From: Miranda Sharpe
Date: 02 Ap 2006
Time: 21:10:38 -0400
Remote Name: 157.252.10.115
Dear Dr. Lloyd,
You are working late. I think you’ve got consciousness on the map for a number of reasons. The first reel of any theory of consciousness has to explain why brains have it and fast laptops do not. I’m just figuring this part out for myself. I don’t know much about neurons, but I don’t think there’s anything special about a single neuron as far as consciousness goes. It’s dumb as a jellyfish, right? — just a squishy transistor. So neurons don’t really matter in particular, but their networks do. This idea of distributed representation. That’s big. Once many many neurons are on the same team, they can play in many many many (many squared?) formations. You need that, because conscious states of mind are so complex, every minute’s a movie, every second a slice of life.
Now you’ve taken some of those many dimensions and some of many possible perceptual states and worked out the plan of the neighborhood. Because you started with patterns in the brain, doing nothing to sort or label them before they squeezed through your multidimensional mapifier process, your map is based on the brain’s innards and nothing but. That’s the right warm-up for a theory of consciousness too. The brain is stuck inside the head after all. It can’t step outside to compare its current neural state with outer reality. All it can do is measure itself against other states of itself. So, for a sensation of touch to be felt as touch can only mean that it is like other sensations of touch, and unlike sights and sounds. That “likeness” of course is not a single scale, but measured along who knows how many dimensions. Your map shows these likeness/difference relations in three condensed dimensions.
If consciousness really is a brain process, then states of the brain will show the distinctions among states of consciousness.
So you’ve got the right approach. And, as I cruise the space, it works out well. At first I thought the senses clumped up, but it’s more complicated than that. I do see seeing, hearing, and touching in their own territories, more or less. But it isn’t the sense that matters, it’s what you use each sense for. For example, several experiments involve rhythm and repetition. Some you watch, some you hear, some you see. These hang out at the bottom of the cube. Others involve language, and hover towards the top and back, even though some are reading, and some are hearing. As you float from left to right, the focus shifts from passive perception to active or thoughtful response. And so forth.
So I asked, why is that? The answer comes from phenomenology. One of the thrills of phenomenology is the point that two experiences can come in through the same sense, but be radically different. Even good old Aristotle had this one right, when he pointed out that we hate to see dead bodies in reality, but actually get excited about the prospect in a good movie (well, for him, stage play). (I know what he means.) The exact same visual sensations can have radically different meanings. At the same time, two experiences can come in from different senses and have similar, overlapping phenomenology – reading words and hearing them, for example.
So you wouldn’t want them in a clump, suggesting that everything visual (for example) was more or less the same. But you wouldn’t want them scattered all over at random either — then it would mean nothing to us that they were visual — there wouldn’t be anything in common to all our moments of seeing. But a region in brainspace allows for some sort of common ground among types of experiences. They can be close, not quite the same, but not absolutely different either.
There are always differences between what we see and how we see it. And what we do next. But these can never be detached from each other. They are all part of the consciousness of any object or situation.
As I reread this, I see that I’ve loaded a lot of phenomenology on you.
The point is, we can see ourselves as comets moving through your VR world. The map you’ve made is a theory of consciousness — a theory you can hang on your wall, like the periodic table. Or at least it's on the way. It bridges the two realms of mind and brain in a point for point mapping of one to the other. The map of the brain is the map of the world of experience (the only world there is). Oz at last. (You can tell I’m a graduate student, no? :) )
I don’t know quite how to get this down yet, but it seems to me, after a really long day, that what the mind is doing is multidimensional scaling. Think about it. A vast shifting pixel pretzel of sensory inputs, news from all over. That’s a lot of dimensions. Zillions. Just to get along in the world you’d have to shrink the space.
The beauty of this beast of multidimensional scaling is that it’s a shell game. You start with one world and a pigpile of variable dimensions. Fold, massage, pound, and presto: a map of a new world. But in a critical way, it’s like the old world. That’s because all the objects in the old world have counterparts in the new world, and counterparts are related to each other in the same way in both worlds. What was alike in the old world remains alike in the new. MDS preserves the structure of the world, in its now. It ignores origins and consequences, describing each object only as it is, and through each new scaling hangs onto the structure of the world, the skeleton of Being.
Each state of consciousness is what it is. Literally, each one is a pattern of agitated neurons, a point in hyper space — like you say in your paper. The pattern is all there is. What it is like to be that pattern can’t be tagged by anything from outside, because for consciousness nothing is outside. And since the pattern is nothing but antsy neurons there is nothing that is intrinsic to it that would lead you to interpret it one way over another. What there is, is just other patterns, and all the patterns taken together fall in a structure, a map, the map of the world. And like any map, there are blank spaces; there must be, or the space could not be shrunk. We interpret each pattern only by exploring its similarities and dissimilarities to others. What it is like to be in a certain state of consciousness equals where that state falls on the map equals where that state sits in the space that is the world.
Multidimensional mapmaking is a process that morphs states of consciousness into new states, new maps with fewer dimensions. But it is true to the basic structure of the world, cinema verité, because the structure stays nearly the same through every morph. The structure is the grid on which the new properties hang, the properties that are not given by eyes and ears but are put there by the mind. These new properties can be anything: the specs for a bodily response; the prediction of the next second or the next decade; a picture; a thousand words. Reality itself. Any of these could a subset of the shrunken map. (Shrink too much and you get a stress fracture. That is, you get it wrong. A pie in the face.)
Not only that, it seems to me that a recurrent network really could do big, fast multidimensional scaling. It can compare the present in the past, locating you on the map by telling you where you’ve been. The brain could do it.
There’s more to be said. Hope to hear from you,
Miranda S.