Concepts, Categorization, and Language in Cognitive Psychology

Conceptual Knowledge

enables us to recognize objects and events and to make inferences

about their properties.

Concepts

Mental representations of categories

Categorization

The process by which objects are placed in categories.

Concepts are definitions

A particular object has to have the defining features of a category to be a member of that category. Membership is all or none

Problems with concepts are definitions

1) Categories do not have clearly defined boundaries (borderline members)

2) Some categories are difficult to define (e.g., "game")

3) Category membership is not all-or-none.

Typicality and category membership

Typical items are judged category members more often

Typicality and Speed of categorization

Speed of categorization is faster for typical item

Typicality and learning

-Typical members are learned before atypical ones

-Learning a category is easier if typical examples are provided

Typicality and Language comprehension

In language comprehension, references to typical members are understood more easily

Typicality and Language Production

-In language production, people tend to say typical items before atypical ones (e.g., "apples and lemons" rather than "lemons and apples")

What makes an item typical?

Feature frequency

Family resemblance

Members of a category have a family resemblance to each other. Membership is a matter of degree

Ideal Member

Has most or all of the most common features

Atypical member

Has some of the common features

Not a member

Very few or none of common features

Prototype Theory

Potential members of the category are identified by how closely they resemble the prototype

Prototype

average of members of a category that are commonly experienced It is a summary description - a set of weighted features

Posner & Keele (1968)

First phase: Exemplars with feedback.

Test phase: old and new exemplars, new prototypes

Results: Old exemplars approximated new prototypes in percent correct

Shows that people create accurate prototypes when given exemplars

Account of typicality for prototype theory

comparing to a concept’s central tendency

Problems with Prototype Theory

1) Fails to capture boundaries of concepts

2) Typicality is context-dependent

3) Typicality and category membership do not always go together

Exemplar-Based Reasoning

exemplars are stored in memory, rather than a prototype or a rule. To categorize a new instance, you match it to the stored exemplars

Exemplar

representative example

Superordinate category hierarchy

Broadest terms (mammal, fish)

basic (category hierarchy)

default •

intermediate level of specificity (e.g., chair, rather than desk chair)

easy-to-explain commonalities

subordinate (category hierarchy)

specific types of basic categories (brown trout, cocker spaniel)

Evidence that basic categorization is privileged

Preferred level for naming objects •

-Objects at basic level are recognized faster • -

Easier to state commonalities •

First learned (labels and categories) by children

cross-cultural and individual differences in categorization

Which level is the basic level is not universal. Cross-cultural and individual differences exist

Armstrong et al

Some even numbers were rated as better examples of the category than others, even though mathematically it is absurd.

Typicality and category membership don't always go hand in hand

Shows that typicality is not always 100% correlated to category membership because if it were, all numbers would be judged the same.

Category membership is independent of resemblance

There can be membership without resemblance (flattened striped lemon) and resemblance without membership (counterfeit money)

Psychological essentialism

the belief that members in a category have an unseen property that causes them to be in the category and to have the properties associated with it.

Keil's Raccoon study

Preschool children believed you could turn a coffee maker into a pot but not a skunk into a raccoon.

Knowledge-based approach

Our knowledge of the world is used in learning and thinking about concepts.

donker vs. blegdav

When we learn new concepts, we try to connect them to the knowledge we already have. Donkers are easily connected to knowledge we already have but Blegdav has conflicting information.

Issues with psychological essentialism

-Essentialism is not applied to all properties of a category (e.g., we can be a doctor without having doctor parents).

-Essentialism is not our only belief that plays a role, but our knowledge in general matters

Concepts as theories

Concepts are theories (i.e., bodies of knowledge about a particular domain) that describe the facts/beliefs about categories and why their members cohere.

Inferences based on theories

Categorization is important as it allows us to make inferences (we can apply knowledge to new cases; draw broad conclusions). Inferences about categories are guided by typicality and our background knowledge that relates to the concept

Principle of Cognitive Economy

Properties are stored at the highest possible level • Concepts below inherit these properties

Sentence verification task

It takes longer for people to respond the higher the level is

Semantic network

consists of nodes representing concepts, joined together by pathways that link related concepts

Hierarchical Model

As applied to knowledge representation, a model that consists of levels arranged so that more specific concepts, such as canary or salmon, are at the bottom and more general concepts, such as bird, fish, or animal, are at higher levels.

Knowledge Network problems with typicality

Predicts that two of the same type of something will be the same time but

A robin is a bird. < An ostrich is a bird.

Knowledge Network problems with hierarchical structure

Predicts that a pig is an animal would be slower than a pig is a mammal but the opposite is true

Knowledge Network problems with association effects

Predicts that a peacock and robin having feathers would take the same amount of time but peacock < robin

Nonredundancy may not hold

Predicts that shark can move < a fish can move < an animal can move but actually all take about the same amount of time.

Communication system

Transmission of a signal (sound, motion, etc.) that conveys information

Language

People are not hard-wired to just learn the language of their biological parents. •

-can talk about anything (things that are not present, lies, about language) •

-can create an infinite number of words and sentences (productivity

Productivity

We can create an infinite number of utterances by combining a finite number discrete linguistic units in different ways.

Phoneme

smallest unit of sound that distinguishes words

Morpheme

Smallest unit of meaning

Kanzi

A bonobo who received linguistic attention and consequently developed a remarkable ability to communicate using lexigrams and to understand spoken English.

Lexigram

a geometric shape used as a symbol for a word

Differences between humans and apes in language

Universal acquisition for children; variable acquisition in apes

-Differ in ease of learning

-children experiment. Apes copy

-differences in usage

Nativist view of language

-Innate learning device

-thought that language input we receive is by itself too poor to learn language

- language and cognition are independent faculties

Anti-Nativist

general-purpose learning device •

-thought that we receive enough information in our language input to acquire language if we are actively engaged in our environment.

-language and cognition are interlinked

Broca's Aphasia

nonfluent/expressive.

left inferior frontal cortex •

dysfluent agrammatic speech •

good comprehension but troubles with grammar

Wernicke's Aphasia

fluent/receptive

temporal lobe

Poor comprehension;

Speech production is fluent but often meaningless speech with word-finding problems

Williams Syndrome

genetic anomaly

-cognitive disability

-hypersociability

-better vocab but poorer syntactic processing compared to mental age controls

Developmental language disorder

typical cognitive abilities, language impaired at all levels

-development is deficient for no apparent reason

-more common than Williams Syndrome (7%)

Statistical Learning

Sequences of syllables occur more often together within words than across words,

Transitional probability

the likelihood that one particular sound will follow another one to form a word

Is statistical learning the general-purpose language mechanism?

We are born with it, it is domain general (works for tones and visual patterns), BUT it is not unique to humans.

Mental lexicon

the mental dictionary of words and their meanings

Visual-World Paradigm

Tracking eye movements to study sentence processing.

Flanker task

an experimental procedure in which participants respond to the direction of the central arrow in an array, regardless of the direction the other arrows are pointing

Benefits of being bilingual

better cognitive control

Bialystok (2004)

Age-related decline in cognitive control is less in bilinguals than in monolinguals.

BUT: Others failed to replicate some of these effects.

Spivey and Marian (1999)

Russian-English bilinguals looked more at the "marker" than the distractor. For bilinguals, words from both languages can compete for recognition even when just listening to one language!

lexical competition

all words stored in our mental lexicon compete for recognition to the degree that they temporarily match the input

Cohort competition effect

People will look more at cohort rather than distractor

Rhyme competition effect

People will look more at rhyme rather than distractor