Week-5 The competing CSs.....

We read about classical conditioning and its way of association. In this post, we can look at Kamin’s blocking paradigm, which shows how the association can occur when there’s more than one conditioning stimulus. When there’s more than one conditioning stimulus, there will be a competition of CSs, and the degree of association can be varied. Kamin’s paradigm describes a way to understand this degree of association of CS to the US. This degree of association can depend upon salience, but here we discuss the case when one stimulus is learned before the other. In such a situation, the second stimulus is not likely to get associated. For a stimulus to associate with the US, it has to bring some new information, but if a previous one already helps in predicting the outcome without error, then there is no new information associated with it.

fig 1: representing Kamin's blocking paradigm

Leon Kamin trained rats to predict shock with a CS -light and then coupled the light with a tone. He noticed that the rats learned less about the tone because the tone didn’t provide any new information as CS. The prediction error was less with the light as CS, so they didn’t learn any association with the CS tone. From fig 1, we can see the difference between a control group and an experimental group, where the experimental group is pre-trained with light as CS. The control group shows medium CR for either of the CS, but the experimental group shows little or no CR for the tone CS.

The next paradigm which explains the degree of association is the Rescorla- Wagner model, which explains the connection strength in the case of association. This model explains conditioning through the error correction principle, where the association takes place to reduce the prediction error. The previous blocking paradigm can be explained using the Rescorla-Wagner model as follows:

fig 2 : blocking in Rescorla-Wagner model

Here the desired output is 1, so the system learns till the actual output is also 1. In the first case, the tone predicts the CR with an actual output equal to the desired output (association strength= 1). Then the tone is coupled with a light stimulus, where the association strength of light doesn’t increase from 0 (no learning). In the third, the light CS is present alone, and it doesn’t give an output of 1, which shows that the system didn’t learn that association, and now the error is 100%; the system will update the weight of association of light CS to predict the output (Gluck et al., 2008).

References

fig 1 and fig 2 retrieved from :

Gluck, M. A., Mercado, E., & Myers, C. E. (2008). Learning and memory: From brain to behavior. Worth Publishers.