A Description And Critical Evaluation Of The Rescorla-wagner Theo

  • Category: Psychology
  • Words: 1490
  • Grade: 100
The Rescorla-Wagner theory was the first and most prominent attempt to explain how, in classical conditioning, animals are able to evaluate which of many conditioned stimuli is the best predictor of the unconditioned stimulus (US). The main problems facing psychologists attempting to account for this centre around the mechanisms by which animals are able to observe and remember all the conditioned stimuli they are faced with and then calculate which conditioned stimulus (CS) has most frequently predicted the US. Problems have also remained surrounding explanations of phenomena such as blocking and overshadowing in the conditioning process and the theory sought to explain these and develop a formula by which psychologists can predict the behaviour of subjects in new conditions.
Central to the Rescorla-Wagner theory is the assertion that conditioning strength is the key factor in determining which CS is successfully paired with the US to produce the conditioned response (CR). In their 1972 article, "Variations in the effectiveness of reinforcement and non-reinforcement," Robert Rescorla and Alan Wagner investigate the levels of conditioning strength along the various points of the conditioning curve. They point out that in the early stages of training levels of conditioning strength increase considerably and, as training progresses, these differences reduce to a point at which they stop - the asymptotic level of conditioning.
Rescorla and Wagner explain the changes in increases in conditioning strength partly by means of the amount the subject has learned and has yet to learn about the task. In the initial stages of training, the subject has had no prior training from the task and so the conditioning strength increases dramatically. As the experiment goes on, and the subject's prior training level increases, thus the increase in the level of conditioning strength decreases as the subject has less to learn. Rescorla and Wagner also noted that the rate of conditioning varied depending on which US and CS was being presented and the level of previous conditioning to other CSs with the US. The latter factor was explained by the assertion that the US can only be condition to a certain level after which it can improve no more, even if another CS is introduced; with the implication that if more than one CS is presented the sum of the associative strengths of each stimulus must be no more than the maximum associative strength sustained by the US.
These observations were used for a mathematical formula designed to predict behaviour using the values of associative strength, prominence of CS and US and maximum level of conditioning. These values are entered into an equation conforming to the laws described in the previous paragraph, that is, that the change in rate of conditioning strength is equal to the nature of the CS and intensity of the US multiplied by the maximum level of conditioning minus the level of conditioning already achieved with the CS and other stimuli.
This formula explains animals' ability to be conditioned when a CS consistently predicts the CS rather than in random presentations by highlighting the conditioning strengths of the other stimuli in the environment at the time of conditioning. These strengths block the condition of the CS to the US when the stimuli are presented randomly. When the CS reliably predicts the US, however the associative strength is such that it will be successfully conditioned to the US providing that the US has not already been conditioned to the asymptotic level.
The phenomena of blocking is explained similarly by the blocking stimulus acquiring an associative strength near enough to the maximum that the US can not support enough conditioning for the CS to be associated and in overshadowing, the associative strength acquired by the most salient stimulus is too great for there to be enough left to condition to the less salient stimulus.
Evidence for the Rescorla-Wagner has come mainly from tests of their assumptions of maximum associative strength. An Experiment by Randich and Ross in 1985 attempted to test the hypothesis that pre-exposure of the subject to US would result in the surrounding stimuli, or context, acquiring an associative strength of their own and therefore blocking any attempted CSs from being successfully paired with the US as most of the maximum conditioning strength would be taken up.
The experiment involved two contexts, one involving a fan and a relatively neutral environment and the other with no fan, but a smell of pine, a light and black-and-white striped walls. Randich and Ross found that when a group of rats were pre-exposed in the first context they were conditioned to the CS (a light) with the US (a shock) much slower than a group who had been placed in the first context but not received the shock. Rats pre-exposed in the second context, however were conditioned rapidly when the re-exposure to them of the shock US was paired with the light CS. The results of this study are consistent with the idea that the conditioning strength had reached a higher level after US pairing with a fan stimulus than with the less salient stimuli and that reaching this level would be responsible for the slower changes in conditioning strength during the second exposure, however it does not provide convincing evidence for the entire theory as does not wholly demonstrate the theory, it is merely consistent with it.

P. Durlach carried out another piece of evidence relating to the theory in 1983, testing the assertion by Rescorla and Wagner that context stimuli block the attempted CS during random presentation. Two pigeons were used in a conditioned key pecking experiment where both groups had the CS and US paired randomly, but in the second group, a tone was paired with food in order to use up some of the associative strength quotient and therefore block the context from being associated with the context. The results showed this to have succeeded and the experiment was used as evidence to support the associative strength theory against theories involving mental calculation as such processes would have resulted in the context being successfully paired with the US as it was equally predictive. This is not wholly convincing, however, as computational theories can be adapted to postulate limited ability of animals to recognise numerous predictors. Experiments on a wider range of animals with different intellectual capacities may be useful in appropriately interpreting the results of this study.                        

The aforementioned studies have shown results consistent with some predictions of the theory. However in order for the theory to be accepted, all of the assumptions related to the values used in the formula must be shown to be correct. A problem with the theory passing this test has been the assumption, inherent in having a value for the saliency of the CS, that the saliency of the conditioned variable cannot be altered during the course of the training. In the book "Psychology of Learning and Behaviour," Barry Schwarz and Steven Robbins cite a number of studies, which they claim demonstrate this is not the case. They point out that "the reliable findings in experiments of this type [involving rats pre-exposed to the tone (CS) is that pre-exposure to the CS significantly slows conditioning." This "latent inhibition effect", along with other findings such as subjects learning that the CS no longer predicts the US deals a powerful blow to this aspect of the formula itself, but does not undermine the assertions regarding the maximum strength of conditioning to a US and, as Schwarz and Robbins point out, an alteration of the formula to account for changes in CS salience would be enough to make this evidence compatible with the theory.
The most challenging evidence to the Rescorla-Wagner theory comes from research that finds alternative explanations for the results seen in evidence supporting the theory. Baker's retrospective processing theory, for instance, found evidence to support his claim that evidence conditioning found in previous experiments such as Durlach's may have been due to factors other than blocking, such as recognising the infrequency of the CS to predict the US. The Comparator theory and the within-compound analysis have also found alternative explanations of phenomena predicted by the Rescorla-Wagner formula.

The Rescorla-Wagner theory can therefore be seen to be a flawed but highly important explanation of classical conditioning. Experiments highlighting the variance in saliency of the CS have made alterations to the formula necessary but these would not contradict the main thrust of the theory, the importance of the availability of conditioning strength. Although there have been other explanations of the processes involved in classical conditioning, the Rescorla-Wagner theory remains the best predictor of events while remaining true to the reductionist simplicity of Pavlovian explanations of behaviour.

Word count (excluding references): 1440 words

Klein, S.B. (1996) Learning: Principles and Applications (pp101-111) US: Mcgraw-Hill.

Schwartz, B, & Robbins, S.J, (1995) Psychology of Learning and Behaviour. (pp.113-g 129)

ad 4
Copyright 2011 EssayTrader.net All Rights Reserved