Rescorlas Contingency Model of Classical Conditioning

Pavlov viewed conditioning from the perspective of a physiologist, leading him to form a mechanistic interpretation of the cognitive and emotional dynamics governing the process. Rescorla questioned Pavlov's contiguity theory of classical conditioning and posited an alternative account that emphasized the importance of contingency:

The notion of contingency differs from that of pairing in that it includes not only what events are paired but also what events are not paired. As used here, contingency refers to the relative probability of occurrence of US in the presence of CS as contrasted with its probability in the absence of CS. The contingency notion suggests that, in fact, conditioning only occurs when these probabilities differ; when the prob ability of US is higher during CS than at other times, excitatory condition occurs; when the probability is lower, inhibitory conditioning results. Notice that the probability of a US can be the same in the absence and presence of CS and yet there can be a fair number of CS-US pairings. it is this that makes it possible to assess the relative importance of pairing and contingency in the development of a CR. (1968:1)

Rescorla interprets conditioning from a cognitive viewpoint attributing both predictive and informative properties to the CS. The model places equal importance on the presence as well as the absence of the CS in relation to the occurrence of the US. According to Rescorla, associative conditioning depends on a predictive contingency (both positive and negative) holding between the CS and US. If the US occurs regardless of the presence or absence of the CS (i.e., the US occurs independently of the CS), then in spite of many chance pairings between the CS and US (all being offset by an equal number of US events occurring without the CS), no effective conditioning takes place. Under conditions in which the US occurs indepen dently of the presence or absence of the CS, the CS is neutralized (Rescorla, 1967). Rescorla's important discovery suggests that classical conditioning is a contingency-based process in which the CS functions as a statistically informative signal about the probability of the occurrence or nonoccurrence of the

As a supplement or correction to the contiguity theory, the contingency theory provides a coherent and elegant way to describe what takes place during classical conditioning. Besides predicting the occurrence of the US, the CS also provides information about the type and size (magnitude) of the anticipated UR, as well as various significant contextual relations between the occurrence of the CS and CR. But, as Rescorla writes, "It is not only temporal and logical relations among events that are important to conditioning. Conditioning is also sensitive to relations involving the properties of the events themselves" (1988:153).

Formulating predictions about such information requires that the CS be somehow as-sociatively linked with the US eliciting the UR. The so-called stimulus-stimulus (S-S)

Contingency Theory Given

Fig. 6.4. Probability (p) space describing excitatory and inhibitory conditioning. CS, conditioned stimulus; US, unconditioned stimulus.

Fig. 6.4. Probability (p) space describing excitatory and inhibitory conditioning. CS, conditioned stimulus; US, unconditioned stimulus.

theory of classical conditioning asserts that the connection between CS and US events is mediated by control centers in the brain, perhaps corresponding to Gray's septal-hip-pocampal comparator system, "a system which, moment to moment predicts the next likely event and compares this prediction to the actual event" (Gray, 1991:112) (see Chapter 3).

Predictions about the size of the US are estimated along an excitatory-inhibitory dimension. If the CS underestimates the size of the pending US, excitatory learning takes place (acquisition). If the CS overestimates the size of the US, inhibitory learning occurs (extinction). If the CS accurately estimates the size of the US, no additional learning takes place (steady state or homeostasis). Classical conditioning is acquired, maintained, or extinguished on the basis of a variable correlation between a predictive CS and a corresponding US. Acquisition or extinction occurs when a dog's expectation of a pending event is different from what actually happens. Regarding this relationship, Rescorla and Wagner write,

Organisms only learn when events violate their expectations. Certain expectations are built up about the events following a stimulus complex; expectations initiated by that complex and its component stimuli are then only modified when consequent events disagree with the composite expectation. (1972:75)

This cognitive view of conditioning is in sharp contrast to the emphasis traditionally placed on factors such as repetition and forward contiguity between associated CS-US events. Although factors like these are important, they are not sufficient alone to explain the laboratory findings reported by Rescorla and other contemporary investigators studying classical conditioning.

Information Provided by the Conditioned Stimulus About the Unconditioned Stimulus

As already discussed, more information is derived from the regular concurrence of the CS and US than simply the probability of the US. Besides calculating event probability, clas sical conditioning also yields information about the size and type of anticipated stimulation. According to Rescorla, the size or magnitude of the CR depends on the associative strength acquired by the CS together with the stimulus intensity of the original US. For instance, a CS paired with an electric shock will yield a stronger avoidance response than a similar CS paired with a light slap on the hands. Additionally, the magnitude of the CR is influenced by the salience of the eliciting CS. For instance, a softly spoken reprimand will yield only a small response from a dog, whereas the same signal spoken more loudly will elicit a correspondingly larger effect.

The context or situation where the CS occurs has a significant bearing on the magnitude of the CR elicited. Dogs, like children, can easily discern that "No" in one situation does not necessarily mean the same thing as it does in another. Dog owners exhibit predictably different behavior regarding the application of punishment, depending on the social milieu current at the time of the offending misbehavior. Dogs learn that "No" when guests are around only infrequently leads to the actual occurrence of the threatened outcome—an event that would more likely occur if guests were not present. Under such conditions, guests represent a safety signal informing dogs that the warning will not likely be followed by actual punishment. The lesson dogs learn here is that displaying unwanted behavior in the presence of guests is safe. Such mixed messages and differential treatment lead dogs into a frustrating and confusing game of probabilities and risk.

An interesting effect of context can be observed by comparing the speed and ease of acquisition taking place in a familiar environment versus an unfamiliar environment. New learning is most easily introduced within a familiar environment. However, at the point where the learning curve begins to flatten, further (sometimes dramatic) progress is easily achieved by moving the training activity into less familiar surroundings. This observation supports the opinion of many professional trainers that introductory training should be carried out first in the home and subsequently reinforced in a group setting.

Assumptions Derived from the Rescorla-Wagner Model

Classical Conditioning (S-S Theory)

Defined: Learning about stimuli or signals predicting the occurrence or nonoccurrence of significant events.

Three possibilities exist for each presentation of the CS:

1. The CS becomes excitatory.

2. The CS becomes inhibitory.

3. The CS exhibits no change.

The model attributes significance to an animal's expectations regarding anticipated stimulation, especially with respect to predictions about the occurrence or nonoccurrence of the US. However, the CS also makes predictions about the impending US, including its relative salience or intensity:

1. If the US is larger (i.e., more attractive or aversive) than expected, then excitatory conditioning of the CS occurs.

2. If the US is smaller than expected, then inhibitory conditioning of the CS occurs.

3. If the US is identical to the animal's expectation, then no additional conditioning takes place.

These predictions generate the following hypotheses concerning the S-S theory of learning:

1. An animal's ability to form accurate expectations regarding the size or intensity and type of the US event presumably entails that the CS and US are centrally linked through associative and cognitive processes. Through conditioning, a neural link or pathway is produced between the CS center (e.g., auditory center in the case of tone stimuli and visual center in the case of light stimuli) and the US center (appetitive center in the case of food and fear center in the case of aversive stimulation).

2. The strength of association between the CS and US is relative to the size or intensity of the expected US. For example, the word "Good" (CS) paired with a large and delicious portion of food (US) will generate a stronger associative link between the CS

"Good" (auditory center) and US food (appetitive center) than if the US presented were a small bit of stale bread. Of course, the relative effect of US size and type on associative strength will depend on the animal's degree of deprivation or satiation, as well.

3. The size or intensity of the US ultimately determines the strength or weakness of the CS-US association. When conditioning is complete (asymptotic), the strength of the association will be directly proportionate to the size or intensity of the US.

Example 1: CS (light) is paired with shock (US)

Characteristics of the US: The associative strength (S) supportable by the US at asymptote is arbitrarily denoted as superscript 1 (i.e., the amount of shock delivered). S1, therefore, represents the actual size of the US (shock stimulus) presented.

Characteristics of the CS: The expectancy (E) is derived from the associative strength existing between the CS and US, that is, between light (L) and shock (S1). E(L) represents an expectation that has been formed by the association of the CS (light stimulus) occurring regularly and contiguously with the US event. Over the course of conditioning, predictions made by the animal [E(L)] will gradually come to approximate or match the actual US event (S1).

Example 2: Pairing a compound CS (light and tone) with a US,

E(L) = the associative strength of the light stimulus

E(t) = the associative strength of the tone stimulus

Over the course of several conditioning trials in which E(L) and E(T) are presented together in the presence of shock, both stimuli will increase in associative strength. However, neither the light CS nor the tone CS will independently progress to the associative strength supported by shock (S1). In the case of compound conditioning, the sum of the two, that is, E(L) + E(T), upon reaching asymptote, will approximate the associative strength supportable by shock.

1. If the auditory CS (tone) and the visual CS (light) are equally salient at the onset of conditioning (i.e., both stimuli elicit an equal orienting response), then the respective associate strengths E(L) and E(T) relative to the US will increase at an equal rate as conditioning progresses.

2. If one CS is weaker or less salient (e.g., a dim light versus a loud tone), the stronger of the two stimuli will obtain more associative strength relative to the US. Nonetheless, at asymptote, the sum of E(L) and E(T) will approximate, but not exceed, the value of S1.

Acquisition, Extinction, and Asymptote (Fig. 6.5)

1. Acquisition occurs when S (associative strength supportable by the US) is greater than E (CS expectancy)—that is, the US is underpredicted by the CS, resulting in excitatory conditioning (the CS increases in associative strength relative to the US).

2. Extinction occurs when S is less than E—that is, the US is overpredicted by the CS, resulting in inhibitory conditioning (the CS decreases in associative strength relative to the US).

3. Asymptote occurs when S is equal to E—that is, the US is well predicted by the CS, resulting in no additional conditioning

Fig. 6.5. Relationship between expectancy and classical conditioning. CS, conditioned stimulus; US, unconditioned stimulus.

(the associative strength of the CS is verified relative to the US).

100 Dog Training Tips

100 Dog Training Tips

100 Dog Training Tips EVERY Dog Owner Should Know. Utilize These Tips to Train Your Dog to Be Obedient and Smart.

Get My Free Ebook


  • hannes
    What are contiguity theory and contingency theory?
    8 years ago
    What warning does rescorla have about learning curves?
    8 years ago
  • joanna
    What is classical conditioning Rescorla's modern based on?
    7 years ago
  • dieter
    7 years ago
  • maarit
    What modern conceptualization of classical conditioning is rescorla idea based on?
    4 years ago
  • alison
    What is contingency model rescola?
    3 years ago
  • halfred sandyman
    What does robert rescorlas contingency model state?
    2 years ago
  • darnell
    How does Robert Rescorla's model for classical conditioning differ from Ivan Pavlov's?
    2 years ago
  • elsie
    What does contingency modelhave to do with classical conditioning?
    11 months ago
  • luam
    What was rescorla's addition/contributin to classical conditioning?
    5 months ago

Post a comment