Theories Of Conditioning

Question 1

The Rascorla Wagner model

Answer 1

• The Rescorla-Wagner model focuses on the surprise value of the US to explain learning
• The Rescorla-Wagner model ‘predicts’ the typical profiles of acquisition, extinction and inhibitory learning

Question 2

The makintosh model uses

Answer 2

Attentional variables

Question 3

Pearce hall model deals with

Answer 3

CS salience

Question 4

Pavlov temporal contiguity theory

Answer 4

Closeness in time certainly helps animals to form associations between events

But temporal contiguity has to be inadequate as a complete theoretical account of classical conditioning
And studies of taste aversion show it isn’t necessary either
(see my last lecture: 3. Mechanisms of conditioning)

Question 5

Pavlov temporal contiguity theory

Answer 5

Conditioned taste aversion over long delays

Smith and Roll 1967

Question 6

Kamins blocking experiment

Answer 6

Design and results

Question 7

Rescorla Wagner 1972

Answer 7

• Originally focused on the surprisingness of the US; or discrepancy between obtained and expected reinforcement
• Pairing CS with unexpected US will increase strength of connection between them to a maximum determined by the magnitude of the US
• In later version, the surprisingness of the CS was also taken into account

Question 8

7. Rescorla-Wagner (1972)

Answer 8

• Subsequent capacity of CS to activate US representation held to depend on strength of association between them (measured as strength of CR)
• On each CSUS pairing, variable increase in associative strength, depending on current associative strength of CS and max. supportable by US
• So learning should be most rapid early in conditioning, rate should level off later

Question 9

Rescorla-Wagner (1972)

Answer 9

Equation used to express theory formally:
∆V = (-V)
V: strength of CS-US association
∆V: change in strength of association on a particular trial
: set by magnitude of US and sets maximum strength CS-US association can reach
: reflects salience of CS and is invariant throughout conditioning

Question 10

Rescorla-Wagner (1972): ∆V = (-V)

Answer 10

Acquisition
Assuming  is set at 0.20 and  at 100:
Trial 1
For novel CS, V is zero because the CS has no associative strength, so
∆V = 0.20(100-0)
∆V = 20
Trial 2
After trial 1, V is now 20, so
∆V = 0.20(100-20)
∆V = 16

Question 11

The effect of US magnitude on learning

Answer 11

Lower asymptote cf. sensory preconditioning - behaviourally silent

Question 12

The effect of CS salience on learning

Answer 12

Overshadowing, CS intensity relative to background, depressed acquisition function resembles that seen with a conditioned inhibitor

Question 13

Rescorla-Wagner (1972): ∆V = (-V)

Answer 13

Extinction
lis now zero because the US is absent and (in our example) the associative strength of the US reached 100:
Trial 1
∆V = 0.20(0-100)
∆V = -20
Trial 2
∆V = 0.20(0-80)
∆V = -16

Question 14

Conditioning with compound stimuli

Answer 14

Rescorla Wagner 1972
US surprisingness will now depend on how well this event is predicted by all the available stimuli
VALL = algebraic sum of all CSs present on any given trial
So for CSA on a single compound conditioning trial, ∆VA = A( - VALL)

Question 15

Conditioned inhibition (CI) aka inhibitory learning

Answer 15

Phase 1: CSL -> US; VL approaches value  = 1
Phase 2: (CIN + CSL) -> no US;  = 0;
ΣV (sum of associative strengths of L and N) approaches zero

Question 16

Rescorla-Wagner model: Conditioning of inhibition

Answer 16

Safety signal in avoidance learning

Question 17

The take home message from Rescorla-Wagner theory

Answer 17

V: strength of CS-US association

∆V: change in strength of association on a particular trial

: magnitude of US, sets maximum strength CS-US association can reach

: salience of CS, invariant throughout conditioning

Question 18

Wagner (1981)

Answer 18

Proposed modification of Rescorla-Wagner (1972) to include CS factors
Notion that CSs must also be surprising
Animals respond to novel stimuli with an orienting response (OR)
With repeated presentation, the OR declines (habituation)
And the associability of the stimulus declines (latent inhibition)

Question 19

Mackintosh (1975)

Answer 19

Animals pay attention to, and so condition most readily to, stimuli established as good predictors of significant events like food

Lawrence (1949, 1950) studies of discrimination learning – ‘the acquired distinctiveness of cues’ – associability is not a fixed factor dependent on intensity (see Mackintosh, 1974, 1983)
●As the associative strength of CS and US increases, the associability of that same CS with other outcomes should increase
●The associative strength of poor predictors should decline…

Question 20

Hall and Pearce experimental design; (B) Phase 2

Answer 20

• Notice that the phase 1 training wasn’t very effective (tone SR = 0.45)

Hall & Pearce (1982) reported similar experiment in which the phase 1 training was more effective (tone SR = 0.32)

Question 21

Pearce-Hall (1980)

Answer 21

Animals should not waste time processing events with known consequences:

1) Stimuli with established consequences are processed in automatic mode and cannot enter new associations
2) Stimuli with unknown consequences are processed in controlled mode and can enter into new associations