Simple & Complex Discriminations

Question 1

SIMPLE DISCRIMINATION

Answer 1

• Pavlov onwards; learning theorists experimented w/providing US (classical conditioning)/reinforcement (operant conditioning) in presence of 1 stimulus (CS+, S^D, S+) but not in presence of another (CS-, S^TRIANGLE, S-)
• differential responding can be obtained aka. animal discriminates 2 stimuli
• early exps; stimuli normally simple; differed on some obvious physical dimension ie. dif pitch tones/dif colour lights

Question 2

PROCEDURES

Answer 2

SUCCESSIVE DISCRIMINATION
SIMULTANEOUS DISCRIMINATION
CONDITIONAL DISCRIMINATION

Question 3

SUCCESSIVE DISCRIMINATION

Answer 3

• present 1 of stimuli
• see how animal responds (respond/ignore)
• reward stimulus (S+) VS other (S-)
• S- response -> no reward + short timeout = no other reward opportunity

Question 4

SIMULTANEOUS DISCRIMINATION

Answer 4

• present 2 stimuli
• see which animal approaches/chooses (ie. pressing 1/2 keys/pecking stimulus
• S+ = reward; S- = timeout
• stimulus randomly swap sides to avoid position based response (tendency in rats/pigeons)
• easier than successive discriminations
• both modelled by simple learning rules (ie. RW)

Question 5

CONDITIONAL DISCRIMINATION

Answer 5

• reinforce dif responses/stimulus-response associations in presence of dif stimuli (ie. Colwill & Rescorla exp)

Question 6

APPARATUS TYPES

Answer 6

• discrimination boxes (mazes w/discriminative stimuli)
• Lashley’s jumping stand
• Harlow’s Wisconsin General Test Apparatus (WGTA)
• Skinner boxes in many variants
• nowadays video/computer displays ie. touch screens

Question 7

COMPLEX DISCRIMINATIONS

Answer 7

HERRNSTEIN & LOVELAND (1964)
- pioneering experiment; modern work concentrates on discrimination between stimuli sets usually defined in human concepts (ie. person VS non-person; fish VS non-fish)/artificial concepts defined by specified multiple features
- aka. categorical discriminations frequently learned quickly
- most discussion centred on whether animals need to possess concepts to perform categorical discriminations aka. what it means for animal to “possess concept”

Question 8

ABSTRACTION TYPES

Answer 8

PERCEPTUAL CATEGORIES
LOGICAL CATEGORIES

Question 9

PERCEPTUAL CATEGORIES

Answer 9

• all cats; abstraction = prototype
• classification/responding on stimulus similarity = categorisation; performed using learning algorithms (ie. RW) in simple conditioning

Question 10

LOGICAL CATEGORIES

Answer 10

• all fours; abstraction = concept (?)
• number ^ abstract concept > categories
• possible for perceptual similarity of number instances = low; RW would struggle to explain it
• animals seem unable to (easily) learn this w/o special training

Question 11

PIGEON VISUAL SYSTEM SPECIAL FEATURES

Answer 11

• 2 foveas in each eye; 1 forward (binocular); 1 lateral
• 2 visual systems w/dif functions/psychophysical responses
• classes of cone differ by oil-droplets filtering light NOT visual pigment
• 3+ cone types
• very wide view range
• UV light detected; affects colour matches
• plane of polarisation of light discriminated

Question 12

PERCEPTUAL CATEGORIES: HERRNSTEIN & LOVELAND (1964)

Answer 12

• higher order concept formation in pigeons
• pigeons learn to peck in presence of person’s picture; withhold pecks in presence of presence of empty picture
• stimuli (holiday slides) varied greatly in people/posture/whole/clothing number
• transfer trials show correct response to new stimuli post successful learning

Question 13

PERCEPTUAL CATEGORIES: CONTINUED WORK

Answer 13

• other concepts (ie. fish/trees/cats/dogs/Bach VS Starvinsky etc.)
• some ecologically valid concepts for species (ie. individual conspecifics/prey items/locations)

Question 14

CATEGORY DISCRIMINATION THEORIES

Answer 14

ROTE LEARNING
- shown to happen via getting pigeon to learn arbitrary slide sets
- BUT not as easily learned as sets respecting categories
MULTIPLE LINEAR FEATURE MODEL USING RW
- predicted prototype effect that doesn’t always occur (BUT oft does)
- sometimes difficult to demonstrate control by multiple features
CONFIGURAL/EXEMPLAR MODELS ALSO EXIST

Question 15

ARTIFICIAL POLYMORPHOUS CATEGORIES

Answer 15

DENNIS ET AL. (1973)
- what makes stimulus a member of group A?
- touch problem for humans; each category defined by values on 3 features/dimensions; easy to see 2 (shape/filled VS empty) not the last (symmetry)
- symmetrical filled circles = group A
- 2/3 = in category corresponding to majority
- pigeons pick this up fine!

Question 16

MULTIPLE COMPLEX FEATURES

Answer 16

FERSON & LEA (1990)
- classification problem based on multiple features using pigeons
- more features BUT easier
- all features controlled beh (eventually post special training)
- reversal on stimuli subset transferred to other stimuli (instance -> category generalisation) BUT not to other features => no evidence of coherent concept
- multiple linear feature model describes data well aka. basically RW model using units to represent features