Muscle control and the computational power of the subconsciousness

This week’s subject was muscle control in brain and spinal cord. The lecture and the book gave lots of information, but one simple anecdote greatly astonished me: details of walking are controlled by the spinal cord, not by the brain. I can consciously decide to walk, and also where to walk and how fast, but usually I make no decisions about and pay no attention to the actual movements of my feet. Thinking about this while actually walking on the break in the middle of the lecture made me feel a mixture of awe and horror (yes, sometimes I am a bit dramatic). In a sense, I felt I was lacking control over myself!

Many movements get more detailed instructions from the brain, but even there, conscious mind is mostly interested in the big picture and not the details. There are many brain regions that work on movement control. The motor programs of cerebellum seem to be pretty important for the details of movement.

The authors of the course book (Bear, Connors & Paradiso) really like using baseball as an example of motor control. I don’t know much about baseball, but I think there are players whose job is to catch a flying ball and players whose job is to hit it with a bat. So, the get changing visual data, from which they have to calculate the nearest point where the ball is going to pass them. Then they will calculate the suitable movements for each muscle in their bodies in order to catch or hit the ball. This is a very complicated mathematical process, and the players seem to achieve the solution in less than a second. How is this possible?

The comparison of brain to a computer is as old as computer science itself. Apparently, even Countess Ada Lovelace considered this metaphor while mentally simulating programs for Charles Babbage’s analytical engine in the early nineteenth century. This comparison has its limitations, but also benefits, and seems very suitable for the discussion at hand. (By the way, I recommend John von Neumann‘s posthumously published book The Computer & the Brain. It is unfinished and partially outdated, but has some interesting points and of course historical significance. It’s a short book, I read it completely yesterday.)

Why is doing complicated maths on Python interpreter or Java virtual machine much less efficient than doing them on a C program? Why is the C program still slower than an optimized Assembler program? The difference is in the abstraction level. The case of Java virtual machine is the most obvious one, since the answer is in the name: it is a computer simulated on another computer. Obviously, a computer can’t simulate (in real time) another computer faster than itself. In each case of a higher level system, the code has to be somehow interpreted on a lower level. To put it simply, this interpretation process takes some part of the efficiency.

The brain can be seen as a very powerful computational machine (but not one having the von Neumann architecture on which almost all artificial computers are based). The conscious mind, then, is a program or an operating system running on this machine. It is a complicated piece of software that has a very high abstraction level. Thus, it also has quite limited mathematical capacity. It is often said that human working memory has capacity for about seven variables at a time (but because of the high abstraction level, these variables can have quite different amounts of information).

So, it is quite understandable that our subconsciousness has a much greater computational power than the conscious mind. That’s why we can safely automate big part of our everyday life, such as walking.

I’m ending on a side note this topic reminds me of. A course mate recently said me something like “we can image brain function but we have never seen a thought”. Well, that is because of many reasons (none of them requiring any supernatural assumptions). In addition to the limitations of modern brain imaging technologies, our interpretational capabilities are limited. How could a conscious mind with its limited mathematical power and working memory even begin parsing a thought from the action potential frequencies of the millions of neurons producing it?

Computers, of course, can help. Using them to interpret all aspects of a human thought would be an enormous task, but they have already been used to reconstruct some human-understandable material from brain activity.

Posted by Mikko Luukinen

This entry was posted in Uncategorized. Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *