Wired for transcranial magnetic stimulation

Transcranial magnetic stimulation (TMS) is one of the most frightening tools used by neuroscientists in the course of their research into the function of the brain. The basic idea is that rapidly oscillating magnetic fields applied through the brain will induce currents in the neurons through which the magnetic field passes, altering their firing properties. It's sort of like magnetoencophalography in reverse. However, these currents are completely uncontrolled and have a spatial scale of centimeters, whereas a single neuron is a few micrometer in diameter. TMS is thus very similar to electroconvusive therapy; the main difference is that the currents are entirely internal to the brain. Unsurprisingly, TMS can induce seizures, and I have yet to see convincing evidence that it does not do any focal damage to the stimulated area.

Some reasonably reputable people use TMS to disable small areas of the brain and thus identify their function. For instance, they apply TMS to a specific region of the brain while a subject counts up from one. If an area related to short term memory or language is targeted, subjects will stop mid-count and lose their place. If the visual areas of the occipital lobe are instead subjected to the induced current, the subjects may report seeing flashes of light, but their count will continue undisturbed. From a scientific perspective, this is relatively reasonable. If you inject a whole bunch of noise into a robust and stable computational system, you would expect to disrupt the local computation without severely impacting more distant computational modules. Experiments like this need not assume that the stimulation has any particular effect on computation or learning, just that it scrambles the local state of the brain.

Less reputable people claim that TMS can be used to enhance cognitive function or treat psychiatric disorders. Wired seems to lap up these claims like a dog guzzling antifreeze: they're sweet going down, but ultimately toxic. I just stumbled on no fewer than three separate articles on this unproven and potentially dangerous technique. Of course, the real crazy reports come not from the popular media, but the academic fringe. Allan Snyder is particular notable in this respect. Anyone who thinks that "the left fronto-temporal lobe" is a localized brain area or that eleven is a sufficient sample size for statistically significant conclusions in a single-blind protocol where the experimenter must subjectively evaluate the quality of hand-drawn images needs to have his head examined.