categorisation and concept formation

Question 1

Basic level concept

Answer 1

consists of the most obvious and neutral way of identifying objects: chair, dog, boat, party, doctor, etc.

Perhaps surprisingly, the basic level is not at a very high level of generality, like animal, plant, or machine, which they called superordinate categories.

Question 2

Superordinate concept

Answer 2

“Child” is the superordinate of “girl” and “boy.”
Most basic level

Question 3

Abstract concept

Answer 3

does not refer to individual entity, but to some property, relation or state (e.g., sameness, truth).

Question 4

Bhatt, Wasserman, Reynolds & Knauss, 1988

Answer 4

• Pigeons in a chamber with choice of four response keys.
• Shown pictures of flowers, cars, people and chairs
  Birds learned to peck different keys for exemplars of each of the four categories of picture
• Then tested them with some new exemplars, that they had never seen before…
• They also were able to respond correctly to the new exemplars, that they had never seen before.
• Suggests birds had formed a “concept” of flowers, cars, people and chairs.
• However, performance was more accurate with the training
• stimuli (80%) than with the novel, test stimuli (60%).

Question 5

Exemplar theory

Answer 5

Learn about every instance independently. Classify novel exemplars via similarity to learned instances

Question 6

Prototype theory

Answer 6

Abstract prototype corresponding to central tendency of training exemplars

Question 7

Prototype Model

Answer 7

Category judgements are made by comparing a new exemplar to the prototype

Question 8

Exemplars Model

Answer 8

Category judgements are made by comparing a new exemplar to all the old exemplars of a category or to the exemplar that is most appropriate

Question 9

Predictions of two theories

Answer 9

• exemplar theory predicts classifying a novel item always worse than one you have seen before
  prototype theory predicts classifying a novel item can be better than one you have seen before

Question 10

Humans compared to animals in prototype vs exemplar

Answer 10

• Conclusion: Both humans and animals retain information about the training items/exemplars (consistent with exemplar theory) but show the prototype effect (consistent with prototype theory)
• So which is best?
• It turns out that a variation of exemplar theory can explain the prototype effect!
• The two theories not so different after all.

Question 11

Aydin & Pearce, 1994. The prototype effect in pigeons

Answer 11

The birds trained on three-element displays, created by distorting the prototypes (swapping one prototype element for one from the other category)

• Birds taught the three positive patterns always paired with food, the three negative patterns were not. Birds pecked more at positive than negative patterns.
• the test of prototype theory is whether they are more accurate with prototype they have never experienced
• This is evidence of a kind of prototype effect
• (though not everyone thinks this evidence is good enough - pigeons fail to learn more complex prototypes)

Question 12

Whittlesea, 1987

Answer 12

Lists 1, 2 and 3 all differ from prototype FURIG by two letters EQUALLY SIMILAR

BUT examples in list 1 more similar to examples in list 2 than list 3

• Prototype predicts List 1 = List2 = List 3
• Exemplar predicts List 1 > List 2 > List 3· Score is improvement from pretest (high scores = good):
  ○ 1 1.07
  ○ 2 0.80
  ○ 3 0.51

Question 13

How can exemplar theory explain the prototype effect?

Answer 13

our categorization judgments are influenced by the similarity between the object and the exemplars stored in memory, with objects that closely resemble the prototype being more easily categorized due to their similarity to a larger number of stored exemplars

• This explanation is actually viewed as a new theory - “feature theory”

Question 14

Question: category learning depend on associative learning:

Answer 14

Blocking: pairing only produces association between X and Category if Category surprising

Feature X –> Category then
- Feature X+ other –> Category feature X? Small CR
- Feature X → Category feature X? big CR

Question 15

Experiment by Shanks (1990; cf. Gluck and Bower,1988)

Answer 15

• Subjects given trials in which medical symptoms paired with a disease diagnosis
• Subjects must predict disease from symptom
  · Symptom -> disease association
• Two diseases, one common (e.g. flu), one rare (e.g. neuroscience allergy - NA)
• Three symptoms:
  · one target symptom: (a – headache)
  · and two nontarget symptoms:
  · b – a runny nose
  · c – rash
• The following is a simplification of Shanks’s experiment:
• 24 runny nose à flu 6 runny nose & headache à flu
• 6 rash à NA 6 rash & headache à NA

Question 16

Which does headache predict more – flu or NA?

