Induction and Deduction

[SillySapienne]

Algorithmic versus heuristic thought processes.

[Evan]

Originally Posted by reason

dissonance,

Think in terms of natural selection. What good would a brain be if it were a perfect deduction machine? Perhaps it would have some use, but would have a major disadvantage--all the right premises woud have to be "built in". Imagine that this kind of brain evolves, but some mistakes creep into the system i.e. occassionally the information copying processes make mistakes a create something novel (analogous to genetic mutations). Sometimes these ideas will be flawed, even hazardous, but from time to time they might actually be improvements, good ideas, and worth keeping. Eventually natural selection begins selecting for brains which make these "mistakes", and moreover, for cognitive apparatus which deal with and evaluate them.

Some billion years later, brains are systematically making "mistakes", and calling them inductions, abductions, or more accurately, good guesses.

Looks like you're finally saying a similar thing to what I'm saying.

It doesn't look like you entirely got my point, though.

I don't know if I have the energy to fully explain it without gesture and other non-verbal communication. That's why I'm always frustrated with these threads. Bah.

One last try: those "mistakes" are still reached through deduction. It's our premises that are contradictory to each other; the conclusions are always logically valid, just not always sound.

How our consciousness interprets this process is an entirely different story, not one I'm talking about.

[reason]

Originally Posted by dissonance

One last try: those "mistakes" are still reached through deduction. It's our premises that are contradictory to each other, the conclusions are always logically valid, just not always sound.

The "mistakes" need not arise from any contradiction, for example, inductive inference is invalid, but not contradictory. In any case, I am not sure what you mean by 'those "mistakes" are still reached through deduction'; they are created by a process, but since the content of the conclusion goes beyond the premises, there does not seem to be anything deductive about them (that is, there is extra content which cannot be reduced to anything which came before). Indeed, without some method of generating new ideas (mutations), learning (evolution) could not take place.

[Evan]

Originally Posted by reason

(that is, there is extra content which cannot be reduced to anything which came before)

That "extra content" is not actually extra. It's included in the premises, we just aren't consciously aware of all of the premises. So we call it induction.

My point is that conclusions are ALWAYS reducible to premises, even if we don't consciously know what those premises are.

There is nothing non-deterministic in our brains; how then, could they generate "mutations" or anything like that?

Again, a valid deduction is:
A->B
A
Therefore B

This is deduction (or call it something else if you have problems with the definition) whether or not we are consciously aware of the premises. Say "A->B" is stored unconsciously and "A" is stored consciously. We will (deductively) conclude B, and call it a "guess" or "inductive" (because we do not know that we know "A->B").

[Aimahn]

I think CC had a good point. I find that a lot of the times the stronger my deductive abilities are the more accurate my inductive reasoning skills become. Granted I haven't taken any philosophy classes so pardon me on some of the more technical terms, but it seems the more i'm aware of the "unconscious premises" dissonance talks about the stronger and tighter my inductive reasoning.

I think I've always sort of had questions similar to dissonance especially in regards to the computational deductive processes of computers and how that makes it hard to replicate human deterministic thought because of induction. I think the trouble is we often times valuate the importance or weight of these unconscious premises in differing and hard to trace ways and then form conclusions based on this. I think that the synthesis analogy is like trying to recreate these premises in a way that new conclusions can be drawn. The pool of premises we choose from whether conscious or not is still vitally important, we just choose to deduce based on the weight we place on each one.

pardon me if that wasn't very clear, its still a topic i'm trying to become more familiar with.

[Evan]

Originally Posted by Aimahn

I think CC had a good point. I find that a lot of the times the stronger my deductive abilities are the more accurate my inductive reasoning skills become. Granted I haven't taken any philosophy classes so pardon me on some of the more technical terms, but it seems the more i'm aware of the "unconscious premises" dissonance talks about the stronger and tighter my inductive reasoning.

I think I've always sort of had questions similar to dissonance especially in regards to the computational deductive processes of computers and how that makes it hard to replicate human deterministic thought because of induction. I think the trouble is we often times valuate the importance or weight of these unconscious premises in differing and hard to trace ways and then form conclusions based on this. I think that the synthesis analogy is like trying to recreate these premises in a way that new conclusions can be drawn. The pool of premises we choose from whether conscious or not is still vitally important, we just choose to deduce based on the weight we place on each one.

pardon me if that wasn't very clear, its still a topic i'm trying to become more familiar with.

