lesson 4

Question 1

Rescorla-Wagner Model (RW)
Q1: What is the purpose of a learning theory in psychology?

Answer 1

To explain or generate learning behavior using a set of rules, often expressed mathematically.

Question 2

What type of conditioning does the Rescorla-Wagner model explain?

Answer 2

Pavlovian (classical) conditioning only.

Question 3

What is the main idea of the RW model?

Answer 3

What is the main idea of the RW model?

Question 4

What does the RW equation describe?

Answer 4

: It describes how associative strength (V) changes over time as learning occurs.

Question 5

What is associative strength (V)?

Answer 5

A measure of how strongly a conditioned stimulus (CS) is associated with an unconditioned stimulus (US).

Question 6

What does ΔV represent in the RW model?

Answer 6

The change in associative strength; how much learning happens on a given trial.

Question 7

What does α (alpha) represent in the RW equation?

Answer 7

The salience of the CS – how noticeable or attention-grabbing it is.

Question 8

What does β (beta) represent in the RW equation?

Answer 8

The salience of the US – typically set to 1 in most experiments.

Question 9

What does λ (lambda) represent in the RW equation?

Answer 9

The intensity or value of the US – what actually happens in the world.

Question 10

What is the equation used in the RW model?

Answer 10

ΔV = αβ(λ − V)

Question 11

What does the term (λ − V) represent in the equation?

Answer 11

Prediction error – the difference between what happens (λ) and what is expected (V).

Question 12

When does learning occur according to the RW model?

Answer 12

When prediction error is large; the more surprise, the more learning.

Question 13

What happens when λ = V in the RW model?

Answer 13

Prediction error is zero; ΔV = 0, so no learning occurs.

Question 14

What does the RW model say about multiple CSs?

Answer 14

Each CS has its own associative strength (e.g., VA, VB, etc.), which is updated individually.

Question 15

What does the term “asymptote of learning” mean in this context?

Answer 15

It’s the maximum level of associative strength (V) the CS can reach, which equals λ.

Question 16

What is the performance rule in RW?

Answer 16

The stronger the associative strength (V), the stronger the conditioned response (CR).

Question 17

Jimmy has lived his entire life in a tiny village where there are no traffic lights. He has now moved to the metropolis of Waterloo to attend university. He is standing at a junction with a traffic light when it turns red and all the cars stop. Jimmy is very surprised, because he did not know that red lights mean stop.
Jimmy’s prediction error is?

Large

Small

Zero

Infinity

Answer 17

large

Question 18

As a result of this experience, Jimmy will

not learn anything

learn a little

learn a lot

Answer 18

learn a lot

Question 19

In Jimmy’s brain, the strength of the association between red lights and stopping will

Increase

Decrease

Become zero

Give him an aneurysm

Answer 19

increase

Question 20

V =

Answer 20

associative strength

Question 21

salience of the us

Answer 21

B

Question 22

intensity of the US

Answer 22

upside down y

Question 23

delta V

Answer 23

change in associateive strength

Question 24

a

Answer 24

salience of the CS

Question 25

Associative Strength

Answer 25

The strength of the association between a Conditioned Stimulus (CS) and an Unconditioned Stimulus (US). Represented by the variable V.

Question 26

Asymptote of Learning

Answer 26

The maximum associative strength that a CS can reach. This is often determined by the intensity or value of the US (λ).

Question 27

cue competition

Answer 27

Phenomena like blocking and overshadowing, where multiple CSs present during training interact and affect the learning about each other. The RW model was specifically developed to explain these.

Question 28

Overshadowing (in terms of RW)

Answer 28

When two CSs (A and B) are paired with a US, they compete for associative strength. The presence of one CS (A) reduces the prediction error associated with the other (B), leading to less learning about B compared to when B is trained alone.

Question 29

Blocking (in terms of RW)

Answer 29

Prior learning about one CS (A predicting the US) reduces the prediction error when a second CS (B) is added to A and paired with the US. Because the animal already expects the US due to A, it learns less about the new CS B.

