Associative Learning

Question 1

Associative learning

Answer 1

Learning that occurs through pairing events

Question 2

4 types of conditioning

Answer 2

Delay, trace, simultaneous and backward

Question 3

Blocking

Answer 3

Prior learning re. stimulus can block learning re new stimulus

Question 4

Rescorla-Wagner theory

Answer 4

Rescorla-Wagner theory: The change ∆ in the associative strength V = the salience of the CS (α) the salience of the US (b) and the amount of surprise (the difference between the strength of the US (λ), and how much US you expected (ΣV)

Question 5

What is the term for the max strength of the US?

Answer 5

Asymptote (λ)

Question 6

What is a conditioned inhibitor in terms of R-W?

Answer 6

Stimuli acquire negative strength until they predict absence of the US

Question 7

Explain overshadowing and blocking in terms of R-W…

Answer 7

Overshadowing = 2 CS’s at once. A limited amount of associative strength means that the stimuli inhibit learning about the other

Blocking = stimulus 1 already reaches asymptote meaning there can be no further learning about stimulus 2

Question 8

Is only a pairing adequate to form an association?

Answer 8

No, because we won’t be able to differentiate between learning and coincidental incidents if it was.

Question 9

What is delay conditioning?

Answer 9

If the CS immediately precedes the US

Question 10

What is trace conditioning?

Answer 10

If the CS takes time to impact the US

Question 11

What is simultaneous conditioning?

Answer 11

If the US and CS are simultaneous

Question 12

Latency means?

Answer 12

Duration of time

Question 13

Mahoney and Ayres (1976) found that?

Answer 13

A forward relationship between the CS and the US is good for associative learning. Highest learning is observed in trace and delay conditioning compared to simultaneous. No learning is observed in backward conditioning.

Question 14

What did Rescorla (1968) find?

Answer 14

Pairings are less important than correlation. Two stimuli won’t be associated with one another unless they occur together.

Question 15

What is blocking?

Answer 15

It refers to the situation where a conditioned stimulus (CS) that has already been associated with an unconditioned stimulus (US) blocks or prevents the formation of an association between a new CS and the same US when they are presented together

Question 16

Besides pairing, what’s essential for learning to occur?

Answer 16

The unconditioned stimulus needs to be surprising.

Question 17

Rescorla and Wagner (1972) found that

Answer 17

Association strength increases in each trial.

Question 18

What determines changes in associative strength according to the RW model?

Answer 18

The discrepancy between the actual US outcome and the organism’s expectation or prediction based on the presence of the CS. If the US is surprising or unexpected given the CS, more learning occurs.

Question 19

What is learning?

Answer 19

Persistent change in behaviour in response to experience

Question 20

What is habituation?

Answer 20

Repeated stimuli => weakening of response

Question 21

What is evaluative conditioning?

Answer 21

Neutral stimulus gaining positive or negative valence

Question 22

Shaping (what type of conditioning?)

Answer 22

Instrumental conditioning. Behaviour gradually taught through reinforcing successive approximations of target behaviour

Question 23

Autoshaping (what type of conditioning?)

Answer 23

Classical conditioning. Subject naturally begins to respond to conditioned stimulus as if unconditioned, even without training.

Question 24

Second-order conditioning vs sensory preconditioning

Answer 24

SOC = S2 -> S1 pairings precede S1 -> US pairings
SPC = S1 -> US pairings precede S2 –> S1 pairings

Question 25

Mental representations

Answer 25

Mental state accompanying experience of stimulus

Question 26

Response representations

Answer 26

Mental state accompanying performance of a response

Question 27

Conditioned inhibition

Answer 27

Stimulus predicts omission of expected US

Question 28

Discriminative stimulus

Answer 28

Signals when response will be followed by US

Question 29

What is a sign stimulus?

Answer 29

Part of a stimulus that is sufficient to evoke a behavioural response in an animal

Question 30

Fixed action pattern

Answer 30

Stereotyped response triggered by sign stimulus. MCQ could be, what type of response is typically triggered by a sign stimulus…

Question 31

Supernormal stimulus

Answer 31

Stimuli more effective than naturally occurring stimuli e.g. red pencil better than a beak

Question 32

Thorndike law of effect. Describe how it differs from more modern views/

Answer 32

Thorndike’s law of effect argues that a positive stimulus will lead to a response whereas aversive stimuli will lead to avoidance responses.
Believed the US acted as a glue to reinforce S-R association i.e. child would continue to tidy up without praise as praise had strengthened the association between parents instructions and tidying up. However, these days, it is believed that the US does play a role – animal responds to get its value.

