quiz 3

two properties of language

1. hierarchical

2. rule based

3. unique to humans

creation of psycholinguistics

Chomsky: language is productive (produce new combinations), comprehension and representation are partially innate

levels of langugage analysis

1. phoneme (smallest unit, eg. [th] in thin

2. morpheme (combinations of phonemes, eg. )

3. word

4. phrase

5. sentence

language development milestones

1. babbling @ 7 mo

2. multi word phrases @ 2 yo

Goldin-Meadow 1972

children will create their own language (deaf)

productivity/generative nature of language

1. produce infinite combinations

2. completely novel sentences

phoneme

• tested with minimal pairs (bed vs. bag —> ed and ag change the meaning, therefore meaningful)

• every language has toolbox from 200 phonemes

• vary across speakers (region, gender, age, etc.)

• vary across contexts

coarticulation

• overlap in phonemes across contexts makes you pronounce it differently

• eg. p in pot vs. spot

• look @ spectrographs to determine presence of coarticulation

categorical perception

• only able to percieve distinctions in the sounds that we speak

• eg. ta vs. tha in Hindi

speech segmentation

• how we separate individual words without actual pauses

• insertion of pauses based on our prior knowledge (top-down)

phoneme restoration effect (Warren & Warren 1970)

• insert relevant phoneme depending on context

• It was on the __eel (wheel, heel, peel) depending on the different cue

morpheme

• smallest unit of sound that has meaning

• eg. re in repay, recall

• means to REDO something

what is parsing, conceptually?

• take a sequence of words

• build structured rep of relation of words

• sentences are more than word list

• incremental and iterative, parses as you go

• choose between competing structures

• more than syntax —> semantic, context matters too

sound-meaning mapping is

random,

lexicon

LTM for words

lexical decision task (meyer and schevendalt 1971)

• determine if a presented word is indeed a word

• POE vs. familiarity (lexical access)

frequency effect (Rayner et. al 2003)

• eye movement measures how fast items were retrieved from lexicon

• more frequent words = faster rxn time

• look longer @ more frequent words

ambiguity of words

• homonyms

• polysemes (same word, many meanings —> diamond - baseball diamond, gem, shape)

how do we understand polysemes?

• biased/balanced dominance

• in the context, what is most likely? depends on FREQUENCY

biased dominance

• tin as a metal is more common than tin as container

• therefore more frequent use = first to come to mind

balanced dominance

• cast as in plaster vs. cast as in actors

• both are equally common therefore you need context to determine which is which

biased vs. balanced dominance, experimentally

• w/o context —> you take more time with balanced dominant words, less with biased

• w/ context —> dominance + context both play a role, takes more time to determine less common word

lexical priming

• semantic context changes interpretation of next word

• eg. they all rose vs. she held a rose

• probe = flower (response is faster for she held a rose vs they all rose)

phrase structure

• hierarchical org of sentences

• helps us parse sentences

• syntactic (noun + verb)

incremental nature of parsing

• helps circumvent limitations in WM

• supported by Aaronson 1979

• final words in the PHRASE not hte sentence take longer

• that’s bc/ we pause and think

garden path model of parsing

1. syntax (grammar, structure)

2. semantic (meaning)

first step of garden path parsing

• structural analysis based on heuristics

• rules of thumb

• top-down

late closure

• attach new words on ongoing phrase (you don’t close the gate early enough)

• avoids new structure

• While Susan was dressing the baby played on the floor

minimal attachment

• build the simplest structure

• fewest “nodes”

• The burglar blew up the safe with the dynamite

constraint based parsing

• parallel parsing

• simultaneously using many types of linguistic and contextual info to create interepretation

• not just linguistic

examples of constraint based parsing cues include

• word meaning

• scene context (VW paradigm — eye movements)

• override ambiguous structure with word meaning

  • The spy saw the cop with a revolver

  • we know you can’t see a cop through a revolver

• memory load/prior experience

garden path sentences are determined by

• word meaning

• the horse raced past the barn fell

example of how scene context impx. parsing

• tannenhaus 1995: apple, pencil, towel

• ambiguous sentence was interpereted differently based on different image

• one apple vs. two apple

how do memory load/prior exp impact parsing

• some constructions of sentences can be more/less familiar

• subject-relative construction

• objective relative construction

subject-relative construction

• the senator [who spotted the reporter] shouted

• describing the subject with a phrase inserted right after

object-relative construction

• the senator [who the reporter spotted] shouted

• creating a subject-verb, then inserting an object phrase in between

constraint based parsing cues are

• memory load/prior exp

• scene/story context

• word meaning

• all are activated in parallel, weighted differently

situation models (Zwaan 1996)