Furthermore, the question of the"weights" we put on the premises (if you want to put it that way), can also be thought of as deductively chosen as long as other premises lead to the choosing of them.

As in, there is a hierarchical web of premises that pick the ones "lower" on the web.

You can also think of learning in the same way. There are learning algorithms that determine where new concepts are put into the web -- those can also be thought of as a set of premises applied deductively to the new concept.

The way I think about it is, you're genetically programmed with a few basic premises (including learning mechanisms -- premises to be used in application to new concepts, rules of metaphor if you will). Each piece of data you are exposed to is fit into the web of concepts you have so far (a piece of data can even be a new learning algorithm, it's like a Universal Turing Machine). Throughout life, you keep building this concept web until it's ridiculously complex (not that it doesn't start complex, you start with mechanisms for vision and stuff like that that are incredibly intricate).

Anyway, some of the data you will be exposed to is inconsistent with the current structure of the web, so there also have to be learning mechanisms for resolving stuff like that.

But the whole system is entirely deterministic; it runs exactly like a computer program.

Each "new" idea (induction, intuition, guess, whatever you want to call it) is an exact result of a deductive calculation in your brain somewhere. "Incorrect deductions" are not logically invalid, they are only logically unsound. As in, the deductive process MUST go according to laws -- the conclusion must follow from the premises. The premises are the actual problem -- they can be inconsistent or untrue (don't correspond to reality).

Our consciousness is merely an observer of this phenomena -- it makes up a story for what it sees. The story may or not correspond to exactly what's happening on the hardware level. Some of the premises are entirely hidden, in other words. So instead of calling it "deduction" with premises we don't have access to, we come up with a different term to describe it. Intuition, or guess, or whatever.

The consciousness usually assumes that the fact that conclusions are sometimes incorrect means there's something fuzzy about the calculations. There's some imperfection in the brain, or there's something fundamentally different between the brain and any other information processing machine. This is not accurate. The consciousness program just doesn't have enough inputs or computing power to accurately explain other brain phenomena.

Meh.

My point is, there is nothing fundamentally different between brains and computers. They are both machines capable of deduction and deduction only. The labels "induction" "guess" and "intuition" are just our ways of describing deductive processes that we can't fully see.

[Aimahn]

That makes sense to me. The question that's really interesting to me would be how do you account for the benefits of what we call "inductive" or "intuitive" reasoning on a computer level. Basically, there are a lot of creative and ingenious ideas that are borne of that type of thinking and to me it would seem hard to replicate on a computer level. How would we sort of put the barriers( the deductive processes that we can't see) in a computers reasoning so they can also reap the benefits of creative thought and ingenuity.

To go back to CC's point how do you sort of put a value on the ability to synthesize, we all know computers can deduce or analyze with the best of them. It seems to me more valuable and challenging to synthesize information in to something novel than to analyze something to perfection, which is what we would be doing if we could see all the deductive processes that occured.

[Evan]

Originally Posted by Aimahn

That makes sense to me. The question that's really interesting to me would be how do you account for the benefits of what we call "inductive" or "intuitive" reasoning on a computer level. Basically, there are a lot of creative and ingenious ideas that are borne of that type of thinking and to me it would seem hard to replicate on a computer level. How would we sort of put the barriers( the deductive processes that we can't see) in a computers reasoning so they can also reap the benefits of creative thought and ingenuity.

To go back to CC's point how do you sort of put a value on the ability to synthesize, we all know computers can deduce or analyze with the best of them. It seems to me more valuable and challenging to synthesize information in to something novel than to analyze something to perfection, which is what we would be doing if we could see all the deductive processes that occured.

Right, that's the problem. It seems obvious that there is some computational process going on in things like induction or intuition, but it's gotta be insanely complex (tons of lines of code, and an insanely giant and intricately linked data structure).

That's sort of what computational cognitive scientists are trying to do, but we're quite far off. (I'm in the cognitive science major at Cal with some of the top names in the field, and have heard lectures from a bunch of them...)

But yeah, that's the problem right there. All we can do is try some code out and see if it matches human behavior, then modify in the right direction and repeat. Theories in cognitive science make it much easier, but we still don't have very well defined ideas of how all these things relate to each other.

[SillySapienne]

CC Aimahn!!!

[SillySapienne]

Oh, and dissonance, when I am not feeling so under slept and groggy, I most certainly will have a thing or two to say about your theories on the human brain.