent research emphasizes the fact that conditioned stimuli evoke whole systems of physiological, emotional, and behavioral responses that help the organism prepare for the significant event. Basic research on classical conditioning has many other implications for understanding the development of clinical disorders, including (but not limited to) anxiety disorders and drug dependence, as well as their therapy.

Classical conditioning paradigms, which involve the presentation of two stimuli within close temporal proximity to each other, have proven very useful for advancing our understanding of the neural substrates of learning. In particular, the use of eyeblink classical conditioning in concert with a variety of neuroscience methods has generated an impressive array of data concerning where and how in the brain plasticity occurs to create changes in behavior. Available data demonstrate that discrete regions of the cerebellum and associated brainstem areas contain neurons that alter their activity during conditioning – these regions are critical for the acquisition and performance of this simple learning task. It appears that other regions of the brain, including the hippocampus, amygdala, and prefrontal cortex, contribute to the conditioning process, especially when the demands of the task get more complex. Given the relatively precise delineation of the neural circuitry involved in this type of learning, eyeblink classical conditioning has been increasingly used in translational human research to study behavioral and brain correlates of clinical disorders.

3.2 Classical Conditioning

Classical conditioning is recognized as the simplest form of associative learning. An association between a signaling stimulus (conditioned stimulus or CS) and a response producing stimulus (unconditioned stimulus or UCS) forms when the CS is presented shortly before UCS onset. The CS gradually comes to elicit a response (CR) similar to that evoked initially by the UCS. A considerable body of research beginning in the 1930s (see Patterson 1976) attempted to demonstrate that spinal reflex circuits show the associational learning of classical conditioning. While beset with theoretical and methodological difficulties, the evidence supported the ability of spinal circuits to support long-lasting (days) changes due to temporal association. Other data (e.g., Beggs et al. 1985) indicate that classical conditioning procedures produce a variety of long-term neural alterations closely approximating associative learning in the intact animal. There is some suggestion that the ability of the spinal cord to sustain this neural plasticity decreases for several days after spinal transection, but may return within a few weeks, presumably after neural reorganization.