• mental reps of events in a text or a movie

• easier to identify the obejcts that are consistent with our situation model

• eg. orientation of hammering nail into wall vs. floor — faster to find the floor oriented nail

N400

• semantic ERP component @ 400 ms

• highly sensitive to congruency (how well does the item match its context)

• spikes when unexpected stimulus shows up

  • eg. cats won’t eat vs. bake

  • bake spikes more than eat bc/ we don’t expect cats to bake

N400 is evidence of

• contextual processing evidence

• it explains how each word fits in the context semantically

spiking order of N400

• unrelated spikes most

• related spikes middle

• expected spikes LEAST

P600

• mismatching syntax causes ERP drops @ 600 ms

• eg. cats won’t eat vs. cats won’t eating

P600 shows

• sensitivity to semantic (n400) and structure (p600) congruency

language development ordering

• cooing @ 6 weeks

• babbling @ 7 mo - consonant vowel sequences (eg. mama)

• telegraphic speech @ 2 years - words without unnecessary sentences (eg. want food)

statistical learning

• regularities in speech

• allow predictions of morphemes (6-7 months)

• later, use prosodic cues as well (8-9 months)

holophrastic stage

• one word stage

• over/underextend

example of overextension

every tall man = dada

example of underextension

only pet = doggie

3 theories of language development

• behaviorist (skinner)

• nativist (chomsky)

• interactionist (vygotsky)

behaviorist theory of language acquisiton

• all learned through operant conditioning

• 3 elements:

  • imitation

  • reinforcement

  • practice

nativist theory of language

• innate and physiologically determined

  • believed in innate grammar

• LAD (language acquisition device) is preprogrammed to extract grammar rules

support for nativist theory of language

• poverty of the stimulus

  • don’t need too much exposure to language to be able to pick it up

  • children do not hear every single possible sentence, but they can still make new ones anyways

• critical period: a certain time period within which they pick up most of their language, biologically determined time = innate

language neural correlates

• language is mostly left centered (hancoke and hubble) (70-80% split)

• ventral and dorsal stream

ventral stream of language

• map phonological (sound —> meaning) to concepts

dorsal stream of language

• map phonological to motor/sensory

• frontal lobe

what is mental imagery

sensory impressions without a sensory input

can be audio, haptic, olfactory

one of Wundt’s 3 pillars of conciousness

premise of imagery debate

is imagery spatial or propositional (language based)

Kosslyn image scanning task

• drawing of island with hut, tree, rock, etc

• farther from target = increased RT

• evidence for spatial based imagery bc you can easily find things closer, shows they were holistically scanning

• IV: distance from target

• DV: RT (reflects mental rep distance)

Pylyshyn

• propositional view

• abstract, symbolic propositions

• list of relationships, not just a picture

• eg. LEFT OF (hut, tree), RIGHT OF (tree, lake) etc

  • we see things relative to one another

what did Pylshin believe about spatial values

epiphenomena (a feature of the relationship) between the two

imagery vs. perception

• mental images can be manipulated

• eg. imagine elephant and rabbit, then rabbit and ant

consensus about imagery and perception

• interaction between the two

• letter flashing task (Farah 1985)

  • create image of letter

  • flash —> letter or letter —> flash

  • accuracy is higher when the letter matches

neural evidence of mental imagery

• imagery neurons in MTL (medial temporal lobe) - kreiman 2000

  • firing rate of neurons were equal for real and imagined objects (baseball vs. face of friend)

• V1 (striate cortex) activity responds to real and imagined stimulus

• Kosslyn topographic mapping

  • larger imagined object = more frontal visual cortex activation

imagery beyond visuals

• auditory cortex activated in schizo pts. hallucinations

• receptive language activated during hallucinations

Kosslyn rotation task

• proved that imagery can also be motor based

• physical rotation then imaged rotation

• prior physical rotation activated new areas compared to only imagining rotation

double dissocation

• RM — occipital and parietal lobe damage

  • recognize + draw objects

  • not from memory

  • could not make judgements about the objects (eg. grapefruit bigger than orange)

• CK — visual agnosia

  • recognize parts of objects

  • draw objects from memory

  • could not integrate

individual differences suggested by

• galton

• 19th century

• varying degrees of vividness

Kozhevnikov 2005

• questionaire to determine visualizers vs. verbalizers

• gave visualizers two types of imagery tests

  • spatial (paper folding task)

  • object (VVIQ - rate vividness)

• lower paper folding score had higher VVIQ score

• high paper folding score had more people with low VVIQ score

degraded pictures/mental rotation

• degraded pic: low quality pic, what’s the # —> object imagers

• mental rotation: same or mirror image? —> spatial imagers

cart/magnet problem

• spatial imagers better at answering that B would see it more forward

• concluded spatial imagery needed for many physics problems

conclusions from Kovzhenikov

• some people are better at one type of imagery vs. another (spatial vs. object)

properties of a problem

• obstacle between current and goal state

• needs 3 things:

  • goal

  • current affairs

  • actions/operators that can get you around it

gestalt approach

problem solving is just restructured mental approach

how does restructuring happen per gestalt

• insight

• sudden realization (eureka type)

Metcalfe and Wiebe

• insight vs. algebra

• participant judgements of how close they were to a solution (warmth)

• warmth increased sharply (exponential) as they got closer to answer

obstacles to problem solving

• fixation

  • functional fixedness

  • mental set

functional fixedness

only see the usual function of objects

mental set

stuck on previous working strategy

examples of functional fixedness

• Maier - two strings, pull together with a chair

example of mental set

• water jug problem

• how many steps/how do you get the easiest solution

Newell-Simon

• info-processing theory, stepwise

• problem space = problem states + operators

• problem states = series of intermediate states between current and goal

features of Newell-Simon’s info processing theory

• iteratively searching through problem space to identify possible operators and how to get to the next step

• means-end analysis

means end analysis

• reduce difference between intl. and goal states

• create subgoals (tower of hanoi)

• intermediates that are closer

• create more subgoals that are more feasible (eg. operator can be applied)

problem solving strategies

• insight

• random search (randomly try the next operator)

• hill climbing (choose the action that gets you closest)

• analogical (transfer one experience to the other)

2 stage process of analogical transfer

• observe relation between source and current domain, map them

• apply knowledge from og to this

Radiation problem analogy

• fortress vs. tumor, mined roads vs. surrounding tissue

• participants figure out the same solution by mapping one problem to the other

expertise

• more experience with a certain type of problem

• does not demonstrate BETTER problem solving skills, just faster

• spend more time in problem analysis

expertise and memory (chase and simon)

• chess grandmasters could better memorize board when meaningful possible positions

• not advantaged over novices bc/ no chunking

expertise and organization (Chi and Glaser)

• expert vs. novice physicists

• novice grouped by visuals

• physicists group by conceptual underpinnings

expertise is

domain-specific

creativity emphasizes

• divergent thinking - solutions with open ended solutions

• seen as 4 step process

4 steps of creativity

1. problem finding

2. problem formulation

3. problem solving

4. solution implementing

creativity and knowledge

• existing knowledge blocks creativity

• Smith 1993: providing examples = influences the nature of solutions

• Ward 1994: imagine aliens — many had animalistic features

role of incubation

• circumvents fixations

• unconcious problem solving

• release from blocking (eg. functional fixedness)

• aka forgetting fixation

remote association task (smith/blankenship)

• 3 seed words (arm, coal, peach)

• locate target word (pit)

• measures convergent thinking — unify many items into a single rep

• incubation only impx fixation group (group that had the misleading hint)

creativity neural correlates

• Chi and Snyder — solve 9 dot problem

• deactivate left ATL, activate right ATL w/ tDCS

  • left inhibits out of the box thinking

• same level of problem solving as people who were given hint (40%)

right ATL role

creative and problem solving skills improved

left ATL role

streamline categories

AUT (alt uses)

• name as many alt uses for a brick

• divergent thinking

• DMN (ACC - anterior cingulate cortex) more active in idea generation

  • related to mind wandering

• ECN (selective attention)

  • directs attention during tasks

EEG activity in RAT

• frontal activity before insight solutions

• occipital lobe activity increases before non-insight solutions