Thursday, November 03, 2005

A Cornucopia of New Intelligence Research   posted by Jason Malloy @ 11/03/2005 07:14:00 AM

Via my co-blogger and pro psychologist, Alex, I learn that the International Society for Intelligence Research is holding its 6th annual conference this December in Albuquerque, New Mexico, and by the looks of it, some very interesting research will be presented. Scientists exploring this important, difficult, controversial concept known as intelligence from many different disciplines and research paradigms - differential psychology, economics, sociology, neuroscience, evolutionary biology, paleoneurology, cognitive psychology, evolutionary psychology, artificial intelligence, genetics, animal cognition - are increasingly working together and combining their discoveries to powerful theoretical and empirical effect. For those interested, a PDF is available here with about 50 new abstracts for almost 60 presentations. Below the fold is a summary of the ones that I find the most interesting sorted by subject.

The Flynn Effect

Probably the most puzzling issue in psychometrics today is the Flynn Effect - the phenomenon of obsolete test norms as IQ scores continue to rise. The effect has caused problems or has had uncertain implications for any number of psychometric issues, including the authenticity of the rise itself, the black-white test score gap, the stability of intelligence with age, the effect of adoption studies, the actual IQ of the earlier generations of Asian Americans, and the existence of dysgenic trends.

Alexander Beaujean and Steven J. Osterlind follow recent reports of the Flynn Effect finally stopping, and even reversing, in Europe (see Alex's post here) in a presentation called Assessing the Lynn-Flynn Effect in the College Basic Academic Subjects Examination (Alex and David Burbridge have previously debated the nomenclature of the rise on GNXP here). Alex finds evidence, using special methods of test analysis based on Item response theory, that the Flynn Effect has started to reverse course in America as well.

And Jelte Wicherts, in Flynn Effect in the Woodcock-Johnson Cognitive Ability and Achievement Tests 1976-1999, uses Structural equation modeling techniques to explore whether Flynn gains have been an artificial score inflation (as Alex and I believe) or genuine gains in ability. Equally important, Wicherts attempts to explore whether the FE occurs between cohorts or to everybody at time of measurement (i.e. if you took an IQ test when you were 30, 40, and 50 would it be higher at each successive age (time of measurement), or would it remain stable for individuals across their lifespan, but three brothers age 6, 10, and 15 would be progressively stupider (cohort). Which one it is has meaningful implications for what's causing the rise (e.g. heterosis - a genetic cause - is only consistent with cohort). Previous evidence, such as the stability of IQ across 50+ year longitudinal studies, seem to falsify the TOM model.

On a related issue, Jan te Nijenhuis, et al. look into another area that might be plagued by "hollow" or fake IQ gains. IQ scores go up after retesting and through training programs. In Score Gains on g-Loaded Tests: No g the team demonstrate that these kinds of IQ boosts are not on the g factor (the business end of IQ), and suggest their results have implications for experiments that show schooling increases IQ. These increases may well be hollow.

Measurement and Mental Chronometry

One way to maybe better overcome this problem of "fake IQ gains" is getting at the underlying physical reality of intelligence. For instance, to determine if intelligence gains were "real" or not, we could measure the areas, tissues or structures of the brain that indicate intelligence or monitor more primal functions, such as things like how quickly or efficiently the brain registers simple stimuli through brain waves or glucose metabolism during problem solving. Issues of test bias and the potential of artifactual test gain would no longer be a problem because we could just clock the brain itself. Simple, more accurate, more efficient intelligence measurement is desirable and was Francis Galton's original vision for intelligence testing that ultimately lost out to Binet's measurement method.

One researcher who has helped revive Galton's model is Joseph Fagan, through his work with infants and young children. Habituation is a method for determining what’s going on in the heads' of infants less than one year old. By monitoring a baby's time spent discriminating stimuli, cognitive psychologists have determined innate or early notions of number and causality, evolutionary psychologists have determined culturally neutral standards of beauty and sex differences in social and object interests, and differential psychologists have tracked intelligence differences to the earliest months of life. In The Prediction, from Infancy, of Adult IQ and Achievement, Fagan et al. find that the correlation between the IQs of a sample of 20 year olds and their intelligence measured before the age of 1 is about .60. This is similar to earlier studies, which found the correlation between age 1 and age 11 was about .50.

Moving to the brain itself, a number of presentations come from Richard Haier and Rex Jung. Along with lead author Richard Colom in Correlated Vectors, g, and Gray Matter: A Frontal-Parietal Network and the Einstein Hypothesis they test the theory (named for the enhancement of this particular region in its namesake), through a number of lines of evidence, that the frontal-parietal network is key to individual differences in intelligence. They also review the current regions and structures of the brain known to be associated with intelligence.