Question 30

Conditioned Inhibition (in terms of RW)

Answer 30

A CS (B) that is paired with the absence of the US in the presence of another CS (A that predicts the US) develops a negative associative strength (V B ​ <0). This negative value represents that B predicts the absence of the US.

Question 31

RW Equation for A+US trials (Conditioned Inhibition Training)

Answer 31

ΔV A =α A (λ−V A )Where λ>0 (US is present).

Question 32

RW Equation for AB [+ nothing] trials (Conditioned Inhibition Training)

Answer 32

For CS A: ΔV A =α A (0−[V A +V B]) For CS B: ΔV B=α B (0−[V A+V B]) Where λ=0 (no US is present).

Question 33

Extinction (in terms of RW)

Answer 33

Repeated presentation of a CS without the US (λ=0) leads to a decrease in the associative strength (V) of that CS. The model suggests that the animal is "unlearning" the association until V approaches 0.

Question 34

RW Equation for Extinction

Answer 34

ΔV=α(0−V) Where λ=0 (no US is present), and V is the current associative strength of the CS.

Question 35

Limitation of RW in explaining Extinction

Answer 35

The RW model incorrectly predicts that extinction is simply the "unlearning" of the original association, leading to a CS becoming like a neutral stimulus again (V→0). It fails to account for phenomena like spontaneous recovery, suggesting that the original association is not entirely lost.

Question 36

RW Prediction for Extinction of a Conditioned Inhibitor

Answer 36

When a conditioned inhibitor (B with V B ​ <0) is presented alone without the US (λ=0), the RW model predicts that its negative associative strength will increase towards 0. This is because the animal expects "negative US" and receives no US, resulting in a positive prediction error.

Question 37

Behavioral Findings on Extinction of a Conditioned Inhibitor

Answer 37

Contrary to the RW model's prediction, conditioned inhibitors often do not extinguish readily when presented alone without the US. Their inhibitory properties tend to persist.

Question 38

Latent Inhibition - Phase 1 (Experimental Group)Latent Inhibition - Phase 2 (Experimental Group) Latent Inhibition - Control Group

Answer 38

Repeated presentations of a Neutral Stimulus (NS) followed by nothing (NS → nothing). The same stimulus from Phase 1 is now paired with a US (NS → US). Does not receive Phase 1. In Phase 2, the same NS is paired with the US (NS → US).

Question 39

Result of Latent Inhibition Experiment

Answer 39

The Experimental group learns the association between the NS and the US more slowly in Phase 2 compared to the Control group.

Question 40

RW Prediction for Latent Inhibition - Phase 1

Answer 40

According to RW, no learning should occur in Phase 1 because there is no US (λ=0) and the initial associative strength of the NS is V=0. Therefore, ΔV=α(0−0)=0.

Question 41

Limitation of RW in explaining Latent Inhibition

Answer 41

RW fails to predict latent inhibition because it assumes that learning only occurs in the presence of a prediction error driven by the US. In the absence of a US (Phase 1), there is no prediction error and thus no learning, even though the animal is clearly learning something about the CS (that it predicts nothing).

Question 42

Latent Learning (as a broader concept)

Answer 42

Learning that occurs without any obvious reinforcement and is not immediately expressed in behavior. RW struggles to account for such learning.

Question 43

Sensory Preconditioning

Answer 43

A phenomenon where prior pairing of two neutral stimuli (A-B) can lead to learning about one stimulus (B-US) affecting the response to the other (A) later. RW does not predict this.

Question 44

General Limitation of RW regarding Inhibition

Answer 44

RW has difficulties explaining several phenomena related to inhibition, including the persistence of inhibition during extinction and the initial learning that occurs when a stimulus predicts the absence of a US (as seen in latent inhibition).

Question 45

Core Idea of the Rescorla-Wagner Model

Answer 45

Learning depends on the surprisingness of the US. Learning occurs when there is a prediction error (the difference between what is expected and what is received).

Question 46

How Associative Strengths Combine in RW

Answer 46

The associative strengths of all CSs present on a trial are summed to determine the total expectation of the US.

