20 December 2012

Does IQ Have Irreducible Components?

A Brief History Of The IQ Concept

The concept of IQ was invented when psychologists noted that across a wide variety of tests, a single common statistical factor, "g", explained a large share of the outcome of all of them.

Early tests designed to measure the "g" component which was distinct from the variance in test results particular to each test instrument were called "intelligence quotient" tests, because the final scaled score on the test when administered to children was expressed as one hundred times a ratio of the average age of someone performing as the test taker did to that person's actual age.

Eventually, the way that IQ tests are scaled was changed. Instead, the results from a sample used to validate the testing instrument were fit to a normal distribution with an arbitrarily chosen mean of 100 and a standard deviation of 15 points (or sometimes 16 points), as that scaling provided approximate back compatability with the age ratio system.

Twin studies, adoption studies, and other psychological approaches have determined that IQ as measured by traditional IQ tests has a strong hereditary component (which is particularly strong in middle class and more affluent people), is relatively stable over time, and is a reliable predictor of success in a wide variety of endeavors including involvement in criminal activity, educational attainment, vocabulary and lifetime earnings.

A synthesis of this data that explains the differences in the extent to which IQ is hereditary by social class, describes the underlying significantly inherited mental abilities that IQ tests seem to measure as basically potential academic ability in the absence of negative environmental factors.

It is also widely acknowledged that high IQ does not appear to be derived from a small number of common genetic variants. While there are some well know and not uncommon genetic variants that are associated with particular types of mental retardation, there is no short list of particular genes that make you smart.

In the case of IQ, the "missing heredity gap" between the strong hereditary component of IQ and the lack of identified "IQ genes" is usually taken as a sign of support for the working hypothesis that the IQ is the product of a massive number of genes, each of which makes a tiny contribution to overall IQ, and that the genetic variants that are associated with high IQ are often rare ones.

To simplify the current hypothesis to toy model form, there might be 100,000 genes that each play roughly equal roles in determining IQ, and high IQ people might have several thousand variants of some of those genes that are not present in average people. But, the thousands of particular rare variants of those genes present in one high IQ individual are likely to differ greatly from the thousand of particular rare varients of those genes present in another high IQ individual.

In this model, the difference between someone with moderately below average intelligence, someone with average intelligence, someone with above average intelligence and phenomenal geniuses are all basically differences in degree, rather than differences in kind. In contrast, the vast majority of people who have greatly below average intelligence have a specific developmental disability, often genetic in origin (although certainly not always) that is different in kind from the hereditary factors that play an important role in distinguishing a "C" student from an "A" student.

A New Study Identifies Three Irreduciable Components To Traits Measured By IQ Tests

A new study, analyzing the results from having 100,000 self-selected volunteers from around the world, who formed a diverse but not random sample, who took a battery of 12 psychometric tests on the Internet that mirror traditional IQ test materials.

The study argues that the psychologists who invented the concept of "g" made too much of a leap in boiling the statistical factor "g" (for the "general" factor applicable to all tests) down to a single number (when in fact it takes more than one factor to meaningfully capture what they tried to capture with the concept of "g") and that they also overreached in then supposing that there is some single meaningful real world mental trait that provides a source for the statistical correlations that were captured by the "g" factor in the tests used to devise and validate the IQ concept.

[T]he researchers asked respondents to complete 12 cognitive tests tapping memory, reasoning, attention and planning abilities, as well as a survey about their background and lifestyle habits. . . .
The results showed that when a wide range of cognitive abilities are explored, the observed variations in performance can only be explained with at least three distinct components: short-term memory, reasoning and a verbal component.
No one component, or IQ, explained everything. Furthermore, the scientists used a brain scanning technique known as functional magnetic resonance imaging (fMRI), to show that these differences in cognitive ability map onto distinct circuits in the brain.
The press release version of the study quoted above from Science Daily doesn't engage with how the researchers addressed the problem of having a self-selected sample made up only of people with Internet access (which is not to say that there aren't methodologically valid ways of using a large convenience sample of that kind to make certain kinds of statistically valid conclusions), or the magnitude or statistical signficance levels of the conclusions that were reached.

For example, while the researchers conclude that no one component explained the entire range of performance on the twelve cognitive tests, the story does not make clear how much, if any, correlation there was between those three factors, or which, if any, of the cognitive tests which most heavily influenced by which of the three components.

Of course, the survey test, like all survey based tests, also suffers from a lack of validation against other means of evaluating the individuals tested with different kinds of instruments. For example, there is no way to compare these survey scores against verified standardized test scores, or to be certain that information on the questionaires that was reported was accurate.

The University of Western Ontario study is Adam Hampshire, Roger R. Highfield, Beth L. Parkin, Adrian M. Owen. Fractionating Human Intelligence. Neuron, 2012; 76 (6): 1225 DOI: 10.1016/j.neuron.2012.06.022 The abstract of their study states:
Highlights
    ► We propose that human intelligence is composed of multiple independent components ► Each behavioral component is associated with a distinct functional brain network ► The higher-order g factor is an artifact of tasks recruiting multiple networks ► The components of intelligence dissociate when correlated with demographic variables
Summary
What makes one person more intellectually able than another? Can the entire distribution of human intelligence be accounted for by just one general factor? Is intelligence supported by a single neural system? Here, we provide a perspective on human intelligence that takes into account how general abilities or factors reflect the functional organization of the brain. By comparing factor models of individual differences in performance with factor models of brain functional organization, we demonstrate that different components of intelligence have their analogs in distinct brain networks. Using simulations based on neuroimaging data, we show that the higher-order factor g is accounted for by cognitive tasks corecruiting multiple networks. Finally, we confirm the independence of these components of intelligence by dissociating them using questionnaire variables. We propose that intelligence is an emergent property of anatomically distinct cognitive systems, each of which has its own capacity.
Serious Researchers Have Never Claimed That IQ Is The Only Mental Trait

The headline and lede conclusion the researchers make "that the notion of measuring one's intelligence quotient or IQ by a singular, standardized test is highly misleading," takes down a straw man argument.

