Categorization and concept formation:
Concept: The mental representation of something. Many concepts are based on the sharing of common properties by items in a class.
Concept formation: The induction of concepts that divide items into classes according to their shared properties (categorization).
Note some concepts don't have necessary or sufficient defining features -- polymorphous
Types of concept:
(i) Basic level concept -- based on similarity of perceptual qualities
(ii) Superordinate concept -- groups of basic level concepts; not based on perceptual similarity
(iii) Abstract concept -- does not refer to individual entity, but to some property, relation or state
Can animals form basic level concepts? superordinate concepts? abstract concepts?
If so, how? Do animals form concepts in the same way as humans?
Basic level concept formation in animals: Bhatt, Wasserman, Reynolds & Knauss, 1988
Pigeons have choice of four response keys. Shown pictures of flowers, cars, people and chairs
Learned to peck different keys for each category of picture. Then tested with new exemplars...... and could respond correctly. Had birds formed “concept” of flowers, cars, people and chairs?
But more accurate with the training stimuli (80%) than with the novel, test stimuli (60%).
how do they do it?
(i) Exemplar theory: Learn about every instance independently. Classify novel exemplars on the basis of similarity to learned instances.
(ii) Prototype theory: Abstract a prototype corresponding to central tendency of training exemplars. Classify novel exemplars on basis of similarity to prototype.
Animals store information about training exemplars, because are more accurate with them than with the novel test stimuli. This supports exemplar theory.
But humans show the prototype effect (e.g., Homa et al., 1981) -- categorize prototype more accurately than training stimuli, even if never seen before - fits prototype theory better.
So are humans are storing a prototype while animals only store exemplars? Are humans and animals forming basic-level concepts in different ways? Can animals store a prototype?
Aydin & Pearce, 1994. The prototype effect in pigeons
Positive and negative prototypes defined as ABC and DEF...
Birds trained on three-element displays created by distorting the prototypes:
+ + -
But do humans store information about the training items...? Whittlesea, 1987
Lists 1, 2 and 3 all differ from the prototype by two letters; but List 2 differs from study items by one letter, and list 3 differs from the study items by two letters.
FURIG
1 FEKIG FUTEG PURYG FYRIP KURIT Subjects tested with all stimuli (30ms
2 FYKIG FUTYG PUREG FERIP PURIT followed by mask; then had to write
3 FUKIP PUTIG FURYT FYREG KERIG down as many letters as they could)
Then list study 1; tested with lists 1, 2 and 3.
Prototype theory says lists 1, 2 and 3 equally easy as equally similar to the prototype.
Exemplar theory says list 1 easiest as studied, list 2’s items differ a little from studied list, and list 3’s items differ a lot from the studied list And that is what they found...
1: 1.07 2: 0.80 3: 0.51
Numbers are mean increase in the number of letters correctly identified relative to the pretest stage. Training on list 1 helped identification of similar items, rather than prototype..
Conclusion: Both humans and animals retain information about the training items/exemplars. So what about prototype theory? Exemplar theory can even explain the prototype effect! The two theories not so different...
How does exemplar theory explain the prototype effect? Let’s go back to Pearce’s experiment:
POSITIVE PROTOTYPE POSITIVE TRAINED STIMULUS
Here prototype effect explained by assuming that each stimulus composed of set of elements, or features, each of which can become linked to the relevant category; generalisation to stimuli that share features. This is feature theory; it is closely related to exemplar theory. Both say you store something about the stimulus that is presented on each trial.
cf. Exemplar theory: stimulus is a configuration, generalisation between similar configurations
Some argue that categories are actually formed by means of associative learning: Shanks (1990):
Subjects given series of trials in which combinations of symptoms and a diagnosis are presented. Subject must estimate extent to which a symptom is associated with a particular disease.
Two diseases, one common (e.g. flu) and one rare (e.g. pig-disease). Three possible symptoms: one target symptom: (a -- e.g, a headache) and two nontarget symptoms: b, -- e.g., a runny nose, and c -- e.g., grunting
6 abàflu 24 bàflu 6acàpig 4càpig
This is a simplification of Shanks’s experiment: Note:
(i) lots more b-->flu pairings than c-->pig pairings
(ii) a always presented with b on flu trials, and with c on pig trials
(iii) same number of ab-->flu and ac-->pig pairings
If number of pairings critical then subjects would, given symptom a, diagnose flu as often as pig-disease (same number of pairings -- (iii))
Associative learning makes a different prediction:
Subjects given symptom a, and must say which disease it predicts, flu or pig. Critical factor is strength of the a--->flu and a-->pig associations -- which is stronger?
Symptom a paired same number of times with flu as with pig, so (things being equal) should be associated equally with both. But things not equal: a is in compound with another symptom when it is paired with flu and pig:
ab-->flu and ac-->pig
Thus b has been paired many more times with flu than c has been paired with X, so b-->flu association is stronger than c--> pig association.
Rescorla-Wagner model says associations only form if the second event surprising -- blocking:
Associative theory predicts a strong b-->flu association will block formation of the a-->flu association because b is a good predictor of flu.
But few pairings of c and pig disease, so only weak c-->pig association forms, which cannot block formation of the a-->pig association effectively
Exemplar theory predicts that, given symptom a, subjects equally likely to predict flu as pig.
Associative theory predicts that, given symptom a, subjects more likely to predict pig.
Proportion common disease (flu) diagnoses: .37
Proportion rare disease (pig) diagnoses: .63
Superordinate level concept formation in animals
Superordinate categories have members that are not necessarily physically similar to
each other, but share a common associate. Can animals form categories of this kind?
Wasserman, De Volder & Coppage, 1992
Pigeons trained with slides of people, chairs, cars and flowers; reinforced for making R1 to either people or chairs, and for making R2 to either cars and flowers. Thus people and chairs in one category, and cars and flowers in another.
Then trained to make R3 to people, and R4 to cars. Finally tested with chairs and flowers, and given a choice of R3 and R4. R3 to chairs and R4 to flowers counted as correct responses.
Looks like superordinate category -- are treating
people and chairs as equivalent because they
were both paired with the same response in
the first phase. This is a more complex type
of categorization because category
members are not physically similar.
Is this the same as how people do it?
Little work has been done on abstract concepts in animals. That studied most is same/different, usually using a Match-to-sample technique (MTS). Birds shown a sample key (red) then given choice of two comparison keys red and green. Task is to peck the same colour as sample. On other trials get green sample, and must peck green comparison.
Birds could master this, but were typically poor at transferring to skill to different colours
(e.g., blue and yellow). Suggests had not really learned concept of same.
However, more complex training techniques seemed to produce better results:
Wasserman, Hugart & Kirkpatrick-Steger, 1995. Pigeons shown complex stimulus displays, and given a choice of a red and a green key. A same display comprised 16 identical icons, and a different display 16 different icons. Were reinforced for pecking the red key on same trials, and the green key on different trials. Then tested with a further set of same/different arrays involving an unfamiliar set of icons. Responded appropriately. Had they learned the same/different concept? Or is there another explanation?
References
Bouton, M.E. (2007) Learning and Behavior. Sinauer Associates.
Lea, S.E.G. (1984). In what sense do pigeons learn concepts? In Animal Cognition
(Roitblat, H.L., Bever, T.G., & Terrace. H.S. (Eds.) (pp.263-276). Lawrence Erlbaum Associates.
Pearce, J.M. (1997). Animal Learning and Cognition. Lawrence Erlbaum Associates.
Shanks, D.R. (1995) The Psychology of Associative Learning. Cambridge University Press.
http://www.pigeon.psy.tufts.edu/avc/huber/
LECTURES AND HANDOUTS ON MY WEBPAGE
WEBQUIZ AT http://www.sinauer.com/bouton (use )