After Phrenology

Will we ever understand the brain?

MIT Scientists Declare There Is No Science of Brains!

Have neuroscientists been wasting their time?

In an article published yesterday in the Proceedings of the National Academy of Sciences (http://www.pnas.org/content/early/2011/08/31/1112937108.abstract), scientists from the Kanwisher lab at MIT report finding regions of the brain that are specialized for language!  

What, you aren't excited?

Admittedly, this will not seem like news to the general public, despite the press release from MIT. In the popular imagination, local specialization--what we in the neurosciences call selectivity--is just how the brain works: this bit for vision, that bit for language, this one for motor control.  So the paper reports exactly what we already know. Dog bites man. (Yawn)

But in fact, a series of recent studies appearing in Behavioral and Brain Sciences, Journal of Cognitive Neuroscience, Nature Methods, Proceedings of the Cognitive Science Society, and Trends in Cognitive Science, among many other places, have painted a rather different picture.  According to this alternate model of brain function, individual regions of the brain are not dedicated to performing tasks in specific cognitive domains, but instead offer their computational services in multiple contexts.  What differentiates the cognitive domains like language and motor control is less a difference in which brain regions are assigned a task, and more a matter of which partners a given region works with.

Find a Therapist

Search for a mental health professional near you.

This finding is illustrated in the following pictures, generated using data from my own lab.  The spheres represent different regions of the brain, arrayed in a 3-D anatomical space, and lines between the spheres indicate that the regions cooperate under some specified condition. The pictures below highlight the functional partners of precentral gyrus when people are engaged in attention tasks, motor-control tasks, and language (semantics) tasks. 

Attention

Action
Semantics

Precentral gyrus is involved in supporting tasks in all of these various domains, but cooperates with a different set of partners in each. In general, emerging research results indicate that the brain is an organ that achieves its function by assembling coalitions of neural partners, and that each partner may serve in different coalitions at different times.

Naturally, not everyone is a convert to this new model of brain function, and MIT has been home to some of the staunchest defenders of regional selectivity as a fundamental principle of the brain's operation. But lately the data have not been going their way, and scientists only have two choices when that happens: modify (or abandon) their theory; or question the validity of the data.  In this paper, the Kanwisher lab has elected the latter strategy.

What they argue, in essence, is that all the studies that purport to show overlaps in the neural tissues dedicated to different functions are in fact mistaken, because they rely on data that has been averaged over many participants.  If anatomical differences shift the "language" regions a bit forward in one brain, and a bit back in another, then averaging these will put it in the middle, where neither brain actually computes language.  Looking at a third and fourth brain in another study, the same kind of averaging might make it appear that "attention" functions are being computed by the very same neural tissue that computed the language functions, even though this would not be true in anyone's individual brain. In neuroscience, the authors conclude, you have to study things one brain at a time.

Hence the title of this post, for if this is true, there would appear to be no science of brains! 

But surely I exaggerate. Well, yes and no. Certainly it is true that individual differences are of interest in their own right, and can be studied scientifically.  But science as we normally conceive of it moves forward by finding general truths--the laws of motion; the pressures of natural selection; the contingencies between behavior and reinforcer in operant conditioning.   Discovering these truths requires adopting the right abstractions and guiding idealizations that allow us to see the common characteristics that unify otherwise disparate things.  Regional selectivity had long been one of the central guiding idealizations of neuroscience.  But recent findings had led many of us to give that up, and to replace it with the notion of stable, identifiable functional networks of regions that brains seem to have in common.  Adopting this perspective, we have so far discovered an attention/salience network, a motor network, and the resting state network--and we expect there is much more to come. For history has shown that once a science finds the right abstraction, the right guiding idealization, progress can become very rapid, indeed.

So it is somewhat misleading to suggest--as one of the authors of this paper did in the interview with the MIT press office--that the choice we face is between a brain with selective regions, or "one big mushy multifunctional thing in there." Instead, the choice is between a brain with selective regions and a brain that assembles functional coalitions--and between the study of idiosyncratic anatomy, and a general science of the brain. 

This is not to say that their research program should not be pursued. In science, we have to let the data lead us where they will. But I know how I am placing my bets.

UPDATE 9/3 -- Further thought and conversation suggests the following, more sympathetic perspective on the research direction being taken by this group: For the past twenty years or so it has been standard practice to map functional responses into a common anatomical space, and to average between participants within that anatomical space. But an alternative is to define an abstract functional space, map individual anatomy into that space, and to average over participants in the abstract functional space. So, putting aside the rhetoric about regional selectivity and the discovery of language specific areas that dominates the paper reviewed above, perhaps it is better to think of the work as a preliminary exploration of an alternate abstraction for the neurosciences.  Put in that light, things look alot more interesting and potentially fruitful, and I would look forward to seeing futher exploration of that possibility.

Michael Anderson, Ph.D., is an assistant professor of psychology at Franklin & Marshall College.

more...

Subscribe to After Phrenology

Current Issue

Let It Go!

It can take a radical reboot to get past old hurts and injustices.