Please enable JavaScript.

Coggle requires JavaScript to display documents.

W7 Conceptual Knowledge, little gain of info, huge gain of info - Coggle…

- - - - typical member of category
      - generalised, average, abstract representation of category
    - - :arrow_up: typicality --> similar to prototype
      - :arrow_down: typicality --> not similar to prototype
    - - calculate how quickly people respond to questions about category membership
      - Typicality Effect
        
        Objects with :arrow_up: prototypicality --> faster RT
        
        :apple: faster than pomegranate
    - - Rosch
        
        primed with color word (eg. "green")
        
        judge whether 2 colours are same
      - faster responses for prototypical colours :green_apple: :green_apple: than nonprototypical colours :pear: :pear:
        
        suggests participants create prototypical images in response to color names
    - - people name prototypical members before nonprototypical members when creating lists
  - - - actual member of category
    - - object similar to more exemplars --> faster classification
    - - able to account for individual atypical cases
        
        eg. flightless birds can be exemplars for comparisons with penguin
        
        not lost in the overall averaged prototype
  - - - works well for strict technical categories (eg. geometric objects)
    - - Family Resemblance
        
        category members resemble each other in various ways
        
        allows for flexibility
- - - - Basic
        
        Specific / Subordinate
        
        kitchen table, Ford
        
        subordinate => more concrete
        
        eg. table, Car
        
        names usually taken from here for things
      - eg. furniture, vehicle
      - superordinate => more abstract
    - - list similar features for each level
        
        Basic level
        
        psychologically special
        for most people
        
        9 common features
        
        Global level
        
        3 common features
        
        Specific level
        
        10.3 common features
      - :one: show participants category label that is global (vehicle) or basic (car)
        
        :two: show picture of object and ask if picture is member of category
        
        faster RT when basic level category is used
    - - special level is different for everyone
        
        experts in category focus on low-level features in specific level
      - Tanaka experiment
        
        asked bird experts and nonexperts to name objects in pictures
        
        experts used more specific categories :arrow_right: named birds by species
        
        nonexperts used more basic categories :arrow_right: named birds as birds
      - scope of expert and basic knowledge can differ across cultures and languages
- - - - https://drive.google.com/open?id=17Rl0lU45z3-9HSSloSFQtRC_gSILn_Qu&usp=drive_fs
      - Features
        
        Link
        
        connects nodes to other nodes
        
        Node
        
        indicates concept :black_circle:
        
        properties of concepts indicated here
        
        Hierarchical Model
        
        Concepts sorted into levels
        
        specific concepts in bottom levels
        
        generic concepts at higher levels
      - Retrieving information (canary)
        
        go to superordinate level (bird) and access general properties at the parent concept
        
        Cognitive Economy
        
        storing shared properties at higher level nodes
        
        :arrow_up: efficient storage
        
        lower level nodes inherit properties
      - not meant to represent neurophysiology but how concepts are mentally represented
      - Sentence Verification Technique
        
        Canary is bird VS Canary is animal
        
        Canary is bird HAS faster RT
        
        Canary is bird :arrow_right: traverses 1 link
        
        Canary is animal :arrow_right: traverses 2 links
        
        Longer RT with larger separation in network
      - Spreading Activation
        
        activity that flows from activated node to other connected nodes through links
        
        activated nodes become primed
        
        Variation of Lexical Decision Task
        
        presented w pairs of words
        
        judge if at least 1 item in pair is nonword
        
        faster RT for associated words (bread & wheat) vs weakly associated words (chair & honey)
        
        recalling one word triggered spreading activation to related nodes
      - Criticism
        
        Typicality Effect
        
        model predicts equal RT for canary and ostrich --> one node away
        
        Sentence Verification
        
        Pig is mammal --> :snail: RT
        
        Pig is animal --> :zap: RT
        
        Model predicts "pig is animal" should have :snail: RT because it is 2 levels away
        
