Thursday, August 13, 2009

What is Knowledge?

I've been delighting in meeting new people on the other side of the world. Open Parachute and MandM are a couple of New Zealand blogs that seem to maintain a surprisingly intelligent level of discourse. Open Parachute is an atheist/non-theist site and MandM is run by a Christian couple.

There's a hot thread happening on Open Parachute right now. It started with a book review but it has turned into a discussion of the nature of knowledge. I'm not expecting that many pearls of wisdom to start flying (if they do, watch out for pigs!), but it's a pleasure to hang out with people who debate the nature of knowledge for the fun of it.

I have nothing to contribute to the debate right now, so I'm retreating to a neutral corner to think about the physical basis of human knowledge. I'm more-or-less committed to the concept that human knowledge has a physical basis. All my ideas have a physical component to them--there's a certain number of neurons in some particular state. I'm not saying that's what knowledge "is," but every idea in every human brain has a physical aspect.

I'm interested in abstracting this physical aspect of human knowledge out of its biological, neurological context. It would seem that a fully-developed-technology could replicate the entire neural network of a living human brain in a medium besides protoplasm. I've long thought that one could represent electronic neural nets as a multi-dimensional phase space. (I use the term "hyperdimensional" to refer to any phase space big enough to include a separate dimension for each of about 100,000,000,000 neurons.) Just to make things interesting, I see no reason to limit my hyperdimensional space to any single human brain. Two heads are better than one--and a phase space of 200,000,000,000 neurons makes just as much sense as a phase space for one hundred billion.

But why stop there? My real interest is in the phase space composed of every human neuron out there. The topology of that space should be filled with fascinating features. What does the word "banana" look like, when you say it to the whole human race? There are so many neurons firing all at once--neurons associated with "sweet," "yellow," "shopping cart," "colonial exploiters," "fruit flies," and a million other connotations. The whole human experience of a "banana," put all together, is different from any actual banana, just as it is different from any individual's understanding of the word--yet each individual has a real understanding of a real fruit.

I can't visualize a phase space with 600,000,000,000,000,000,000 (six hundred quintillion) dimensions, of course, so I just imagine little dots in outer space. Each time somebody interacts with a banana, I light up a little spot in space. As more and more people interact with more and more bananas (banana splits, banana peels, banana boats, etc., etc.) more dots light up. I picture these dots beginning to cluster. Eventually, there should be enough data points that one could say, "This cluster is 'banana' in human conceptual space."

"Banana" is a trivial example, of course. Nobody needs to go through this much work to talk about bananas. The concept of conceptual space gets more interesting when we move on to words like "beauty," "justice," "truth," "love," or "God." Would a hyperdimensional map of these terms produce a meaningful pattern, the way "banana" does--or would it be a chaotic smear with no distinct features?

No comments: