Archive for January, 2008

The Brain-More Fantastic Than any Computer

Friday, January 25th, 2008

I just got back from my Green Hills Rotary Meeting and we had an emeritus professor from Vanderbilt speak. His speciality is the biology of the brain. The storage capacity and processing capacity of the brain is phenomenal beyond belief.

As a point of comparison, a gigabit of hard drive space means that there are approximately 1 billion on and off “switches” on the device. A computer running at 1 giga herz means that it can process 1 packet of information every 1 billioneth of a second. A 32 bit bus can process 32 bits with each packet. Now each action that you take, from typing a word to calculating a sum of two numbers takes multiple steps and requires a series of packets and steps to achieve.

Please realize that my specialty is not hardware, but software. I do not have the exact details at the tip of my tongue or the time to look them up. What I am calling a switch could easily contain more information or be a slightly different beast.

The point is that I cannot really comprehend just how many pieces of information are floating in my computer and the speeds at which it travels and processes. By way of camparison, China has, I believe, around a billion people. A billion is an abstract number that I cannot concretely fathom. But I work with things every day that depend on the number 1 billion.

Each of these 1 billion swithces has effectively two states, on and off and it exists in a two dimensional plane (meaning flat like a piece of paper.) When I was in college in the late 1980’s, there were material scientists trying to make a huge leap forward in storage capacity by trying to get storage to more dimensions than 2 by looking at the molecular structures and allowing inputs as more than the single input and output that we currently use. Another thing that they have probably considered is differientiating states by storing differnt amounts of electrical energy in each switch and then trying to read 2 or three different levels of energy to get a different storage amount. But 2 state systems (on and off) are much easier to for our electronics to interpret because there is much, much less room for error.

Now lets go back to the brain. A human brain has hundreds of thousands if not millions of nerves. A million cells in a two state system like our computers, could not do as much as our current computers, since a billion is 1000 times as much as a million. However, each cell can make thousands of connections. Let us assume 1000 input connections per cell and 1000 out put connections. That means that any two cells have at least 2000 thousand connections to separate neurons.

But we are only getting started. Each nueron accomplishes a task by transmitting an electrical charge from one location to another. So when a nueron gets an impuls from one of its 1000 input connections, it has to choose which of its 1000 output connections. So if the relationship is forced to be one to one, meaning an input can only go to one output location, then there are at least 1,000,000 possible connection states or routes for each neuron since each of the 1000 inputs can go to any one of 1000 outputs.

Yeah, that is 1 million possible connections per nueron and as far as I know there is no requirement that the connections have to be one to one and that a neuron can’t receive multiple inputs at one time.

This number or concept is mind boggling enough, but growth of this system is exponential, not linear. When I add a second neuron to the calculation, I do not add just a million connections; I might add 3 million possible connections/routs.

To see how this is possible, lets make some assumptions: 1. the second neuron is added so that it receives input from the first neuron and 2 it makes only one connection with the first neuron. Now one of the 1 thousand connections in my first neuron has 1,000 additional outputs final output destinations that it did not have before we added the second neuron. Now each incoming input has 999 out put destinations on the original neuron and 1000 on the new neuron. So now we have 1,999,000 possible connections counting only the first neuron’s possibilties.

But the second neuron has up to 999 free inputs other inputs of its own going to its 1000 outputs. (Remember that travel is only one way.) That provides another 999,000 result combinations within the second neuron and a new total of 2,998,000 possible connections.

If add a third neuron to the first neuron as an input (rather than as an output like number 2) I add 1000 new input locations that can go to 999 outputs on this first neuron, 999 outputs on the middle neuron, and 1000 outputs on the last neuron in the series. 999 free inputs on the middle neuron can travel to 999 outputs on that neuron and 1000 outputs on the last neuron in the series. The free 999 inputs on the last neuron on our series has 1000 possible free outputs. The total number of outputs is now 5,994,001.

Consequently, we can see that the growth rate as we add each neuron is not simple addition. We started with 1,000,000 possible connections/routes and after adding 2 neurons we have 5,994,001 possibilities.

I have no idea how to calculate the final result because eventually some of the routes have to come back to one of the inputs on our starting neurons. However, with even a few hundred thousand possible neurons, the storage capacity is well beyond my comprehension.

Now just to make it even more stunning, lets look at the possibility that each input on a neuron can choose to go to 0, 1, or 2 final destinations on that neuron (You know, like the office manager that has sometimes has to communicate the client’s requested changes to 1 employee and sometimes to 2 employees and sometimes to no one) …

    Please come back when we have finished our re-branding and have republished this site