Intelligence
The ability to generalize (Memory); flexibly use knowledge (Concepts) to solve new problems (Problem solving)
* Learning from experience and adapt that learning to your environment when making decisions
* Acting purposefully
ChaptGPT solves ______ decision-making and reasoning tasks similarly to humans
Ambiguous
- Maybe chatgpt can reason beyond rules ? but counter to that : we feed Chatgpt all our data, and problems like the Linda one are very famous : it could be pulling from our responses
ChaptGPT can do these tasks
ChatGPT cannot do these tasks
IQ tests have been designed to measure general intelligence differences but …
Other factors underlie differences on these tests aside from “intelligence”
Psychometrics
Standardization
IQ tests scores
Highest recorded IQ
Reliability
Consistency across instances of testing
* IQ scores have high test-retest reliability
* Score at age 6 correlates with scores at age 18
* But can change with environment and education, which can dampen down the reliability
Validity
The test is measuring what it is intended to measure
IQ scores should have predictive validity :
* Should predict performance on something requiring intelligence
* Yet, low correlations ( r = .5, low to mid) with job performance
* Intelligence will vary across context and culture
Francis Galton (1822 to 1911)
one of first thinkers to suggest intelligence could be measured
* Founded the eugenics movement
* Racially-motivated view of how to “improve” society
* A dark start to intelligence testing
Alfred Binet
Developed a test in response to a request from the French government
* Identify children that needed special education in school
Binet viewed intelligence as as important for:
* Practical life, adapting to circumstances judging and reasoning well
Binet thought his test measured ______, not intelligence
Academic output
Simon-Binet Test
30 questions of increasing difficulty (easy → hard)
* Easy items: Follow a light beam, answered easily by young children
* Difficult items: Describe abstract words, for older more educated children (e.g. state difference between abstract words)
* Some questionable items
Standardization : A child’s mental age was calculated by comparing the score /30 to the score of a group of children the same chronological age
- Platform for the current IQ test
The Stanford-Binet Test
Based on the Simon-Binet test
Item for a 4 years old
“Repeat the following numbers: 3 6 7”
Item for an Adult
“Describe the difference between misery and poverty”
IQ (intelligence quotient) as we know it :
* IQ RATIO scores : (Mental Age (MA) / Chronological Age (CA)) * 100
* If MA > CA, ability is above average of peers (gifted)
* If MA < CA, ability is below average of peers (delayed)
IF MA = CA : IQ = 100 (average)
Wechsler Tests
Components of full scale IQ (FSIQ) in Wechsler tests
Pros and cons of Weschler tests
Pro :
- Total of 14 different test : good tests because they acknowledge different types of abilities
Cons :
- culturally biased and they’re affected by our language function (e.g. not all cultures play cards, use the same words)
Raven’s progressive matrices
Shown patterns with a missing section
* Asked to determine the missing piece from a set of options
Non-verbal assessment :
* Relatively free from linguistic influences and thus free from cultural biases
Factors affecting IQ performance
Genetics and IQ
The Flynn effect
In 1980s, for all the tests, Americans’ IQ scores increased 3 points per decade over 100 years
Flynn effect: Education