1. Flynn effect
1.1. average IQ scores within the US have increased about 10 points over 30 years (similar in other countries)
1.2. increases primarily in low IQ groups; high IQ groups have remained stable or even are slightly decreasing
1.3. score increase may be in areas other than those explained by the "g factor"
1.4. tests are recalibrated so that the average IQ remains 100
1.5. also forces reinterpretation of supposed decline of intelligence with aging
1.6. reasons are unclear
1.7. it may be ending: in Europe, IQ scores stopped going up in the 1990's and have been declining in mathematical ability
2. multiple intelligences
2.1. Howard Gardner (1983)
2.2. idea
2.2.1. there isn't one intelligence
2.2.2. there are multiple ones
2.2.3. learning and teaching should be adapted to the strengths of each child
2.3. intelligences
2.3.1. logical-mathematical
2.3.2. spatial
2.3.3. linguistic
2.3.4. bodily-kinesthetic
2.3.5. musical
2.3.6. interpersonal
2.3.7. intrapersonal
2.3.8. naturalistic
2.3.9. existential
2.4. objections
2.4.1. empirically, there is a high correlation between these "different" intelligences
2.4.2. validation of educational approaches based on the theory has been limited
2.5. alternative
2.5.1. skills
2.5.2. differences in preferences and skills
2.5.3. learned differences (rather than innate factors)
2.5.4. physiological differences other than intelligence
2.6. NB: There are pretty clearly differences in how humans learn, what they are good at and what they are bad at, etc. The question is whether those differences fall under the definition of what we might call "intelligence".
2.7. fundamental challenge
2.7.1. Does this even meet our definition of intelligence?
2.7.2. Psychometrically, we define intelligence as the "g factor", so this is everything but the "g factor".
2.7.3. These measures are still very useful, but theyprobably shouldn't be called "intelligence"
3. psychometrics
3.1. psychometric approach
3.1.1. test performance on a variety of mental tasks
3.1.2. see whether there is a common factor that can account for correlations among performance for all of them
3.2. IQ
3.2.1. intelligence is measured as an "IQ"
3.2.2. intelligence quotient
3.2.3. quotient of what?
3.2.4. original definition
3.2.4.1. give tests to a variety of individual at different ages
3.2.4.2. performance increases with age
3.2.4.3. for a given performance, find the average age at which that performance is achieved
3.2.4.4. this yields the "mental age"
3.2.4.5. divide the mental age by the physical age
3.2.4.6. hence the "intelligence quotient"
3.2.5. modern definition
3.2.5.1. for a given population and age group, the mean performance on a test is scored at 100
3.2.5.2. one standard deviation within the population is a change in 15 IQ points
3.2.5.3. intelligence scores are distributed normally, so this lets you easily translate the scores into population frequencies
3.2.5.4. intelligence scores are distributed normally because they are the aggregate of many individually random decisions (central limit theorem), both at the question level and possibly at the level of many different individual skills
3.3. mental capabilities being measured
3.3.1. Cattell-Horn-Carroll theory
3.3.2. fluid intelligence
3.3.3. crystallized intelligence
3.3.4. quantitative reasoning
3.3.5. reading/writing ability
3.3.6. short term memory
3.3.7. long term memory
3.3.8. visual processing
3.3.9. auditory processing
3.3.10. processing speed
3.3.11. reaction times
3.4. what correlations do exist?
3.4.1. correlations among school performance
3.4.2. correlations among intelligence test subtests
3.5. conclusions so far...
3.5.1. There does seem to be something like a "g factor" (general intelligence).
3.5.2. What does it mean?
3.6. explanations
3.6.1. causal models
3.6.1.1. how does one variable cause another one?
3.6.1.2. i.e., if we intervene and change one variable, how does the other variable behave?
3.6.2. factor models
3.6.2.1. how can variation of a quantity be explained as the variation of component quantities?
