Exam 3 Ch 9 Flashcards
Category
A set of things in the world that have something in common
Concept
Mental representation of a category
Categorization
when we place an object into a category
5 Theoretical Approaches to understanding human mental representation of categories [Placeholder]
.
Definitional (Classic) view
There are Strict rules/characteristics that define category membership
- The characteristics/rules are both necessary and sufficient for objects to be part of that category
Implication:
- All-or-none category membership
- All members are equals
- Clear categorical
Examples?
- Triangle (necessary because 3 sides, 3 corners)
- Bachelor (sufficient since no strict rules)
- Cat? (Sufficient since no strict rules meow, zoomies, purrs, furry usually, whiskers, carnivore, paws)
Problem: not how we (humans) actually think… This is called definitional view
Prototype
Lies at the center of a concept
- Family resemblance
- An imaginary ideal thing that has ALL of the most typical features
- A mental representation
Categorizing
Graded membership
- An object is a member of a category to a degree that it resembles the prototype
Prototypes build up with experience
The Typicality Effect
Graded membership
Members closer to the prototype are “better” members of that category (mostly birds)
Categories center around prototypes, but the boundaries are ambiguous
Typicality Ratings
Rate how typical an example of each item is of the category 1 (low) - 7 (high)
- Some category members are considered better than others
ex: An apple is considered more typically a fruit than a coconut
Sentence Verification Task (RT)
Read a sentence if the sentence is true smack the desk
Rate how typical an example of each item is of the category 1 (low) - 7 (high)
[[[Typicality Effect]]]
- The more typical an item is (high number), the faster we verify it. (interpretation it’s closer to your mental representation of a prototype of a fruit)
Sentence Verification Task (RT)
Read a sentence if the sentence is true smack the desk
- The more typical an item is, the faster we verify it. (Interpretation it’s closer to your mental representation of a prototype of a fruit)
Production Task (naming)
- List as many members of the following category as you can (for example birds)
Typicality Effect shows here: the more typical an item is, the more likely we are to output it first
Ex: For birds I would first think of robins, cardinals, bluejays
Priming [covered in book]
Occurs when the presentation of one stimulus facilitates the response to another stimulus that usually follows closely in time
Categorization [Might delete this]
We categorize things by comparing them to our mental representation [concept] of the category
Actual members of the category that we encountered in the past
- Explains the typicality effect by proposing that objects that are like more of the exemplars are classified faster
Prototype theory is not context-sensitive and does not handle variability among category members
Prototypes don’t capture our knowledge of category variability
ex: there is a prototype for cars but it does not tell us how many members of that category there are
Exemplars
Actual members of categories a person has encountered in the past
We store all of the category members (store individual examples)
- Can explain typicality effects just as well we prototype theory
It can do what prototype theory cannot
- Handles variance within categories
- Handles atypical cases better (have exemplars stored of penguins and ostriches)
- Handles ad hoc categories better
– Things to grab from your house if there’s a fire
– Things you take to the beach
– Things you take to the beach
– Things in your backpack
– Things you wouldn’t tell your mom
Prototypes vs Exemplars
If there were 4 objects protype theory would average those objects into 1 thing
Exemplar would store the actual examples
Semantic Network (aka Associative Network) [placeholder]
.
Nested Concepts
Super-ordinate (EX: animal, furniture)
Basic-level (EX:, bird, chair)
Subordinate (EX: warbler, wooden desk chair)
Where are the features most likely stored?
- basic level: middle level of abstraction for categories.
basic levels concepts are first to be learned and is the most economical place for most features to be stored, natural level at which objects are named,
highest level at which the objects all share the same parts and overall shape. (ref: Rosch, Mervis, Gray, Johnson, & Bayes Braem, 1976)
Evidence about basic levels being special? (Two experiment examples) 1st
- Rosch, E., Mervis, C. B., Gray, W. D., Johnson, D. M., & Boyes-Braem, P. (1976). Basic Category Levels and Production Task
what about EXPERIENCE? Rosch’s idea of PROTOTYPES was thought to be built up from experience, and thus each person’s prototype of a category could be different.
would hierarchical structure of category levels be the same way?
Evidence about basic levels being special? (Two experiment examples) 2nd
- Results of Tanaka and Taylor’s (1991) “expert” experiment. Experts (left pair of bars) used more specific categories to name birds, whereas nonexperts (right pair of bars) used more basic categories. Role of Expertise
the “novices” were actually dog experts who had no experience birdwatching
cool thing: some evidence that as people gain expertise, what were previously subordinate levels become their basic levels! like warbler, or poodle, or real-time strategy games.
Parts of a semantic network
Nodes - Major concepts
Links - represents relation
Features - Properties/features are associated with each concept
Things about a semantic network [placeholder]
.
Symbolic
The nodes and links don’t correspond directly to brain physiology
Hierarchical [skip it means exactly what it says]
… obvious
Inheritance (cognitive economy)
Property info is stored as high up the hierarchy as possible to minimize redundancy
We do not still information on everything that has two legs, we take this as a given
Spreading Activation
When one node is “activated” (perceived. or retrieved from LTM), activation spreads (It primes the other nodes)
Lexical Decision (Online Experiment) [Showed semantic priming, understand the results]
You were shown one word then a second word then you had to decide if the second word was real
IV: related words vs unrelated words
DV: reaction time
When primed by a semantically related word the second word was guessed faster
This experiment is evidence that supports the theory of spreading activation
Sentence Verification Task (Collins & Quillian 1969) [Understand the experiment and what the RT mean]
Sentence Verification Task:
respond Y/N is the sentence (proposition) true
This research supports the evidence that verifying information further away in the semantic network it takes longer
DV: RT depends on distance in semantic network
(slide 22 of 50 ch 9 part 2)
A canary can sing. [0levels]
A canary can fly. [1 level]
A canary has skin. [2levels]
A canary is a canary. [0 levels]
A canary is a bird. [1 level]
A canary is an animal. [2 levels]
Problems for semantic network view
- Cannot explain the typicality effect
– Robin is a bird is verified faster than Penguin is a bird even though they’re the same number of links to be traversed
Connectionist Network (aka neural network)
- Inspired by brain physiology, AND effort to model a computer program
- A concept is represented by pattern of activity distributed across many units
- Individual “units” are like neurons, that don’t have any meaning by themselves
Symbolic network (aka semantic network)
- a concept is represented by a node
- not tied to brain physiology
Connectionist Network Parts
Input Units (like sensory neurons)- receive input from outside of the system
Hidden Units (like interneurons) - majority of neurons in your brain that do not interface with the outside world
Output Units (like motor neurons) - Movement
Links- Connections between neurons: axon and synapse. link weights vary and representation long term potentiation
long term potentiation- neurons that fire together wire together
Parallel Distributed Processing (Connectionist Network)
Def- spreading activation all at once between nodes across weighted connections
- LEARNING: process of changes connection in weights
The network starts with all equal link weights
You can try to activate it with a word like robin and the first time the output will be nonsense
Our brain tells the network what is wrong with the output and then the network makes singles propagate backwards through the network to adjust link weights
After many trials the link weights adjust correctly to what a robin can actually do as well as other things using all of the same set of inputs
General Representations [4 coming up also this is a placeholder]
.
Schema
mental representation of what is typically expected in a particular situation
Script
Schema with order
Concept
mental representation of a category of things, their features
Stereotype
mental representation of a group of people, their characteristics
