December 4, 2006

Aristotle believed that what made objects what they were was something called essence. So a table has table essence. It also has four legs and a flat surface, but those are incidental properties. What makes it a table is its essence, not its properties. What makes water water is that it has water essence; that it’s made up of H2O molecules is an incidental property.

This is absurd. It implies that someone could show you an object with four legs and a flat surface and you’d say, “Oh, that’s a table,” and the person could reply, “Oh no, it just has the properties of a table, but it doesn’t have table essence, so it’s not a table.” An oracle could show you a liquid and you could analyze it down to its molecules to find that it’s water, and the oracle might insist that it’s not because it doesn’t have water essence.

This way of thinking dominated philosophy for over a thousand years. Eventually philosophers realized (painfully) that what makes an object is its attributes. They’re not incidental, they’re by definition what defines an object. If every observable property of an object implies that it’s water or a table, then it is. There’s no internal unobservable “essence”.

We can all sit around here and laugh at stupid Aristotle and his followers for believing such an inane idea, but the concept of essence hasn’t completely died away. We still carry some of it with us. I’ll give you two examples.

Ninety years ago Einstein considered the following two scenarios. In the first, you’re sitting on earth in a room. An object falls because of the force of gravity—the mutual attraction between that object and the earth. When you let go of an object, it goes downwards. In the second, you’re in the middle of space, in the same room, and someone is pulling upward on the room at a steady acceleration. In fact the acceleration matches what we see on earth due to gravity. An object falls because it keeps going at the same speed it was going when you let go of it, but the room accelerates upward until the floor catches up with the object. When you let go of an object, the room moves upward to meet it.

Einstein found that there were no observable differences (gravitationally) between the two cases. There was no gravitation experiment you could run in either room to determine which room you were in. Most people would have declared that, fine, there are no observable differences, but the two rooms are clearly different, so they’re different situations. Each situation has an essence, which although unobservable, remains and distinguishes the two.

At the heart of the general theory of relativity is Einstein’s decision to get rid of this essence. The two situations are the same. The second scenario (in space) is well understood (while gravitation is not), so he picked that as the “true” situation and modified the other: on earth, we are perpetually going up. When you let go of an object, it keeps going up at the same speed and the room accelerates upwards to meet it. Literally.

The literal interpretation is important. If you follow it, you’ll discover all sorts of things that you wouldn’t have in Newton’s force-based interpretation. For example, light is attracted by gravity, gravity is attracted by gravity, and spinning objects warp gravity around them. (That last one is being tested by a satellite right now. We should get results in maybe six months.)

That was the easy example. Physicists have had ninety years to get used to the space-time view of gravitation, and now it’s pretty much accepted. Here’s a harder case, mostly because it’s fairly new. It’s also harder because it raises questions that are hard for us, as humans, to answer with honesty.

You’re sitting at a computer terminal with an instant-messenger window open to someone else. You type, “I think you’re a great person.” They reply, “Why thanks, that makes me feel happy.” You then type, “Too bad you’re fat.” They reply, “What was that for? You hurt my feelings. I’m crying.” You would conclude that the other person has emotions.

Now replace the other person with a program that can simulate this conversation. The program parses English sentences, keeps track of some internal state of emotion, and replies with sentences that reflect this emotional state. I wrote such a program when I was eleven. I showed it to my father, claiming, “Look, my computer has emotions!” He pointed out that the computer doesn’t have emotions, it merely says that it has emotions. I immediately realized he was right; my computer’s emotions were fake and algorithmic.

But now we’re back to essence. The difference between the two cases (in the instant-messenger example) is unobservable. We can in theory write a program that will react emotionally just like a person. If you claim that these two cases are different despite being observationally indistinguishable, then you’re revealing (as most people would) this last shred of belief in the concept of essence. It was painful to let go of it a thousand years ago with tables and water, ninety years ago with gravity, and now with computer simulation of the human brain.

You’ll come up with all sorts of objections. “Emotions are chemical.” That’s an implementation detail, actually. And if you don’t think it is, then I’ll just simulate the chemicals in the computer. “The instant-messenger case is artificial and sterile.” Okay, I’ll build a mannequin that blushes and smiles and squirts saline out of its eyes. “The person feels the emotions, but the computer doesn’t.” How do you know? With the person or the computer. All you know is that your mind makes you feel as though you have emotions. But you don’t know this about anyone else. All you have is their output, just like the computer’s.

Your argument, ultimately, will dissolve to, “But only humans (or animals) can feel emotions, by definition.” And here you lose, as does everyone who resorts to winning by definition.

There is no difference. Either both the human and the computer have emotions, or neither does. There’s no essence. One hundred years after computers are able to copy our emotions, people will think that this is evident. And they will mock us like we mock Aristotle and his tables.