Intelligence: A Very Short Introduction – Ian J. Dury

Summary: Ian Dury outlines the history of the study of intelligence, as well as the current scientific understanding of it. While intelligence is an abstract concept, its utility has been well established: it can be used to predict many things, such as career success and longevity, and can give us insights into how the brain works. Intelligence is heritable and fairly stable over the course of one's life.

Thoughts: I found this book to be quite interesting. Dury supports each of his main points by looking at several scientific studies, and I appreciated that all of these were either metastudies or robust studies with very large sample sizes. With fewer than 200 pages, it wasn't a huge investment of time. This was my first Very Short Introduction, and I expect it won't be my last!

(I took more notes than usual on this book, so it should come fairly close to a full summary.)

Dury, Ian J. 2020. Intelligence: A Very Short Introduction (2nd Edition). Oxford UP.

Chapter 1 - Is there one intelligence or many?

The Wechsler Adult Intelligence Scale IV

• Among many intelligence tests/batteries, the WAIS-IV is among the most widely used
  • The WAIS consists of 15 sub-tests
  • when the results of pairs of sub-tests are compared, every correlation is positive, and every pair is moderately- to strongly.
    • This was a test of 1800 American adults, English-first-language
    • (These are averages: individuals may still score highly on some traits and low on others)
  • Certain subtests with high correlations cluster together:
    • Verbal Comprehension
      • Similarities, Vocabulary, Information, Comprehension
    • Perceptual Reasoning
      • Block Design, Matrix Reasoning, Visual Puzzles, Figure Weights, Picture Completion
    • Working Memory
      • Digit Span, Arithmetic, Letter-Number Sequencing
    • Processing Speed
      • Symbol Search, Coding, Cancellation
• since all correlations between sub-tests are positive, g, or general intelligence, can be identified as a factor in predicting how well people will do on sub-tests.
  • g can predict about 40% of the variation among people taking a specific subtest
• similar results are obtained using other tests
• Charles Spearman was the first to observe g.
• summary: there exists a general intelligence factor, and there also exist more specific cognitive capabilities.

Carroll’s ‘Human Cognitive Abilities’ Survey

• John Carroll’s 1993 metastudy of factor analyses, Human Cognitive Abilities: A Survey of Factor Analytic Studies
  • compared 400 high-quality studies
  • found positive correlations among tests, clusters of skills that were more positively correlated
  • basically, confirmed the results of the WAIS, with the existence of specific mental capabilities
  • "When people obtain a good score on a mental test, it is for at least four possible and non-exclusive reasons:
    • they are good at mental tests overall (they have high general intelligence);
    • they are good at that sort of test (they have a high ability for that cognitive domain);
    • they are good at that specific test (they have high ability for that specific cognitive task);
    • and they had the wind behind them on the day (chance favoured them on the testing occasion)."

‘Intelligence’ beyond intelligence tests

• The findings above stand somewhat in opposition to Howard Gardner’s theory of multiple intelligences.
  • exploring this a little further, Wikipedia suggests that the theory of multiple intelligences is backed by little (if any) empirical evidence
• it’s important to recognize what scores on intelligence tests don’t recognize (some of which relate to Gardner’s theory) - differences in personality, creativity, wisdom, etc.

Chapter 2 - What happens to intelligence as we grow older?

Salthouse’s Virginia Studies

• Different cognitive domains tend to change differently as people age:
  • Processing Speed holds steady until ~30, then declines sharply
  • Reasoning, Memory and Spatial Visualization decline gradually from young adulthood to ~60, then decline sharply
  • Vocabulary Knowledge increases until 60, then begins to decline
• This suggests that intelligence can be usefully divided into two categories:
  • Fluid Intelligence
    • assessed using unfamiliar materials
  • Crystallized Intelligence
    • assessed using knowledge built up over the course of one’s life
• Fluid Intelligence and Crystallized Intelligence are positively correlated

Does intelligence ‘all go when it goes’?

