4.3: Theories of Conditioning

Section Learning Objectives

• Describe Pavlov’s stimulus substitution theory.
• Describe the preparatory-response theory.
• Describe the compensatory response theory.
• Describe the Rescorla-Wagner model.
• Describe Mackintosh’s attentional model.

In this final section of Module 4, we will cover five theories of conditioning that identify the processes that underly respondent conditioning. They include the stimulus substitution theory, preparatory-response theory, compensatory response theory, Rescorla-Wagner model, and the attentional model.

Stimulus Substitution Theory

According to Pavlov (1927), respondent conditioning is a matter of substituting one stimulus with another, or the CS acts as a substitute for the US. A connection or association is established in the brain between CS and US, and when the CS is activated alone, following acquisition, it automatically activates the US portion of the cortex. The CR, therefore, is identical or nearly identical to the UR since the connection between US and UR is hardwired or innate.

As such, the presentation of food (US) to a dog activates the food center in the cerebral cortex. This, in turn, activates the salivation center in the brain which leads to the behavior of salivation (UR). Then if we introduce an NS such as a bell ringing, it activates an area of the brain responsible for processing the sound and then is followed by an US and UR as described above. This happens over a few trials (the conditioning phase). Learning has occurred if after the presentation of the bell (CS) the area of the brain which processes the sound of the bell activates the area responsible for processing the food, which activates the area responsible for salivation, and then salivation (CR) occurs. It is the simultaneous activation of the brain areas responsible for the CS and then the US that causes a new functional neural pathway to form between the active areas.

It should be noted that Pavlov was incorrect and the process is more complex than he made it seem. Timberlake and Grant (1975) tested Pavlov’s theory by conditioning rats to expect a food pellet after a brief interval across two situations differing in terms of what type of CS was utilized. In one situation, a woodblock was secured to a platform and was the CS for food, while in the other situation a live rat was secured to the platform and was the CS for food. Utilizing the stimulus substitution theory, they predicted that the rats would approach and bite the CSs that were paired with the food. The results showed that rats in the woodblock condition bit the block CS as predicted but this did not occur when a live rat was the CS. Instead, rats groomed the live rat CS. They concluded that the nature of the CS influenced the topography of the CR, contradicting the stimulus substitution theory. The specific response that was displayed was related to the form the CS took.

Recall that in the stimulus substitution theory, the CR is identical or almost identical to the UR. We know that is not always the case though. Consider a man who has panic attacks (US) which can be quite fear-evoking (UR). Flying is an NS which yields no response. If the man is flying on a plane (NS) and has a panic attack (US) which causes him a fear (UR), then in the future the mere thought of flying in a plane (CS) will cause not fear necessarily, but definitely anxiety (CR). Hence, in this example, the CR is not the same as the UR, and not even close. Fear and anxiety are different physiological and emotional reactions. Hence, the UR and CR being identical, or even close, is not always the case as Pavlov’s theory suggests.

Preparatory-Response Theory

It might be that the CR exists to prepare the organism for the presentation of the US such that a dog salivates (CR) when it hears the bell ring (CS) to prepare for the arrival of the food (US). This is called the preparatory-response theory (Kimble, 1967; 1961) and it makes up for the shortcoming of the stimulus substitution theory in terms of the UR and CR not having to be identical (or close). In the example just given, the CR and UR are virtually the same. But consider a rat that is shocked (US) and displays fear (UR). If a light (NS) signals the presentation of the shock (US) causing fear (UR), then the rat will display a freeze behavior (CR) when the light turns on (CS) as it expects the shock to follow.

Compensatory Response Model

Not only can the UR and CR be different, the CR can be the direct opposite of the UR. In the compensatory-response model, and building off the opponent-process theory of emotion (Solomon & Corbit, 1974), a CS that has come to be repeatedly associated with the a-process or primary response to a US will with time, elicit a compensatory response or b-process. Evidence for this process comes from Siegel (1972) who gave rats repeated injections of insulin which reduces the level of glucose in the blood. He tested the CR by giving the rats an injection of saline in place of insulin. The results showed that a strong CR did occur, but it was the opposite of the reaction to insulin. The rats showed an increase in blood glucose levels (hyperglycemia CR). The CR and UR were not the same and the CR was compensatory.

Consider drug tolerance. Morphine, as a US, causes the UR of analgesia, or a reduced sensitivity to pain. Siegel et al. (1978) found that the CR to stimuli paired with morphine such as lights or tones, is hyperalgesia, or an increased sensitivity to pain. In the study, Siegel placed a rat’s paw on a hot plate and measured latency in terms of how long it took the rat to pull its paw off the plate. He found that rats injected with morphine took longer to remove their paws compared to rats which did not receive the injection. The rats which had a stimulus such as a tone paired with morphine removed their paws quicker than rats that had a stimulus not paired with morphine (the US).

Rescorla-Wagner Model

Robert Rescorla and Allan Wagner (1972) developed an associative model of respondent conditioning built on the idea that a given US can only support so much conditioning and must be spread out among the CSs that are present. Four main ideas are captured in this model.

1. There is a maximum associative strength that can develop between a US and CS. This is determined by the US and different US support varying maximum levels of conditioning. Stronger stimuli, therefore, support more conditioning such that if a US is a favorite food such as chicken, it will produce more salivation in an organism than a less preferred food such as Brussel sprouts.
2. Associative strength goes up with each trial, though the amount of associative strength gained on a trial is a function of the level of prior training. In general, more associative strength is gained in early trials.
3. Associative strength will accumulate quickly to some stimuli and slowly to others and some USs will produce rapid learning compared to others.
4. A specific US can only support a certain degree of conditioning even when paired with more than one stimulus. The addition of each stimulus beyond the first means that the US must share associative strength across all stimuli. Let’s say a US has 15 maximum associative strength units. If 11 of these units are distributed to Stimulus A, then only 4 can be shared with a Stimulus B. This would be particularly true in the case of a compound stimulus. A would obviously be more salient than B, or recall from earlier, A overshadows B.

The Rescorla-Wagner model also does a good job explaining blocking. Recall our earlier example of a dog presented with a tone (NS) and food (US) leading to salivation (UR). Acquisition is complete when the tone (CS) causes salivation (CR). The US of food has a maximum associative strength of 8 units which is transferred in full to the tone (CS) upon acquisition. If we now introduce an NS of a green light presented simultaneously with the tone (CS) to form a compound stimulus followed by food (US) which causes salivation (UR), then when the tone and green light are tested separately, the tone will cause salivation as it has the associative strength of 8 units assigned to it and the light will cause no response since no associative strength is left to be assigned to it.

Mackintosh’s Attentional Model

Nicholas Mackintosh (1975) presented a very simple theory of conditioning centered on the concept of attention. The attentional model states that how much attention an organism will give a CS is dependent on how well the CS predicts the US. If it is a good predictor, we will pay attention to the CS. If it is not a good predictor, our attention will decline. When attention is high, learning will be high as well. Organisms will pay more attention to the best predictor of the US and less attention to weaker predictors during conditioning. The theory explains blocking by saying that though the US was able to bring about learning to both the light and the tone, the animal paid less attention to the green light than it did to the tone.