The Literary Life

Life as a Piece of Text

This is not a philosophical argument against physicalism, but it is an argument showing that physicalism can lead you into some very unphysical places. It bears a resemblance to Searles' argument of the Chinese Room.

First let me describe a Turing Machine. This is an idealized computer, made up by Alan Turing, mathematician, pioneer computer theorist, and World War Two hero (for his code-cracking work). It consists of a very simple processor with an indefinitely long tape running through it.

The tape is divided into spaces, containing ones or zeroes – binary code. The processor can do the following things with the tape:

The processor is run by an "action table," a list of states the processor can be in. The entries in the action table's list read things like:

state 17:
if the tape reads 0, write 1 and go to state 85;
if the tape reads 1, shift right and go to state 42

In modern terms, the action table is the computer program and the tape is the working memory.

A Universal Turning machine is a Turing machine with an action table that lets it mimic any other Turing machine. The other machine's action table is encoded on the Universal Turing machine's tape. This is even more like a modern computer, which uses the same memory for data and programs.

A Turing machine, universal or otherwise, is by no means a practical computer. It is enormously cumbersome. But that's not why Turing made it up. It can be described mathematically, and therefore it can be used to do mathematical reasoning about the nature of computing.

It can move data around on its tape. It can do arithmetic and logical operations. Therefore, it is almost universally believed by people who do computing theory that a Turing machine can compute anything that is at all computable.

Now, what is the most popular metaphor for mind in the present day? "The mind is the software of the brain." The brain acts like a digital computer, people say. Therefore, if a mind is the program that a brain runs, that program can also be run on a Turing machine.

Now, think about the subtlety and complexity of a human mind—quite literally the most sophisticated thing conceivable, because it's the thing that's doing the conceiving—and compare it with the Turing machine that's hypothetically able to enact it. At any given moment, all the machine is doing is writing a 1 or a 0, or moving its tape right or left. A washing machine has more subtlety—an old washing machine, with no electronics in it.

So whence cometh all the depth psychology and sophistication of a human-level A.I. running on a Turing machine? It must reside in the data in the action table and the tape. And, if it is a Universal Turing machine, its action table is the same, whether it is running the A.I. or a game of Pong or a cash register. In that case, everything particular to the mind resides on the tape.

This is why I've titled this essay "The Literary Life." The hypothesis is that we have a conscious mind, a living soul, that is nothing but a text string.

I am not going to argue, here, that it's impossible for conscious intelligence to arise from a tape being chewed over by such a blatantly unconscious machine (though I do think it should give any physicalist pause). I am going in a different direction with this.

Consider a different bit stream, a novel coded in ASCII. Print it out. Read it. If it is at all worth your trouble, it exhibits spelling and grammar, imagery, metaphor and simile, characters, plot, and features like humor, horror, pathos, and themes like love, hate, honor, and shame, in dozens of delicate shades.

To get there from the bit stream takes repeated acts of abstraction—from patches of ink on paper to letters, letter groups to words, word groups to sentences, sentences to images, collections of images to characters and events. What there is, physically, is a random-looking array of magnetized patches in the computer, or white paper spattered with black ink. The story, though, floats high above that, on a skyscraper of repeated abstraction.

The Turing A.I. is like the novel.

(Actually, it is much worse, The novel produces its effects on a sapient reader. The tape bearing the Turing A.I. somehow produces its effects on itself. Have you ever used a book to read yourself to sleep? This is a book that reads itself awake. But let that pass for now.)

The Turing A.I. is like the novel because all its significant features—every flicker of sensation, memory, conception, emotion, or volition—are reached by ascending a similar skyscraper of abstraction – more likely one much taller. Physically, there's just the bit stream. Those bits have to be interpreted to become logical operations, which must be related to each other to emulate (say) neurons, which are related to form ganglia, which do ... something or other (of which we have not the slightest idea) to produce those basic moments of awareness that group into ideas and intentions, which form a personality.

The A.I.'s personality is as "imaginary," if you like, as that of a fictional character, no more concrete than a literary style. It is removed by multiple layers of interpretation and abstraction from physical reality.

The catch is that the same is true of us, on the physicalist hypothesis. The Turing A.I. has logical operations and raw bits under its neurons; we have molecules and atoms under ours. That's the only difference.

So physicalism leaves us with a world in which we are not physical at all, and the life we know most immediately and concretely is not "concrete" in the least, but as fictional as Pooh's Hundred Acre Wood.

If this is not a problem for a philosophy that holds that nothing exists but matter, it is at least a very great irony.

Return to Introduction to Essays
Return to Wind Off the Hilltop

Copyright © Earl Wajenberg, 2011