Intuitions
can be deceptive. It is not intuitive that we share a common ancestor with
fruit flies. It is not intuitive that the apparently stable ground below our
feet moves with a velocity of 1666 km/h. Interestingly enough, not
only facts discovered by but also facts about
science can be rather counterintuitive. And often it is history which helps us
to unearth such facts. This is why my research seeks to combine philosophical
with historical methods. My research topics include
Prediction.
We have very strong intuitions that a theory that predicts phenomena
which we
then go to discover, should receive a major boost in our acceptance of
the
theory. Yet, as it turns out, this intuition is not warranted by the
historical
record of science. As S. Brush (inspired by pioneering work by J.
Worrall) has
shown in a number of case studies, the successful explanation of
already known
phenomena by a particular theory was valued much higher in the
scientific
community than the prediction of novel facts by that theory (e.g.
Einstein’s
explanation of Mercury’s perihelion vs. the prediction of light
bending). This can be explained thus: 1. really novel phenomena are
often
hard to control with the devices, which had to be invented for the
discovery of
those phenomena and which themselves are subject to a number of
uncertainties;
2. novel phenomena might be explained equally well (or even better) by
other
(yet unknown) theories than the one that predicted them. On the other
hand,
phenomena that resisted explanation for some time before eventually
being
accommodated by a new theory will count very much in favour of it.
Those, who have taken these findings seriously (by no means the majority
of philosophers!), have tried to save at least some form of the concept of
novel predictions. In particular, Zahar and Worrall have tried to
defend the notion of so-called use-novel predictions, according to which all
those facts count as novel which are not only (i) derivable from a theory T,
but which also (ii) were not used in the construction of T. The latest attempt
to lend support to such a notion (Scerri and Worrall 2001) I criticised in a
recent publication (Schindler 2008). Not only are there well-known conceptual
difficulties with such an account (how can we ever know for sure that certain
facts were or were not used in the construction of a theory by certain
scientists and, perhaps more severely, why should such information be relevant
for the assessment of theories in the scientific community to which such
information is likely to be inaccessible), but I also think that the heuristic
account is hard to justify historically.
Explanation.
Intuitively,
scientific explanations that are not true cannot be scientific
explanations
proper. However, particularly in the context of scientific models, it
has been
appreciated for some time now that scientists deliberately simplify and
distort
real systems, for instance, in order to be able to apply higher level
theory to
the world and to facilitate mathematical treatment. Yet, not many
philosophers would agree that false
postulates about a real system can serve as the basis for explanations
that are
on a par with explanations that rely on postulates that are true.
Especially
causal-mechanistic accounts of explanation explicitly make the contrary
assumption. I believe that this assumption is entirely dispensible.
Scientific discoveries and data
reliability judgments.
Intuitively, we discover the
furniture of the world by conducting experiments and by making
observations.
How else could we learn about the world other than empirically? But how
do we
know that the results produced by our experiments are trustworthy? For
instance
by repeating experiments, by comparing the results gained with one
experimental
technique or instrument with another, by calibrating our instruments
(instrument A produces the same results in a certain spectrum of the
phenomenon
as instrument B, which already has proved its reliability), and so on.
All this is very intuitive. And yet again, a
closer reading of history may raise doubts about the above strategies.
As
N. Rasmussen has shown, in the 1950s and 60s robustness arguments
pointed
to the existence of so-called “mesosomes“, but they later were
nevertheless deemed
to be artefacts of the preparation techniques. Thus the most prominent
criterion for probing the reality of experimental results is not
infallible. This would only prove problematic if there were no other
strategies
for probing experimental results. But of course, A. Franklin has
proposed a whole
list of such criteria that scientists allegedly use in their
day-to-day-practices. Yet, as I shall argue in a forthcoming paper,
Franklin’s
list of “epistemological strategies” is too weak for securing the
reliability
of experimental results.
|