Chapter 9

Concept

mental representation of a particular type of thing

Difference between prototype and exemplar

-A prototype is an average — there is only one prototype -Prototypes may work best for larger categories: sandwhiches -A exemplar is just one example — there are many exemplars •A new object is compared to each exemplar — what is the summed similarity across all exemplars -Exemplars seem necessary for small categories: presidents But likely use both!

Exemplar Approach

-Concept is represented by multiple examples (rather than a single prototype) -Examples are actual category members (not abstract averages) -To categorize, compare the new item to stored examples

_______ is a representation of the "typical" or "average" member of a category.

A prototype

Spreading Activation

Activation is the arousal level of a node •When a node is activated, activity spreads out along all connected links •Concepts that receive activation are primed and more easily accessed from memory

The prototype approach to categorization the prototype for a category is based on

All category members that have been encountered in the past

Olin and Bob are neighbors. Olin loves birds and his father works for the zoo. He has been to a dozen bird sanctuaries, and he and his dad go on bird watching hikes once a month. In contrast, Bob doesn't think much about birds. His only contact with them is in his backyard. It would be correct to say that Olin's bird-related knowledge structure likely has

All category members that have been encountered in the past.

Semantic Network Model

Assumes concepts are arranged in networks that represent the way concepts are organized in the mind •Node = category/concept •Properties are attached to nodes •Relationships among Concepts are represented in links between nodes

Criticisms of Semantic Network Models

Cannot explain typicality effects •"ostrich is a bird" is verified more slowly than : •"canary is a bird" (the typicality effect) •even though the same number of nodes are traversed •Some sentence-verification results are problematic for the model •"pig is an mammal" is verified more slowly than "pig is a animal"

Which of the following statements is NOT a reason why categories are useful?

Categories allow you to know an object's name the first time you see it.

high prototypicality

Category member closely resembles category prototype Example: For category "bird" = robin

The Connectionist Approach

Computer models for representing cognitive processes (also called Parallel distributed processing) •Knowledge represented in the distributed activity of many units •Weights of the connections determine how strongly an incoming signal will activate the next unit

Definitional Approach:

Determine category membership based on whether the object meets the definition of the category

o Definitional Approach:

Determine category membership based on whether the object meets the definition of the category

Exemplar approach explain findings

Easily account for Low-prototypicality members (low summed similarity) •Explains typicality effect in sentence verification •Explains priming effects

Basic level

For each category list as many features as you can that would be shared by all members of the category •Do this for different levels •Global (e.g., furniture) •Basic (e.g., table) •Specific (e.g., kitchen table) Going above basic level results in a large loss of information •Going below basic level results in little gain of information

Based on what we know about the different category types, jumping from _______ categories results in the largest gain in information.

Global level to basic level

basic level is special

Going above basic level results in a large loss of information •Going below basic level results in little gain of information Regular people used basic level labels for the birds •But the experts used specific level labels for the birds

Advantages of connectionist approach

Graceful degradation: disruption of performance occurs gradually as parts of the system are damaged •Generalization of learning: because similar concepts have similar representations, the properties of one exemplar are naturally transferred to a new exemplar

If activation spreads in the network, lexical decision should be faster for those pairs

Meyer and Schvaneveldt (1971) •The prediction was verified — evidence for spreading activation

Evidence for spreading activation

Meyer and Schvaneveldt (1971) used the Lexical Decision Task •Participants read stimuli and are asked to say as quickly as possible whether the item is a word or not •Two words shown at once •"Yes" if both strings are words; "no" if not

In the semantic network model, a specific category or concept is represented at a

Node

Hierarchical Organization

Organization of categories in which larger, more general categories are divided into smaller, more specific categories. •These smaller categories can, in turn, be divided into even more specific categories to create a number of levels

Levels of Categories

Rosch observed that although there are several levels of categories that people understand, the basic level is special (psychologically privileged) Specific Basics Global

Collins and Quillian (1969) •Predict that the amount of time it takes a person to retrieve information about a concept should be correlated with the distance that has to be covered in the network. •use the sentence verification technique for different sentences

Sentence verification time increases as the number of nodes that must be traversed increases

Cognitive Economy

Shared properties are only stored at higher-level nodes •Exceptions to higher-level properties are stored at lower nodes (penguins can't fly) •Inheritance Lower-level items share properties of higher-level items

People playing the parlor game "20 Questions" often use hierarchical organization strategies. One player asks up to 20 yes/no questions to determine the identity of an object another player has selected. The player's questions usually start as general and get more specific as the player approaches a likely guess. Initial questions asked by a player are often one of three questions: "Is it an animal?" "Is it a vegetable?" and "Is it a mineral?" Each of these three questions describes which level of categorization?

Superordinate (global)

•Remember repetition priming? •Doesn't have to be exactly the same item repeated: •Processing one stimulus facilitates later processing of similar stimuli (bread -> butter)

Tested if reading a color word could prime detection of actual colors They were slower if the color patches had Low-prototypicality -What happens if there is no prime? •Book leaves out a critical control condition

The connectionist network has learned the correct set of properties for a concept when

The error signals are reduced to nearly none and the correct properties are assigned.

Solution: Connectionism

These results suggested the semantic network we've been talking about is not correct •But it doesn't mean there is no semantic network at all •Perhaps it is just organized differently •Connectionist networks allow for more complex organizations

Back Propagation

When the network gets the wrong answer, an Error Signal is generated •Difference between actual activity of each output unit and the correct activity •Back-propagation: error signal transmitted back through the network •Indicates how weights should be changed to allow the output signal to match the correct signal •The process repeats until the error signal is zero

Category

composed of all possible examples of a particular concept ex: animals is category- dogs, cats, etc.

If you wanted a network to have cognitive economy which option would you choose? The property ______ is stored at the ______ node

can fly; bird

low prototypicality

category member does not closely resemble category prototype •Example: For category "bird" = penguin

Not all of the members of everyday categories have the same features. Most fish have gills, fins, and scales. Sharks lack the feature of scales, yet they are still categorized as fish. This poses a problem for the ______ approach to categorization

definitional

Conceptual Knowledge

knowledge that enables us to recognize objects and events and to make inferences about their properties

Categorization

process by which things are placed into categories

Smith et al. (1974) — evidence that prototypicality influences thinking Used sentence verification

shown a sentence and say yes/no to indicate if it is true Subjects given sentences about category membership With items that varied in prototypicality "An apple is a fruit" "A blueberry is a fruit" "A pomegranate is fruit"