Question 47

RW Explanation of Excitatory vs. Inhibitory CSs

Answer 47

Excitatory CSs acquire a positive associative strength (V>0), predicting the occurrence of the US. Inhibitory CSs acquire a negative associative strength (V<0), predicting the absence of the US.

Question 48

Influence of Background Context in RW

Answer 48

Background contextual stimuli also acquire associative strength and contribute to the overall prediction of the US, influencing learning about other CSs.

Question 49

Key Insight from RW-Stimulated Research

Answer 49

Conditioning is not a simple matter of pairing two events. Learning depends on the existing associations and the prediction error.

Question 50

Shortcomings of the Rescorla-Wagner Model (Summary)

Answer 50

List: Incorrectly predicts the loss of inhibition when an inhibitor is presented without a US. Does not explain latent inhibition. Fails to fully account for blocking, suggesting a role for attention.

Question 51

Mackintosh Model

Answer 51

Core Idea: Animals pay attention to relatively good predictors of the US and tune out bad or redundant ones. Learning is influenced by attention to the CS.

Question 52

Pearce-Hall Model

Answer 52

Core Idea: Animals attend to uncertain predictors. Attention to a CS on a given trial depends on how poorly the US was predicted on the previous trial.

Question 53

Hybrid Learning Models

Answer 53

Models that propose both attentional processes (as described by Mackintosh and Pearce-Hall) can operate and play complementary roles in learning.

Question 54

Wagner's Short-Term Memory Model

Answer 54

Expansion of RW: Proposed that learning is influenced by the surprisingness of both the US and the CS, based on whether they are "primed" in short-term memory (due to recent presentation or retrieval from long-term memory).

Question 55

Wagner's SOP Model (Standard Operating Procedure)

Answer 55

Core Idea: CS and US representations in long-term memory can be activated to different levels (A1 - focal awareness, A2 - peripheral awareness). Associations form only when both are in A1 simultaneously. Accounts for timing effects in conditioning.

Question 56

AESOP Model

Answer 56

Extension of SOP: Proposed that USs have both "sensory" and "emotive" nodes, leading to different kinds of responses with different decay rates.

Question 57

Configural Effects (in conditioning theory)

Answer 57

Definition: Situations where organisms respond differently to a CS when it is presented alone versus when it is presented in a compound with other stimuli. Some extensions of SOP attempt to explain these.

Question 58

Overall Significance of Conditioning Theories

Answer 58

They provide frameworks for understanding the principles of learning and can account for a wide range of experimental findings in Pavlovian conditioning. SOP (with extensions) is highlighted as a particularly powerful theory.

Question 59

Jimmy has severe mental health issues (because of all the craziness we put him through in earlier lessons). He goes to a therapist who tries to help him. Try to think through all the following scenarios with RW in mind. The therapist first wants Jimmy to be calm when in the office. So, Jimmy is allowed to go into the office and spend time there with nothing happening – to learn that it is a safe space. Later, when scary memories are brought up in there, Jimmy will be less scared than he would have been elsewhere. What kind of experiment is this? Would RW predict this effect or not?

Answer 59

This is a latent inhibition experiment. The office is a CS which is first presented with no US (nothing scary). Later, when it is paired with scary things, the association forms more slowly. RW would not predict this effect. In the beginning, there is no reason for Jimmy to be scared of the office (it is an NS), and so RW predicts he will learn nothing.

Question 60

The therapist gives Jimmy toys to hold onto when reliving scary experiences. Sometimes he is given a teddy bear. Other times, the teddy bear is not available, and Jimmy is instead given both a doll and a stress ball. According to RW, which of these three things will Jimmy associate most strongly with negative memories? Why?

Answer 60

This is an overshadowing experiment. RW says that the teddy bear is associated with an aversive US alone, so it will gain lots of associative strength. The doll and the stress ball are always presented together, so they will compete with each other for strength, and will overshadow each other. So, Jimmy will learn less about both of them than about the bear.