Jung et al. also turn to the neurochemistry of intelligence in Biochemical Markers of Individual Differences in Cognitive Functioning, and "highlight the importance of white matter structural and chemical integrity to intellectual performance” which supports "the "neural efficiency" hypothesis that suggests optimal brain organization underlying individual differences in cognitive processes".

Finally, in Investigating the Cortical Temporal Dynamics of the Speed-Intelligence Relationship Using Magnetoencephalography (MEG), Robert Thoma uses MEGs to monitor the activity of the brain, to show that the regions involved in reaction time (RT) experiments (how quickly you lift your finger off a panel to turn off a light) are the same areas that show activity during complex intellectual tasks, suggesting RT is a simple and accurate measure of intelligence (something disputed by NJ Mackintosh in IQ and Human Intelligence, where it is argued that RT is a very nongeneralizable mental ability, not the same as g).

Group Differences

Those interested in HBD will not be disappointed, there are a number of presentations on race and sex differences, including a biggie from Greg Cochran and Henry Harpending titled The Evolutionary Biology of Human IQ Diversity: Some Current Directions and Hints. Fresh from their Ashkenazi notoriety, they discuss the far more contentious issue of Eurasian intelligence, with a new theory that fingers the Neanderthals!:
Modern humans apparently left Africa ca. 40,000 years ago and appeared soon afterwards in western Eurasia and in Australia. The southern arm peoples arrived with middle Paleolithic technology that persisted unchanged for tens of millenia while the northern arm peoples were host to the famous "creative explosion" of the upper Paleolithic with elaborate tools, worked bone, beadword and other adornment, sculpture, and painting. We discuss the hypothesis that incorporation of Neanderthal genes led to elevated intelligence (or something closely related) in the northern arm. We will mention some likely examples of such assimilated genes . . . We discuss the appearance and spread of an ASPM variant, one of the microcephalin complex genes, as an example. A puzzling pattern among candidate genes for elevating IQ is that they seem not to have spread in Africa

Additionally, two more talks add to and work off of Lynn and Vanhanen's Important book. In IQ & Wealth of Nations: Prediction of National Wealth, Deborah Whetzel and Michael McDaniel replicate Lynn and Vanhanen's findings and also find that education spending per student provides no incremental prediction of GDP beyond IQ. They examine a set of the highest predictors of GDP and find that economic freedom, health spending per capita and IQ explain 90% of the variance in national wealth (See also the Jones & Schneider paper and Garett Jones' newer paper (PDF)). Earl Hunt and Werner Wittmann also replicate Lynn and Vanhanen's finding in Relations Between National Intelligence and Indicators of National Prosperity, and add to it by examining cross-national student scholastic achievement from international datasets like PISA and TIMSS. They find a strong relationship between intellectual competence and economic indicators. In Criteria For Studies of Race and Intelligence, Earl Hunt also makes the case for agnosticism about genetics being the source of these (real and important) differences, with discussion of research ethics and study design for racial behavior genetics.

In Race Difference in General Intelligence g in Relation to Blood Pressure, Body Proportions, Hormones, and Personality, Helmuth Nyborg et al. test and reject another theory for the black-white IQ gap - that greater black hypertension plays a role in the difference.

Of course these kind of theories are DOA. The real problem is how to test the factor X theory, that the black-white IQ gap is due to something unique to the black environment that affects all blacks equally but is completely absent from the white environment in a way that could evade all detection thus far. A team of researchers published a paper (PDF) in 2003 responding to the well known argument that high within-group white and black heritability has no implications for their between-group IQ difference, showing that a common factor model "approach clarifies that absence of measurement bias implies common sources of within- and between-group variation" (earlier, the late David Rowe similarly showed, through an ingenious method of structural equation models using blacks and whites and their full and half siblings, that the source of the within group differences was also causing the between group difference). In Factorial Invariance and the Representation of Within-Groups and Between-Groups Differences: A Reconsideration Keith Widaman disputes their argument, but offers his own methods "for study design that will enable a test of the hypothesis that sources of within-group differences are also responsible for between-group differences".

On the sex difference front, Paul Irwing presents Sex Differences in General Cognitive Ability: A Reexamination of the Evidence, which seeks to question whether the 100 year old position in psychometrics - that there is no mean difference between men and women on IQ - is correct (as Jensen also concluded in 1998's The g Factor), or whether Richard Lynn's newer (1990>) position - that men have somewhat higher average IQs - is correct. Irwing concludes from large SPM samples that Lynn is correct (as was reported in the media a few months ago). Furthermore, he finds no evidence that this is due primarily to the male advantage in spatial visualization. Also, he finds that some research previously presented to show that there are no sex differences shows exactly the opposite.

David Puts et al. also present a meta-analysis titled Possible Organizational Effects of Early Androgens on Human Spatial Ability: Meta-Analyses of CAH and Digit Ratio Studies Showing that females that get a heavy dose of male hormones in utero also perform higher on spatial tasks like men do.


David Puts' finding has implications for the ridiculous Larry Summers' witch-hunt earlier this year as well. Rose Mary Webb et al. study the geniuses among us in Spatial Ability: A Neglected Dimension in Talent Searches for Intellectually Precocious Youth, and find using longitudinal data that children with high spatial abilities (who - a la Puts - are going to be disproportionately male) have higher levels of interest in math and science ("theoretical" endeavors) and that the ability can be used to predict which gifted students follow scientific or humanities pathways. Similarly, Summers' critics have tried to rebuff the annoying fact that many more men score at the top of the ability distribution by asserting that IQ/test score ability stops being meaningful at the higher levels of ability anyway (that is, conveniently, at the precise point where the male/female ratio starts to get ridiculously incongruous. e.g. 7 to 1 in the top 1%). In Creative Accomplishments Covary With Ability Even Among the Top 1%, Jonathan Wai et al. show this is nonsense; IQ keeps discriminating between levels of creative and career achievement (such as earning a math-science PhD, securing a patent, and achieving tenure at a top 50 U.S. university), even at the very skinny right tail of the ability distribution. (and it keeps on going: IQ even distinguishes those in the top .0001%!)

Two additional presentations on those at the right tail, An Examination of Spearman's Law of Diminishing Returns by Christopher Condon and David Schroeder and A Test of Spearman's Law of Diminishing Returns in the Kaufman Assessment Battery for Children, Second Edition by Matthew Reynolds and Timothy Keith, test Spearmen's idea that the g factor is more important at low IQ levels, and that lower order factors are more independent and important at progressively higher ability levels. Using different data sets both studies find support for this. "Multiple intelligences" (and not the fictional Gardner ones) are only something for the very smart, for people at the left tail it's all about g.

The Evolution of Human Intelligence

A number of presentations attempt to tie general intelligence into the framework of human evolution. James Lee's The Evolution of General Intelligence in the Primate Clade, for instance, talks about research showing that mental testing across primate genre best distinguishes them by a single general factor (see also the latest issue of Behavior Genetics for a detailed exploration of the g factor in mice). The correlation between brain size and the g factor across 25 primate genre is .77. Lee discusses this in the context of Bruce Lahn's recent papers.

David C. Geary uses material from his fascinating new book in The Origin of Mind: Evolution of Brain, Cognition, and General Intelligence. Geary argues that "The primary dynamic that has driven and is currently driving human evolution is competition with other people and groups of other people for resource control" and that intelligence involves the ability to form more accurate "mental models" of the outside world, and that the systems to build these models "are known as general fluid intelligence, working memory, and attentional control" and that "The combination of these systems and folk knowledge is the foundation upon which human intellectual and cultural advances have been built".

In contrast to Geary, Linda Gottfredson argues in Innovation, Fatal Accidents, and the Evolution of General Intelligence that competition with other people hasn't been the primary engine of human intellectual evolution, but instead that each boost in intelligence led to new technologies which created new dangers that continuously pruned off those at the bottom of the spectrum. Innovation then would boost IQs through increased selection, which would thus lead to even more innovation, and the cycle fuels itself. (her paper, by the way, is one of the few that can already be found online. Here (PDF).

Kim Hill presents The Adaptive Function of High Cognitive Ability in Hunter-Gatherers: Feeding Niche or Social Complexity?, where an IQ study of modern hunter-gatherers is discussed. In support of David Geary and the modern consensus of "Machiavellian intelligence", it was found that higher IQs in this group were associated with higher social status. Contradicting the old "Man the Hunter" theory of intelligence, Hill reports that there was no association between hunting prowess and measured intelligence.

In Mutual Mate Choice for Intelligence as a Fitness Indicator, Geoffrey Miller uses research from evolutionary psychology to support his theory that sexual selection was responsible for driving up human intelligence. And in a related lecture, Intelligence and Mate Choice, Mark Prokosch explores the assumptions behind the sexual selection theory (e.g. how important is intelligence in long and short term mates? How accurately do females judge intelligence?) and tests some of them directly.


A number of other presentations deal with the issue of "Emotional Intelligence" (empty) and "Life History Theory" (all the presented papers fail to support it). I do not find those issues interesting, but the abstracts are in there for you. A few more deal with technical issues, such as test refinement, which are no doubt important to the field but less fun to talk about. The last one I'll leave you with, though, might be of interest if you've ever had the misfortune to encounter Internet ads, and says all it needs to just in the title: Web-Based IQ Tests: A Concept Whose Time Has Not Yet Come.