Apr 10 2012

Simple Brain Wiring

As a general rule it’s a safe bet that things in nature will turn out to be more complex than we initially imagine. We seem to pass through several general phases in our understanding of any phenomenon. First we have no idea what’s really going on and essentially invent fanciful superstitious ideas that are in line with our prejudices and desires. Then when we actually study the phenomenon scientifically we begin to see some regular patterns and some basic rules present themselves. We then are at high risk for assuming that we actually understand the phenomenon because we have developed simple rules to explain them, and certainly we know much more than we did in the pre-scientific superstitious phase.

But then as we make further observations and experiments those simple theories break down and we discover there are layers of depth and complexity to the phenomenon. We may feel for a while that our ignorance is growing more rapidly than our knowledge, as every discovery leads to yet more questions. Khunians may describe this as a period of scientific crisis leading to paradigm shift, when we are searching for radically new theories to solve the growing problems with existing theories.
This is a familiar story, but science does not always progress in this manner. Sometimes our observations and experiments lead us to an elegant simplicity, rather than a hoary complexity. As scientists we like to find regular patterns, and I particularly like finding meta-patterns – ways in which the scientific process itself tends to operate. What, then, do phenomena that appear to yield to elegant simplicity have in common? One feature, I suspect, is that the phenomenon is the result of emergent complexity.

In other words, there is an underlying phenomenon that is deceptively simple, but complexity can emerge from it. Bees constructing a nest appear to be the result of individual bees following very simple rules. The spiral structure of a nautilus shell emerges from very simple developmental rules.

What about the wiring of the brain? Perhaps the brain fits both patterns at once – it is both more complex and more simple than we had imagined. That it is more complex is something I have discussed many times before. There are layers of complexity to the functioning of the brain that we are beginning to see as we develop new tools to image brain anatomy and function. We do not yet know how deep this complexity goes, but we are beginning to get some idea.

I would emphasize at this point, however, that deepening our scientific knowledge does not necessarily invalidate more superficial understandings. Simpler theories may still be correct as far as they go, but do not account for the deeper levels of complexity or underlying causes. My favorite example is that no matter how many layers of complexity we uncover to genetics, DNA will still be the primary molecule of inheritance.

Recently neuroscientists published the results of the most detailed scan of brain wiring to date. They are taking advantage of a new MRI installed last fall at the Massachusetts General Hospital (MGH) – the Connectome diffusion magnetic resonance imaging scanner – the most powerful MRI scanner of its kind that can apparently image brain wiring with 10 times the resolution of other MRI scanners. Van Wedeen, M.D., who is heading this research, says that while older scanners could see about 25% of the brain’s wiring, this new more powerful scanner can see about 75%. Also, this scanning can take place in minutes instead of hours, which is much more practical for scanning a living creature.

What the researchers found is that the primate brain’s wiring appears to be much more simply organized than suspected. Rather than a complex tangle of connections, they found a regular grid pattern. The grid is three dimensional – axons from brain cells seem to turn at right angles only, either left-right or up-down. There are no diagonal fibers. The NIH reporting of this research notes:

As the brain gets wired up in early development, its connections form along perpendicular pathways, running horizontally, vertically and transversely. This grid structure appears to guide connectivity like lane markers on a highway, which would limit options for growing nerve fibers to change direction during development. If they can turn in just four directions: left, right, up or down, this may enforce a more efficient, orderly way for the fibers to find their proper connections — and for the structure to adapt through evolution, suggest the researchers.

This is a very interesting finding. It implies that the brain applies simple developmental rules that allow for complexity to emerge, much like those bees building a nest. This does not mean that the brain is not horribly complex in its anatomy and function. It just means that this complexity is partly emergent. This makes sense in light of what we know about brain development.

DNA does not contain a blue-print of brain anatomy. We know this partly because the brain contains much more information that all our DNA, let alone just that part of the DNA that codes for brain proteins. As the brain develops it encodes greater information than was required for its development. This is partly due to the various types of mapping that occur as the brain develops – brain connections follow sensory input or muscle control, for example.

So, rather than being a blueprint, DNA is more like a set of instructions or algorithms, not unlike the behavioral rules followed by bees. Follow the instructions and brain anatomy, with all its complexity, emerges. It is therefore not all that surprising that this type of regular grid pattern would be seen in the wiring pattern of the primate brain.

I have to emphasize that this does not mean the brain is not complex, it just tells us something about how that complexity comes about. I do think this bodes well for our understanding of brain function – it is always helpful when there is an elegant simplicity to nature there for us to discover. Our brains like elegant simplicity. This will provide something for us to hold on to when we try to unravel the mind-boggling complexity.

This type of discovery may be especially helpful to our efforts to reverse engineer the brain with virtual models – building functioning computer models of brain wiring. I believe that it is likely we will succeed in building artificial intelligence first by copying brain wiring, even without fully understanding that wiring. Discoveries like this may make that outcome more likely. If we discover a set of relatively simply algorithms by which the brain develops, then we can copy those algorithms and run them in a virtual simulation and hope that a virtual brain emerges as a result.

We may then end up with a virtual brain that functions much like a human brain, and even results from a similar developmental process, long before we grasp the full complexity of that brain. This, in turn, creates the potential for such virtual computer brains to evolve and form even greater complexity, completely outstripping our ability to keep up and understand those virtual brains.

9 responses so far