• Elliot Tucker-Drob’s studies suggest that ~60% of individual differences in ageing-related cognitive changes are shared across domains
• This doesn’t mean, however, that cognitive decline is not caused by a single root cause

The Scottish Mental Surveys of 1932 and 1947

• In 1932, all Scottish students aged 11 +/-0.5 took a took the Mental Survey Test
• In 1947, a similar test took place.
• In 1998, Ian Dury and team ran the same, 1932 test for people who originally took the test
  • People tended to do better at age 77 than age 11
  • People who scored well at age 11 tended to score better at age 77
  • similar results obtained by comparing 11-year-olds with 90-year-olds
  • about 50% of a person’s performance on the later test could be predicted by looking at the test score during their youth; about 50% must be accounted for elsewhere

Preventing some age-related cognitive decline

• The evidence is not conclusive at this point, but there are hints that:
  • genetics can predict the steepness of cognitive decline (in specific, a genetic predisposition to dementia)
  • smoking may lead to steeper cognitive decline
  • exercise and cognitive engagement may prevent cognitive decline
  • eating a mediterranean diet or lots of fruits and vegetables, taking part in leisure activities, and being in a more professional occupation (J: what does this mean?) may prevent cognitive decline

Chapter 3 - Are there sex differences in intelligence?

• due to confounds, it’s difficult to study this unless the entire population is surveyed

The Scottish Mental Survey 1932

• Results of the Scottish Mental Survey indicate that IQ is the same among boys and girls
• The standard deviation among boys, however, is wider than that of girls
  • there will be more extremely sharp boys than girls, and more extremely dull boys than girls
  • there are more “average” girls than average boys

The National Longitudinal Survey of Youth 1979

• looked at boy-girl sibling pairs to control for confounds - 14-22 years old
• a series of 10 tests
  • boys scored higher on some tests and lower on others - less than 1 point difference in IQ (negligible)
  • same results as Scottish Mental Survey - boys had a higher standard deviation than girls

The Cognitive Abilities Test 3 sample

• CAT3 tests three cognitive domains
  • verbal reasoning
  • quantitative reasoning
  • non-verbal reasoning
  • each of these has three sub-tests
• girls tend to score higher in verbal reasoning (~2 IQ points) and g (~0.75 IQ points)
• boys’ standard deviations were higher in all three cognitive domains (though in verbal reasoning, there are more above-average girls than above-average boys)

Chapter 4 - What are the contributions of environments and genes to intelligence differences?

Twins

• Dury explains the statistics behind studies of mono- and dizygotic twins

10,000+ twins, from three continents

• Drawing on the research of Claire Haworth
  • By young adulthood, about 2/3 of people’s differences in intelligence can be explained by genetics, while about 1/3 can be explained by environmental influences
  • environmental factors express themselves more in the intelligence of children than in adults - that is, as a child ages, genetics play more and more of a role in explaining intelligence

DNA

• describes structure of DNA, how mutations work

DNA from 300,000 people in fifty-seven studies

• Results of a study by Gail Davies et al. of 300 000+ people of European ancestry:
  • Intelligence is a polygenetic trait
    • SNPs related to intelligence were found within multiple genes, and also in non-coding portions of the genome
  • Genetic traits related to intelligence are often also related to health
  • so far, DNA studies have found lower correlations between genetic factors and intelligence than twin studies; the gap will probably narrow in the future
  • predictions of intelligence based on DNA are better than chance, but currently, less than 10% of the variation in intelligence between people can be predicted using DNA
  • among people of lower SES, environmental effects play more of a role than in people of high SES

Chapter 5 - Are smarter people faster?

• does intelligence depend on an underlying factor, for example processing speed?

The West of Scotland Twenty-07 Study

• 1987-2007 - longitudinal study of three cohorts of people, aged 15, 35 and 55 to begin.
• Compared reaction times to scores on the Alice Heim 4 test:
  • faster reaction times in an identification test (i.e. press the matching button) are correlated with higher scores on the AH4 test
    • the correlation is medium
    • it holds across different age groups
  • faster times in a simple reaction test are correlated with higher scores on the AH4 test
    • but among people with lower AH4 scores, there is more variability in reaction times

The Lothian Birth Cohort 1936

• involves a test where a simple image is flashed on a screen for a short amount of time, and then the respondent can respond at their leisure
• Results: compared to results from a subset of the WAIS, people who did better on the image processing test scored better on the IQ test

Speed and other cognitive processes

• these tests present results that are hard to explain based on book learning or social class

Chapter 6 - What do more intelligent brains look like?

The Lothian Birth Cohort 1936

• Greater brain volume is correlated with increased intelligence
• Thicker grey matter is correlated with increased intelligence
• Better white matter integrity is correlated with increased intelligence
• Fewer white matter hyperintensities is correlated with increased intelligence
• None of these correlations are particularly strong (+/- ~0.2 to ~0.3)

Brain volume and intelligence

• Jakob Pietschnig conducted a meta-study of studies of brain volume and intelligence - the results were all fairly similar, with correlations of (0.25 to 0.3).
• No one knows why/how intelligence should be correlated with brain size

Intelligence beyond brain structure

• less robust findings here
• if interested, look up Richard Haier’s book The Neuroscience of Intelligence

Chapter 7 - Does intelligence matter in the school and the workplace?

• Historically, measures of general intelligence have been overzealously applied at times

English National GCSE examination results

• tens of thousands of 11-year olds’ IQs were measured using the Cognitive Abilities Test (the CAT)
  • results between subtests were strongly correlated - g could account for about 70% of the variation in test scores
  • girls tended to score better (by about 4 IQ points) on the verbal reasoning section
• these results were then linked with their performance on the General Certificate of Secondary Education (GCSE) test at age ~16
  • results: performance on CAT was strongly correlated (between 0.7 and 0.8) with performance on the graduation test
  • across all subjects, girls tended to do better on the GCSE, and the results can’t be explained based just on girls’ better verbal reasoning skills
    • i.e. if a boy and a girl scored the same on the CAT, the girl was likely to score higher on the GCSE
• suggests that better standardized testing results could be obtained by working to improve people’s IQs earlier in life
• suggests that while there are other factors affecting how well people do on standardized testing, intelligence has a lot to do with it.

Job selection and job performance

• Looks at the work of John Hunter, Ronda Hunter and Frank Schmidt
• Ian Dury asks, “among other things, would it be useful for a hiring committee to know applicants’ IQs in knowing the best/most productive person to hire?”
• Hunter, Hunter and Schmidt looked at thousands of studies of hiring practices to see which were the best predictors of later job performance. The most informative practices:
  • Work Sample test
  • Psychometric Intelligence Test (i.e. IQ)
  • Employment interview (structured)
  • Integrity Test
  • (of these, the most robust is the IQ test - has been looked at in many more studies than any of the others.)
• IQ tests are easy to apply in any situation (compared to, say, a work sample test)
• IQ predicts the effectiveness of training, and the ability of the employee to learn on the job. The more complex the job, the better IQ is at predicting job performance
• if IQ is already a factor in the employee evaluation, an integrity test adds the most predictive power, followed by work sample and structured interview.
  • the more approaches are combined, the better the predictive power
• Schmidt and John Hunter, in a different metastudy, concludes that:
  • general intelligence is a better predictor of performance than specific aptitudes
  • general intelligence gets slightly better at predicting performance as job experience increases
  • the mechanism of IQ’s predictive power is probably that intelligent people acquire information more quickly

Success in life beyond intelligence

• Dury underlines that IQ is not the only thing that matters in schooling and employment, but that it can play a very large role.

Chapter 8 - Does intelligence matter for health and longer life?

• a look at the new field of cognitive epidemiology

The Scottish Mental Survey of 1947

• Research by Catherine Calvin, looking at IQ at age 11 and seeing who had survived to age 79 - sample of ~70,000 people
  • if people hadn’t survived, also looked at what they had died of
• Results:
  • High IQ at age 12 led to increased likelihood (by 20%, for a difference of 15 IQ points) of surviving to age 79.
  • Similar results were obtained for looking at many more specific causes of death. Strongest association were with respiratory disease
  • Adjusting for SES made little difference in the results
  • Adjusting for smoking shrinks the effect sizes, but cannot fully explain the association

Intelligence in youth and all-cause mortality

• Metastudy by Catherine Calvin
• Main result: an increase of 15 IQ points was associated with a decreased likelihood of dying at any timepoint by 24%

The National Longitudinal Survey of Youth 1979

• Analyses by Christina Wraw
  • higher intelligence correlated with decreased chance of being diagnosed with a range of diseases at age 50
  • possible explanations: increased income, increased education, increased occupational safety. They account for some of the differences, but none of them individually, nor all of them together, could fully explain the results
  • intelligent people are more likely to adopt healthy habits

Chapter 9 - Is intelligence increasing generation after generation?

