Advanced Theories of Learning and Behavior
Associative Learning Models
The Rescorla-Wagner Model
The Rescorla-Wagner model revolutionized classical conditioning by formalizing the concept of prediction error. Learning only occurs when a US is surprising. The change in associative strength is defined mathematically as:
Variables in the Rescorla-Wagner Model
Which phenomenon does the Rescorla-Wagner model successfully explain by demonstrating that a previously conditioned CS prevents learning to a novel CS presented in compound?
Blocking occurs because the original CS already perfectly predicts the US, reducing the prediction error to zero, thus preventing new learning to the novel CS.
Correct answer: Blocking
What term describes the discrepancy between the expected outcome and the actual outcome in learning models?
Prediction error is the core driver of learning in associative models; it is the difference between what an organism expects to happen and what actually happens.
Correct answer: Prediction error
Rescorla-Wagner vs. Pearce-Hall Models
Rescorla-Wagner Model
- Learning is driven by US processing (prediction error).
- Assumes constant CS associability.
Pearce-Hall Model
- Learning is driven by CS processing (attention).
- Assumes CS associability changes based on past predictability (more attention to uncertain predictors).
Machine Learning Connections
The principles of prediction error formulated in the Rescorla-Wagner model directly laid the groundwork for Temporal Difference (TD) learning algorithms used in modern reinforcement learning and artificial intelligence.