Docstoc

Resisting the Pessimistic Induction

Document Sample
Resisting the Pessimistic Induction Powered By Docstoc
					Resisting the Pessimistic
        Induction




      Kareem Khalifa
  Department of Philosophy
     Middlebury College
                            Overview
   No Miracles versus Pessimistic Induction
   Carving up Laudan’s Hit List
       Genuine versus spurious empirical success
   Psillos’s Divide et Impera
       Explaining genuinely empirically successful, but non-
        referring theories
            Theoretical constituents
            Essential contributions
   Objections to Psillos
         No Miracles Argument
   Scientific theories are empirically
    successful.
   The approximate truth of these theories
    best explains their empirical success.
   So scientific theories are approximately
    true.
   The Pessimistic Induction attacks the 2nd
    premise.
             Pessimistic Induction
   If a theory’s approximate truth best explains its
    empirical success, then there should be few/no
    empirically successful but false theories.
   But there are! Recall the Hit List!
       …or see the next slide…
   So a theory’s approximate truth need not be the
    best explanation of its empirical success.
                   The Hit List
   The crystalline spheres of ancient and medieval
    astronomy
   The humoral theory of medicine
   The effluvial theory of static electricity
   ‘Catastrophist’ geology, including the Noah’s Flood
   The phlogiston theory of chemistry
   The caloric theory of heat
   The vibratory theory of heat
   The vital force theories of physiology
   The electromagnetic aether
   The optical aether
   The theory of circular inertia
   Theories of spontaneous generation
        Important Details in the
      Pessimistic Induction (S307)
A. Current successful theories are approximately
   true.
B. If current successful theories are truthlike,
   then past theories cannot be.
     Ex. If the oxygen theory is truthlike, then phlogiston
      is not.
C. These characteristically false theories were,
   nonetheless, empirically successful.
 Thus, a theory’s approximate truth need not
   be the best explanation of its empirical
   success.
                          I use the DIVIDE et IMPERA
                          MOVE to get rid of the rest by
Psillos vs. the Pessimistic Induction
                          showing how their genuine
                          empirical success depended on
B. If current
                          the extent to which they
   successful theories    anticipated the current theory.
   are truthlike, then
   past theories
   cannot be.
C. These
   characteristically
   false theories were,
                          Many realists whittle down this
   nonetheless,           list by restricting it to theories
   empirically                 with genuine empirical
   successful.                       successes…
    Whittling down the Hit List (S307)
   “Any theoretical framework can be made to fit
    the phenomena-and hence to be 'successful'-by
    simply writing the right kind of empirical
    consequences into it.”
      That shouldn’t count as genuine empirical
       success though!
   Genuine empirical success = novel predictions.
   This takes care of most members of the hit list,
    since they are not genuinely empirically
    successful in the sense of yielding novel
    predictions.
                         But…

   There are some false theories that yielded
    novel predictions, and thus were
    genuinely empirically successful.
       Fresnel's theory of diffraction novelly
        predicted that if an opaque disk intercepts the
        rays emitted by a light source, a bright spot
        will appear at the center of its shadow
     A brief digression on Fresnel
   Fresnel believed that light travels as a wave
    through the luminiferous ether
   From this, he made the novel prediction that if
    light hit an opaque disk, the disk will cast a
    shadow with a very bright center
   But we now know that there is no ether, and
    that light behaves like both a wave and a
    particle…
   So this would seem to suggest that novel
    predictive theories need not be approximately
    true…
Enter…
                     The strategy
   Take a past theory that made successful novel
    predictions
   DIVIDE its theoretical/unobservable entities into:
       Essentially contributing theoretical constituents; and
       Idle theoretical constituents.
   CONQUER the novel prediction by showing that
    they are the result of essentially contributing
    theoretical constituents, and the latter have
    been retained in our current theories.
                More on dividing
   A theoretical constituent is a particular law or
    mechanism posited by a theory, but not the whole
    theory
   A theoretical constituent H essentially contributes to a
    novel prediction P if:
      H together with another set of hypotheses H' (and
       some auxiliaries A) entail P;
      H' and A alone cannot yield P; and