        Cognitive Economy
        
        people may store general properties at nodes at lower levels
        
        "can fly" stored at canary node
        
        Circular reasoning
        
        network structure can only be derived experimentally
    - - semantic network may develop very early
  - - - Features
        
        parallel distributed processing
        
        Weights
        
        relate to synapses
        
        decide how signal strength :arrow_up: or :arrow_down: at next unit
        
        Edges
        
        relate to axons
        
        connections that transfer information between units
        
        Nodes
        
        relate to neurons
        
        input :arrow_right: hidden :arrow_right: output
        
        Hidden Layer
        
        contains different number of nodes than input and output layers
        
        forces restructuring of information
        
        significantly increases possible input/output relationships
        
        internal representations of concepts
      - https://drive.google.com/open?id=17d7_rooeYp3dXnp1Nvt3lcN3ban2TURA&usp=drive_fs
      - represents stimuli or knowledge with patterns of activity distributed across many units
    - - Features
        
        Relation units
        
        ISA
        
        A canary ISA :bird:
        
        is
        
        A canary is :yellow_heart:
        
        can
        
        A canary can fly
        
        has
        
        A canary has wings
      - Concept Representations
        
        :one: Activate unit (canary) + relation unit (can)
        
        :two: spreading activation to some representation and hidden units
        
        :three: Weights cause some units to have :muscle::skin-tone-4: activation and others weaker
        
        pattern of activations in representational, hidden and property units
      - Training
        
        :one: Weights are uncalibrated initially
        
        :two: Generate input and check output
        
        :three: Error signal
        
        Back propagation
        
        error signals sent back through network from property units
        
        error signals at hidden and representation units provide feedback on adjustment of weights
        
        spreading activation to unrelated property nodes
      - https://drive.google.com/open?id=17iBaXH-b7VxlAs497I67NFn8xtyoqRA2&usp=drive_fs
    - - similarities between connectionist networks and real-life neurological phenomena
        
        Graceful Degradation
        
        partial damage to network (lose one node in hidden layer)
        
        selective deficits in functioning
        
        gradual disruption of performance
        
        Generalised learning
        
        similar concepts have similar ativation patterns
        
        can predict properties of novel concepts that are related to old concepts
- - - - patients with memory loss due to encephalitis
        
        suffer from category-specific memory impairments
        
        unable to identify animals but can identify furniture or tools
    - - system that distinguishes sensory attributes (stripes, spots) :arrow_right: living things
      - system that distinguishes functions :arrow_right: artifacts (non-living things)
    - - patient with sensory deficit
        
        poor perceptual test scores
        
        identify animals better than artifacts
      - patients who identify different types of artifacts
        
        artifacts not homogenous category like stated in hypothesis
  - - - asked to rate strength of association of object with a feature
        
        how much do you associate door with brown?
        
        features included colour, motion, performed actions (actions with object)
      - Animals
        
        :muscle::skin-tone-4: association with color OR motion
      - Artifacts
        
        :muscle::skin-tone-4: association with performed action
      - Music instruments and mechanical objects (machines, vehicles)
        
        :muscle::skin-tone-4: association with colour, sound, motion, performed actions
    - - alternative explanation for category-specific impairments
        
        difficulties distinguishing objects with similar featuresm (animals)
      - members of categories share many similar properties
        
        animals have eyes, legs, freedom of motion
  - - - aimed to create category map by measuring ffMRI data from listening to spoken language
        
        https://drive.google.com/open?id=17pw7pdx8t5nbbsjRJDzgCJ-qiybhokPp&usp=drive_fs
        
        area in back of brain processes words related to violence
        
        responses can also be distributed over several areas
        
        facial identification dependent on areas of emotion processing, facial orientation, attractiveness, FFA
  - - - mirror neuron fires when monkey performs action or observes action
    - - measured fMRI activity when participants
        
        read action words "kick :footprints:", "pick :hand:", "lick :tongue:"
        
        moved feet, index fingers, tongue
      - :arrow_up: activation for movements
      - Semantic somatotopy
        
        activation for movements and words associated with same body part occur in similar brain areas
  - - - Semantic Dementia
        
        damaged anterior temporal lobe
        
        general loss of knowledge for all concepts
        
        equally bad at identifying living things and artifacts
    - - Damage to spokes :arrow_right: specific deficits
        
        unable to identify artifacts
      - Damage to hub :arrow_right: general deficits
        
        semantic dementia
    - - measured RT for naming pictures or living things or artifacts
        
        TMS inactivation of inferior parietal lobe
        
        TMS inactivation of ATL
      - Parietal inactivation
        
        :snail: RT for manmade objects but not living things
        
        :snail: RT for high manipulatable objects :hammer_and_pick: but not non-manipulatable objects
      - ATL inactivation
        
        affected everything equally