Dopamine

Question 1

factors that affect learning

Answer 1

1. Probability of US presentation.
2. Rate of US presentation.
3. Contingency.

Question 2

Probability of US presentation

Answer 2

100% CS+ training produces faster acquisition and stronger behaviour than 50% training. However, 50% training is more resistant to extinction.

Question 3

Rate of US presentation

Answer 3

Temporal duration of cues matter.

10s CS generates stronger CR than 40s CS.
25% CS+ of 10s CS generates similar conditioning as 100% CS+ of 40s CS.

Question 4

Contingency

Answer 4

Correlation matters. It is not just about the raw pairing frequency.

Question 5

3 evidence that insight-like behaviour is gradual

Answer 5

1. Presolution reversal.
2. Speed of responding.
3. S+ advantage.

Question 6

Presolution reversal

Answer 6

Reversal of reinforcement schedule prior to solution shows impairment in learning.

Question 7

Speed of responding

Answer 7

Animals respond faster to VR+ than to HR- even while animals are still responding to R (suggests influence of V).

Question 8

S+ advantage

Answer 8

After animals stop responding to old cue R, they are more likely to respond to VR+ than other values (learning about V throughout training).

Question 9

Shape of learning curve

Answer 9

Sigmoidal function (diminishing returns).

Question 10

Clark-Hull model of learning

Answer 10

ΔH=α(λ -H).

The change in habit is salience * (the total possible amount of learning - the current strength of habit).

Question 11

Problem with the Clark-Hull equation

Answer 11

The model simply states that associations are acquired with everything (lots of noise). Does not distinguish between relevant vs. irrelevant cues.

Question 12

Kamin’s blocking

Answer 12

Animals fail to learn about CS2 if they have already associated CS1 with the same US.

Question 13

Rescorla-Wagner equation

Answer 13

ΔV=αβ(λ-ΣV). The change in associative strength is determined by the sum of all cues present on a trial (allows for multiple CSs).

Question 14

Eippert et al. (2012) blocking in humans

Answer 14

Demonstrated blocking. Learning seems to be mediated by connectivity between amygdala and PFC.

Question 15

Overshadowing

Answer 15

Two CSs followed by US. The more salience CS forms a stronger association (less salient CS ‘overshadowed’). Learning depends on cue salience, not just co-occurrence.

Question 16

Relative learning

Answer 16

Learning = learning rate x (how much you could learn - how much you know already).

Question 17

Delay conditioning

Answer 17

CS stays on until US appears.

Question 18

Backward conditioning

Answer 18

US appears before CS (often inhibitory).

Question 19

Simultaneous conditioning

Answer 19

CS and US start and end together.

Question 20

Relative validity

Answer 20

AX+, BX-. Even though X in both, animals learn more about A and B as they are more predictive.

Question 21

3 evidence that dopamine signals RPE

Answer 21

1. Earliest predictive cue acquires the response.
2. Stimulus generalisation (A+ then test AX. AX elicits weaker dopamine signal than A alone). Less certain about reward.
3. A- exposure elicits dopamine response as long as animal orients to stimulus (not a pleasure signal).

Question 22

Steinberg et al. (2003) unblocking

Answer 22

Stimulation of dopamine neurons prevented blocking and extinction (despite no prediction error present).

Question 23

Waelti et al. (2001) dopamine in blocking

Answer 23

No dopamine response for blocked stimulus.

Question 24

Tobler et al. (2003) dopamine in inhibition

Answer 24

A supressed response for inhibitors (stopped dopamine neurons from firing to excitor).

Question 25

3 arguments against dopamine's role in learning

Answer 25

1. Dopamine may encode perceived salience rather than error correction. 2. The timing of dopamine is for preparing motor responses rather than error correction. 3. Other effects like superconditioning and overexpectation may not be due to error correction.