The Rescorla-Wagner Model
Error-correction learning updates associative strength according to prediction error.
When Prediction Leaves Nothing New
Suppose an animal has already learned that one cue predicts an outcome. Later, that familiar cue appears together with a new cue, and the same outcome occurs. Will the new cue automatically become a strong predictor? The Rescorla-Wagner model says that the answer depends on what remains surprising. If the familiar cue already accounts for the outcome, the compound produces little new prediction error. The new cue therefore receives little or no additional associative strength.
What do you think happens?
A familiar cue already predicts the outcome. What is likely to happen when a new cue is added and the same outcome still occurs?
Reveal answer
Answer: The new cue receives little or no additional learning.
The familiar cue already predicts the outcome, so the actual outcome creates little or no new prediction error. In the Rescorla-Wagner account, low error means little reason to increase the new cue's associative strength.
Learning Through Prediction Error
The Rescorla-Wagner model treats conditioning as error-correction learning. On each trial, an animal has a prediction based on previously learned associative strengths. The actual outcome is then compared with that prediction. The difference between them is prediction error. That mismatch determines how much further updating is justified. A large mismatch provides a basis for substantial change. When the prediction already accounts for the actual outcome, the error can be zero or low, so little additional learning occurs.
A Conditioning Trace Across Trials
Consider a simple generated scenario. At the beginning, a cue has little or no prior conditioning history. An outcome occurs after the cue, so the outcome is not yet fully accounted for by the cue's prediction. The mismatch creates a basis for increasing the cue's associative strength. On later trials, the cue contributes to a stronger prediction. If the same outcome continues to occur, the remaining error becomes smaller, and the amount of further learning becomes smaller as well.
From Initial Learning to Reduced Error
Trace what happens when a cue is repeatedly followed by the same outcome.
Initial trial: The cue has limited prior associative strength, so the outcome is not fully predicted. The difference between prediction and outcome creates prediction error.
Update: Because error is present, the cue's associative strength is adjusted in the direction that reduces the mismatch.
Later trial: The cue now supports a stronger prediction of the outcome. If the outcome is unchanged, less error remains.
Consequence: With less prediction error, the rule provides less basis for additional learning.
Learning is strongest when the outcome is not already accounted for, and it becomes limited when prediction and outcome increasingly agree.
Kamin Blocking in a Compound Cue
Kamin blocking is a two-stage conditioning situation. In the first stage, one stimulus is repeatedly involved in conditioning with an outcome. In the second stage, that previously conditioned stimulus appears together with a new stimulus, followed by the same outcome. The new stimulus is the added component of the compound. According to the Rescorla-Wagner explanation, the prior cue already predicts the outcome. The compound therefore introduces no new surprise, or only a small amount. The prediction error is zero or low, so the new component gains little or no additional associative strength.
Error Correction and LMS Learning
| Rescorla-Wagner model | Least Mean Square learning rule |
|---|---|
| Associative strengths are adjusted according to prediction error. | Weights are adjusted to reduce error. |
| Learning depends on the mismatch between the predicted outcome and the actual outcome. | Learning depends on an error signal used to guide weight adjustment. |
| A well-predicted outcome produces little new learning. | A reduced error produces a smaller basis for further adjustment. |
The connection to the Least Mean Square learning rule is conceptual and structural. In the Rescorla-Wagner model, associative strengths are adjusted to reduce prediction error. In LMS learning, weights are adjusted to reduce error. Both accounts therefore make the error signal central: the current prediction is compared with what actually occurs, and the adjustable quantities are changed in response. The source does not provide a formula or numerical update rule here, so the comparison should be understood as a correspondence between learning ideas, not as an equation to calculate.
Surprise, Reinforcement, and Historical Context
A central idea associated with the Rescorla-Wagner model is that animals learn when they are surprised. This emphasis on surprise helps explain why adding a stimulus does not always produce new learning. If earlier conditioning has already prepared the animal for what will happen, the added stimulus may contribute little or no additional learning. Kamin blocking provided an important historical basis for this way of thinking.
