A paper calling into question brain scan data (the functional mri methods and statistical analyses used to correlate brain areas with emotion or behavior) is a good reminder, at least to me: this science is still in the very early stages.
I won’t go into all the details, but see Neuroskeptic for a really great synopsis and commentary of what Ed Vul et. al.’s “Voodoo Correlations in Social Neuroscience” reports, and the surrounding kerfuffel. This is the best analysis I’ve found because it has pictures. These are particularly helpful if you slept through most of your statistics classes, and just sort of have a fuzzy memory of “r”)
And so, let me say this about that: the field of social neuroscience is a rats nest of confounding factors.
Yet, the brain scans are so tantalizing. I mean, who doesn’t like a nice clear brain outline with a big orange puff on it somewhere? (Ah, the freedom in not being a scientist, I can use the technical term “orange puff” and not have to worry about colleagues calling me out for it. . . )
For diagnostic or clinical purposes, imaging can be one of the best things ever. I’ve done a teeny little bit of work commercializing some of the newer imaging contrast agents and they’re very groovy, the practical device development is pretty far along. On an individual, you can get really accurate data for some of the more diagnostic imaging.
For plain vanilla fmri, involving basically measuring the resonance of water, the trouble comes in drawing comparisons between and among individuals, as well as getting any kind of statistical significance out of the results. And no wonder — time in the imaging department is expensive, you have to book it and make sure the imaging researcher you want is on duty, and get the people you want imaged to actually show up. I worked on a project some time ago where there was a little teeny rodent mri machine that even had a waiting list at an academic institution. It’s a miracle that papers get published at all sometimes.
And so, a solid minority of papers do have a valid conclusion as to fmri data and behavior. But, a majority of published, refereed papers showing fmri’s actually only demonstrate ambiguous results. (See Neurocritic for another excellent review and terrific reader commentary).
The analytical methods at this point may not be appropriate for the questions we are asking, like, oh, say , “why do people think the way they do?” Although I put on my waders and slog through the scientific abstracts, I don’t critically review the materials and methods usually, and just take at face value of whatever’s published. I’ll try to put more disclaimers in the posts. But the brain scans are data. I wonder if the “more is better” approach to getting all sorts of data will yield less ambiguous results.
My dream world web site would include some kind of user interface sort of like the OMIM — where you click on the behavior (such as, oh, say, sending fake anthrax in the mail to financial journals because you’re pissed off that you lost money), and up pops the relevant biology. This would not only include anatomical brain regions, but also white matter, axons, dendritic shapes, genetic and epigenetic influences, and physiology, such as confounding tumors or infections or other conditions. (My guess is that there will be a fairly large non-Mendelian genetic component that is temporal, not only by age, but also by environment.)