Concepts and knowledge Flashcards
Concepts
- General knowledge of a category
- mental representation of a category held in semantic memory
Categories
Items that are grouped together according to concept, often concepts and categories are used interchangeably
Exemplars
Individual items within a category
Our multi-dimensional way of organizing concepts allow us to access them at different ____
Levels
3 levels of concept specificity
- Superordinate level
- Basic level
- Subordinate level
Superordinate level
General terms for concepts (e.g. fish)
Basic level
- Gives more specificity to the concept (e.g. shark is a fish)
- How we usually talk about things (we have a bias toward basic level in communication)
Subordinate level
- More detailed specific representation of a concept
e.g. this shark is a hammerhead
Order of levels in the development of concepts
Basic, superordinate, then subordinate concepts
Semantic dementia patients level of concepts
- Early in disease, basic level concepts are accessed
- A dog is a dog
- As the disease progresses, use general concepts
- A dog is an animal : accessing basic level is impaired, subordinate level instead
Cognitive economy
Use the simplest terms that is still meaningful for the situation
E.g.
General public : piano
Piano players : Casio privia
Graded concept organization
Grades how well a certain item can belong to a concept (e.g. trout is more representative of a fish than a shark)
Concept inclusivity
Ability to refer to a single object on more than one level
E.g. violin and instrument
Generalization
- Process of deriving a concept from specific experiences
- See examplars that seem to be linked together and figure out their communality to class new items as belonging to this concept category
Classic approach to concept learning
Concepts involve forming rules about lists of features
Defining features (classic approach to concept learning)
necessary and sufficient for category membership
e.g. a dog has to be a living thing
Characteristics features (classic approach to concept learning)
Common but not essential for category membership
Solution to the problem that not all examplars that we know are part of a concept have defining features (classic approach to concept learning)
- Feature comparison between encountered items and list
- Refines what a defining features is for a concept
The classic approach to concept learning does not work well for…
- Complex concepts that are subject to variability (e.g., a fur-less dog)
- Ambiguous concepts: ‘student’; a ‘bachelor’ … a ‘hot dog’
The Cube Rule
- Rules to define what food is based on starch position
According to this rule, a hot dog is a taco
Similarity-based approach
You use the similarity between items rather than any sort of explicit rule rule to figure out if something new you see, an exemplar instance, belongs to their concept.
According to the similarity-based approach, items are defined by their resemblance to …
A collection of features (more than just defining features like in the classic approach)
Fuzzy boundaries
Items are, more or less, part of a category
* An item can be categorized into more than one category
E.g. a sled can be a vehicle or toy depending on the context
Steps of concept learning
- We start off trying to learn simple rules and list of features to define a concept.
- We’ll test hypotheses to try to refine those rules.
- With more knowledge and exposure, then we might start to use more similarity based representations of concepts (network of similarity)
_______rule/similarity based representation is more representative of how we store concepts in our semantic memory.
Similarity
In a network of similarity, concepts that resemble a concept more are…
Closer to the concept (e.g. robin is closer to bird than penguin)
What do the prototype and exemplar theories of concept learning have in common ?
They do not have one required defining feature or attribute like in the rule-based approach
Difference between the exemplar and prototype theories
What we hold in memory to represent a concept to define similarity : more stored elements in Exemplar theory
- Exemplar based models assumes that concepts are represented by many specific exemplars
- Prototype theory assumes that concepts are represented by a central tendency, a prototype, which is an abstracted representation that comes from a bunch of exemplars that we’ve encountered.
Prototypes
Prototypes are formed from the overlap of exemplars that are extracted from experience : they represent the most common features across members
- This abstracted prototype is stored in memory to represent the category
Test of prototype theory results
- Participants learned to classify dot patterns that were variants of a prototype
They did not see this prototype
* Participants classified the studied patterns, new patterns and the prototype into groups
- Worse at classifying new compared to old patterns
- Equally able to classify prototype and old patterns
- They acted like they saw the prototype before : suggests they had a false memory of a prototype
- Evidence that people are forming prototypes in their mind when learning about concepts
Prototype mental network
The prototype is the center of a category network
* Other category members (exemplars) resemble the prototype to different degrees
* The more obscure members are farther away from the prototype in the network
How do we identify examplars as belonging to a conceps according to Prototype theory ?
You compare an examplar in the world to the prototype of a concept : if it corresponds to it enough, you decide it is a member of the concept
Typicality effect
A preference for processing items close to the prototype
- faster to recognize robin as bird than penguin
- People name more common members of a category that are closer to a prototype first when asked to list members of a category (robin before penguin)
Category name primes typical exemplars test results
Ask people to say if something is a word or not, and look at reaction time
Typical trial : Gave typical examples of a fruit like an apple after showing the fruit label (prime)
Non-typical trial : less typical and familiar fruit like papaya
- Fruit primed the reaction time for apple more than it did for papaya, which is less familiar.
- You’re faster to identify an apple as a fruit when it followed the word fruit than papaya because of the typicality effect.
According to prototype theory, concepts are context ______dependent/independent
independent (there is evidence this is not the case)
How context shifts what we think is a relevant member of a concept
What is seen as a typical exemplar depends on the context
E.g.
Is this a harmonica a typical musical instrument?
* Differs if you are around a campfire or at a concert hall
Context affects ______ effect
Typicality
The typicality of an exemplar is the result of …
Our concepts, the context that we’re in and our past experiences
What were the results of this test on typicality of exemplar depending on context ?
word chicken with picture of city vs farm
People rated chicken as a more typical bird after they saw an image of a farm compared to an image of a city
Exemplar theory
- There is no single abstract prototype for a concept
- Every instance of a category is stored in memory, not a prototype
How do we identify examplars as belonging to a conceps according to Exemplar theory ?
- Retrieve some or all exemplars of category members
- Compute similarity to new item at the time of concept determination (this computation happens at the time that you’re retrieving them)
- Conjure up all recent dog images from similar environments, and then pattern match those to what I see.
- If there is a strong overlap with one of my past encounters : it is a dog
Role of context in Examplar theory
Context can influence concept representations
- Experience and situational context are used to form concepts at retrieval
Knowledge-based theories
Explanation rather than similarity-based view of concept categorization
- Implicit intuitive knowledge used for categorization (not similarity or rules)
e.g. this is a fruit because it is fruit-like
Essentialism (knowledge-based theories)
The idea that certain categories have an underlying reality or true nature that one cannot observe
Embodied view of concepts
- Concepts are not rigid abstract representations
- Concepts are created in our environment to meet goals
How do we use our brain differently depending on what we are accessing about a concept?
- Our brain’s sensory, motor and perceptual systems change depending on what we are accessing about a concept
- We use cognitive capacities at different levels depending on what we need to know about a concept
- We simulate or re enact a concept in different areas of our brain depending on what we need to do
Ad-hoc categories
- A category concept that is invented for a specific purpose or goal
- Bringing together dissimilar members into a temporary category
- Related to creativity
(e.g. wood and hair are things that can catch on fire)
Knowledge is stored as _______ neural representations across our brains
Sensorimotor
Perception and conceptual knowledge are linked as …
‘perceptual symbols’
Activating a concept will engage certain ______-perceptions to engage mental simulation for the goals of the current task
Sensory
Concepts are not abstract, but we actually engage our _______ processes to create simulations of a concept for a given task.
Perceptual
E.g. Watermelon
* Is a watermelon tasty? taste regions
* Is a watermelon round? visual cortical regions
Property verification task: Does this percept matches the object?
- Participants are shown a concept paired with a perception and say if the perception matches the object
Participants are faster to respond to this task if a previous trial asked about a feature from the same sense/percept (percept match trials) than another perceptual feature (priming effect)
* Evidence that concepts are represented via senses/perceptions
Brain representations evidence of perceptual stimulation of concepts
Different patterns of activity in the brain when people read 3 words (pick, kick, lick) that were similar to activity when people actually use these body parts.
- Evidence that concepts are really rooted in motor and sensory actions, and they’re not just abstracted ideas in our brains
Neuropsychological case studies evidence of category specific deficits
Some have selective impairment in naming living things
* Cannot name animals but can name tools
Some have selective impairment in naming non-living things
* Cannot name tools but can name animals
According to sensory functional theories, living things defined by ______ features (in terms of perception)
Visual
- Visual processing regions engaged when accessing living concepts
According to sensory functional theories, inanimate objects (tools) are defined by _______ features
Functional
Motor cortical regions (related to action) engaged when accessing inanimate concepts
Damage to motor cortex could cause difficulty accessing ______ but not ______ concepts
Only difficulty accessing inanimate concepts
Superordinate concepts are ______, while subordinate concepts are ______.
distinctive but not informative; informative but not distinctive