Cognition Final

Concepts

Mental representations of a class or individual

Category

All possible examples of a particular concept

Conceptual Knowledge

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

What additional questions does conceptual knowledge help answer

• How do we tell items in our environment apart? (Apples and oranges)

• What criteria do we use to categorize an item? (What makes an apple and apple?)

• What are the various kinds of ‘things’ that exist in the world?

Why are categories useful?

• Pointers to knowledge

  • Items categorized allows us to know a lot

  • Allows focusing our attention on what is unique about the item

• Helps us to understand items being encountered for the first time

• Lightens the cognitive load, allows us to create an extension of our current body of knowledge

• See something an make inferences about of its properties

Pitfalls

• Heuristic-based thinking

• Leads to overgeneralization, stereotyping, etc.

Family Resemblance

Idea that ‘things’ in a category are related through overlapping similarities

• Doesn’t require a single shared feature

• Captures ‘lumpiness’ of categories

• Features correlate across instances of the category

Prototype

‘Typical’ member of a category

• Based on an average of common category members

Typicality

The degree that a particular item is representative of its category

Typicality [Rosch (1975)]

High prototypicality: a category member that closely resembles the category prototype

• “Typical” member

• For the category “bird” = robin

• High family resemblance

Low prototypicality: a category member that doesn’t loosely resemble the category prototype

• For the category “bird” = ostrich

• Low family resemblance

Typicality Effect [Smith et al. (1974)]

Participants are faster at identifying objects as belonging to a certain category that are high in prototypicality

Prototype Approach [Rosch(1975)]

Highly prototypicality members of a category are more affected by a priming stimulus than members with low prototypicality

Summary of qualities of highly prototypical objects:

• High family resemblance (i.e., share many features with other members of a category)

• Categorized/recognized rapidly

• Named first in a category-member recall task

• Primed better

Priming

Presentation of one stimulus facilitates the response to another stimulus

What are the problems with the prototype model?

• We’re only storing the prototype. What about all the members from which we created the prototype?

• To categorize something, there’d need to be existing prototypes for every conceivable category

• What about ad hoc categories? (Categories without an existing prototype)

Exemplar

An actual member of the category that has been encountered in the past

Exemplar Approach

Category membership is determined via comparison to many ‘exemplars’

• Explains typicality effect

• Easily accounts for atypical cases (flightless birds)

• Easily deals with variable categories (games)

Exemplar Approach and Prototype Approach Comparison

• Similarity to prototype approach:

  • Representing a category is not defining it

• Difference from prototype approach:

  • Representation is not abstract, assumes storage of individual exemplars

Is exemplar or prototype approach more accurate?

• Exemplar: the more you know of something the better you can identify outliers

  • Exemplar is reinforced with experience

Hierarchical Organization

Organization in which general ‘umbrella’ categories area divided into smaller, more specific categories

Levels of categories in hierarchical organization

Global (superordinate)

• Furniture

• Vehicle

Basic

• Chair; table; bed

• Car; truck; bicycle

Specific (subordinate)

• Kitchen; dining room

• Single; double

• Ford; Chevy

• Pickup; van

• Road; trail

Semantic Network Approach [Collins & Quillian (1969)]

A type of semantic model for how concepts and properties area associated in the mind

• Hierarchical model

  • Node = concept or category

  • Nodes (concepts or categories) are linked

• Cognitive economy

  • Common features (properties) aren’t repeated, but rather placed at a higher-level nodes

    • Exceptions of properties found at lower-level nodes (e.g., property of ostrich, “can’t fly”)

• Evidence

  • Activate robin will activate other things in category by going up to bird and back down to canary

  • Reaction time for identifying and associating birds will vary depending on the bird

Connectionism

The brain can be studied using brain models or neural networks that perform sophisticated computations with simple neuron-like elements

Connectionist Approach

Most models based on two principles:

• Our experiences rooted in patterns of action potential’s activity

• Our ability to remember the past on long lasting modifications of synaptic connections

Hebb proposed the idea of a cell assembly as a functional unit of the nervous system

McCulloch-Pitt Neurons

• Simplified model—a neuron fires an action potential when a sufficiently large number of excitatory synapses on it are activated simultaneously

• Utilizes Boolean logic

• 0 = false, 1 = true

Connectionist Approach [McClelland & Roger’s (2003)]

• Theory as to how different concepts might be represented. Such as ‘canary’

• Stimuli represented by the pattern of activity across the network

• This model requires training to obtain this activation pattern

Connectionist Approach—Training a Network

• The first presentation of a concept unit may randomly activate many units across the network evenly

• Training occurs when error signal are sent back through the network (via back propagation)

