lecture 11 - language and thought Flashcards
Does the language we use affect the way we think?
- Sapir-Whorf Hypothesis
- Does language mean thinking?
- Turing Test
Searle’s Chinese room
Sapir Whorf Hypothesis
- Linguistic relativism
- Linguistic differences mirrored by nonlinguistic differences
- Linguistic determinism
- People think differently because of differences in language
Linguistic relativity
- People think differently because of differences in language
- Eskimos have 7 words for snow
(or 4? Or 100s?) - English only have 1 word for snow
(but “powder” “hardpack”?)
Experts often have more terms for concepts than novices
categories and concepts
for different breeds of rabbit to a novice they are just rabbits but to an expert they are satin rabbit, amami rabbit, swamp rabbit and a brush rabbit
Words that are difficult to translate say something about other cultures
- Salmon Rushdie
As the concept that’s important for one society/ culture doesn’t exist in the other
eg Waldeinsamkeit is German for the feeling of being alone in the woods
culaccino is Italian for the mark left on the table by a cold glass
sombremesa is Spanish for the time spent after lunch or dinner talking to the people you shared the meal with
Linguistic relativity
- Shouldn’t be ethnocentric
Shouldn’t assume world is carved up in the way reflected by our language
Linguistic determinism
- We think in a particular way because of our language
- Whorf claim: language of the Hopi has no expressions for time
- No past, future, present tenses
- No units of time
- “persistence” “announcing”
Test this experimentally?
Colour terms and processing
- Languages have different numbers of colour terms
- Classical Greek no distinction between blue and black
- Do colour names determine perception of colour?
Different languages have numbers of colour names
Heider (1972)
- Studied Dani tribe New Guinea
- Only two colour terms
- mili (light), mola (dark)
- Recognition memory task for colour chips - told them to remember them and tested them on it and asked if they had seen it before
- Presented colours
- Tested colours
Focal colours vs non-focal colours
images in notes
Results
* Dani better on focal colours
* Could also distinguish between colours within a colour category
* Colour perception not determined by colour terms in language
Evidence against linguistic determinism
Hard vs soft determinism
Does language influence how easy it is to distinguish colours?
Kay & Kempton (1985)
- Colour continuum green -> blue
- Three chips. Which is odd one out?
Which two are most similar? - English vs Tarahumara
Tarahumara language does not distinguish between blue and green
which is the odd one out
3 green
2 green one blue
3 blue
English chose most extreme when triplets within single colour name
But more likely to pick “different” colour when chips across colour name boundaries
=> Colour names influencing perceptual choice
Tamahurmara pick most extreme for all three types of triplet
=> Absence of green/blue categories affected perception
Kay & Kempton (1985)
* Colour names influenced colour perception (similarity judgments)
* Support for “soft” determinism
Language influences thought
Does language mean thought?
- If something could use language, would it think?
- Does language = thought?
Can computers think?
- Alan Turing (1950)
“I believe that at the end of the century…one will be able to speak of machines thinking without being contradicted”
the Turing test 1950
- In one room… a machine
answers questions posed by
the interrogator. - In another room… a human
control answers questions
posed by the interrogator.
3.In a third room… a interrogator engages in teletyped conversation with
the contestants as judges look on. If a machine can trick the judges that
it is human, then that machine has passed the Turing test
diagram in notes
Turing test
* Language is everything
* If it communicates, then it thinks
- Science fiction that a computer could pass the Turing test, right?
Wrong. See ELIZA, ChatGPT
Computers can’t think because …
- They are programmed to do things
- They are not creative
- They can’t have emotions
- They don’t have brains
- They can’t learn.
- Computers are programmed to do things
* How can they choose what to do?
Humans are programmed
* Nature has programmed us to behave in certain ways
* Evolution has determined what will we do
- Computers are not creative
* They have no originality or creative powers
* They can’t surprise us
They are creative.
Computerized Haiku (e.g., Masterman, 1971)
ALL GREEN IN THE HEAT,
I HEAR WHITE COWS IN THE SHADE.
HUSH! THE CHILD HAS BURST.
They are creative.
2. Computerized Jazz
Philip Johnson-Laird (1988)
The Jazz Generator
(music by Band-in-a-box)
- Computers can’t have emotions
* Machines can never be in emotional states
* Emotions are necessary for thought.
* Therefore, computers can’t think.
Why are emotions necessary?
* No emotions, no thought?
* Is Spock conscious?
What are emotions?
* Emotions help us satisfy goals
* Emotions are the solution to a design problem
* Having “love” means we are more likely to procreate…
- But what about the FEELING in my heart?
- Computers can’t learn?
* Computers can be programmed to learn
* Current industrial applications- speech recognition software
- games e.g., chess
- They don’t have a brain.
* Computers do not have organic, biological matter as a brain
* organic stuff generates consciousness
therefore computers cannot think
deep brain stimulation
Turing Test
- A computer can think if it can communicate
- Sufficiently to “fool” a person
Nothing else matters
- Sufficiently to “fool” a person
John Searle (1932-)
Chinese Room argument (1980)
* passing the Turing test doesn’t mean thinking
using language doesn’t mean being conscious
people outside the room ask questions in Chinese and a non-chines speaker in the room uses a Chinese langauge instruction book in English to answer the questions in Chinese. so the Chinese speaker thinks the person in the room is Chinese
To the native speaker the man in the room understands the language and the world but they don’t - passes the turing test
The Chinese Room
* understanding = consciousness
* appearing to understand = appearing to be conscious
* passing the Chinese room test = passing the Turing test
to the native Chinese speaker, the man in the box understands Chinese
- Something that passes the Turing test is like the man in the Chinese room
Just as the man can’t understand Chinese, the computer can’t think.
Searle (1980)
- … “.. could something think.. solely [by] being a computer with the right … program? … no …
- “Because the formal symbol-manipulations by themselves … are quite meaningless; … they don’t symbolize anything. Any [meaning] is solely in the minds of those who program them.”
Note that we are equating understanding with speaking chinese
Replies
1. The systems reply
2 - The brain simulator reply.
- Searle - man memorizes all the rules! Rule book, everything.
- Searle (1) if you need a robot with perceptual apparatus, then thinking isn’t all about manipulating symbols
(2) Put man inside the robot, takes symbols in through TV cameras - Construct program that simulates a brain.Searle - Brains don’t matter. Simulating only structure of brain, not the relation between the brain and the outside world. Symbols need to be about something in the real world.
summary
- Turing (1950) – if something appears to think, it thinks.
- Searle (1980) disagrees.
- appearance isn’t everything (Chinese Room)
- Language and thought interact
- Anything that has language must think
Next lecture: Candice Morey and Memory