Psychologists Have Never Claimed That A Single Standardized Test Perfectly Captures "g"

Psychologists have since the very begnning recognized that any particular standardized test will be an imperfect measure of the theoretical concept "g" that is what is usually meant when someone uses the term "IQ". Indeed, the entire concept of "g" evolved from the recognition that different standardized tests were measuring underlying traits that were similar but not identical and defining the similarity statistically.

Also, esssentially every widely accepted IQ test already produces a composite result from subscores involving test items that test different domains of mental ability that are believed to be influenced by "g" to a greater or lesser extent.

Standardized IQ tests are designed to strike a balance between having test results that are usually very similar to an estimate of "g" for the test taker based on a wide variety of tests, not requiring undue time or expense to conduct, being comparable when individuals taking IQ tests on a variety of instruments are compared, and being culturally neutral - so that knowledge that only people with certain cultural experiences would have doesn't make someone smart who hasn't been exposed to that culture look dumb.

There Are Widely Accepted, Well Defined Mental Traits Other Than IQ That Matter In Life

There are many mental traits, including all of the "Big Five" personality traits, whose outcomes appear to be unrelated to IQ. For example, in the famous Tremain IQ study, the distribution of personality traits of the people in the sample limited to high IQ individuals were not significantly different from the distribution of personality traits in the general population, even though personality traits were shown by the same study to have a non-neutral impact on lifetime economic success.

There have also been popular and serious academic efforts to identify other stable mental traits that are independent of IQ. Some of the traits that have been explored in this effort, other than the Big Five personality traits, include structural visualization, creativity, "emotional intelligence", and musical ability. A number of congenital mental traits sometimes classified as "disorders" or as forms of "neurodiversity" including sexual orientation, schizophrenia, bipolar disorder, phobias and anxiety disorders, and ADHD, are likewise usually conceptualized as unrelated to IQ. Many of these traits also have strong hereditary components, sometimes at least as strong or stronger than IQ.

Finally, some mental traits of adults, while quite stable, are widely acknowledged to be completely or predominantly learned traits, rather than traits that are hereditary or at least congenital. Examples include birth order effects on personality, the language and dialect of that language that you speak as a native speaker, your religious affiliation, and the cultural preferences associated with one's social class.

The Limits Of fMRI Data

While certain kinds of thinking, like short-term memory, reasoning and verbal ability are localized in particular areas of the brain that can be identified using functional MRI imaging studies, the conclusion that the factors that distinguish high IQ individuals from low IQ indviduals are localized in these parts of the brain is not a consensus view of researchers looking at the neuroscience of IQ.

The conclusion the high IQ represents high performance of several independent and localized types of brain function is one legitimate hypothesis about where IQ comes from, but it isn't the only one.

An alternative view, which ties into the notion of a general factor in IQ, is that IQ is mostly a function of brain features like connectivity and neuron activity rates or processing rates across the brain, and that it just so happens that these general and distributed features of brain organization most visibly manifest themselves on certain kinds of tasks like short-term memory tasks, reasoning tasks and verbal tasks that localize themselves in particular areas of the brain. In this view, the neurological factors that are involved in high IQ are distributed but psychologists simply haven't been as good at devising good ways to test it for brain functions localized in other parts of the brain.

Other Observations Made By The Study

The large online survey that the researcher conducted also made other notable observation:
Regular brain training didn't help people's cognitive performance at all yet aging had a profound negative effect on both memory and reasoning abilities. Intriguingly, people who regularly played computer games did perform significantly better in terms of both reasoning and short-term memory. And smokers performed poorly on the short-term memory and the verbal factors, while people who frequently suffer from anxiety performed badly on the short-term memory factor in particular.
A decline in memory with age is normally conceptualized as having roots in one or more of a variety of specific chronic disorders associated with the aging process that differ in kind from the source of a person's IQ.

The factors associated with computer games could be a product of self-selection, with the fact that the test instrument was computer based and that gamers were more comfortable with that means of interacting, or social class (since this was not a random sample).

The link between smoking and some of these factors could be a consequence of smoking or could be a proxy for social class effects since smokers have a lower socio-economic class on average than non-smokers and social class is associated (causally) with IQ.

The link between anxiety and short-term memory is perhaps the most interesting of the connections mentioned above. Some of the leading theories about the neurology of anxiety disorders associate them with defects in the amygdala, a part of the limbic system that isn't particular important for short-term memory. Another leading theory about the neurology of anxiety disorders related to individual variation in the way that individual brains produce and reuptake neurotransmitters such as seretonin, out of the brain. I am not aware of any studies previously linking anxiety disorders to either IQ or to short-term memory.

No comments: