13 June 2013

Neuroscience doesn’t need a grand theory to advance

In a new post on the planned BRAIN Initiative, the National Science Foundation characterizes neuroscience as:

Desperately seeking a theory

Not just seeking, but desperately seeking.

In other words, scientists lack a basic, overarching theory about healthy brain function that would explain how memories, thoughts and behaviors emerge from dynamic activities in the brain – any brain.

That doesn’t sound like a theory of the brain. That sounds like what is desired is a theory of consciousness. This does not surprise me; it’s the big hairy audacious goal for many neuroscientists. I have a message for my fellow neuroscientists about this.

People, chill out. You don’t need a theory to make excellent progress in understanding how any of those things work.

Let me draw a parallel. Consciousness is a network property of certain combinations of matter. Life is also a network property of certain combinations of matter. We want to explain what are the conditions necessary for those network properties to appear. So, the task of understanding consciousness for neuroscientists is very similar to the task of understanding of life for biologists.

We do not have a theory of life in biology.

By this I mean no theory predicts what configurations of matter are capable of life. Could you have a silicon based life form? A life form that exists in liquid methane instead of water? We have no idea. We have been surprised by extremophiles on Earth that live in conditions that were generally predicted not to be able to support life.

In my estimation, one of the last hopes for a unifying theory that separated life from non-live was vitalism: the idea that all living things had an “essence” that non-living ones did not. But we now know that there is no clear distinction between animate and inanimate matter, and vitalism is dead.

Of course, we may develop a theory of life, particularly if we can ever discover other independent origins of life, whether it be “shadow life” on our planet, evidence of life on other planets, or develop artificial life.

It’s not that we lack theories in biology; we do have them. Evolutionary theory and cell theory are the two main (some would argue only) theories in biology. But neither of these do the job of predicting what configurations of matter have what properties of life, and which don’t.

Similarly, neuroscience does have has theories: neuron theory, for instance. It doesn’t explain the “consciousness, but then again, its biological relative, cell theory, doesn’t explain “life,” either.

Yet this lack of a theory has not prevented an explosion in our understanding of biology. Inheritance and the development of complex multi-celled embryos from single cells were once viewed as great mysteries. Work on DNA and stem cells and gene regulation and so much more means that we have extremely good understanding of these processes. We did all of that without a “theory of life,” and there is no end in sight for biological discoveries. Biologists are not complaining that not having a “theory of life” is limiting their research. Almost nobody in biology is worrying about it.

I am not saying it wouldn't be nice to have a grand unified theory for consciousness. Theories are wonderful things to have. With the Society for Neuroscience meeting being one of the biggest scientific meetings in the world, it seems that neuroscientists are not being limited in making discoveries by their lack of an overarching theory.

Hat tip to Erin McKiernan.

Related posts

When is neuroscience not neuroscience? When it’s neurobiology
Nominees for the Newton of neuroscience

External links

Prying open the black box of the brain

Photo by FunGi_on Flickr; used under a Creative Commons licence.


John Hewitt said...
This comment has been removed by the author.
John Hewitt said...

Agree. In searching for theories in neuroscience though, it seems we may need theories about life just to get started. forget building it first in AI as a way to understanding. While a paramecium or amoba isn't conscious, it sure has some complex autonomous behavior. Unfortunately one can learn all there is to learn about a individual termite yet be unable to predict termite mound dynamics when you put a bunch together. For a small paramecium and amoeba, mitochondria are not so central, for brains however, understanding them is the best path forward to understanding the structure, and hence function of the brain.

Thomas Raab said...

I basically agree: What we need is not a grand theory of the brain, which would only limit the diversity of research.

What we do need,however, are theories of the brain, that is partial explanation complexes that focus on different aspects of the brain and allow us to describe certain aspects of the brain separately.

What we need to look for, I think, is diversification rather than unification - like the many different ways to do biology.

We already have many such theories, of course. But it can't hurt to make clear that a grand theory of the brain might as well be a theory, that sufficiently explains part of the functions of the brain, rather than all of it (which is probably still way off).

neuromusic said...

I think you set up a bit of a straw man here with the whole "consciousness" bit.

I see no reason why asking "how memories, thoughts and behaviors emerge from dynamic activities in the brain" implies a search for consciousness... indeed, all of these can be addressed in models which are typically not considered to have "consciousness" such as c elegans.

Except maybe the "thought" part. But that depends on how one experimentally operationalizes "thought".

Still, I think you are right that the theoretical challenges that neuroscience faces are more akin to those in biology rather than physics.

In my reading, the call here for a "unified theory" of brain function is largely a justification for pursuing neuroscience in model organisms. That is, if we are indeed only interested in "how healthy [human] brain function explains memories..." blah blah blah, then why waste our time in model systems at all if there is not the hope of discovering a unified theory (or theories or evolutionarily conserved principles).

This isn't news to the 95% of neuroscientists who don't work with the humans, of course. But noting the lack of a "unified theory" is a one-liner way to sell model organism neuroscience to the public... that we aren't just trying to understand THE brain (whatever that is) but (as the NSF quote emphasizes) ANY brain.

Anonymous said...

First off, I think that the comparison between "a theory of life" and "a theory of consciousness" is a poor one, and I think that the leap that "a theory of the brain" must mean "a theory of consciousness" is a big one.

We do in fact have theories for many theories for how pieces of the brain work. People like Simoncelli have theories of vision (including data from invertebrates!) - or at least different parts of it - and clearly Reinforcement Learning is one of the more successful theories that explain what the dopaminergic system is doing. These have enhanced our understanding of how the areas function and operate over and above what the data tells us. Progressively more comprehensive theories can give progressively more comprehensive understanding.

Theories synthesize data. Saying that you're okay without a theory is saying that you're okay without understanding or predictive power.

Biologists are not complaining because we have an overabundance of data that we can get from our tools - for now. Once the data starts becoming harder to get, then you want the theories.

Unknown said...

'Elimination of internal complexity is feasible when the relevant graph is strongly connected...'(J. Gunawardena, PLOS ONE May 14 2012; A Linear Framework for Time-Scale Separation in Nonlinear Biochemical Systems).

Is consciousness simply an emergent property of strongly connected active neurons? Of complex neural signals collapsed into a single dimension of information?

The math seems to add up..