Lecture 3 Popper cont.

1
Lecture 3Popper (cont.)
2
Semmelweis a Popperian story?

• The Vienna General Hospital 1844-1848.
• Division I 8 of women sick Division II 2
  sick. Why?
• H1 Airborne infection
• H1 ? P (P Many women sick outside of the
  hospital)P (Observation) H1
• H2 Rough examination by students
• H2 ? P (P Fewer students, fewer deaths)P
  (Observation) H2
• H3 Overcrowding in Division 1
• H3 ? P (P More women in Div. 2, fewer deaths in
  Div. 1)P (Observation) H3

3
Semmelweis (cont.)

• H4 Position during childbirth
• H4 ? P (P Position changed, fewer deaths)P
  (Observation)
• H4
• H5 The priest effect
• H5 ? P (P Priest changes path, fewer deaths)P
  (Observation) H5
• H6 Doctors and students transfer cadaveric
  material
• H6 ? P (P Wash hands, fewer deaths)P
  (Observation)H6 corroborated!

4
The main problems with Poppers philosophy

• How do we decide that observation statements are
  true?
• Is there a clear distinction between observation
  statements and theories?
• What is falsified theory under test or auxiliary
  hypotheses?
• Is it really theories themselves that are
  non-falsifiable, or is it people who make them
  so?
• Are all scientific statements falsifiable?
• Can we define corroboration without induction?
• Why should we prefer corroborated theories over
  falsifiable theories that have not been tested?
• Indeed, why should we prefer corroborated
  theories over falsified theories?

5
1. The truth of observation statements

• Take a universal statement No one can levitate.
• I observe someone being suspended in the air
  without support.
• Therefore, I falsify the statement No one can
  levitate.
• The problem here, of course, is that I am wrong
  if I think I observed levitation. (Maybe its a
  magic trick, optical illusion, hallucination)
• But then in any situation where a prediction
  derived from a given theory is contradicted by
  observation, there is a possibility that the
  observation is not genuine.
• Falsification can be avoided by rejecting the
  observation, instead of rejecting the theory!

6
2. Observation and theory

• If I see a star through a telescope, is this
  observation?
• Someone might argue that what I really see is a
  yellow dot, which I only interpret as a star.
• In other words, my belief that there is a star
  out there is not based only on observation but on
  much theoretical knowledge (that the telescope is
  properly constructed, that light tends to move in
  straight lines, that lenses are cut in such a way
  to reflect distant objects accurately)
• With electronic microscopes this is even more
  obvious, or with traces of particles in
  accelerators.
• Pure observation (without assuming any
  theoretical knowledge) is very rare.

7
3. The role of auxiliary hypotheses

• To derive a prediction from a theory, one usually
  assumes the truth of some auxiliary hypotheses.
• For example, testing Newtons theory scientists
  derived a prediction about planetary motions. But
  to do that they had to assume many things about
  planets (about their number, masses, speeds,
  positions).
• All predictions were correct except the
  prediction about Uranus. Uranus did not behave as
  Newtons theory predicted. There were two
  possible reasons for that.
• Either (1) Newtons theory was false, or (2) some
  of the things assumed about the planets were
  false.
• The simplicity of modus tollens is lost!

8
3. The role of auxiliary hypotheses (cont.)

• Scientists did not even consider rejecting
  Newtons theory.
• They rejected the belief that there are only 7
  planets Uranus behaved strangely, they said,
  because probably there was another planet (whose
  force perturbed Uranus).
• And the planet was found Neptune.
• But unfortunately Neptune also did not behave the
  way Newtons theory predicted. Refutation?
• No. Maybe there is still another planet?
• And the planet was found Pluto.
• But Mercury also misbehaved. Maybe another planet
  again?
• The planet was even given a name Vulcan.
• But it was never found.

9
4. Are theories non-falsifiable?

• In some sense, Marxism (Poppers main example of
  pseudo-science) was falsifiable. It predicted (1)
  that the revolution will break out in capitalist
  societies (2) that workers will become poorer
  (3) that communism will lead to a more humane
  society.
• All these predictions were shown to be wrong by
  observation, so Marxism was falsified!
• Marxists did not agree about this, of course.
  Faced with false predictions, they reformulated
  their theory in order to avoid refutation.
• Their was nothing unscientific about their
  theory. What was unscientific was the way they
  defended it.

10
5. Non-falsifiable science

• Some statements are part of recognized science
  but it is not clear at all that they are
  falsifiable.
• Some examples
• There are neutrinos (Existential statements are
  not falsifiable.)
• Probability statements are also not falsifiable.
  Take theory T An event of type A is extremely
  unlikely. Suppose you nevertheless observe A.
  Does this observation falsify T?
• No. Remember, falsification works in a deductive
  way, and obviously you cannot derive from T a
  prediction that A will not happen. You cannot
  even derive a prediction that A will not happen
  very often!

11
6. Is corroboration an inductive notion?

• Popper says that theories are more corroborated
  if they are tested in a variety of circumstances
  than if they are always tested under the same
  conditions.
• He says that All metals expand when heated is
  more corroborated if we observe 5 pieces of
  copper and 5 pieces of iron expanding when heated
  than if we just observe 10 pieces of copper
  expanding when heated.
• This accords with common sense but it is unclear
  whether Popper can justify this attitude without
  induction.
• If observing 5 pieces of copper is not an
  inductive reason for generalization about copper,
  then testing the statement about metals on
  another 5 pieces of copper is as risky as testing
  it on 5 pieces of a new metal (iron).

12
7. Why prefer tested theories?

• The acceptance of scientific statements is
  connected with the belief that they are useful in
  practice (for building bridges, curing diseases,
  launching rockets)
• If corroboration tells us only something about
  the past (which tests a theory passed), why
  should we trust these (tested) theories more than
  untested theories?
• Suppose there are two kinds of bridges, A and B.
  We tested a theory Bridges A are stable many
  times in the past by deriving a prediction that A
  will not fall, and we did observe that A never
  fell. But if the past is telling us nothing about
  the future (no induction!), why should we prefer
  A bridges over B bridges (which we never tested)?
  We might as well build bridges, relying on the
  falsifiable but uncorroborated theory All B
  bridges are stable.

13
8. Why prefer non-falsified theories?
It surely seems that Popper can give us a good
reason for preferring tested theories over
falsified theories. After all, falsified theories
are false, and tested theories have at least a
chance of being true. Is it not obvious that hey
are better?
Yes, they are better in some sense, but the
question is whether they are more to be trusted
for our future actions. Without induction, the
answer must be negative. False theories are
equally good as corroborated theories!
Past
Future