[Date Prev][Date Next] [Chronological] [Thread] [Home]

Re: Underdetermination and Scientific Realism

To reply to this message or start a new topic please email: BUPS-DIS@bups.org



Dear Andy,

Thanks for your comments. Thought provoking stuff.

1. ?The proof of any scientific theory is qualified by Popper?s falsification
principle, by which empirical evidence can only ever falsify a theory, never
conclusively prove it?it is not proven but simply adopted through pragmatic
considerations.?

Popper showed that the truth of a theory could not be deduced from the evidence,
but once we see the virtues of a fallibilist epistemology this does not commit
us to saying that there can be no rational defence of a theory based on
evidence. Certain models of inference to the best explanation allow that we can
reason from evidence to theory, so theories don?t just have to be accepted on
pragmatic grounds.

2. ?UT has nothing to do with considerations such as beauty and elegance -
typical reasons for choosing between competing theories.?

I agree with you that UT doesn?t get a grip here since it only applies to
observational evidence. However, I did not mention aesthetic theoretical
virtues since, as you go on to mention, there is no guarantee that they will
lead you to the truth. The truth might very well be rather inelegant. I chose
to mention other virtues, such as coherence with other established theories.
Here I think we can be confident that they will lead us to more approximately
true theories but I do have another concern. When we say that a theory A
benefits from coherence with B, are we not simply saying that A is supported by
the observational data which supports B? If so then we have failed to introduce
any new criteria of confirmation, and in that case, UT is as threatening as
ever.

I think that salvation comes when we recognise that some, but not all,
observational evidence counts as confirmatory. In particular, observations
which are made following certain rigorous methodological procedures. Scientists
impose constraints on their experiments which control for variability and
spurious correlations, etc.  

3. ?The underdetermination of theory by evidence need not necessarily lead to an
antirealism about reality (or the objects that constitute it). David Papineau
points out that there is nothing inherent in the underdetermination of theory
by evidence to say that all competing theories are equally supported or equally
plausible.?

I think you conflate the ontological and epistemological claims of scientific
realism. It?s true that UT does not challenge the existence of the unobservable
world, but this does not contribute anything to our epistemological
predicament. UT does say that many theories will be equally well supported by
the evidence (although not that all theories will be), but this claim rests on
the assumption that all and only positive instances confirm. Once we recognise
that there are observations which do not confirm, and confirmations which are
not observations then Papineau?s point comes into its own. (I recognise that
we?re broadly in agreement here Andy, I?m just seeking to clarify.)

4. You have put your finger on an interesting way of arguing for UT based on the
fact that any theory can be manipulated so that it predicts any particular
piece of observational data. The question is whether this can be done
plausibly. If you turn a tap on and off you will see water begin and cease to
flow. Now I grant you that any theory can be manipulated so that it predicts
the flow of water that you observe, but are any of these others equally well
supported as the common sense conclusion that it is your turning of the tap
which causes the flow? For example, my theory that the aliens are landing can
be supplemented with the conditional that ?if the aliens are landing then water
will be seen to begin and cease to flow at the precise moment you have
observed.? Hence your observation is a positive instance of my theory. But
surely we will all agree that the observed evidence determines one theory ? the
common sense theory ? as better supported than any other?

Any supplementary conditions have to be independently testable, hence the
argument for UT based on the irrefutability of a package theory fails. It fails
because it does not recognise that there are constraints on which observations
count as confirmatory. In addition to being confirmed by observation, for a
theory to be supported, it must not be ad hoc.

Nick


Browse or search the BUPS-DIS archives, or unsubscribe from the mailing list at: http://www.bups.org/mailinglist.shtml