• Think about a thermostat

• Supported by McClelland & Roger’s (2003) computer simulation of training a network (via adjusting connection weights)

Sensory-functional hypothesis

• Living things —> sensory properties

• Artifacts —> functions

Different brain areas may be specialized to process info about different categories

Multiple-factor approach

Objects have many features in common that are shared across other categories

• Crowding: when different concepts within a category share many properties

Embodied approach

Knowledge of concepts is associated with reactivating of sensory and motor processes that occur when we interact with the object

• Mirror neurons: fire when we do a task or we observe another doing that same task

• Semantic somatotopy: correspondence between words related to specific body parts and location of brain activation

Semantic category approach

Specific neural circuits for specific categories

• Hub and spoke model

  • Damage to the anterior temporal results in semantic dementia

  • Damage to the spoke causes specific loss (e.g., artifacts)

  • Damage to the hub causes complete loss

• Overstimulation sends pulses that interfere with brain processes

Social categorization (SC)

Concerned mainly with the way individuals categorize themselves and others

Function:

• Helps make prediction of how people will act

• Generalize behaviors

Social Categorization—Inference

• SC facilitates and constrains two kinds of inductive inference: downward and upward

• Deductive inference: when features of a category are attributed to individual members

• Inductive inference: when creatures of an individual are generalized to the category (i.e., group)

Social Categorization—Intergroup Relations

• Individuals more similar to other members within group than outside

• Both attributions and generalization benefit from this similarity

• SC present in infants and leads to in group favoritism/outgroup homogeneity

Social Categorization—Erroneous stereotypes

Stereotypes can be incorrect and harmful

Mental imagery

Ability to create sensory impressions (recreate the sensory world) even in the absence of actual, physical stimuli

• Visual imagery

Imagery and thinking—Early ideas

• Imageless thought debate: some believe it is possible while others think it impossible

• Behaviorists dismissed the imagery debate altogether, because visual imagery is invisible to everyone except the one experiencing it

Imagery and Cognitive Revolution [Paivio (1963, 1965)]

Paired-associate learning

• Memory for words that evoke mental images is better than for those that do not

Conceptual peg hypothesis (the ‘why’)

• Concrete nouns create images that other words can hang on to, which enhances memory for those words

Dual coding prediction

• Memory will be better for concrete word pairs because they are coded two ways (verbal + imagery)

Mental Chronometry

Inference of cognitive processes by determining the amount of time needed to carry out various cognitive tasks

Mental Chronometry [Shepard & Metzler (1971)]

• Participants would use mental rotation to determine whether two figures were the same

• The time it takes participants to identify two rotated objects as the same is proportional to the degree of rotation

  • Imagery and perception might share similar mechanisms

Mental Scanning Experiments [Kosslyn (1978)]

• Mental scanning: creating mental images and then scanning them in your mind

• Mentally ‘traveling’ a further distance takes longer, which implies that visual imagery is spatial in nature

Imagery—Spatial mechanism

Spatial representation—different parts of an image are described as corresponding to specific locations in space

Imagery—Propositional mechanism

Propositional representations—relationships can be represented by abstract symbols (equation), or statements (“the cat is under the table”)

Imagery [Pylyshyn (1973)]

• Propositional representation

  • Moving around the boat to different points to scan the area of the boat

• Depictive representation

  • Scanning across space and creating an image of the boat

Spatial Representation—Comparing Imagery and Perception

• Moving physically closer to an object (a car), has two effects:

1. The object fills more of your visual field

2. The details of that object are easier to see

Comparing Imagery and Perception [Kosslyn (1978)]

• Participants were asked to picture/imagine different animals, which filled different proportions of their visual field

• Participants were able to answer questions about the animals (e.g., “does the rabbit have whiskers?”) more quickly when the rabbit was pictured as being located more closely

Interactions of imagery and perception [Perky (1910)]

• Participants were asked to mentally “project” an image onto a screen

  • Very dim images projected on the back of the screen

• Participants described “imagining” the images that were being dimly projected

• An actual stimulus (perception) impacted the participant’s mental imagery

Interactions of imagery and perception [Farah (1985)]

• Participants visualizes a letter on a screen. Then two squares flash sequentially on screen, one contains a letter. Participant determines whether the letter appeared in the first or second square

• Accuracy for this task is higher when letter shown is same as letter imagined

• Participant’s mental imagery impacted performance on a perception task (imagery—>perception)

Imagery and the Brain [Kreiman et al. (2000)]

• Single cell recordings from electrodes placed throughout the medial temporal lobe (think amygdala and hippocampus) show that neurons respond preferentially to some objects, but not others

• These same neurons would respond preferentially to imagining some objects, but not to imagining other objects—known as imagery neurons

Primary Visual Cortex [Le Bihan et al. (1993)]

Activity (as measured via fMRI) in striate cortex neurons increases when perceiving visual stimuli and when imagining visual stimuli

Visual Cortex [Kosslyn et al. (1995)]

• Looking at small objects leads to activity at the most posterior part of the visual cortex

• Activity will spread more anterior as the objects being viewed grow larger

• Imagining objects (mental imagery) of different sizes mimics this activity pattern

Whole Brain [Ganis et al. (2004)]

• Is there an overlap in overall brain activity when perceiving an object, as compared to creating a mental image of that object?

• There is almost complete overlap in frontal lobe, but not as much in visual cortex

Whole Brain [Pearson (2015)]

• Activity patterns evoked by visual perception and visual mental imagery are more similar when futrther along the visual stream

Multivoxel Patter Analysis [Johnson & Johnson (2014)]

• Trained a classifier (computer program) on brain activity in response to visual stimuli

• Later feed the classifier the brain activity in response to an unknown (to program) stimulus being (1) perceived or (2) imagined, to see if it can predict the target stimulus

• Classifier is able to predict the image being seen or imagined at a rate higher than chance

Neuropsychological Case Studies [Farah et al (1993)]

• Patient M.G.S.: 

  • Did the mental walk task (imaging walking towards an animal until it overflows the visual field) before and after resection of the right occipital lobe. Estimated stopping at a 15-foot distance before occipital lobe removal, and 35 feet after

  • Her mental imagery experience mirrored that of actual visual perception, where her visual field had in actuality, shrunk

Neuropsychological Case Studies [Bisiach & Luzzatti (1978)]

• When a patient with unilateral neglect imagined a place he had been before (Piazza del Duomo), he neglected to name objects that would have appeared to the left of his mental image

• Parietal lobe associated with orientation of reality is in the where pathway

• Temporal lobe or where pathway deficits affecting visual and imagination

  • Limits our perception in visual field

Unilateral neglect

Patient ignores objects in one half of the visual field

Animal Communication

• Animal communication is quite rigid as compared to human language

• There are lmited sounds/gestures that communicate a limited number of things—all of which are important for survival

• Human language, however, can be used in a countless number of ways

What is language?

• Hierarchical system

• Governed by rules

• Universal

Hierarchical System

Made of components that can be combined to create sentences; sentences are combined to create stories

Governed by Rules

There are specific ways that components of language are to be arranged:

• You can say: “What do you want for lunch?”

• You cannot say: “Want you for lunch what?”

Universal

• All neurotypical humans develop language

• Language development is similar across cultures

• All languages have nouns, verbs, negatives, past/present, and questions

Broca’s Aphasia

Individuals have damage in Broca’s area in frontal lobe

Wernicke’s Aphasia

Individuals have damage in Wernicke’s area in temporal lobe

• Speak “fluently“

B.F. Skinner

Believed that language was learned through reinforcement

• Children are rewarded for “good” behavior (correct language) and punished (not rewarded) for “bad” behavior (incorrect language)

Noam Chomsky

Believed that language was coded in genes (therefore, the underlying basis of language is similar)

• Children produce language that they have never heard (“I hate you mommy!”) that hasn’t been reinforced

Major Concerns of Psycholinguistics

Comprehension

• How do people understand spoken and written language?

Representation

• How is language represented in the mind?

Speech Production

• How do people produce language?

Acquisition

• How do people learn language?

Language Comprehension—Difficulties

• Word frequency effect

• Variable word pronunciation

• Speech segmentation

Lexicon

All of the words we know, our “mental dictionary”

Semantic

The meaning of language

Lexical semantics

The meaning of words

Rainer & Duffy (1986)

• Word frequency effect—tendency to respond more quickly to high-frequency word (home) than low frequency words (hike)

• Tracked durations of fixations for high-frequency vs. low-frequency words. First fixations and overall gaze lasted longer for low-frequency

Pollack & Pickett (1964)

• Variable word pronunciation—not all words are pronounced in the same way (i.e., accents, speed of speech, slurring of words)

• Recorded the voices of the participants while they waited for the study to commence. When presented with the recording of singular words (out of context) the participants could only identify 50% of the words, even though they were the ones that had spoken them

Speech Segmentation

Perception of individual words even in the absence of actual pauses between words

• Understanding of meaning helps segment words

Influence of Speech Segmentation on Word Comprehension

1. How frequently we’ve encountered a word in the past

2. Context around the word

3. Knowledge of statistical regularities (pre-tty ba-by)

   1. Tty more likely to follow pre than precede ba

Lexical Ambiguity

A word can have more than one meaning. Conext is needed to clarify the intended meaning of the word

Understanding Ambiguous Words—Tanenhaus et al. (1979)

• Recall that the first presentation of a stimulus is believed to activate mental representation of the stimulus, enabling a person to respond more reapidly when the stimilus is presented again. Lexical priming

• Participants briefly accessed all meaning of a word before relying on context to determine the accurate meaning

Meaning Dominance

Some words are used more frequently than others (“many words have multiple meanings, but these meanings are not all created equal”)

Syntax

Structure of a sentence, which follows rules for combining words into sentences

Biased dominance

Tin (type of metal) is more common than tin (small metal container)

Balanced dominance

Cast (members of a play) is equally common as cast (plaster cast)

Garden Path Model

• Suggests that syntax drives out initial processing of sentences (late closure). Context and meaning play a role later

• Syntactic process operates heuristically, applying our best guesses first

• We reanalyze syntax/meaning of the sentence until the derived meaning is ‘good enough’

Parsing: Constraint-based approach

Other factors and info in addition to syntax is involved in processing as a person reads/hears a sentence

• Influence of word meaning

• Influence of story context

• Influence of scene context

• Influence of memory load/prior language experience

Influence of Word Meaning

• Meaning of words can influence parsing right off the bat

• Word ‘defendant’ presents two possible meanings of ‘the defendant examined’, while ‘the evidence examined’ does not

Influence of Scene Context—Tanenhaus et al. (1995)

• Visual world paradigm: determining how scene info influences how a sentence is processed

• “Place the apple on the towel in the box” vs. “Move the apple that’s on the towel to the box”

Subject-relative Construction

“The dog that chased the cat is very friendly”

• Main clause: The dog is friendly

• Embedded clause: The dog chased the cat

• Dog is subject of both clauses

Object-relative Construction

“The dog that the cat chased is very friendly”

Which is more difficult: Subject-relative Construction or Objective-relative Construction? Why?

Object -relative construction:

• This sentence structure more demanding of one’s memory

  • We have to hold the early part of the sentence “the dog that the cat” in memory until we find who did the “chasing”

• This sentence structure is more complicated

  • The dog is the subject of the main clause, but the object of the embedded clause

• Object-relative construction is less common in English

  • About 6.5% of relative clause constructions are subject relative

Coherence

Representation of the next in one’s mind that creates clear relations…

• Between different parts of the text

• Between parts of the text and the story’s main topic

Inference

Readers create info during reading that is not explicitly stated in the text

• Anaphoric: connecting objects and people

  • “I take the kids out and we fish. And then, of course, we grill them.”

• Instrument: tools or methods

  • “William Shakespeare wrote (with what?) Hamlet while he was sitting at his desk

• Causal: events in one clause caused be events in a previous sentence

  • Sharon took an aspirin. Her headache went away

Understanding Text & Stories—Stanfield & Zwaan (2002)

• Subjects heard sentences and then indicated whether the picture was the object mentioned in the sentence

• Subjects responded more rapidly with the image that was consistent with the situation model (mental representation) of the sentence

Understanding Text & Stories—Metusalem (2012)

Concert Scenario

• The band was very popular and Joe was sure the concert would be sold out. Amazingly, he was able to get a seat down in front. He couldn’t believe how close he was when he saw the group walk out onto the (stage/guitar/barn) and start playing

Measured ERP response to three different versions of a sentence

• Unexpected responses led to a larger N400 response

  • Things associated with scenario are activated, even if they are unexpected in the context of the sentence

Given-new contract states that sentences should be constructed so that they include two types of info:

1. Given information—info already known to listener

2. New information—info being hard by the listener for the first time

If not followed, sentences are more difficult to comprehend

Having Conversations—Yule (1997)

• Common ground is the mental knowledge and beliefs shared between individuals who are conversing

• Referential communication task—12 cords depicting different abstract figures

• Participants get better (faster) at this task across trials because they’re establishing common ground

Having Conversations—Branigan et al. (2000)

• Syntactic coordination: tendency to use similar grammatical constructions in conversations

• Syntactic priming: increased chance of producing a sentence with the same syntactic construction as was just heard

• Found that 78% of time, participants would use same syntactic construction to describe picture

Other skills necessary for people to engage in effective conversation

• Theory of mind

• Nonverbal communications

• Turn taking

Theory of Mind

Ability to understand what others feel, think, or believe

Nonverbal Communications

Ability to interpret and react to another person’s gestures, facial expressions, tones of voice, other cues to mean

Turn Taking

Ability to anticipate when it is appropriate to enter the conversation

Entrainment

Synchronization between two partners/conversational parties

Problem

Occurs when there is an obstacle between a present state and a goal, it is not immediately obvious how to get around the obstacle