Induction vs. deduction: Ways of knowing

Stephen Barnard (steve@megafauna.com)
Tue, 20 Aug 1996 17:13:50 -0800

It's about time that this thread got a new name.

Marty G. Price wrote:
>
> On Tue, 20 Aug 1996, Stephen Barnard wrote:
>
> >
> > That's right, induction and deduction aren't even remotely similar. Anyone
> > who knows anything at all about logic knows this. Mind you, I'm not talking
> > about "mathematical induction", which is a formal type of mathematical
> > argument. I'm talking about inductive reasoning, and Godel's Theorem has
> > nothing to say about inductive reasoning.
> >
>
> [snipped explanation of the distinction]
>
> Stephen, you're working in the details & missing the great big historical
> development. Induction & deduction developed as mirror images: induction
> working from "instances" to "categories" and deduction from "categories"
> to "instances." In terms of historical development (and likely
> development as human patterns of thinking) they are deeply related.
>

I suppose there must be an intimate historical relationship between deduction
and induction, but I'm not very familiar with it. I'll take your word for it,
Gale. Didn't Socarates establish the deductive method with the syllogism?

My point is that induction and deduction are completely different, and in some
sense opposite, ways of acquiring knowledge. Deduction is the only path
toward rigorous mathematical results (aside from unreliable intuition via
induction), but inductive reasoning is (IMHO) by far the more important way
that we come to discover truth. That's not "Truth" with a capital T, but the
always suspect knowledge that we take as truth until we discover something
better.

There are varieties of inductive reasoning, which I'll characterize as "naive"
induction and "principled" induction. Naive induction says something like,
"We did this dance and it rained, so we'll do this dance whenever we want to
make it rain." Naive induction can lead to useful knowledge, like the
knowledge that various plants can cure certain ills. Principled induction
demands much more. When some regularity in cause and effect is seen, and a
theory based on that observation is constructed, then principled induction
demands that the theory be predictive. If the rain doesn't come at more than
chance frequencies then the dance is considered ineffective. The plants that
supposedly cure ills are subjected to double-blind tests to eliminate bias and
the placebo effect.

There is much, much more to this. For example, principled induction demands
that the theory appeal to deeper knowledge. Why "should" the dance cause
rain? If this thread continues I'll say more about what I think, if anyone
cares.

Steve Barnard