unsubscribe
Arthur Jackson (ajackson@SSA.CO.SANTA-CLARA.CA.US)
Thu, 19 Nov 1998 07:52:55 -0800
Bruce Edmonds wrote:
>
> Thanks for your reply Don, this is much clearer than other stuff I have
> read.
>
> > Don Mikulecky replies:
> > If I can summarize Rosen's "Fundamentals of Measurement" briefly ...
> > ... Reductionist methods seem to
> > eliminate the problem by reducing systems to simple physical models which
> > seem totally compatable with measurement. In a nutshell, what is lost here
> > is the identification of functional entities which have their identity
> > defined by the context provided by the system. reducing the system for
> > measurement eliminates that context and thus makes the system seem simple.
>
> Right, if I may paraphrase: the data gained by measurement is iteself a
> model of the phenomena under study (the 'data model'), the measuring
> apparatus itself forms the encoding relation - an apparatus we have
> designed. The modelling language is (typically) numerical (or an other
> formal language). Such maths necessarily is context-free (thats why we
> use it - so we can do unlimited inference in it, this is its limitation
> and its power). It is this loss of context which causes the loss of
> meaning. Thus the formal modelling we do is already in the realm of the
> artificial (starting from the data model), but we actually do a lot of
> implicit modelling with our measuring apparatus.
>
> There is a slight wrinkle here, for it seems to me that the data is
> reinterpreted in two indirect ways: (1) via a (meaning-laden and
> context-dependent) analogical model and (2) via the predictions of
> future measurements via a phenomenological model of the data (which is
> ususally in some formal inferential system).
>
> The analogical models are important becuase they give meaning to the
> whole enterprise and help use direct our scientific actions (disparingly
> called 'heuristics'). This is part of what is missing from the
> formalism.
>
> > ... Clearly, one does not just "measure".
> > Measurements are abstractions DESIGNED to fit certain formal descriptions in
> > the modeling relation. Thus what we observe is NECESSARILY dictated by our
> > choice of formal system. More subtile is the fact that the encoding and
> > decoding are independent mental activities which are in no way formulated by
> > the formalism.
>
> [I think 'dictated' is a bit over-strong here - I do not think we can
> *totally* dictate what we observe, 'inevitabley constrained and shaped
> by' might be more accurate In my terms the phenomena are
> approximated&fitted to an entity (a datum) in a modelling language (via
> our measuring) so both phenomena and modelling language (and the
> modelling relations implicit in the measurment process) together dictate
> the data. The way these all interact is complicated, no one part
> entirely dominates.]
>
> I think there are some parts of the philosophy of science which would
> interest you (you have probably come accross the second already):
>
> Bogen, J and Woodward, J (1988). Saving the Phenomena. The philosophical
> review, 48:303-352. - Makes a clear distinction between data and
> phenomena, the data are used for prediction the phenomena in
> explanation.
>
> Cartwright, N (1983). How the Laws of Physics Lie. Oxford: Oxford
> Univerity Press. - Shows how, in practice, physicists use a variety of
> types of models including: phenomenological laws (which are true of the
> data but don't explain much) and theoretical laws (which are used in
> explanation but don't predict).
>
> Anyway, to conclude, nobody thinks they *themselves* have a problem:
> most scientists (typically positivist & realist) don't think about it
> much and just do the stuff, philosophically minded reductionists will
> discount the meaning-laden stuff as mere heuristics which will be
> eventually replaced by 'proper' science, the holists don't see the fact
> we only have models of reality a problem - this is how things are.
>
> The 'measurement problem' is something the holists think the
> reductionists *should* have (but the reductionists don't actually
> care). Is that it?
>
> Regards.
>
> PS. Which is the first/most basic Rosen ref. that deals with the
> modelling relation?
>
> --------------------------------------------------
> Bruce Edmonds,
> Centre for Policy Modelling,
> Manchester Metropolitan University, Aytoun Bldg.,
> Aytoun St., Manchester, M1 3GH. UK.
> Tel: +44 161 247 6479 Fax: +44 161 247 6802
> http://bruce.edmonds.name