Tuesday, January 24, 2012

How do you know you have a brain?

RJ Lipton posts at Godel's Lost Letter about the use of slime molds for computation. They are
no more than a bag of amoebae encased in a thin slime sheath, yet they manage to have various behaviors that are equal to those of animals who possess muscles and nerves with ganglia — that is, simple brains.
Scientists have been using slime molds to model various situations, and have been able to demonstrate that they come up with interesting "answers"
One of the driving forces behind the new computational interest in slime molds is a seminal experiment performed by Atsushi Tero of Hokkaido University. He grew the mold Physarum polycephalum on a standard dish, but placed both attractors and obstacles on the dish. The attractors were food—oat flakes if you must know—and the obstacles were bright light. The mold is attracted to food—who is not?—but is camera shy and tries to avoid bright light. Tero for fun arranged the attractors and obstacles to model the major centers in the Greater Tokyo Area. The mold initially filled the whole dish, but over time evolved into a network that connected the centers in a way that closely approximated the actual Tokyo rail system.
No single cell is "aware" of anything except its immediate environment, and yet the creature quickly converges on a particular configuration. Of course, life is full of such phenomena. A single-celled embryo in a reasonably hospitable environment quickly arrives at a "solution": a functioning adult. An adult body which manages to continue to function through all the insults life throws at it.

Or our brains. Like every other part of our bodies, it is descended from the original embryo. At no point is any neuron aware of anything more than its immediate surroundings, and yet the whole organ is capable of rich behavior. Even more impressive is how robust it (and the human body in general) is. The various parts of our bodies vary significantly from one person to another, and yet they are able to work together, to the point where people can have their viscera on the wrong side of the body, and never come to know it.

Bradley Voytek over at Oscillatory Thoughts blogged a fantastic, mind-blowing post about the incredible resilience of the human brain: "Why we don't need a brain". Most of the rest of this post simply plagiarizes what he has written.

He describes a case reported by Professor John Lorber, and includes the CT scan published with the study.
There's a young student at this university... who has an IQ of 126, has gained a first-class honors degree in mathematics, and is socially completely normal. And yet the boy has virtually no brain.
To be more exact
To put this in perspective, at its largest parts, this boy's brain was still only half the size of a normal brain. Total volume appears to be well below that.
Voytek rejects Lorber's theory that this indicates enormous redundancy in the human brain. The question he asks is simple: given what we know about the terrible consequences suffered by victims of stroke, how could this boy be so normal in all important respects?

He then describes another case ( "How much brain is really necessary" ) published by Distelmaier: a young girl born with hydrocephalus that caused a severely underdeveloped brain. Surgeons treated the hydrocephalus with a shunt to remove fluid, and the result was that, at 20 months, her cerebral and cerebellar hemispheres had managed to do quite a bit of catching up and, as he puts it:
At 34 months the authors performed a followup neurodevelopmental examination of the girl. Although she had some developmental delay (particularly motor problems), she appeared to be socially and cognitively well off, especially considering where she started!
He ends with a paper by Desmurget, Bonnetblanc, and Duffau ( "Contrasting acute and slow-growing lesions: a new door to brain plasticity" )
Desmurget and pals were basically trying to reconcile some strange clinical observations: if a patient has a stroke to an "eloquent" part of the brain (basically neurosurgeon-speak for language or motor cortex), there are clear behavioral deficits. That is, damage to eloquent cortex causes speech problems or paralysis/hemiparesis.
Paradoxically, however, patients with low-grade gliomas (a type of brain cancer), could undergo surgical removal of large parts of brain tissue in the eloquent cortex without any noticeable behavioral consequences.
The reason appears to be that a stroke is practically instantaneous. A glioma grows over the years, and can become enormous. In the meantime, however, the brain adapts, other parts of the brain take over from the damaged areas, and when the surgeons come along and excise the growth, the consequences for the person are relatively minor. You can recover from enormous damage to your brain, provided it happens slowly enough.

I obviously find this simply wonderful. How, as some brain tissue dies off, other parts take over and ensure that the organism is able to function. This is a kind of computational problem, similar to that of the slime molds, and I guess the organ solves it by using feedback from the external world to train itself. I assume something similar happens during the "critical period" for vision.

This is all computation of a kind. However, this is obviously very different from how our computers work today, though the Internet, and other more robust networks of computers do show similar capabilities. Fascinating, but I don't know enough about the subject. Time to get a copy of Lewis Wolpert?

No comments: