Saturday, November 6, 2010

"Light, I salute thee, but with wounded nerves"

I expect this picture to become iconic. Certainly the breakthrough it represents is important enough: biologists can now activate individual neurons in a rat's brain with a laser pulse and show that this causes the rat to do something (in this case, get scared and freeze). [The writeup is lucid enough that I won't bother to paraphrase it.] The significance of these techniques is that they could potentially turn neuroscience into a discipline like physics in which one can measure responses to well-defined stimuli: the earlier fMRI work, I think one could fairly say, was more like astronomy in the sense that one mostly looked at correlations between events. The basic advantage of this is that one gets around causation-correlation problems because the causal structure of responses to, e.g., a light pulse is unambiguous. A sort-of-corollary is that data interpretation becomes much less dependent on statistics. Boris Altshuler is reported to have said that if you need statistics to understand your experiment you should make a better experiment; this is hopefully what optogenetics will do for neuroscience.

It is, of course, unlikely that these experiments can or will be performed on humans any time soon; from a scientific point of view, however, this is not an important loss. Humans are pointlessly messy subjects for almost all questions about how the brain works.

No comments: