announcements clarifying universal grammar vs internalized grammatical rules –ug = a theory that...

62
Announcements Clarifying Universal Grammar vs Internalized Grammatical Rules UG = a theory that all languages rely on a set of highly abstract, unconscious rules—innate knowledge of the general form that language can take (Nature) IGR = learning through experience the specific grammatical rules of a particular language but learning the overall “rules” rather than word by word (Nurture) Fast-mapping (the word need not be contrasted with anything, it is quick word to world mapping/pairing through brief exposure or incidental learning rather than direct teaching)

Post on 21-Dec-2015

216 views

Category:

Documents


0 download

TRANSCRIPT

Announcements• Clarifying Universal Grammar vs Internalized

Grammatical Rules– UG = a theory that all languages rely on a set of highly

abstract, unconscious rules—innate knowledge of the general form that language can take (Nature)

– IGR = learning through experience the specific grammatical rules of a particular language but learning the overall “rules” rather than word by word (Nurture)

• Fast-mapping (the word need not be contrasted with anything, it is quick word to world mapping/pairing through brief exposure or incidental learning rather than direct teaching)

Mental State Reasoning:

‘Theory of Mind’

Today’s Plan: • Last of Video on Non-human ‘language’ and symbol use• A little Trivia Exercise• Research on mental state reasoning or ‘Theory of Mind’• How one’s ‘theory of mind’ can help learn language

Theory of mind is the ability to reason about psychological/mental states in one’s self or in someone else.

(e.g. “mind”-reading, not ESP)

How do infants, children, adults, non-human animals (?), understand or reason about the mind e.g. Emotions, Desires, Intentions, Knowledge,Thoughts/Beliefs, etc.

It is one thing to HAVE emotions, thoughts, etc.—it is quite another to be able to THINK about them!

Why do we need a theory of mind?

For example:

• Sarah’s dog is missing.

• Sarah hears a scratching noise in the shed.

• She goes to the shed and opens the door and finds a squirrel.

• She begins to cry.

How do we make sense of behavior? (e.g. triangles moving clip, Stewardess in the aisle that bumps herself on the forehead)

Reason 1: Understanding what a person is thinking, feeling, believes etc helps us predict how they will act and/or interpret their behavior

Why do we need a theory of mind?

e.g. teaching

Reason 3: Learning through social transmission (knowing who knows more, who to learn from)

e.g. deciding who or when to learn from someone

Others reasons: e.g. deception, surprises, tricks, jokes, etc.

Reason 2: Successful communication with another person requires appreciating what the other person knows, doesn’t know, and how knowledge can be acquired.

How to test for a ‘Theory of Mind’?• Historical Background: Research began in 1970s with

chimps—do they reason about the beliefs of other chimps?– Where will Chimp A think Chimp B will look? Location A or

Location B (where the food is)– Chimp A expected Chimp B to look in Location B– Interpretation: Chimp A thinks Chimp B believes the food is in

Location B– Criticism: Could succeed without any understanding of the

mind of the other, to show theory of mind they must reason about a false belief (counter to reality)

• The birth of the classic test: False Belief tasks as a litmus test of a theory of mind

1. Displacement Tasks (aka Sally-Ann or Maxi Tasks)

2. The Unexpected Contents Task (aka the “Smarties” task)

3. Appearance-Reality Tasks

Video Segment: A Change of Mind

• Results from the classic false belief tasks:

–4 year olds succeed at the task, 3 year olds fail!!

• What do the results mean?

–Children younger than four do not understand that people can hold beliefs that are false?

• Or is there an alternative explanation?

False Belief Results

Two Broad Theories

1) Radical Shift Theory—Conceptual Change View – children prior to the age of four are unable to attribute belief

states to themselves and others.

– lack concept of false beliefs—kids have a ‘copy-theory’ of the world (the mind = reality)

2) Processing Demands Theory – More continuous view of development

– young children fail these tasks because they lack the attentional, mneumonic, linguistic, or processing resources, rather than the ability to attribute false beliefs.

Definition: A difficulty appreciating a more naïve perspective as the result of being biased by one’s own knowledge.

A ‘Curse of Knowledge’ Account

ExamplesCar salesmenAnagram Study‘Sarcastic’ passages study

An example of a Processing Demands View

Parallels Between Adults’ and Children’sKnowledge Reasoning

Predicting What Others will Think• Subjects read descriptions of events that could have

various outcomes. In one condition, subjects were told the outcome, in another condition they were not.

• They were asked to judge what others who did not know the outcome would predict.

• Subjects who knew the outcome thought others would be much more likely to predict that outcome. . Fischhoff (1975)

Like the children who claimed Sally would know that her chocolate had been moved, adults claimed that others would share their outcome knowledge.

Recalling Your Own Earlier Thoughts• Subjects were asked prior to Nixon’s trip to China and the

USSR to estimate the probability of the various outcomes.

• 2 weeks - 6 months later they were asked to recall their predictions

• Subjects remembered giving much higher probabilities than they actually had to the outcomes that took place.

Fischhoff & Beyth (1975)

Parallels Between Adults’ and Children’s Knowledge Reasoning

Like children who claimed they knew all along that there were pencils in the box, adults showed a more subtle effect by giving biased recollections of what they had previously predicted.

• Do young children find it easier to assess what someone else will know when they are not “cursed” with knowledge?

• Are younger children more susceptible to the curse of knowledge than older children?

• Is this why young children do poorly on false belief tasks?

Questions of Interest

Familiar Unfamiliar

“Percy’s played with all of these toys before! He brought all of these from home.”

“Percy’s never everseen these toys before! These are brand new.”

Method

No Curse (Ignorant) Cursed (Knowledgeable)

Children were told there wasa special little thing in each one.

Children were shown what wasinside each one before it wasclosed again.

Experimenter: “Does Percy know what’s inside this one?”

PredictionsChildren would judge that Percy knows what is inside

the toys he is familiar with, and will not know what is inside the toys he is unfamiliar with.

But, the curse of knowledge would work against this appreciation, leading children to overestimate Percy’s knowledge when they knew what was inside.

The magnitude of the curse of knowledge would decline with age.

Yes Responses to “Does Percy Know What is Inside?”

0

1

2

3

4

5

6

Age 3 Age 4 Age 5

"Yes

" R

espo

nses

Familiar

Unfamiliar

Significant main effect of familiarity Significant effect of familiarity at each age

p < .05

p < .05

0

1

2

3

4

5

6

3 4 5

Age in Years

"Yes"

Resp

on

ses

Child Knowledgeable

Child Ignorant

“Does Percy know what is inside?” (for Unfamiliar toys)

• 3- and 4-year-olds, but not 5-year-olds, overestimated Percy’s knowledge when they were knowledgeable.• The magnitude of the curse significantly decreased from age 3 to age 5.

Birch & Bloom (2003) Psychological Science

“Does Percy know what is inside?” (for Familiar toys)

• an asymmetry in perspective-taking--No ‘curse of ignorance’• when taking another point of view—it is harder to put aside one’s knowledge than one’s ignorance.

•You get the same asymmetry pattern with adult’s reaction times and trivia judgments

0

1

2

3

4

5

6

3 4 5

Age in Years

"Yes"

Resp

on

ses

Child Knowledgeable

Child Ignorant

Percy’s Knowledge Study: Conclusions

• Children are sensitive to the knowledge states of others, but the curse of knowledge can work against this sensitivity

• The magnitude of the curse of knowledge decreases significantly from age 3 to age 5

• Children’s knowledge assessments are biased asymmetrically and cannot be reduced to ‘egocentrism’.

– They are biased by their knowledge when assessing what someone else knows, but are not biased by their ignorance (see Birch & Bloom, 2003, Psychological Science)

False Belief ExperimentWould adults experience difficulty attributing false

beliefs if they were knowledgeable of the outcome, compared to adults who did not know the outcome?

False Belief Experiment

Participants

• 210 Introductory Psychology Students

Design

• ‘Sally-Ann’ Questionnaire using 4 locations instead 2.

• Probability judgments were measured instead of where will she look

False Belief Attribution

• Vicki places her violin in the blue box and goes outside.

False Belief Attribution

• While Vicki is outside, her sister, Denise, comes in and rearranges all the boxes.

False Belief Attribution

Cursed (standard) Condition = object is moved to “the red” container (the participant knows where it is)

??????

No Curse Condition = object is moved to “another” container (the participant does not know where it is)

When Vicki comes back she wants to find her violin, what’s the probability she will first look in each of the containers?

Red Green Purple Blue

No Curse 23% 3% 2% 71%

Curse 34% 4% 6% 59%

Birch & Bloom (under review), Cognition; see also Birch & Bloom (2004) Trends in Cognitive Science

Mechanisms and Implications

Mechanisms (Why the decrease with age?)• Increasing inhibitory control?• Increasing source memory?

Implications• Decreased empathy? Increased guilt, shame?• Indirect effects (social judgments)?• Education, Law, Business, Politics, all

interpersonal communication

This is Jeff. This is Linda.

Artwork by Susan Chen

Jeff pulls Linda’s hair.

How mean is Jeff?

•There is more to passing the false belief task than theory of mind.

•There is more to theory of mind than passing the false belief task

Keep in mind…

Solving the Reference Problem• Whole Object Bias• Taxonomic Bias• Basic-level Bias• Shape Bias• Linguistic Context (grammatical form of the word)• Syntactic Bootstrapping (grammatical structure of

sentence)• Mutual Exclusivity Bias• Theory of Mind and Pragmatics

So how do they do it?