The (James) Flynn Effect of rising IQ

• IQ tests need to be re-normed every few years, as the average IQ, which should always be 100, rises over time
• Flynn offered 3 possible causes for his observations:
  • a statistical artifact: the sample used to norm the tests is skewed
  • people are more used to answering the types of questions asked on IQ tests
  • people’s intelligence is rising

Flynn’s ‘Massive IQ gains in 14 nations’

• IQ was found to rise more quickly when the tests were “culturally reduced” - where [[Fluid Intelligence]] was tested more than [[Crystallized Intelligence]]
  • so, it reflects an increase in problem-solving ability, not in learned content
• within individual generations, the [[Flynn Effect]] doesn’t compromise the validity of the tests’ results
• also doesn’t compromise the results of genetic effects

‘One century of global IQ gains’

• Metastudy of the Flynn Effect by Pietschnig and Voracek
  • globally, IQ rises by 2.8 points per decade
  • Fluid Intelligence rose faster (4.1 per decade) than Crystallized Intelligence (2.1 per decade)
  • likely not explained by computers - not widely available during the period examined
  • possibly due to people being more comfortable guessing on IQ tests?
  • possibly due to a “multiplication” process, whereby a slightly better environment leads to increased intelligence, which allows the environment to be further improved, etc.
  • among all the possible environmental causes, none came close to fully explaining the effect, none had direct evidence of causation, and they were hard to prove/test - the Flynn Effect is still a mystery!
• Dury concludes that the Flynn Effect is definitely real, but probably not as large as the numbers suggest it is (and certainly it should not suggest that the average person in 1900 was learning disabled); its causes are still one of the big questions in psychology

Chapter 10 - Do psychologists agree about intelligence differences?

The Bell Curve

• 1994 book The Bell Curve by Richard Herrnstein and Charles Murray created a furor and caused a response from the APA to point out what most psychologists agree on regarding intelligence
  • Section II contains some really insightful work

A working party on intelligence’s knowns and unknowns

• initial response to the book was heated, and it was often difficult to pick apart what was scientifically/empirically motivated and what was politically motivated
• The Board of Scientific Affairs appointed a Task Force to state the current state of understanding within the field
  • made up of a wide variety of people studying intelligence in disparate ways - conflicts resolved by discussion, final report had the unanimous support of all on the task force

The APA Task Force on conceptions of intelligence

• people generally agreed that Intelligence/IQ has to do with the psychometric approach, or the measurement/testing of mental processes.
  • This doesn’t mean that all aspects of intelligence can be measured, i.e. the brain is capable of things not evaluated on IQ tests

The APA Task Force on intelligence tests and their correlates

• intelligence is an abstract concept - that is, it can’t be measured from 1 … x like height can
• why, then should people care about intelligence?
  • intelligence is fairly stable throughout life
  • test scores can predict aspects of people’s lives independent of the tests themselves - academic success, employment success, morbidity…
  • IQ scores can lead to a better understanding of what’s going on in the brain

The APA Task Force on genetic and environmental contributions to intelligence

• summarized the results of genetic studies (DNA studies weren’t yet possible); noted the Flynn Effect

The APA Task Force on group differences in intelligence

• different groups, i.e. different sexes and different ethnicities
• Dury doesn’t summarize, but notes this section is worth reading

The APA Task Force’s conclusions

• Dury lists some unknowns that were identified by the task force, e.g.:
  • why is the flynn effect happening?
  • what environmental factors influence intelligence? how does nutrition influence intelligence?

Summaries of intelligence research after the APA Task Force

• Points towards summary by Richard Nisbett et al.
  • state that measures of intelligence are imperfect, but still useful for predicting certain life outcomes

Signing off, and encouragement to read more …

• more resources found in the References and Further Reading

References and Further Reading

Defining intelligence, at last

“The American Psychological Association’s Task Force (Chapter 10) wrote that definitions come at the end of research rather than the start. I agree, and therefore I waited until here before offering one. Many people repeat Linda Gottfredson’s definition: ‘Intelligence is a very general mental capability that, among other things, involves the ability to reason, plan, solve problems, think abstractly, comprehend complex ideas, learn quickly and learn from experience. It is not merely book learning, a narrow academic skill, or test-taking smarts. Rather, it reflects a broader and deeper capability for comprehending our surroundings—“catching on”, “making sense” of things, or “figuring out” what to do.’”