The word reinforcement has changed meaning across historical traditions in animal learning. According to the source, the term first appeared, to the best of its knowledge, in the 1927 English translation of Pavlov's monograph. In that use, it referred to a non-action-contingent case, meaning that the event was not described as depending on an action. Later, Mackintosh proposed using reinforcement for either strengthening or weakening a pattern of behavior. Skinner used the term only for strengthening behavior and treated weakening as something produced by punishment. These differences matter because the same word can point to different ideas in different traditions.
| Account or tradition | What the source establishes |
|---|---|
| Rescorla-Wagner | An error-correction account linking learning to surprise and prediction error. |
| Kamin's blocking work | A historical basis for examining how prior conditioning affects later learning. |
| Klopf | Listed as an alternative model associated with classical conditioning. |
| Grossberg | Listed as an alternative model associated with classical conditioning. |
| Mackintosh | Listed as an alternative model; also associated with a distinct historical use of reinforcement. |
| Moore and Stickney | Listed as alternative models associated with classical conditioning. |
| Pearce and Hall | Listed as an alternative model associated with classical conditioning. |
| Courville, Daw, and Touretzky | Listed as alternative models associated with classical conditioning. |
Mistakes About Blocking and Error
Assuming that every stimulus present during an outcome must gain substantial associative strength.
The Rescorla-Wagner model ties learning to prediction error, not merely to the presence of a stimulus during the outcome.
Fix:
Ask whether the compound creates a new mismatch between prediction and outcome. If prior learning already predicts the outcome, Cue B may receive little or no additional strength.Treating each conditioning trial as an isolated event.
Prior conditioning changes the prediction available on the later trial.
Fix:
Trace the learning history across trials before deciding how much new learning should occur.Describing blocking as proof that the added cue was not perceived.
The source states that the model's logic does not require the new component to be ignored perceptually.
Fix:
Explain blocking through low prediction error: the outcome was already accounted for by prior learning.Treating surprise as a vague emotional reaction rather than as a mismatch between prediction and outcome.
The model explains learning through the difference between the predicted outcome and the actual outcome.
Fix:
Name the prediction, name the actual outcome, and then describe the remaining error.Claiming that the source provides a numerical Rescorla-Wagner or LMS calculation for this topic.
The source presents a conceptual trace and explicitly does not provide a formula or numerical update rule here.
Fix:
Explain the direction and role of error without inventing numerical values or an unsupported equation.
Apply the Trial-by-Trial Logic
A cue has already undergone prior conditioning and predicts an outcome. A second cue is then presented together with the first cue, and the same outcome occurs. Explain what the Rescorla-Wagner model predicts for the second cue. Your answer should identify the prior prediction, the actual outcome, the resulting prediction error, and the expected change in the second cue's associative strength.
Hints
- Begin with the learning history of the first cue.
- Compare the outcome produced during the compound presentation with the prediction already supported by the first cue.
- Use prediction error to explain why the second cue receives little or no additional learning.
A Model Answer for the Blocking Scenario
Explain the result of adding Cue B to a previously conditioned Cue A when the same outcome follows.
Prior conditioning: Cue A has already been involved in conditioning and therefore predicts the outcome.
Compound presentation: Cue A and Cue B appear together, followed by the same outcome.
Prediction error: Because Cue A already accounts for the outcome, the compound creates no new surprise or only a small amount of surprise.
Learning about Cue B: The low prediction error provides little basis for increasing Cue B's associative strength.
Prior conditioning blocks substantial later learning about the added cue. This is Kamin blocking as explained through the Rescorla-Wagner error-correction rule.
Key Takeaways
- The Rescorla-Wagner model explains learning as error correction: associative strength changes according to prediction error.
- Learning is supported by a mismatch between the predicted outcome and the actual outcome.
- When an outcome is already predicted, the remaining error is low, so little additional learning occurs.
- Kamin blocking occurs when prior conditioning of one cue limits learning about a newly added component of a compound cue.
- The model shares a central idea with Least Mean Square learning: adjustable quantities are changed to reduce error.
- The model is part of a broader history that includes changing meanings of reinforcement and alternative conditioning models.
Key Takeaways
- The Rescorla-Wagner model makes prediction error the driver of conditioning.
- A fully predicted outcome produces little new learning because little error remains.
- Kamin blocking shows how prior conditioning can prevent substantial learning about an added cue.
- The model is conceptually related to the Least Mean Square learning rule through error-reducing updates.
- Its emphasis on surprise belongs to a wider history of reinforcement and alternative theories of animal learning.