3.6.3. mathematically formalizable
3.6.4. Spearman's Two Factor Model
3.6.5. Carroll's Three Stratum Theory
3.6.6. note: correlation is highest among people with low intelligence test scores (what does this mean?)
3.7. predictions
3.7.1. intelligence measures are no good if all they do is predict performance on intelligence tests
3.7.2. how do they relate to real-world performance?
3.7.3. academic performance
3.7.3.1. correlation is high in elementary school, about 0.6-0.7
3.7.3.2. correlations drop to about .5 in college, and .4 in graduate school
3.7.4. job prestige
3.7.4.1. very high correlation (.9) between prestige of an occupation and the average IQ of people who take it
3.7.4.1.1. what does this mean?
3.7.4.1.2. correlation with average IQ
3.7.4.1.3. prestige
3.7.4.2. prestige
3.7.4.2.1. occupational prestige (USA)
3.7.4.2.2. occupational prestige (Germany)
3.7.4.3. high correlation (.7) with individual IQs
3.7.4.4. low dispersion of scores in highly prestigious occupation suggests minimum requirements
3.7.5. job performance
3.7.5.1. how does IQ correlate with job performance within each job?
3.7.5.2. generally, good correlation (.55)
3.7.5.3. highest for jobs with high complexity
3.7.5.4. job specific tests are no better at predicting performance than general IQ tests
3.7.6. income
3.7.6.1. correlation between IQ and income is about .4
3.7.6.2. somewhat higher if you look at peak income
3.7.7. other correlates
3.7.7.1. morbidity and mortality
3.7.7.2. crime
3.8. neurological basis
3.8.1. moderate correlation with brain volume (.3-.4)
3.8.2. reasons are unknown
3.8.2.1. maybe "more hardware" makes you smarter
3.8.2.2. bigger brains correlate with various other factors, such as childhood nutrition and body size
3.8.2.3. these other factors may be determining
3.8.3. small correlation with height (.2)
3.8.3.1. may be due to mating preferences rather than causal factors
3.8.4. "g factors" can also be identified in animals (mice, rats, monkeys), correlating problem solving and learning on a variety of tasks
3.8.4.1. what kinds of problems do these animals solve?
3.8.4.2. we'll talk about animal intelligence later
4. capabilities of intelligent systems
4.1. logic
4.1.1. classical view of intelligence
4.1.2. predicate logic
4.1.3. deduction and induction
4.1.4. trying to explain all of human intelligence through logic falls flat
4.1.5. common trope in SciFi ("Mr. Spock")
4.1.6. limitations of logic
4.1.6.1. fails to take into account randomness (but logic is a special case of Bayesian probabilistic reasoning)
4.1.6.2. computations can be very expensive, so the brain may not be able to carry them out
4.1.6.3. traditional logic doesn't deal well with time, incorrect information, belief, etc.; AI has developed more complex "logics"
4.1.6.4. where does it come from? Is it hard-wired? How?
4.2. problem solving
4.2.1. Generically: given some start state and some constraints, reach some goal state.
4.2.2. When it is a mental process, it is often the same as "planning".
4.3. planning
4.3.1. Solve the problem mentally before solving it physically, since physical exploration is more costly than mental exploration.
4.3.2. Long term planning involves decision making about probabilities, expectations, and risks, as well as preferences.
4.4. learning
4.4.1. Retain past experiences and apply them to new situations.
4.4.2. There are two fundamentally different kinds of learning
4.4.2.1. episodic memory
4.4.2.2. skill learning
4.4.3. You can lose episodic memory while still retaiing skill learning.
4.5. symbolic communication
4.5.1. Transform complex physical realities into a stream of symbols, and back.
4.6. perceiving
4.6.1. recognize objects in 3D scenes
4.6.2. recognize objects from the sounds they make
4.6.3. identify objects by touch or smell
4.6.4. understand spoken language
4.7. acting
4.7.1. produce spoken language (far more complex than playing the piano!)
4.7.2. walk upright (small errors mean death)
4.7.3. handle and manipulate tools