Notes on The Computer and the Brain

The nervous system is a computing machine which manages to do its
exceedingly complicated work on a rather low level of precision

John von Neumann, The Computer and the Brain

In 1956, John von Neumann wrote notes for a lecture series on the state of computing and how those computers related to the human brain and nervous system. The lectures themselves never happened because he had a cancer that was too aggressive but his lecture notes were published after he died. They contain a lot of interesting information, and considering that they're only 82 small pages, you should really read those rather than my brief notes here.

The Computer

I shouldn't have been surprised at how relevant the book is. The discussion of “memory-stored control” is engaging and presumably timeless - even though the original design wasn't his, we still call it the von Neumann architecture. Then he discusses maintaining multiple banks of memory with different speeds, which has become no less relevant with our multiple CPU caches > motherboard memory > SSD > network storage. And just when you find yourself wondering what an SSD would seem like to the person writing this 60 years ago, there he is describing the speed advantages of solid state devices.

The most wonderful part of this is to hear him describe working with the historical but futuristic computers. As he tells you what was used for storing and operating on data, you can almost hear the pulse of the machine.

The Brain

Following a brief description of how the human brain might work, he attempts to quantify comparisons between brains and computers. He gives estimates for their sizes, maximum data processing rates and memory capacities.

Von Neumann's own memory and processing speed is legendary - by convention every article about him must contain at least one anecdote of his superhuman cognitive abilities. So when it came to the numbers, I wondered if he was going to provide estimates based on his own ability or everybody else's, which would definitely be more modest. But the only reference to the author's own brain during that calculation is a single-word hesitation. Having arrived at a maximum input data flow rate (i.e. what you can take in with your senses) of $14 \times 10^{10}$ bits per second von Neumann believes that

an estimate for the entirety of a normal human lifetime can be made.
Putting the latter equal to, say, 60 years […] the total required memory capacity would turn out to be $2.8 \times 10^{20}$ bits.

He knew that he was terminally ill as he came to that estimate. The one-word hesitation before assigning a span to a normal human lifetime came months before he himself died aged 53.

He then hints at a statistical mechanics of the mind, based on activation energies and delays of firing neurons. And there the book came to an abrupt end.

Thomas Bewick woodcut of children on gravestones.

Turing Switch

The famous 1937 paper “On Computability” by Alan Turing is cited a couple of times. Strangely the original printing of The Computer and the Brain mentions “English logician R. Turing showed in 1927 that it is possible to develop code instruction systems for a computing machine which cause it to behave as if it were another, specified, computing machine.

Those errors are obviously not von Neumann's and were presumably introduced when his notes were transcribed. Later printings fix the mistakes. It may seem odd that a posthumously published book by the greatest thinker of his time was published uncorrected. But today we can check Turing's dates on Wikipedia, then read - or at least see - the original papers, before verifying that no-one called R. Turing really did publish anything important in 1927.

Read the book and be grateful that Turing and von Neumann's work contributed to a world where we can get information and correct documents in the blink of an imperfect eye.