Alex sent me a link to this study yesterday. The other weekend we were talking about all the bad studies that, like this one, appear much better than they are simply because they involve pretty pictures of the human brain. My initial impulse was to write an extensive piece on why the vast majority of brain imaging studies are complete and utter crap, but quickly realized that to write it properly would require some more care and research than I tend to give these entries. I would like to spend the time to write a good article, for The Colossus or elsewhere, but because this study is such a load of bull, I can’t resist pointing out a few glaring flaws.

To summarize their findings: the researchers made morphometric MRI scans of a sample of experimental subjects, and looked for correlations in brain volume with performance on a standardized IQ test. Morphometric MRI is a noninvasive method for measuring the volume of different brain regions. Now, it’s important to realize there’s a lot of variability between brains, both in terms of size and how the space is allocated. The question is whether that variability means something, and to address that you have to see whether the variability can be explained by another parameter, like intelligence or gender.

Finding those other parameters is the job of statistics, and statistics are only as good as the variables you choose. The independent variable, intelligence, is established by administering a WAIS IQ test. Although this is a fairly standard test, and as little as I hate to jump on the Anti-IQ Bandwagon Of Political Correctness, IQ is a deeply flawed measurement. Even if it is an accurate indicator of cognitive ability, no measure of “cognitive ability” tells us anything about how a person is figuring things out.

The other variable, the dependent or “output” variable, is even more useless. The classic error in statistics is to mistake correlation for causation, and the best way to avoid this mistake is to have a plausible, testable explanation for why one thing causes the other – a mechanism. On such terms, the ratio of gray matter to white matter is totally meaningless. It’s true that gray matter is where the cell bodies of neurons are located, while white matter is composed primarily of axons, the long-range connections between neurons. But this is an anatomical distinction, and not a functional one. Not all connections are of equal value, and they certainly do not all serve the same function. Connectivity does not explain intelligence any more than “more gray matter” equals “more processing power”. Neither “gray matter” nor “white matter” get us any closer to knowing how brain circuits work, how they talk to each other, and how information is processed and utilized.

All that this study has done is to show a statistical connection between an uninformative measurement of intelligence and a couple of arbitrary anatomical categories. They might as well be demonstrating a correlation between “an irascible temperment” and a pronounced forehead, for all the good they’re doing for real neuroscience. What’s sad is that hundreds of poorly-conceived studies just like this one, dealing in arbitrary labels and meaningless correlations, are being funded just because their authors can show donors pretty pictures and talk about “what makes people smart”.