SUO: Re: Re: Re: Question about Example in KR Book
----- Original Message -----
From: "Jay Halcomb" <firstname.lastname@example.org>
Sent: Tuesday, November 25, 2003 06:52
Subject: SUO: Re: Re: Question about Example in KR Book
> Nothing is ever solved to everybody's satisfaction:
This is why I advocate a lattice of theories :)
> > Since you speak of 'summaries ... of observations' and mention
> > Hume, I (like TJ) wonder if you are alluding (at least in part)
> > to Hume's doubts about induction -- the 'justification of
> > induction' problem. I don't know that this has ever been solved
> > to everyone's satisfaction.
> What I am claiming, however, is that Hume was attacking a strawman.
So, there is no problem involved in justifying inductions, then? Are you
perhaps a vindicationalist -- induction is as induction does?
> Scientific method uses induction for forming hypotheses, but that
> is only the beginning. To be accepted as a theory, a hypothesis
> must make testable and tested predictions about the future: i.e.,
> "what will happen if". Induction, by itself, is insufficient for
> realiable knowledge; predictive power is necessary for science.
Well, yes, scientific theories (or pieces of them, or assertions -- one
by one? -- but that's another question) are supposed to be predictive -
testable. Perhaps that's not too controversial.
Don't all inductions 'have predictive power'? Or can be used/tested so? I
it would help if we got mutually clear on what 'induction' refers to.
Another part of your remark above doesn't seem to be quite right as stated:
a hypothesis when formed has not (usually) already been "tested", has it? Or
is that what you really mean? If so, can you explain the process to me?
> To use the car example, every time that you proceed when a traffic
> light is green, you are making a testable prediction about the
> future: cars coming from the cross direction will stop. That
> is not a prediction about physics, but about the training and
> dependability of your fellow human beings. To a large extent,
> people in modern societies have been so well trained that those
> cases when they cause fatal crashes by going through a red light
> are newsworthy events in the local papers.
Your car example is that of an unconscious process, not a rational
reconstruction; while it was Hume who emphasized the role of habit in the
belief-forming processes. His challenge lay in asking if there was anything
more than habit involved in belief formation which could be _rationally_
consciously, explicitly) justified.
> Any hypothesis that has been tested by making successful
> predictions about the future is a sign that it accurately
> reflects some real aspect of the world.
That can't be right as stated. (Or is it supposed to be 'approximately
right'? :) 'Any hypothesis' - even hypotheses that are later strongly
disconfirmed? Refuted? Suppose Bob hypothesizes (because Bob likes cats and
is given to wishful thinking) that some agency kindly to him has turned all
humans but him into cats. Thus, he expects to see cats where he used to see
people. He walks outside and happens to see three cats straight off. Has his
hypothesis been shown in any interesting way to "accurately reflect some
real aspect of the world?" Does it stop doing so once he sees another
Or does it keep doing so anyway? Some 'real aspect' besides the fact that
dogs do exist, and his hypothesis mentioned dogs? Is it on all fours with
'hypothesis' that the world is instead as usual? This is a trivial example,
but there are many important cases similar to this one -- it is commonplace
that hypotheses enjoy some initial success, but are later disconfirmed.
What is an 'unreal' aspect of the world, and when would a hypothesis reflect
that? What would disconfirm a hypothesis?
>Even the traffic light
> hypothesis reflects neural features in the brains of licensed
> drivers -- even though we can't pinpoint exactly which neurons
> cause drivers to stop for a red light.
> That paper you cited is a good summary of the issues:
> Scientific Realism
> But as the author says in the conclusion, scientific realism
> actually depends on a version of metaphysical realism. I
> consider scientific realism to be established beyond any
> reasonable doubt, and I also consider it a sufficient support
> of at least some versions of mataphysical realistm.
Your last clause is interesting. You propose to derive or to support in some
sense, metaphysical theses from science? I'd like to see a sketch of that
> are, of course, many more details to be discussed, but
> they're far beyond the scope of an email note.
> I would, however, like to give my short refutations of some
> challenges to scientific realism that the author of that
> paper mentions:
> 1. Theoretical entities: Many challengers, such as my arch
> nemesis, Ernst Mach and the behaviorists, refuse to accept
> the existence of unobservable entities such as atoms and
> neural features such as drivers' habits of stopping for
> a red light. I would answer that the nature of those things
> might be unknown, but when people are willing to make life
> and death decisions based on them, there must be something
> real underlying them -- even if we don't know exactly what.
Well, people (being what people are) are at times willing to make life and
death decisions based on damn near anything, or on nothing at all
-- even to die for very bad reasons -- so I don't know that I'd put too much
stock in that as a sign of reality. Certainly not a guarantee.
But I think you mean 'sober, reflective' people, so that
narrows the field a little, maybe.
I'm also not too happy with your notion of 'unknown natures' --
the 'something, I know not what, but it's real, damnit, whatever it is'.
I don't think that's quite what nuclear scientists, for example,
are thinking when they're at work. I hope not :)
However, I think that the point of those questioning the reality of physical
entities like atoms, is to ask for some epistemological warrant for
accepting them. And surely that is a serious question, since how are we
supposed to distinguish atoms from, say, quiffles (something I just made up,
but which are the tiny cubes which comprise all matter).
> 2. Kuhn's methodological incommensurability: I admire Kuhn's work
> very much, and I think that his study of the paradigm shifts
> that occur in scientific revolutions is important. But a tested
> theory based on an old paradigm still reflects something real
> for what it predicts. Newtonian mechanics has been replaced by
> Einsteinian mechanics, but Newton's theories are still just as
> effective as they ever were for low-speed events. They still
> reflect something important about reality. Even old theories,
> such as phlogiston, were accurate for many predictions, and
> they also reflected something real, but we now have better
> theories that give better approximations.
I doubt that many contemporary scientists would want to maintain that the
phlogiston theory referred (in any interesting way) to something real, and
still does, to boot. I suspect that most would simply say it was a false
and confused theory, with an inadequate physical and mathematical
foundation. They would say that phlogiston doesn't exist, and never did. Do
you think that there is no conflict at all between the Newtonian and
Aristotelian conceptions of motion? Or between the Galilean and
pre-Copernican views of the planet's motions? Do epicycles refer to
'something important about reality'? Is all science 'approximative'? But
then, approximative to what? Does 'approximative' mean 'probable'? How are
some approximations better than others? Are the propositions of a scientific
theory, approximative? Or true or false? Are any of them true or false?
Yikes! I'm in danger of becoming JA :) But I'm finding it difficult to
understand you, John. So, a little more below.
> 3. Social constructions: Post modernists argue that science is
> "merely" a social construction. I agree. It is a complex of
> signs, and all signs are social constructions. Signs such as
> red lights are social constructions that make reliable predictions
> about the future -- because they reflect something real inside the
> brains of licensed drivers. Any well tested scientific theory is
> a socially constructed complex of signs. And if we are willing
> to stake our lives on its predictions, it passes the life and
> death test of realism.
This doesn't seem like a refutation of social contructivism, but rather an
endorsement. And the author of that Encyclopedia article took 'social
constructivism' to be a challenge to scientific realism, not a support of
it. So there's a tension here...
However, I don't know what it means to say that 'science is a system of
signs'. A scientific theory is communicated using signs or symbols, but I
should say that it is what the signs convey which constitute the scientific
theory: the propositions of a science.
> As I said before, whenever we drive a car, we are making life and
> death decisions based on our predictions about Newtonian mechanics
> and about the behavior of our fellow drivers. Any driver who makes
> such decisions must truly believe that there is reality underlying
> Newtonian mechanics and human behavior. Once you admit that such
> things are real, you are a scientific and metaphysical realist.
I don't think most automobile drivers are very knowledgeable about Newtonian
mechanics or neurons, and I don't think it's a prerequisite for a driver's
license, even for 'nominalists'. Hence, I don't think most drivers (even
'nominalists') are or must be scientific or metaphysical realists.
I think that the more usual general term in these situations is
> Bottom line: Any nominalist philosopher who drives a car is either
> (1) a solipsist, (2) a fool, (3) a hypocrite, (4) somebody who is
> using the word "nominalist" inaccurately, or (5) somebody who has
> not thought through all the issues that really matter.
The usual expression is, "Even philosophers go out by the door and not by
the window." But there still seems to be a very wide gap between your
premises and your invective, John.