The Rescorla-Wagner Model: what it is and how it explains learning.

An explanation of one of the facets of learning, based on classical conditioning.

In 1972, Robert A. Rescorla and Allan R. Wagner proposed a theoretical model very relevant to the psychology of learning. It is the Rescorla-Wagner Modelbased on classical conditioning and the concept of surprise.

Here we will know the characteristics of this model and what it tries to explain.

Rescorla-Wagner model: characteristics

The Rescorla-Wagner Model was presented in 1969 at a series of conferences on classical conditioning in Canada. It is the most influential theory of classical conditioning, and the one that has generated the most research.

The authors, initially was given the name Variations in Reinforcement and Non-Reinforcement Effectiveness, but was later recognized as the Rescorla-Wagner Model (1972).but was later recognized as the Rescorla-Wagner Model (1972).

The model is based on classical or Pavlovian conditioning, beyond associative learning of contingent stimuli. The aim of the Rescorla-Wagner Model is to predict and describe the changes (trial-by-trial) in the associative strength that binds a stimulus (or more than one) to a stimulus. that links one (or more) conditioned stimulus with the unconditioned stimulus.

The central idea of the model is the competition that occurs between various stimuli for association with the unconditioned stimulus. In addition, it highlights other concepts that we will see below.

Associative strength

In the model, conditioning is presented as a variation in the associative strength that relates conditioned stimuli to unconditioned stimuli. The essential parameters are their respective intensities or their "salience". (central concept of the model).

This associative strength is an intervening or intermediary variable, which integrates stimuli and responses. It is mathematically inferred through the measurement of the conditioned response.

On the other hand, the associative strength is limited (values from 0 to 100). Once an EI is no longer surprising because it is already predicted by a CE with 100% certainty (associative strength of 100), there is no need to continue learning, not even to predict it with another CE.

Surprisingness of the unconditioned stimulus.

According to the model, conditioning or learning will take place when the unconditioned stimulus (UI) is unexpected (surprisingness of the UI).. That is, the animal learns about an EC when the EI is unexpected.

Thus, if the IE is surprising, the animal will pay attention to the conditioned stimuli that precede it, i.e., it will learn to be able to better predict in the future that the IE is approaching, based on the ECs. According to Rescorla and Wagner, this learning is a tremendously useful capacity for survival in animals.

In contrast, if the IE is not unexpected (does not cause surprise), it will not be possible to continue learning..

If we relate surprise with the associative strength already mentioned, we know that the more surprising an EI that appears after the EC, the less associative strength that EC has with that EI (because if it surprises us, it is because we expected that the EI was not going to appear). In other words, the EI gives the associative strength to the CE as a function of the surprise.

How does learning occur?

Thus, as we have seen, classical conditioning is produced as a consequence of the quantitative variations in the associative strength between conditioned (CEs) and unconditioned (UIs) stimuli..

Variations depend on a positive or negative discrepancy between the associative strength that the organism has at a given moment and the true association that is present in the environment between the CSs and the ISs.

These variations consist of changes that the greater they are, the more conditioning or learning they will produce.

Later model: Mackintosh's theory

The first theory to emerge as a competitor of the Rescorla-Wagner model was Mackintosh's theory of attention. Mackintosh's theory of attention (1975). This theory does not assume that learning depends on the discrepancy in associative strength between a conditioned stimulus and a constant value of the unconditioned stimulus.

Instead, it posits that the value of the conditioned stimulus is not constant for the organism but changes through experience.

Mackintosh's theory states that pre-exposure to a stimulus hinders its subsequent conditioning (the appearance of the conditioned response). (the appearance of the conditioned response). That is, animals, if exposed to a CE prior to conditioning with the IE, end up "interpreting" that CE as irrelevant.

Mackintosh also suggested that animals try to get information in the environment that allows them to predict the occurrence of biologically relevant events (EI's).

Bibliographical references:

• Pérez-Acosta, A. (2001). The Rescorla model: Wagner at twenty. Theory and foundations. Revista psicologíacientífica.com
• Pérez-Acosta, A, Rozo, J. and Baquero, H. (2003). Milestones of the molar perspective of classical conditioning. Psicología desde el Caribe, 12, 2-12.