BAM! My Thoughts on Big Bucks for Big Brain Science
The latest Big Science project is a proposal (outlined recently in the New York Times) to complement previous efforts to map the anatomy and connectivity of the brain by characterizing the activity of brain circuits that underlie perception and behavior. The details of this Brain Activity Map (BAM) project have not been officially announced, but the scientists at the center of the plan published a paper last year in the journal Neuron (Alivisatos et al.) that seems to be the outline of the plan (though we won’t know exactly until the initiative is officially unveiled). I think the article is reasonably accessible for non-neuroscientists, but at present the full text is locked behind Neuron’s pay wall. Since many people may not be able to download the work, I provided a brief synopsis in my previous post. If the paper is the center of a huge science policy initiative, we have to assume that Neuron will release it at some point. In any case, I’d like to consider some of the specifics of the proposal.
First, I should lay out some of my own biases. As a blogger, it would be great if I thought this idea was a complete waste of money proposed by idiots who somehow insinuated themselves, like inagural ball party crashers, into the elite circles of Obama’s science policy advisors. Then I could bring on the snide sarcasm, add in a little carefully pitched profanity and end up with some holier-than-thou recommendation to give the money to schools or poor people or puppies or something. That blog post would pretty much write itself. Unfortunately for my snark generator, the BAM project is basically a scientific wet dream for me. It’s pretty much the exact sort of research I’ve wanted to do since I started grad school about 10 years ago. In fact, my first project was trying to apply some of the imaging technique used by Rafael Yuste (last author on the Neuron paper) to study seizure dynamics in mouse brain slices. As it happens, I’m not really doing that kind of research now, but I certainly wouldn’t turn down a good offer. So let’s just say I’m not the most impartial observer of this particular initiative, but if my readers can indulge my transparently declared bias, then I will try to consider the merits with as much distance as I can muster.
As I mentioned in the last entry, the basic motivating premise for BAM is the idea that the neural activity that underlies behavior (in the broadest sense) is an essentially irreducible emergent property of neural circuits likely composed of thousands, millions, or perhaps, billions of cells. The authors suggest that the collective behavior of these networks is so complex, and the activity so distributed over the population, that studying only one (or a few cells) is incapable of revealing relationships with perception and behavior. Essentially, I agree. It’s not that we can’t learn anything more from recording the activity of individual neurons, but in many neural systems, really cracking the neural code seems to require a massively multicellular approach. Not only does this strategy demystify systems in which the correlation of any single cell with perception/behavior is weak, but it can help us investigate important questions about how information is processed and maintained in networks amid ongoing neural activity that is presumably accomplishing some other tasks at the same time. They are calling for a systems approach to neuroscience, and while that idea in itself is not particularly new, they are correct in suggesting new technologies make that strategy more feasible now.
However, I do think there are some important questions that need to be asked about this project. Since we only have the Neuron paper to go on, I’ll have to rely on that straw man until more information is available. First, as one of the authors suggests in the NYT piece, the goal of mapping brain activity is much more amorphous than sequencing the human genome or landing on the moon. In their roadmap, they propose scaling up at most to the mouse neocortex (still fantastically complex, but let’s be clear that they are not talking about human brains here. Currently there are no technologies for noninvasively analyzing human brain activity in the kind of detail they are proposing, so we will have to be satisfied with model organisms). As intermediate steps, Alivisatos et al. say that they want to record all the activity of all the cells in discrete neural circuits tied to behavior, but brain networks are so highly interconnected that defining circuit boundaries will have to be somewhat arbitrary. Another squishy element of their plan is that, while their proposal is to analyze neural circuits in freely behaving animals, the most high-volume recording/imaging methods (many 100s of cells) that they describe as proof-of-concept are limited to recording activity in brain slices. In behaving animals, recordings that include tens of cells are more typical with current technology, and these limits have crept up only rather slowly in the last ten years. That’s not to say that millions of dollars couldn’t be radically transformative, like the advances that brought us close to $1000 individual genome sequencing, but let’s just say scaling up current methods is a non-trivial problem.
Another major challenge that the BAM authors address only superficially is the problem of analyzing the data from massive multicellular recording technology. The current techniques have been developed with the help of numerically minded researchers like mathematicians, statisticians and machine learning experts. Many of these techniques are not easily applied to data sets that include even 100s of cells without drastically simplifying assumptions so much that it could eliminate the entire reason for such large-scale recordings. Could additional resources spur more rapid innovation? Sure. Are there likely to be techniques capable of really analyzing the interrelated activity from 1,000,000 cells in 15 years? Not a chance.
Even so, as a general approach to systems neuroscience, I think the BAM project is one reasonable strategy. Not being a science policy wonk or an economist, I can’t really speak to some of the more fundamental criticisms leveled against this plan in the last few hours. If you are against top down Big Science in general, then I don’t really have a counter-argument. If you think the federal government should only fund directly clinical or translational research, then that’s a different discussion. If you think the money could be better spent on puppies (or kittens), then you should probably be reading a different blog. But, if you think that basic science about how brains make sense of the world is not only valuable in its own right, but will also lead eventually to important advances in medicine and health, then perhaps BAM is a program you can believe in.
Update (2/20/2013): Some of the details of the plan are explained in more detail at the Science site.
A. Paul Alivisatos, Miyoung Chun, George M. Church, Ralph J. Greenspan, Michael L. Roukes and Rafael Yuste. The Brain Activity Map Project and the Challenge of Functional Connectomics. Neuron, Volume 74, Issue 6, 970-974, 21 June 2012