Theory of Mind and Word Learning• Pragmatic explanation of Lexical Contrast

• Joint attention and Inferred Intent?

• Sensitivity to when a goal is satisfied (Bucket Study)

• Deciding Who is More Knowledgeable. Who do I want to learn from? (e.g. Tracking knowledge over time)

• Body language, gestures, and paralinguistic info indicative of knowledge/confidence

Mutual Exclusivity (revisited)

Where is the blicket?

“Look at this one. It’s a jop! “Look at this one. It’s nice. See, it’s a jop. This is a jop”. This one is cool. It’s neat.”

“Can you give me the bem?”

Theory of Mind and Word Learning

The speaker wants one of those two objects. If she wanted me to give her the one she called jop she would have done so in some way I would understand (e.g. She knows I know what a jop is). But instead she used a different word--She must intend to refer to something different.

“Can you give me the bem?”

Pragmatic (Theory of Mind) account of Lexical Contrast

“Look at this one. My uncle “Look at this one. It’s nice. gave me this one. My uncle This one is cool. It’s neat.”gave this to me. This is theone my uncle gave to me.”

“Can you give me the one I keep in the kitchen?”Same Speaker Results: The other one--not the one my uncle gave meDifferent Speaker Results: 50/50Not specific constraint to word learning, but specific to the mind of the other person.

Theory of Mind and Word Learning• Pragmatic explanation of Lexical Contrast

• Joint attention and Inferred Intent?

• Sensitivity to when a goal is satisfied (Bucket Study)

• Deciding Who is More Knowledgeable. Who do I want to learn from?

• Body language, gestures, and paralinguistic info indicative of knowledge/confidence

Joint Attention & Inferred intent/desire Eyes as a Window to the Mind

What the eyes can say

Keeping Track of Previously Knowledgeable Speakers

History Phase

“I think that’s a spoon. Yeah, that’s a spoon.I think that’s a comb. Yeah, that’s a comb.” etc.

History Phase

“I think that’s a fork. Yeah, that’s a fork.I think that’s a shoe. Yeah, that’s a shoe.” etc.

Naming Phase

Ben:“I think that’s a ferber.Yeah, That’s a ferber.Do you see the ferber?”

Jenny:“I think that’s a ferber.Yeah, That’s a ferber.Do you see the ferber?”

Testing Phase

“Can you give me the ferber? Where’s the ferber?”

Testing Phase

Different Word Test Condition:“Can you give me the koba? Where’s the koba?”

??

Word Learning: Paralinguistic and Nonverbal Cues

Imitation: Nonverbal Cues

“There’s Jessie!”

Putting Cues togetherExample: Learning a Proper Name

• Syntax, Linguistic Context (Katz, Baker, & Macnamara, 1974)

• Range of reference (Hall, 1996; Hall & Belanger, in press)

• Eye-Gaze, joint attention (Baldwin, 1991)

• Mutual Exclusivity (Clark, 1993; Hall & Graham, 1999; see also Diesendruck & Markson, 2001)

• Animacy (Gelman & Taylor, 1984; Hall, 1994; Katz et. al., 1974)

–Ownership/hierarchy of importance? (Hall)

• Knowing about knowledge, Familiarity, Theory of Mind (e.g. Birch & Bloom, 2002)

Cues to a Proper Name’s Referent

• Do children understand the relationship between knowledge and familiarity? That one needs familiarity or experience with someone to know about unobservable properties (e.g. siblings, proper names)

• Are they sensitive to the speaker’s knowledge state, and can they use this to help them learn the referent of a new word?

That is, if a speaker uses a proper name such as “Jessie”, do they know it is more likely to apply to an individual with whom she is familiar?

Question of Interest

Familiar Unfamiliar

“I brought these from home.I’ve played with all of theseanimals before.”

This bag of animals is ‘discovered’.“Wow! I’ve never seen that dog before.”

Method

Proper Name Condition:

“Where’s Jessie? Can you find Jessie?”

Common Noun Condition: “Where’s the dog? Can you find the dog?”

0%

20%

40%

60%

80%

100%

Adults Age 4 Age 3 Age 2

0%

20%

40%

60%

80%

100%

Adults Age 4 Age 3 Age 2

Proper Name Condition Common Noun Condition

Results

Proper Name Study: Summary

Children as young as two are sensitive to the speaker’s knowledge state when learning new words.

–They appreciate that the speaker must be familiar with an individual in order to know its proper name.

–They appreciate that familiarity with a specific individual is not necessary to know a common noun.

Book Recommendations

‘The Language Instinct’ ‘How Children Learning the Meanings of Words’by Steven Pinker by Paul Bloom