Representing Representation – Mary’s Room

April 6, 2007

Some comments on Pete Mandik’s paper The Neurophilosophy of Subjectivity. Here Peter struggles with some of the main issues I’m interested in albeit more professionally and coherently. An unjust summary of points made in the paper: Phenomenal Raw Feels don’t exist as mere sensory input devoid of higher level mental processes. If this is true, then would it be possible to replicate an “experience” purely at a higher level? Pete answers in the affirmative, at least to an extent. Mary, by her study, could have more what-it’s-like knowledge than a control subject who isn’t a brilliant color scientist.

How higher level representation might represent “phenomenal” representation? Pete remarks,

If a picture is indeed worth a thousand words, then isn’t it cheating to say that whatever a picture P represents, the same thing can be represented by the description “whatever P represents”? Even if the “whatever P represents” move is somehow disqualified, the following move remains open: just add words to the description. Why couldn’t a sufficiently long description represent all of the same things without itself being a picture?

The context of that comment is a discussion he has of Dennett’s Jello box. Tear a Jello box in half, and the tear line is infinitely complex, each side becoming the detector for the other. So it is with our perception of red. Red is what it is because one side of the Jello box is a set of physical properties of the environment, and another side is our sensory input and brains which evolved in an environmental context. I think it’s within Dennett’s observation that if we could perfectly describe one side of the tear, we’d be able to anticipate the other. That’s one way in which higher level representation might represent the knowledge of redness. But is the perfect memory, or imagined concept, exactly the same as staring at a red dot – can it duplicate the Raw Feel?

Mary’s room frames the problem of qualia in terms of knowledge. And solving the problem in the above way has Mary knowing red the same way the rest of us know red while not presently looking at something red. It’s tempting to raise the bar and demand Mary induce the experience of red without looking at something red. Or to “know” red the same way we know red while staring at something red.

Following up on Pete’s comment in this way, how could words, some other way of representation, represent the same thing as whatever is going on while staring at the red dot?

Could, as Pete suggests, words stand in for whatever is going on to create the picture? And in real time, in order to mimick the Raw Feel of looking at red?

The answer I try to convince myself is true, is yes, to both. My strategy is to grapple with the kind of representation going on in a Raw Feel, the kind of hardware necessary to produce the Raw Feel, and ultimately, question the project of representation altogether by confusing the symbol with what’s being represented. Some of this is groping in the dark I admit as I’m not 100% certain of what I’m trying to say.

1) What we convince ourselves we see isn’t necessarily what we’re really seeing. As Dennett argues, we don’t really see every detail in a complex rug pattern. As I understand, he argues that even a Raw Feel, to stare at a field of wheat, doesn’t put one million little wheat grains in our head .GIF style, but rather our Raw Feel works like a .JPEG. This is a consequence of the experience happening between high and low level brain functions. At the high level, all kinds of cultural, language, and other things are working to shape the experience. So the complex picture that words can’t describe, is already in part, a kind of word, or language. Like in a .JPEG, let some symbol or number stand for a whole lot of wheat shafts. Let something else help shape differing shades of brown throughout the picture and the thousands of edges that aren’t really continously and individual “before our minds eye”. In this way, by lowering the bar of what constitutes a picture, we can see more easily how words might be able to access the high level functions involved during a Feel.

2)Hardware. It’s probably ridiculous to believe that by pure mental manipulations, we can give ourselves the same thing as a stare at a field of wheat. Though the brain devoid of real inputs can come close, for those who have had vivid dreams. And those, like me, who have had intense episodes of sleep paralysis and hag dreams, it’s even easier to imagine internal experience with “experiences” that arn’t pain or sound but God knows what. So again, perhaps this lowers the bar of Raw Feels a little more.

But if we still can’t get concepts to jump high enough, we can call upon hardware substitutions to aid, so long as the hardware is sufficiently different and we rely on it for processing power rather than it’s specific function. For instance, if I lost my sight tommorow, and God zapped me with the ability to echolocate, how different would my experiences be? (assuming others couldn’t see me making the noises!)

When I’ve brought up the Cypher analogy (from the Matrix), I’ve intended to cover points 1) and 2) together. Cypher claims that all he sees is “Blonde” and “Red Dress” while he stares at the code streaming by. The hardware substitution here is just beefing up his normal capacity so it makes sense for him to supersonically speed-read and process. The visual aspect of his experience is unrelated to visually seeing red. The code is the words, and the Matrix experience is the picture. So equiped with superhardware (that I’m adding), he’s trained himself to unconsciously translate fine grained language in real time so efficiently that it’s, by his report, no different than seeing “red”. And no doubt as time goes on, simple substitutions work their way in as placeholders for complex information analogous to the way digital compression works in order to form more effective experience. The same perhaps could be acheived with supermemory.

3)The final point is that if Cypher can see red by reading matrix code, red as a standard to be acheived is tied to his history as a sighted person navigating the world. But what is the true experience of the world? It would seem that Cypher and the later Neo both understand the code perfectly well but translate in opposite directions. Neo just processes code when he effortlessly battles the agents at the end instead of seeing a picture. The point at which they forget which mode they’re in, where working with symbol substitutions or filled in pixels begin to run together is where the computational thesis gains force.