CLASSICAL CONDITIONING: BEHAVIORAL PHENOMENA

Classical conditioning involves learning the relations between stimuli. In its simplest form, a neutral stimulus precedes a stimulus (the unconditioned stimulus, or US) that elicits a response (the unconditioned response, or UR). Learning is indexed by the development of a response (the conditioned response, or CR) to the neutral stimulus (which is now a conditioned stimulus, or CS). The interval between the onset of the CS and the onset of the US is called the interstimulus interval (ISI). Stimuli that can become CSs may be discrete or more contextual, and they need not even be external (Bouton, Mineka, and Barlow, 2001). Responses to stimuli (both CRs and URs) may be as simple as an eye blink or more complex, such as approach and withdrawal. Originally thought to be due simply to contiguity between the CS and US, modern conceptions of learning in classical conditioning emphasize that the CS must provide information about the US, and that the CR is both elicited by the CS and anticipates the US.

Excitatory Classical Conditioning

In excitatory procedures, the CS signals that a US will follow. Learning is indexed by the probability, magnitude, and latency of CRs to the CS. In the delay procedure, CS onset precedes US onset and the CS remains on when the US is delivered. In the trace procedure, CS onset precedes US onset but the CS terminates before the US is delivered, leaving an empty "trace" interval between CS and US. Evidence suggests that CS offset, as well as the CS itself, may elicit CRs in the trace procedure. Kehoe and Weidemann (1999) tested rabbits in eyeblink conditioning using a tone CS and a corneal air puff US in a trace procedure. Typically, longer CS-US intervals slow conditioning relative to shorter CS-US intervals. However, across CS-US intervals of 400 milliseconds to thirty seconds, the learning rate was similar, as long as the interval between CS offset and the US was held constant at 400 milliseconds.

Another procedure for producing excitatory conditioning is the discrimination procedure. In simple discrimination conditioning, two CSs that differ along some dimension are used. One CS is always followed by the US (CS+) whereas the other CS is never followed by the US (CS-). Typically, CRs develop to both CSs initially, and then decline to the CS-. This procedure is thought to produce both an excitatory CS (CS+) and an inhibitory CS (CS-, discussed further below). A more complicated type of discrimination is occasion setting. In occasion setting, one CS is used. Another stimulus (the occasion setter) signals whether the CS will be paired or unpaired on a particular trial. This occasion setter is not considered a CS because it does not appear to form a direct association with the US. An example of this procedure, in which a contextual stimulus served as the occasion setter, is provided by Rogers and Steinmetz (1998). In that study, a flashing chamber light was the occasion setter. Whenever the flashing chamber light was on, the context was lit and a tone CS was followed by a corneal air puff US. Whenever the flashing chamber light was off, the context was darkened, and the same tone CS was not followed by the US. Rabbits learned CRs when the chamber was lit and suppressed CRs when the chamber was darkened. The light or dark chamber by itself cannot elicit CRs, so it is not a CS. Rather, the state of the context served to modulate the development of CRs to the tone CS.

Inhibitory Classical Conditioning

In inhibitory procedures, the CS signals that a US will not follow. It is important to note that an inhibitory CS, like an excitatory CS, provides information about the US. Thus, a US must be present somewhere in the conditioning situation for inhibitory conditioning to occur. This is in contrast to latent inhibition (discussed under "Stimulus Preexposure," below). The fact that a US must be present somewhere in the conditioning situation means that some other stimulus in the conditioning situation, either discrete or contextual, becomes a conditioned excitor.

A number of procedures are thought to be able to produce inhibitory CSs. The most common procedure was originally used by Ivan Pavlov (1927). In Pavlov's conditioned inhibition procedure, two CSs are used. On some trials, one of the CSs is paired with a US (CS1-US trials) and becomes a conditioned excitor. On other trials, the two CSs are presented simultaneously (CS1/CS2-alone trials) and no US is presented. The CS that is never paired with the US (CS2) becomes a conditioned inhibitor. Another common procedure for producing inhibitory CSs is the discrimination procedure, in which one CS (CS+) is followed by the US and the other CS (CS-) is not. In this case, CS+ becomes excitatory and CS- becomes inhibitory.

Because conditioned inhibition involves the suppression of CRs, which renders them unobservable, special tests have been developed to detect conditioned inhibition (Rescorla, 1969). In the retardation test, the potential inhibitory CS, when paired with a US, must take longer to elicit CRs than a new CS in order to be considered a conditioned inhibitor. In the summation test, the potential inhibitory CS is presented simultaneously with an excitatory CS. If fewer CRs are observed to this compound than to the excitatory CS alone, the test CS is considered a conditioned inhibitor.

Unlike the widespread acceptance of excitatory associations between a CS and US (but see Gallistel and Gibbon, 2000, for an alternative), inhibitory associations have been controversial (Savastano, Cole, Barnet, and Miller, 1999). The most common view has been that inhibitory associations are simply the opposite of excitatory associations (Rescorla and Wagner, 1972). In this view, a single continuum for association exists, ranging from positive (excitatory) to negative (inhibitory). Other researchers have proposed two separate continuums, one for excitatory associations and the other for inhibitory associations (Pearce and Hall, 1980). These two continuums oppose each other and the presence or absence of CRs is a reflection of this opposition. Finally, Miller and Matzel (1988) have developed a model of classical conditioning in which inhibitory associations do not exist at all, and the presence or suppression of CRs is determined by the inequality of excitatory associations for different stimuli.

Stimulus Preexposure

Exposure to the stimuli by themselves prior to classical conditioning can change the rate at which these stimuli become CSs. The most well-studied example of stimulus preexposure involves preexposure to the CS, or latent inhibition (Lubow and Moore, 1959). In contrast to a new CS, a preexposed CS that is subsequently paired with a US will come to elicit CRs only slowly. Because no US is present during pre-exposure, the CS does not provide information on either the presence or absence of the US during pre-exposure and, therefore, forms neither an excitatory nor an inhibitory association. Rather, the preexposed CS is thought to undergo a loss of salience, and is treated as irrelevant.

Schmajuk, Lam, and Gray (1996) developed a detailed neural network model in which latent inhibition is explained by a reduction in the ability to store and retrieve CS-US associations. Preexposure to a CS reduces both attention to and the magnitude of the representation of that CS. During subsequent CS-US pairings, less attention to the CS slows the storage of an excitatory CS-US association and a less intense representation slows the ability of the CS to elicit retrieval of the CS-US association.

Cue Competition

Multiple stimuli may be paired with a US, but not all of them will develop the ability to elicit a CR. Consider two CSs presented simultaneously and followed by a US. If the CSs are similar in salience, each of the CSs will acquire the ability to elicit a CR when presented separately. However, if the CSs differ greatly in salience (for example, a bright light and a soft tone), only one of the two CSs will come to elicit CRs. The more salient CS is said to overshadow, or disrupt, the formation of an association between the less salient CS and the US. This phenomenon demonstrates that mere contiguity of a CS and a US is not always enough to produce an association between them.

The insufficiency of contiguity between a CS and a US for learning in classical conditioning was forcefully demonstrated by Kamin (1969). Kamin's procedure, which is known as blocking, involves three phases. In phase 1, a CS is paired with a US (CS1-US trials) until it becomes a strong conditioned excitor. In phase 2, the excitatory CS is presented in compound with a new CS, and the entire compound is paired with a US (CS1/CS2-US trials). In phase 3, the two CSs are tested separately (CS1-alone and CS2-alone trials) for the ability to elicit a CR. The typical result is that CRs occur only to the CS used in phase 1 (CS1). This demonstrates that a CS must provide new information about the US in order to form an association with it. Mere contiguity between a CS and a US is not sufficient for classical conditioning.

Contextual cues may contribute to the blocking effect. In essence, it has been proposed that the CS and the context in phase 1 form a combined association with the US and this combined association blocks learning to the other CS in phase 2. Giftakis and Tait (1998) tested rabbits using a tone CS, a flashing light CS, and a periorbital shock US. Rabbits given several sessions of exposure to the context alone between phases 1 and 2, which serves to extinguish any context-US association, showed less blocking than rabbits that did not receive context-alone exposure. In addition, rabbits shifted to a different context between phases 1 and 2 also showed less blocking than rabbits training in the same context in phases 1 and 2.

CR Timing

In classical conditioning, a US elicits a UR and a CS comes to elicit a CR. However, the CR is not just elicited by the CS. In addition, the CR is made in anticipation of the US. In many classical conditioning procedures, such as eyeblink conditioning, this anticipation takes the form of precise timing of the CR to occur just before the onset of the US.

There are several ways to demonstrate timing of CRs. The simplest method is ISI shift. In an ISI-shift procedure, training initially occurs at one CS-US interval. When the CS-US interval is lengthened or shortened, the latency of CRs to the CS lengthens or shortens to coincide with the new CS-US interval. Another procedure involves a dual ISI. In a dual-ISI procedure, the CS-US interval varies between two values across trials. Typically, this procedure produces a double-peaked CR, with the peaks corresponding to the two CS-US intervals. Finally, timing of CRs can be demonstrated with ISI discrimination. In an ISI-discrimination procedure, two CSs that differ along some dimension are used, as in a standard discrimination procedure. However, in contrast to a standard discrimination procedure, each CS is paired with a US, but at a different CS-US interval. Mauk and Ruiz (1992) used tones of different frequencies for the two CSs and a corneal air puff as the US. Rabbits learned a CR to each CS with a latency appropriate to the CS-US interval.

Conclusion

Although classical conditioning involves learning the relations between two stimuli, these relations can be complex. The rate and content of learning in classical conditioning depends upon a number of factors. These include the following: 1. whether the CS is followed by a US or not; 2. how many CSs are present; 3. whether the organism has any previous experience with the CS, the US, or the relationship between the two; and 4. how much time elapses between the delivery of the CS and the delivery of US. In addition, the stimuli used as CSs can be as complex as an entire context. Although the responses examined in laboratory studies of classical conditioning are often simple (such as an eye blink), this is simply a control and measurement issue. Classical conditioning can involve complex learned responses as well, and is believed to play an important role in everyday human behavior.

See also:CONDITIONING, CELLULAR AND NETWORK SCHEMES FOR HIGHER-ORDER FEATURES OF; CONDITIONING, CLASSICAL AND INSTRUMENTAL; DISCRIMINATION AND GENERALIZATION; KAMIN'S BLOCKING EFFECT: NEURONAL SUBSTRATES; NEURAL SUBSTRATES OF CLASSICAL CONDITIONING: DISCRETE BEHAVIORAL RESPONSES

Bibliography

Bouton, M. E., Mineka, S., and Barlow, D. H. (2001). A modern learning theory perspective on the etiology of panic disorder. Psychological Review 108, 4-32.

Gallistel, C. R., and Gibbon, J. (2000). Time, rate, and conditioning. Psychological Review 107, 289-344.

Giftakis, J. E., and Tait, R. W. (1998). Blocking of the rabbit's classically conditioned nictitating membrane response: Effects of modifications of contextual associative strength. Learning and Motivation 29, 23-48.

Kamin, L. J. (1969). Predictability, surprise, attention, and conditioning. In B. A. Campbell and R. M. Church, eds., Punishment and aversive behavior, pp. 279-296. New York: Appleton-Century-Crofts.

Kehoe, E. J., and Weidemann, G. (1999). Within-stimulus competition in trace conditioning of the rabbit's nictitating membrane response. Psychobiology 27, 72-84.

Lubow, R. E., and Moore, A. U. (1959). Latent inhibition: The effect of nonreinforced preexposure to the conditioned stimulus. Journal of Comparative and Physiological Psychology 52, 415-419.

Mauk, M. D., and Ruiz, B. P. (1992). Learning-dependent timing of Pavlovian eyelid responses: Differential conditioning using multiple interstimulus intervals. Behavioral Neuroscience 106, 666-681.

Miller, R. R., and 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, pp. 51-92. San Diego, CA: Academic Press.

Pavlov, I. P. (1927). Conditioned reflexes. New York: Dover.

Pearce, J. M., and Hall, G. (1980). A model for Pavlovian learning: Variations in the effectiveness of conditioned but not of unconditioned stimuli. Psychological Review 87, 532-552.

Rescorla, R. A. (1969). Pavlovian conditioned inhibition. Psychological Bulletin 72, 77-94.

Rescorla, R. A., and Wagner, A. R. (1972). A theory of Pavlovian conditioning: Variations in the effectiveness of reinforcement and nonreinforcement. In A. H. Black and W. F. Prokasy, eds., Classical conditioning II: Current research and theory, pp. 64-99. New York: Appleton-Century-Crofts.

Rogers, R. F., and Steinmetz, J. E. (1998). Contextually based conditional discrimination of the rabbit eyeblink response. Neurobiology of Learning and Memory 69, 307-319.

Savastano, H. I., Cole, R. P., Barnet, R. C., and Miller, R. R. (1999). Reconsidering conditioned inhibition. Learning and Motivation 30, 101-127.

Schmajuk, N. A., Lam, Y-W, and Gray, J. A. (1996). Latent inhibition: A neural network approach. Journal of Experimental Psychology: Animal Behavior Processes 22, 321-349.

John T.Green

Joseph E.Steinmetz