C9 concepts

Question 1

Define concept

Answer 1

A general idea which groups similar things, events, people. Applies to every member of a category.

A category being what is thought about (the concept is the thinking)

Words are used to communicate thoughts/concepts. Some words can be ambiguous - multiple meanings (fair/fare)

Concepts are unambiguous - if it appears to relate to two different categories it’s actually two different concepts.

Everyday concepts can be taken for granted but can develop over a long period of time. Categorisation can be very important - difference in legal status between an adult and child. Clinical diagnosis is an example of categorisation.

Question 2

Categorization

Answer 2

Bruner - the ability to treat individual things in terms of their group membership - implies similar behaviour towards different instances of the same category.

Techniques used to study categorisation -
sorting tasks - some people class eggs as breakfast food, others as dairy produce.
property listing tasks - write down properties of concept (eg dog) frequency indicates how central property is to the concept.

Question 3

Concepts and cognition

Answer 3

Make it easier to remember information

Enable semantic classification - similar to perceptual classification used in recognition

May be units of semantic memory that store facts (compared to episodes/experiences stored in episodic memory)

We store relationships between concepts (cats are animals) and lexical concepts in the mental lexicon may be used to link words with their meanings.

Concepts enable inferences/assumptions to be made

Question 4

Explaining categorisation - theories

Answer 4

Classical theory: things with the same properties belong to the same category;

Prototype theory: things that have “most” of the properties of a category belong to it;

‘Theory’ theory: instances of a category have the same explanations;

Psychological essentialism: members of categories share some essential properties (even if we can’t identify these).

Question 5

Classical and prototype theory

Answer 5

Both classical and prototype theory consider that category membership of an item depends on how similar (to different degrees) it is to other members of the category, i.e. how “typical” the item is of that category;

This approach has been successful to some extent:

Hampton (1998) found that typicality explained most (46-96%) of the variance in how likely an item was to be judged as a member of a category in borderline cases;

However other predictors such as lack of familiarity of the participant with the item/category, as well as “technical membership’ (e.g. a dolphin is “technically” a mammal but judged to be superficially more like a fish) were almost as good predictors.

This suggests that deeper analysis is involved, not just comparison of external properties/features, leading to other explanations such as “theory” theory and psychological essentialism.

Question 6

Classical theory claims

Answer 6

That items belong to the same category if they share certain properties, if they don’t have all the properties they are not members of that category (necessary condition)

If an item has all the properties it must be a member of that category (sufficient condition)

Question 7

Classical theory evidence for

Answer 7

Empirical studies (Hull, 1920; Bruner et al., 1956) showed people do associate common properties with category members

Question 8

Classical theory evidence against - Typicality

Answer 8

Typicality - eg a robin is a more typical bird than a penguin.

Rosch showed all or nothing doesn’t work - people think of some items as more typical than others - verified robin is a bird sentences faster than penguin is a bird

Suggest categories have rich internal structure which classical theory doesnt explain

Question 9

Classical theory evidence against - Borderline cases

Answer 9

when does red become orange? - doesnt fit with all or none

McCloskey and Glucksberg found people rated items (e.g. bookends as furniture or not) inconsistently both as individuals at different times, and between individuals. This might have been due to lack of knowledge, but probably not. However the classical theory would claim that items are category members or not.

Question 10

Classical theory evidence against - Intransitivity

Answer 10

Classical theory claims category membership is transitive, e.g. if As are members of category B and members of category B are also members of category C, then As should also be members of category C;

However empirical studies show that people don’t categorise according to this rule: e.g. Hampton “Chairs are furniture; car seats are chairs but car seats are not furniture” (could be criticised on the basis that by “chairs” people mean “some chairs”)

Question 11

Classical theory evidence against - Lack of definitions

Answer 11

Putnam (1975) claimed that we can retain belief in a category even if its definition changes completely - e.g. we would still call them “cats” if they were proven to really be robots controlled from Mars rather than mammals;

Because the defining characteristics of a category can change as we discover new information, even though we continue to believe that it’s still a category, the classical theory that all members of natural categories can be defined by the common properties they share doesn’t make sense.

Question 12

Prototype theory claims

Answer 12

Concepts relate to the most typical example - the one that best matches a set of properties - not all or nothing like classical.

There is no necessary condition - can have many values but with typicality ratings (red apple more likely that yellow)

Question 13

Prototype theory evidence against - Typicality effects in definitional concepts

Answer 13

Armstrong et al. (1983) found robust typicality effects even for concepts that conform to a definition, not just opinion such as “even numbers” some participants rated 4 as a more typical even number than 34.

Implies categories have an internal structure. So suggest a dual-process model for categorisation…

1. A concept core is used for judgement of category membership (classical theory);
2. A set of identification procedures are used to match possible instances to a category (prototype theory).

Question 14

Prototype theory evidence against - Context sensitivity

Answer 14

Context affects how “typical” an instance is seen to be of a category

Medin and Shoben demonstrated that the influence of “large” and “small” on typicality was affected by the context of “wooden” or “metal” spoons

Prototype theory does not explain this instability effect - enough similar properties should be matched in each case for typicality to be equal, but this suggests that weightings are changing

Question 15

Prototype theory evidence against - Complex concepts

Answer 15

Combinations of concepts (“red car”) are hard to explain as it’s difficult to theorise how prototypes could be combined

Question 16

Common sense theories (the theory theory) claims

Answer 16

Several problems with similarity-based (classical/prototype) theories led to development of theories that deeper analysis processes are involved in categorisation.

Factors like familiarity, “technical” membership were just as good predictors.

Hard to define what similar means. Items can share an infinite amount of things in common… a dog and a planet share the property of not being a unicycle, and neither can type.

Question 17

Common sense theories (the theory theory) evidence - Murphy and Medin

Answer 17

Murphy and Medin suggested that category members share theories or explanations such as why items in the category behave in similar ways, and that categorising items is part of the process of explaining their shared behaviour - e.g. categorising someone jumping into a swimming pool fully clothed as being intoxicated. (seems a bit circular - e.g. “this robin is a bird (which is why it has wings and a beak)

Claim that we may not be as aware of our underlying theories as we are of property similarities, until novel instances appear (robot cats) or in borderline cases. In these situations we may realise that we’re using some underlying theory to categorise.

Question 18

Common sense theories (the theory theory) evidence - Rips

Answer 18

showed participants a pizza and a US quarter, then asked them how they would categorise a third object that was “bigger than a quarter but smaller than a pizza”

This dissociation between category and similarity shows that the underlying theory used is not just similarity-based.

Question 19

Common sense theories (the theory theory) evidence - Kiel

Answer 19

examined developmental aspects of categorisation

Children were told about a hybrid animal that looked and behaved like a zebra but that had a horse’s insides and horse parents.

Younger children (4yo) used appearance to categorise it as a zebra; Older (7yo) children used biological properties to categorise it as a horse;

Gelman (PAPER) raccoon that was disguised as skunk - Younger children said it had changed into a skunk, older ones said it was still a raccoon

Kiel suggested a characteristic-defining shift that occurred at the age where children moved from using appearance to categorise - in contrast Murphy suggests that older children have more knowledge, so older and younger children might use different properties within the same process rather than experiencing some developmental change.

Question 20

Common sense theories (the theory theory) evidence against

Answer 20

explains the role of deeper knowledge in categorisation and not just similarity of properties but doesn’t mean that it’s any more correct than similarity-based theories;

poorly-defined - may just replace the hard-to-define idea of “similarity” with an equally intangible one of “theory”

doesn’t explain how complex concepts are formed

Question 21

Psychological essentialism claims

Answer 21

also assumes deeper principles and explanations form the basis for categorisation;

Gelman (PAPER) People categorise objects according to unobservable shared “essential” properties that make the category members “what they are” (their essence) rather than on superficial properties;
Different from classical theory because while the categorisers know what the superficial properties of items are, they may not be able to define the essential ones that they nevertheless believe exist;
Children are a good source of evidence for this theory because if pre-school children who lack scientific knowledge show evidence of essentialism then it might be innate rather than acquired along with knowledge;

There is a “placeholder” for these essential properties eg when the platypus was found, everyone’s placeholder was empty. Experts came to fill it with “amphibian”, “mammal etc. Eventually it was decided that it was a mammal, so everyone now categorises it as a mammal (the ID in everyones’ placeholders).

Question 22

Psychological essentialism evidence for - innate potential

Answer 22

Innate potential:

Gelman (PAPER) children were told about animals thats biological parents were switched - which parents would it be more like?

Pre-school children reported that a newborn kangaroo adopted by goats would have a pouch and be good at hopping, showing they believed hopping was an innate property of kangaroos.

Children can be more nativist than adults: 5yo children claimed that adopted children would speak the language of their biological parents

Question 23

Psychological essentialism evidence for - Underlying structure

Answer 23

Gelman and Wellman found that young children believed the insides of objects were more important identifiers of an item’s category than the outside - a dog would still be a dog if its outside was removed, but would not be if its insides were.

Question 24

Psychological essentialism evidence for - Inductive potential

Answer 24

Gelman and Markman showed children a leaf, an insect and a leaf-insect:

When told that the labels for these were “leaf”, “bug” and “bug”, children generalised new facts about the insect to the leaf-insect;

This showed that words signify category membership even when instance properties are not obvious and that this happens with children as young as 1-2yo;
By 4-yo children use labelling to assess category membership but also use other criteria, such as whether instance features are permanent or temporary.

Question 25

Psychological essentialism evidence against - Influence of context and perspective

Answer 25

The amount of H₂O in “water” should be an essential property but Malt (1994) found people were more influenced by the source of water: tears were judged to be “not water” although containing 89% H₂O while pond water containing only 79% was judged to be water

“Tibby is a cat though we were wrong about her being a mammal” vs. opposite: “Tibby is not a cat, though she is a robot controlled from Mars”;

Suggests essentialism does not hold but that both context and perspective affect the consistency and contradiction of categorisation

Question 26

Psychological essentialism evidence against - role of expert opinion

Answer 26

Malt - Given the choice, more people would “ask an expert” to help categorise borderline natural cases (e.g. a tree that is half oak, half maple) than would ask for expert help with borderline artefacts (a half boat/half ship)

Suggests natural categories may involve psychological essentialism more because people recognise the additional information that experts bring

However Braisby (2001) found only half of participants changed their categorisation of biological categories when told it conflicted with that of an expert and quarter of them changed it when told it conflicted with the opinion of a non-expert (“a shopper”). This undermines psychological essentialism as it suggests only 25% of people changed due to expert opinion.

Question 27

Psychological essentialism evidence against - indirect evidence

Answer 27

Some of this evidence needs other inferences: e.g. the belief that removing a dogs insides stops it being a dog depends on the assumption that insides = essence in some way;

“Essence” is not required to explain this finding anyway (Strevens, 2000);

Also other common criticisms e.g. failure to explain complex concepts can be applied to this theory too.

Question 28

So which theory

Answer 28

All these theories appear to be flawed

Classical: necessary and sufficient conditions only identifiable for very few categories;

Prototype: doesn’t explain context sensitivity or complex concepts;

‘Theory’ theory: imprecise, no definitions, can’t explain complex concepts;

Psychological essentialism: mixed evidence, much of which doesn’t map onto essences.

Another possibility is that categorisation involves all of these (and possibly others too).

Question 29

Is all categorisation the same

Answer 29

It is possible that each type of approach is used under different circumstances

However if the way we categorise is completely context-dependent the results of categorising things would appear to be quite random and this is not the case - the empirical results show that we categorise systematically as the outcome is generally very similar between and within individuals.

So maybe we use a number of different techniques depending on the situation…

Classical: might be used to classify according to a definition (“a person is an adult if over 18-years of age”);

Prototype: might be used for fast matching or fuzzy matching (when all properties are not available);

Theory-based judgement might be used for creating explanations for category membership;

Psychological essentialism might be used to integrate scientific knowledge into categorising items.

Or maybe a dual-process of categorisation:
a similarity-based mode using classical//prototype techniques and a rule-based mode that uses theory/explanation based methods.

Malt cites the example of a container for shampoo that was not a bottle, yet was labelled a “shampoo bottle”, because it fulfilled the same purpose as a shampoo bottle would have.

People could also categorise differently depending on their knowledge and their goals (e.g. what they might use the items for). Medin asked taxonomists, tree maintenance workers and landscapers to categorise 48 tree names; first two used scientific methods, landscapers used use factors.