      No other available hypothesis H* which is
       consistent with H' and A can replace H without loss
       in the relevant derivation of P.
   Otherwise, H is idle.
                   Fresnel Again
   Most of the properties attributed to ether were
    idle with respect to Fresnel’s novel prediction
   The essentially contributing constituent was
    Lagrangian dynamics, which is still used in
    contemporary physics
       We now attribute Lagrangian properties to
        electromagnetic fields rather than ether.
   So the novel empirical success of ether theories
    is best explained by Lagrangian dynamics
   So Lagrangian dynamics is approximately true.
    Objections to Psillos: Overview
   Must current theories be true?
   Where have all the good theories gone?
   Are essential contributions really essential?
   Are realists good historians of science?
              Objection 1:
 Must current theories theories be true?
 Recall Psillos’s characterization of the
   Pessimistic Induction:
  A. Current successful theories are
     approximately true.
  B. If current successful theories are truthlike,
     then past theories cannot be.
  C. These characteristically false theories were,
     nonetheless, empirically successful.
 However, the antirealist needs something
   much weaker than A, i.e.,
     If any theories are truthlike, our current
     successful theories are it.
   Implications of this for the
     Pessimistic Induction
If any theories are truthlike, our current
successful theories are it.
Thus, even if divide et impera is
otherwise flawless, we would only be
entitled to the following:
• If our current theories are truthlike, then
  our past theories had truthlike theoretical
  constituents.
But we still haven’t proven that our
current theories are truthlike.
            Objection 2:
Where have all the good theories gone?
   To whittle down the Hit List, realists make the
    criterion of empirical success more stringent,
    e.g., using novel predictions.
   However, the more stringent the criterion, the
    greater the number of irrational/unsuccessful
    scientific theories.
   Thus, if novel prediction is the mark of genuine
    empirical success, then even realists should
    accept theories that are technologically,
    observationally, retrodictively, and/or
    explanatorily successful can be false with no
    penalty.
   But that’s most theories! Look at the Hit List!
   So we can be antirealists about most theories.`
      Objection 2 continued

 Furthermore, why is it that only
  theories yielding novel predictions
  require realist explanations of their
  success?
 If it’s merely to “stump the
  antirealist,” then this looks ad hoc.
       Objection 3: Are essentially
    contributing constituents essential?
   Recall: A theoretical
    constituent H essentially       Underdetermination: We
    contributes to a novel          could have a different set of
    prediction P if:                hypotheses and auxiliaries
      H together with another
                                    that would yield P!
       set of hypotheses H' (and
       some auxiliaries A) entail
       P;
      H' and A alone cannot
       yield P; and                 This could just speak to our
      No other available           lack of imagination. The fact
       hypothesis H* which is       that H* is unavailable
       consistent with H' and A
       can replace H without loss   doesn’t mean it’s false,
       in the relevant derivation   which is what realism
       of P.                        requires.
      Objection 4: Are realists good
          historians of science?
   The realist claims that a novel prediction
    happened because a past theory used the same
    theoretical constituents that current theories
    use.
   However, “with hindsight, we can rather easily
    work it out so that the theoretical constituents
    that 'contributed' to the successes of past
    theories turn out to be those which were, as it
    happens, retained in subsequent theories”
    (S311)
    Psillos’s Reply to Objection 4
   With their own current theories, scientists
    take different attitudes towards whether
    something is essential or merely
    speculative/tentative.
   So realists can do this too. It is possible to
    identify essential contributors without the
    benefit of hindsight.
                Rebuttal to Psillos
   But even very good scientists aren’t great at
    this. Look at Maxwell on the status of ether:
       Whenever energy is transmitted from one body to
        another in time, there must be a medium or
        substance in which energy exists after it leaves one
        body and before it reaches the other.
   In effect, this just produces another pessimistic
    induction over the course of scientists’
    judgments about what is an essentially
    contributing theoretical constituent.
                        Recap
   The Pessimistic Induction can be
    challenged on the grounds that:
       Many members on Laudan’s Hit List are not
        genuinely empirically successful.
       Divide et Impera: The ones that are can be
        explained by their having truthlike theoretical
        constituents.
   There are at least four objections to these
    challenges to the Pessimistic Induction.

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:1
posted:6/24/2012
language:
pages:24