L14 - Perceptual Learning and Discrimination

Question 1

Discrimination Learning

Answer 1

Learning to behave differently to different stimuli. This is critical for mastering most tasks.
- With time and practise we can discriminate

E.g. Obvious traffic lights
E.g. Less obvious Wine tasting

Question 2

How to teach discrimination learning in lab terms

Answer 2

T - US
T + L - NO US

In the presence of both together overtime responding diminishes

Question 3

What does S+ and S- Mean?

Answer 3

S+ used to note that it is followed by a US

S- US will not be delivered

Question 4

4 factors affecting the rate of discrimination

Answer 4

1. Similarity between discriminative stimuli (S+/S-) (fingerprints)
2. Prior exposure to discriminative stimuli (shapes and taste)
3. Salience and valence of consequences (taste and stress)
4. Contingencies that differentiate the stimuli

Question 5

What is radiology?

Answer 5

Detection of abnormalities implicit for doctors, built up over time

• Formal training, explicit feedback, a tone of exposure

Question 6

Tangen, Thompson and McCarthy (2011) Fingerprint Experiment

Answer 6

Fingerprint expertise

• Experiment looking at novices and experts given fingerprint from a crime scene
• Novices not too bad at MATCHING and NOT MATCHING but experts are extremely good
• Where they really differ is noticing subtle differences when they are SIMILAR
  - Subtle differences are
  key
  
  • Novices are worse
    than chance (44%)

Due to exposure to stimuli

Question 7

Gibson and Walk (1956) Perceptual learning and development

Answer 7

Got young rats passively exposed to circles and triangles outside of the cage

• Control group were not exposed for the 90 days
• Then given a task to go through 2 doors, one has a circle one has a triangle above it
• One door has reinforcement behind it
• Experimental group learn it much more quickly

Rats who have cutout shapes visible from their
cages are better able to learn subsequent discriminations involving these
shapes than rats who are exposed to other shapes

Question 8

Mackintosh, Kaye and Bennett (1991): Perceptual Learning of flavours

Method

Answer 8

1. Pre-exposure involved the rats to consume 2 fluids, complex tastes made up of saline and lemon (AX) and given access to sucrose + lemon (BX all g).
2. Rats allowed to drink both
3. AX is paired with sickness - conditioned taste aversion
4. Then give rats salty lemon they wont drink it, they retrieve feeling of illness
   a. What extent will it generalise to BX
   b. If you discriminate well you shouldn’t averse to BX as they know it is different to AX

Question 9

Mackintosh, Kaye and Bennett (1991): Perceptual Learning of flavours

Results

Answer 9

Results:
- Novel group don’t like either solution (cannot distinguish), consumption is low

• Pre-X drink considerable more of BX than salty lemon, they still drunk more of the bad one but clear distinction

Question 10

How can pre-X lead to faster discrimination? Given the laws of LI

Generalisation of Common Elements

Answer 10

Mere exposure leads to a reduction in salience (mechanism responsible for latent inhibition), which will affect the stimulus elements that the two stimuli share in common more than the elements that are unique to just one stimulus

Better at discriminating as pre-exposing to stimuli you reduce the salience of what they have in common

Question 11

generalisation of common elements using experiments (Shapes and tastes)

Answer 11

• Imagine stimuli composed as elements
• Salt, lemon in common, sugar (ven diagram)
• Shapes example, circle, things in common to all black shapes, triangle

The idea is that if you have conditioned A and B, if the X conditioned have been lead to aversion you will sack both, if A is involved as main, less aversion to B will occur

Question 12

Bennet et al (1994) pre-exposure to common element X that impacts discrimination

Answer 12

First phase
- 4 groups of rats different Pre-X
: nothing, BX, BY or X (where B - Sucrose, X - lemon)

• All then receive AX, made to feel ill with this
• Test solution is BX (how much they consume after AX aversion learning)

Question 13

Results of Bennet experiment

Answer 13

Results:

• Control won’t drink much
• X alone will drink more
• B and X will drink the most
• BY (random flavour) show similar to control

Question 14

Implications of Bennet experiment

Answer 14

Exposure to the common elements is enough to produce a perceptual learning effect

Suggests reduction in the perceptual salience of the common elements results in reduced generalisation between similar stimuli

The novel unique elements are more salient: learned about faster, capture attention more readily

Question 15

Yerkes and Dodson

Stress performance

Answer 15

Inverted U-shape stress performance

• Right door all good, wrong door shock
• How many trials does it take to work out the discrimination, (colour of door)

Question 16

Results for Yerkes and Dodson

Answer 16

3 different difficulties (Black Vs white) (grey vs dark grey)
- More punishment with easy discrimination do it faster

• Harder, too much stress doesn’t learn anything
• Shows the optimal level of punishment relates to difficulty of the task

Question 17

How do contingencies that differentiate the stimuli impact discrimination learning

Answer 17

They learn discrimination faster when contingencies are stronger

By in large humans find it easier to learn discrimination when stimuli independent features predict the outcome

When non-linear discriminations where no single feature predicts the US or its absence it can be learned but it is more difficult

Question 18

Linear discrimination example

Answer 18

A- / AX+

A+ / AX -

Question 19

Non-linear discrimination example

Answer 19

Negative paterning
A+ / B + / AB-

Biconditional discrimination
AB+ / BC- / CD+ / DA-

Confused at first and then learning is much more gradual

Question 20

3 consequences of discrimination learning for generalization.

Answer 20

1. Less generalisation between trained S+ / S-
2. Sharpening of generalisation gradient
3. Asymmetries & biases in responding and preference

Question 21

Peak Shift

Answer 21

Displacement of the peak of the gradient away from the S+ in the opposite direction of the S-

It occurs when the strength of a particular response is directly proportional to the magnitude of a somewhat simple perceptual cue. Exaggerate that cue, and you can exaggerate the response it elicits.

Question 22

Zebra Finches and beak colour discrimination

Answer 22

Weisman et al (1994) found that beak colour discrimination is learning during bird ‘infancy’ and affects alter preference.

Females choose mates with same colour beak as father, only when beak colour of the mother and father differed. They need to have defined masculinity.