So Thorndike thought: Sd => R (US)
Whereas these days it is believed that US => R (Sd)

Question 33

Types of reinforcement

Answer 33

Positive/negative = adding or taking away
Punishment/reward = whether it’s nice or nasty

Question 34

Passive vs active vs signalled avoidance

Answer 34

Passive = refraining from response
Active = active response to avoid aversive stimulus
Signalled = a specific signal such as a warning shock leads to a response

Question 35

Sidman avoidance

Answer 35

An aversive stimulus e.g. shock comes after a specified time period e.g. 5s. However a response e.g. pressing a lever will delay that shock e.g. by 10s. Key idea is that animal is learning based on time passing rather than a specific stimulus.

Question 36

Kamin (1969) did what experiment?

Answer 36

Terminating a warning signal. Response led to either a tone or a buzzer or both being stopped. Animals learned most when both the tone and the buzzer were stopped together. Both = CR’s.

Question 37

Signalled inhibitor

Answer 37

A stimulus that signals the absence of the US. E.g. bring a friend on a night out who ruins the fun. Can lead to extinction e.g. if buzzer ceases to signal shock, however this can take many trials (e.g. Solomon’s dog => 490 trials).

Question 38

What is V? (in the RWM equation)

Answer 38

Strength of the association between the CS and the US. A strong V means the association is very strong. Delta V is the change in association after each pairing.

Question 39

What is alpha? (in the RWM equation)

Answer 39

Alpha is the salience of the CS. Its intrinsic perceptual intensity.

Question 40

What is beta? (in the RWM equation)

Answer 40

Beta is the salience of the US, usually the “lure” of the food. For instance, rats love peanut butter, so it will have a high beta value.

Question 41

What is lambda? (in the RWM equation)

Answer 41

Lambda is the quantity of the unconditioned stimulus.

Question 42

The amount of learning that occurs in each pairing depends on:

Answer 42

1. How noticeable your CS is
2. How noticeable your US is
3. How big your US is

Question 43

What is sigma V?

Answer 43

The sum of the associative strength of everything else present in the trial. Basically how much the US was predicted.

Question 44

What is lambda - sigma V?

Answer 44

How surprising the US is. The quantity minus how much it was being expected.

Question 45

What is sigma V if a stimulus has never been presented before?

Answer 45

Zero

Question 46

What happens if the stimulus gets less surprising? What happens to delta V?

Answer 46

The amount of learning reduces. Delta V reduces and goes to zero.

Question 47

What is an asymptote?

Answer 47

The maximum amount of associative strength that completely offsets the size of the US. No more strength can be acquired. The maximum associative strength equals lambda.

Question 48

How does lambda go up and what does this do?

Answer 48

By increasing the quantity of food. This increases learning.

Question 49

Mackintosh (1976)

Answer 49

Mackintosh (1976) did some experiments where he compared conditioning with a light, light + small sound, and light + big sound. He found that if you just condition the light on its own, the animals are very afraid. If you condition them with a small noise, then there was some learning about the noise, but not much more than with the light alone. However, a lot of learning occurred when the sound was loud.
The more salient the second stimulus, the more learning it “takes” from the light.
The surprise is going to be determined by how much strength is already present.

Question 50

What are Q competition effects?

Answer 50

The competition for associations with a particular US.

Question 51

What is inhibition?

Answer 51

The taking away of an association. For instance, the CS paired with a US will only get associated if the US is surprising. If you present the CS without the US, the association goes away.

If you take away the US, the value of lambda will become zero, because there’s nothing present. The lack of US is surprising, so the content of the lambda-sigma V becomes negative. The associative strength goes down again.

Question 52

Give an example of negative associative learning

Answer 52

If you associate a light (CS) with food (US) until asymptote and then have further trails with light (CS1), a noise (CS2), and no food, negative associative learning will occur, which is shared between the light and the noise. As they reduce, the strength of the light will reduce but not up to zero (because it started so high), but the associative strength of the noise will go down to negatives.

Question 53

What are conditioned inhibitors?

Answer 53

Stimuli with negative associative strength

Question 54

What happens to avoidance responses once a shock is taken away?

Answer 54

Avoidance responses go on for a long time after the shock is taken away and it takes much longer to extinguish. The fact that the avoidance response (inhibitor) is present tells the animal that the expected shock isn’t going to happen. You’re pairing a CS that predicts shock with an inhibitor for shock.

RWM explains that what you’re extinguishing is the warning signal with an inhibitor that has negative associative strength, so the lack of shock isn’t surprising. Because the inhibitor and the CS cancel each other out to zero.

Question 55

What is prediction error and how is it represented?

Answer 55

If the US is surprising, the CS will not adequately predict it, leading to prediction error. The difference between what the US is and what the CS predicts describes the size of the error (lambda - V).

Question 56

What is the overexpectation effect?

Answer 56

When there are two CSs used to predict one US, they overpredict it, and the association thus reduces.