Over the past couple of days I’ve listened to extensive arguments between Manapatra and Freelance concerning whether it is like something to be a bat, and whether Mary the psychologist can figure it out from studying the brain anatomy and connection diagrams of the bat’s nervous system. There was a lot of hand-waving going on. There was some quoting of Dennet (which always works to destroy the logic of any question). There were some sparks. Not surprisingly, nobody’s opinions were changed by the discussion, and nobody was able to prove their point to the other party’s satisfaction. I think that’s good. If we could change each others’ opinions just by making some quick and extemporaneous arguments, we would probably all be victims of TV advertisements, too. And that’s clearly not the case.
But Dennett is wrong on this one, and Nagel has a good point. The biggest con game in the whole philo of mind business is the assumption that minds are one-to-one encodings of brains (mind states = brain states) and nothing could be further from the truth.
For one thing, the relation of mind to brain is similar to that of program to computer. Programs involve a level of semantics that can’t be encoded in machine language, and that is stripped out and discarded by the compilation process. In other words, the meanings of symbols are never the symbols themselves, and if the brain is all there is to human cognition, then it is, and must be, inherently meaningless. Which is ridiculous.
The question of where to put the meanings of symbols is always difficult to resolve. For years, people have been putting them in brains, saying that it’s all just in our heads. But when we begin a study of heads and brains, we get symbols that don’t speak for themselves: we need someone to interpret them, and barring asking somebody what theyre thinking, the only alternative is to ask the brain scientist what his subject is thinking. Except that that just interposes an interpreter, saying the meaning of somebody’s brain lies in someone else’s. That’s ridiculous, too.
The only chance for a non-mystical resolution of all this is to suppose the meanings of brains lie in the society where that brain lives, just as language is relative to the people who speak it. We can then find definitions for symbols, expanations for peoples’ behavior, and contexts to contain such abstract objects as careers and scientific theories, by looking about at the relations between people that define these things and create their frame of reference. Take a brain out of its society and you have nothing but signals.
It isn’t my job, at this particular juncture, to explain how societies are any more material than the meanings that can never appear explicitly in brains, but then, I never claimed materialism was the sufficient answer to all questions, either.
In conclusion, the reason why Mary can’t figure out what it’s like to be a bat, is because the answer isn’t in the bat’s brain.
April 29, 2008 at 1:39 am
The biggest con game in the whole philo of mind business is the assumption that minds are one-to-one encodings of brains (mind states = brain states).
Rightly stated. It’s assuming a lot of the things about the way our brains are hardwired that we simply do not know yet. Like I mentioned in chat, it seems to be a reductive use of logic to me. “What is x cannot be non-x at the same time”, bla bla.
May 1, 2008 at 12:02 am
You get positively Lacanian at the end there.
{];)
May 1, 2008 at 12:03 am
So.. the smiley with a hat doesn’t work properly here I see
May 1, 2008 at 12:50 am
Kevin, nope, apparently not. It looks like the little guy is winking.
I don’t know much about Lacan. I’ve heard his views were along the lines of individuals being social constructs. It’s a bit objectified, but I can see how we would arrive at such similar theories.
My basic problem is that I want to describe some sort of Noos, some universal space of ideas, almost Platonic, and yet that’s too vague (controversial it may be, but worse, it’s just very vague), and so, at the moment, I’m looking at how we might the definitions of our meanings in the space between people.
Some people will claim there are only heads, and nothing in the spaces between. Very materialistic, but unsatisfying. A theory of isolated heads only makes the problem deeper.
At any rate, I think it’s kind of funny you see a parallel with Lacan, because I got there completely independently.
May 3, 2008 at 1:52 pm
You said: “Programs involve a level of semantics that can’t be encoded in machine language, and that is stripped out and discarded by the compilation process.”
I don’t see how this is so at the most basic level. The code the programmer writes specifies how the program behaves. The code is compiled, at which point the source may be completely lost. But the compiled machine code is still faithful to the programmer’s intentions (if the compiler is any good), and the program will run just as the programmer specified. Moreover – it is often possible to decompile the binary into the source – this has been done in Python and .NET.
On a higher level, any programming language can be no better than a Turing Machine, and any source file can be the input of such a machine (in fact it is). Moreover – our brains are also apparently Turing-compatible. So any computation we carry out can in principle be carried out by a Turing Machine or its equivalent. (Even less so, since our memory is bounded.)
So, what is lost here?
As for meaning. I think that meaning of symbols comes from the things they symbolize, and from the things you can do with those outside objects. The word “snow” signifies actual snow, that slowly falls out of the sky and covers the ground. And so on. I don’t see how society is inherently necessary – or rather, it is necessary, but only in the context of our evolution. I can imagine a brain which would explore its own surroundings and form meaningful statements about them without recourse to other brains. Or minds. Take your pick
May 3, 2008 at 2:54 pm
Just read you were a computer programmer, made me feel very silly and embarrassed going into the workings of a computer and Turing Machines. My apologies! Still I don’t understand – what is being lost after compilation?
May 4, 2008 at 11:40 am
Vladimir, no need to feel silly. It’s a good question. Of course, we (the programmer community) have been dealing with decompilers of machine code for a long time, almost since the birth of compilers. Not to speak of reading machine code directly; I’ve been in many situations over the course of my career where I had to read machine code. So it’s not as if it isn’t possible. Machine code does say something, and it is a program.
The problem, though, is most apparent when trying to backwards engineer something. Now, it should be clear that compilation does discard information, otherwise the compilation process would not be necessary, and there would be no such thing as optimization. The source code and machine code are not equivalent.
The most important thing lost, though, is the intent of the program. This is important, because the programmer’s intentions are the only thing that allow us to debug a program. The intentions tell us what the program is supposed to do. The machine code only tells us what it does do. This is apparent, for instance, in the ambiguity of meaning of memory cells, such that the source code may label a variable as a “rate of change”, but at the machine code level, this label is lost. The computer doesn’t need to know what the variable represents, in order to add 5.0 to it, so we discard that in compilation.
This is important at the symbolic level also, when analyzing a piece of text, or someone’s remark. We know the words. They stand on the page, or are easy to hear. The question is, what is their intent? What does the speaker mean by them? And this is always what happens when we form a representation of type S->M for some symbol S. The token stands for the meaning. The token is explicit. The meaning is not.
And of course this implies to brains, too. It’s why we would need some scientist to say what a brain scan means: the brain scan doesn’t say it, itself. It’s just a token.
May 7, 2008 at 7:13 pm
I’m reminded of a program I used to cheat in games. I gave it a quantity – such as remaining ammo – and it looked it up in the program’s memory. Then I changed the quantity and fed it to the program again. Fairly soon it would find the place in memory which changed uniquely after my modifications, which I could then directly alter – and see the result on the screen, and in the way I could interact with the game. I have found the remaining ammo variable without knowing a thing about the source code, without it being available at all.
A person reverse-engineering a machine-code executable might take the approach of choosing one variable and tracking down its changes and effects. This quickly becomes difficult as the program grows, but this is a problem of practice, not theory. It is the same with translation – not many can read Plato in the original, but does that affect the meaning of his writings?
When you learn a new word, you do not just learn its grammatical usage – it is not enough for the token to be explicit. For proper learning, you must learn what the word signifies, and how that which it signifies behaves. When you speak the word to those in the know, you evoke in them reasoning about the-thing-which-the-word-signifies, otherwise you wouldn’t be able to communicate.
The brain scan itself is just a token, but neurologists can work with these tokens just as easily as we can work with words such as “apple” and “cat”. They mean something exactly because they signify something “out there”, because they imply causes, effects, actions. Isn’t it what it means for any word to *mean* anything at all?
Likewise, the token “rate of change” is meaningful to humans only because it evokes in the brain relations, causes and effects within the program and outside it. And this is exactly the information the programmer has to code into a computer!
So I would argue that the intent of the program is not lost when the source code is lost, just as the intent of the author is not lost with a good translation. Studying and altering its behavior becomes more difficult, very much so for a human. But the “true” meaning of the program is in its effects, not in its tokens.