Lecture 10 Scientific realism

1
Lecture 10Scientific realism
2
Observable and unobservable

• Some scientific statements talk about (a)
  observable things (stars, rivers, animals,
  test-tubes, thermometers), while others talk
  about (b) unobservable things (protons,
  electromagnetic waves, viruses).
• With a-statements, it is clear how we can know
  whether they are true or false (youve got it, by
  observation!).
• But b-statements seem problematic. How can we
  know anything about unobservable things?
• Some philosophers of empiricist orientation think
  that science should be interpreted in such a way
  that it does not say anything about
  unobservables.
• It only appears to say something about
  unobservables.

3
Operationalism and instrumentalism

• Two ways to get rid of the statements about
  unobservables (1) to show that they are not
  about unobservables, or (2) to show that they are
  not statements.
• OperationalismTranslate statements about
  unobservables into statements that talk only
  about observable things.
• InstrumentalismShow that talking about
  unobservable things should not be taken as making
  claims that have a truth value. Rather, it is
  just a useful instrument for organizing and
  predicting experiences.
• The translation enterprise was abandoned, and the
  same happened with instrumentalism.

4
Scientific realism vs. anti-realism

• Scientific realismScience aims to give us, in
  its theories, a literally true story of what the
  world is like, and acceptance of of a scientific
  theory involves the belief that it is true.
• Constructive empiricism (Van Fraassen) Science
  aims to give us theories which are empirically
  adequate, and acceptance of of a scientific
  theory involves the belief that it is empirically
  adequate.
• True means true as a whole (i.e., about
  observables and unobservables).
• Empirically adequate means true in the segment
  that is about observables).
• What is the goal of science truth or empirical
  adequacy?

5
Inference to the best explanation

• Inference to the best explanation is one of the
  main weapons of scientific realists.
• Suppose we observe an event E, and that E can be
  explained by three different hypotheses H1, H2,
  and H3. Suppose H1 is the best explanation of E.
  Then if we use the inference to the best
  explanation (IBE), we would conclude that H1 is
  probably true.
• Example the presence of a mouse is the best
  explanation of the disappearance of cheese during
  the night and scratching in the wall. Therefore,
  I infer that there is a mouse in my flat.
• Application to science if the truth about
  unobservables (e.g. atoms) is the best
  explanation of why science correctly describes
  observables, then probably science is true even
  about unobservables.

6
Problems with IBE

• What does the best explanation mean?
• First, how do we compare explanations? Second,
  what if we have just an arbitrary and small
  number of possible explanations? Third, what if
  the best explanation we have is not very
  impressive?
• As Peter Lipton said, the rule should be more
  correctly called inference to the best available
  explanation, when the best is good enough.
• Does the best mean that the explanation is (1)
  loveliest (that it explains the phenomenon better
  than other explanations), or that it is (2)
  likeliest (that, under the circumstances, it is
  more probably true than other explanations)?

7
Lovely or likely?

• Neither answer is without problems.
• If it means likely, it sounds trivial. Of
  course, we should infer the truth of an
  explanation that is most probably true!
• In that case, we should just forget about
  comparing explanations, and just apply the
  familiar Bayesian formula p(H/e) p(e/H) x p(H)
  / p(e). The part in red is giving the probability
  of the explanandum given the truth of explanans,
  and is doing all the work for us.
• If best means loveliest, we have to explain
  what it means that one explanation is lovelier
  than others.
• Besides, and more importantly, we have to explain
  why we believe that lovely is connected with
  likely. Why should one think that explaining
  better is in some way linked to probable truth?

8
Van Fraassen against IBE

• Van Fraassen does not criticize IBE in general.
  Rather, he just wants to show that its use in the
  argument for scientific realism begs the
  question.
• Why not reinterpret IBE so that it becomes also
  acceptable to an anti-realist? Infer that the
  best explanation is empirically adequate, rather
  than true.
• It works well with the mouse example if we infer
  that the mouse explanation is empirically
  adequate, this means that a mouse exists in the
  flat (because the mouse is observable).
• But it does not work with electrons if we infer
  that the explanation in terms of electrons is
  empirically adequate it does not mean that
  electrons exist (because electrons are not
  observable).

9
How far does the evidence reach?

• The above picture represents the content of a
  hypothesis H, divided in two parts, T and O. O
  includes all statements about observables, while
  T contains only statements about unobservables.
• Now Van Fraassens argument is this. Our evidence
  comes from a small segment of O and this evidence
  e can be a reason to believe that O is true. But
  if e supports the truth of O why should we think
  that it also supports (TO)?
• In this way we would just expose ourselves to a
  greater risk, for no reason.
• Van Fraassen In epistemology, we do not accept
  the maxim You may as well hang for a sheep as a
  lamb.

10
What is observable?

• Every anti-realist must make a distinction
  between observable and unobservable. But can it
  be made really?
• Is what we see through a telescope observable?
  Microscope? Electronic microscope? Eyeglasses?
  Where is the boundary?
• Van Fraassen accepts that there is no sharp
  boundary. But he says we use many concepts that
  are vague. For him it is enough that there are
  clear cases, easy to classify.
• Another question is about the epistemological
  relevance of the distinction between observable
  and unobservable.
• If it is OK to believe what science tells us
  about small, barely visible particles of sand,
  why not then also about particles that are 10 (or
  20, 100 times smaller)?
• Why would the size matter? Why would it matter
  whether we (humans) can see something (a purely
  accidental fact)?

11
Pessimistic induction

• A popular argument against scientific realism is
  so-called pessimistic induction from the history
  of science.
• Take a currently well supported hypothesis H that
  realists take as being approximately true and
  describing really existing unobservable entities.
• But consider the historical record. Practically
  all past theories were shown to be false, and the
  unobservable entities that they posited were
  abandoned (crystalline spheres, caloric,
  phlogiston, electromagnetic ether).
• Therefore, H will most probably be refuted as
  well. We dont have good reasons to believe that
  its unobservable entities will have a different
  fate.

12
Different theories, different attitudes

• Past theories were not so highly developed and
  tested in so many domains as some theories in our
  current science.
• For example, is it reasonable to expect that some
  day science will tell us that atoms and molecules
  do not exist? How about DNA, chemical elements,
  and proteins?
• At some level, it seems, we do not expect more
  surprises.
• The Thales problem has been solved. We now know
  the constitution of things around us (like
  trees, water, plants, animals, atmosphere,
  earth). This knowledge is stable.
• At deeper levels, things are less certain. We
  believe that quarks exist or that the big bang
  happened, but it is easier here to imagine that
  we are actually wrong.
• Our epistemological insecurity does not coincide
  with the unobservable domain.