Answer 16

• 24 runny nose à flu 6 runny nose & headache à flu
• 6 rash à NA 6 rash & headache à NA
• same number of pairings
• BUT when headache paired with flu, runny nose is also present
• and when headache paired with NA, rash is also present
• and runny nose predicts flu much better than rash predicts NA
  so flu less surprising when paired with headache than NA is when paired with headache

Question 17

Nonassociative theory

Answer 17

predicts that, given headache, subjects will be just as likely to predict flu as NA (pairings important) flu = NA

Question 18

Associative theory

Answer 18

predicts that, given headache, subjects will be more likely to choose rare NA than common flu (surprise important) flu < NA

Question 19

Superordinate level concept formation

Answer 19

• Superordinate categories have members that are not necessarily physically similar to each other, but share a common associate

Question 20

Wasserman, De Volder & Coppage, 1992

Answer 20

• Pigeons trained with slides of people, chairs, cars and flowers
• The birds reinforced for making Response 1 to either people or chairs, and for making Response 2 to either cars and flowers.
• people and chairs in one category, cars and flowers in another
• Then made Response 3 to people, and Response 4 to cars, and tested with chairs and flowers, with choice of Response 3 & Response 4
  Response 3 to chairs & Response 4 to flowers was correct
• Birds formed superordinate category

Question 21

Abstract concept formation in animals

Answer 21

The one that has been studied most is same/different, usually studied using a match-to-sample technique (MTS)

• Birds shown a sample key e.g. red; then given a choice of red and green.
• Must peck the same colour as the sample – i.e. red. On other trials the bird gets a green sample; then task is to peck the green comparison
• Birds could master this, but were typically poor at transferring to skill to different colours (e.g., blue and yellow). This suggests they had not really learned the concept of same.

Question 22

Wasserman, Hugart & Kirkpatrick-Steger, 1995

Answer 22

Pigeons shown complex stimulus displays, and given a choice of a red and a green key

• They were trained on arrays with one set of specific icons
  Rewarded for pecking red on same trials, green on different trials

Some evidence they could generalise this abstract concept.

Conclude: animals can form basic, superordinate, and to a limited extent abstract concepts

Question 23

what is the key importance of concept formation?

Answer 23

to allow categorisation of novel stimuli

Question 24

what does it mean if a concept is polymorphous?

Answer 24

concepts are not always defined by specific features… sometimes do no have necessary or sufficient features

what is the defining feature of a game?

Question 25

describe the difference between prototype and exemplar theory of concept formation?

Answer 25

in prototype theory, category judgements made by comparing a new exemplar to one prototype

whereas in exemplar theory, category judgements made by comparing new exemplar to all old exemplars of a category or to exemplar that is most appropriate

Question 26

how do exemplar theory and feature theory differ?

Answer 26

exemplar theory - store each whole exemplar, new stimuli classified on similarity of whole stimulus to stored exemplars

feature theory - learn about/store component features of each exemplar, new stimuli classified on basis of sharing features with stored exemplars

Question 27

explain how associative theory predicts results of Shanks (1990) study into blocking for category formation?

Answer 27

associative theory predicts that, given headache, subjects will be more likely to choose rare NA than common flu (surprise important) flu < NA

suggests associative learning best explanation of performance on this categorisation task in human subjects

Question 28

describe results of Wasserman, Hugart & Kirkpatrick-Steger (1995) study into abstract concepts?

Answer 28

• testing training stimuli = almost 90% correct
• testing familiar stimuli = above chance
• testing unfamiliar stimuli = 75%
• argued good evidence pigeons can learn same and different