[Return to Dmitry Sepety’s Personal Page]
What
Falsification Comes to
by Dmitry Sepety
The contention that
scientific epmirical theories (universal statements checkable by observational
reports) aren’t verifiable but are falsifiable is the core of Karl Popper’s
solution of problems of induction and demarcation.
This contention was contested
by other philosophers, who argued that no theory is really falsifiable. One of
their arguments is that it is always possible to contrive some ad hoc
hypothesis that would save the theory under testing. Another is famous Duhem’s contention
that predictions (which on Popper’s account may be contradicted by
observational statements, so that such contradictions constitute a
falsification of the theory from which predictions are deduced) are never (or
almost never) deduced from the theory alone: they are deduced from a set of
premises one of which is the theory under examinations (others are initial
conditions relevant). So, if observational statements contradict predictions,
we can put the blame on any of premises. Besides, on Popper’s account
observational statements are also hypothetical, which makes open another escape
– to deny the truth of observational statements.
On some later fashionable
accounts (Kuhn, Lacatos etc.) the "normal" activity of scientists is
the opposite of what it must be on Popper’s account: it is not attempts to
falsify adopted scientific theories, but attempts to save those scientific
theories which are foundational for their scientific community-school
(paradigms ore cores of scientific research programs) by such contraptions.
These objections were anticipated
by Popper before they were advanced (in "The Logics of Scientific
Discovery", where he first formulated the theory of falsificationasm).
Popper explained that
his theory is not descriptive – it is not mere "description" of what
scientists do (for scientists do quite a many different things: sleep, eat,
make money and career, defend their reputation etc.). Though his theory has
descriptive features, its "descriptions" are wery selective and cover
only those aspects of scientific activity which are positively relevant for
scientific progress − the progress toward theories which are more
explanative and nearer the truth then their predecessors. His theory is
explanation of what there is in scientific activity which pushes scientific
progress, and, so, is normative – it explains what scientists are to do if their aim is to continue and intensify scientific
progress. Surely, if scientists will pursue, primarily, other aims, they
will probably do, primarily, other things.
About denial of the
truth of observational statements meant to test a theory, it surely is always
possible; but we won’t resort to it, if our aim is truth. Surely, this doesn’t
mean that we are to believe any observational statement whatever, and to renounce
a theory whenever someone stated that he have observed some events which contradict
it. To be proper tests, observations must be reproducible, repeatable,
intersubjectively checkable. Now if a certain effect is regularly observed
under certain conditions, if scientists who did check it agree on what is
observed, then observational report is to be accepted as true. "Basic
statements" used to test theories are observational statements which
commands a consensus of competent observers-scientists. To quote Popper, on his
solution of Fries’ problem (accentuations mine):
"Every test of a theory, whether resulting in its corroboration or
falsification, must stop at some basic statement or other which we decide
to accept. If we do not come to any decision, and do not accept some basic
statement or other, then the test will have led nowhere. But considered from a
logical point of view, the situation is never such that it compels us to stop
at this particular basic statement rather than at that, or else give up the
test altogether. For any basic statement can again in its turn be subjected to
tests, using as a touchstone any of the basic statements which can be deduced
from it with the help of some theory, either the one under test, or another.
This procedure has no natural end. Thus if the test is to lead us anywhere,
nothing remains but to stop at some point or other and say that we are
satisfied, for the time being. It is fairly easy to see that we arrive in
this way at a procedure according to which we stop only at a kind of statement
that is especially easy to test. For it means that we are stopping at statements about whose acceptance or rejection the
various investigators are likely to reach agreement. And if they do not
agree, they will simply continue with the tests, or else start them all over
again. If this too leads to no result, then we might say that the statements in
question were not inter-subjectively testable, or that we were not, after all,
dealing with observable events. If some
day it should no longer be possible for scientific observers to reach
agreement about basic statements this would amount to a failure of language as
a means of universal communication. … In this new Babel, the soaring
edifice of science would soon lie in ruins." (Popper 2002a, 86)
As for possibility of
saving the theory by contriving ad hoc
hypotheses Popper admitted it, but argued that we must adopt methodological
rules which are to prevent it, if our aim is scientific progress. This
methodological rule must forbid defending theories by contriving such ad hoc hypotheses which aren’t
themselves empirically testable. Popper wrote that ad hoc hypotheses are permissible and useful for science if, and
only if, the resulting system, which consists of the theory under examination
and ad hoc hypothesis is more
empirically testable then (or, at least, as well empirically testable as) the
theory by itself.
Really, the question of
permissibility of ad hoc theories can
be considered as a special case of Duhem’s problem; they change the set of
premises (initial conditions) from which, together with the theory under
examination, predictions are deduced.
As for Duhem’s problem
generally, Popper’s answer is that other premises may be independently tested.
Popper also make the
point that some posiible observational statements are incompatible with a
theory whatever other premises (initial conditions) are. Theory strictly
forbids some events, and if such events happen, the theory is falsified.[1] For example, Popper writes that
Newton’s gravitational laws forbid that planets moved in rectangular or
triangular orbits. Hense, if rectangular or triangular planetary orbits were
observed, then Newton’s gravitational laws would be falsified.
I think this point is
very weak, on two reasons.
First, Newton’s
gravitational laws don’t forbid absolutely
strictly that planets moved in rectangular or triangular orbits. Imagine
that scientists have discovered some anomal planet which moves in rectangular
orbit. Now someone can propose untestable ad
hoc hypothesis that though Newton’s gravitational laws hold, there are some
additional forces of unknown nature which make the planet to move in
rectangles. As A.Musgrave explains:
"Newton’s laws (his laws of motion and
gravitation) do not exclude the possibility that a body might travel in a
square – think of a snooker ball moving in a square due to collisions with
other balls. Only Netin’s laws together with auxiliary assumptions about
panets, assumptions which entail that a planet does not behave like a snooker
ball, can exclude the possibility of a planet going in a square orbits"
(Musgrave 2009, 193-194).
Second, even if we admit
that Newton’s gravitational laws strictly forbid rectangular and triangular
planetary orbits, it is too poor and uninteresting prohibition. Even without
Newton’s gravitational laws we would hardly expect to discover a planet which
moves in rectangles or triangles. The qreat multitude of other purported
gravitational laws, very different from Newton’s, can be proposed which
"forbid" rectangular or triangular planetary orbits just as well.
In most cases
(well-controllable reproducible experimental observations) questions about
initial conditions are solved in the same way as questions about the truth of
observational "basic" statements. In a well-organized experiment, most
initial conditions usually are themselves well checkable observational
statements. But still, there always remains a possibility that some factors
which exert a considerable influence on observed events have escaped our foresight
and didn’t got into the set of premises used for deduction of predictions. All
efficient empirical tests presuppose isolation of the system checked, −
in the sense that there are no "external" (i.e. unaccounted) factors
which considerably influence observed events. Scientists painstaking contrive and
organize such experiments which are expected to achieve such an isolstion. But there
is no absolute guarantee that it is achieved. That is where ad hoc hypotheses enter.
Now the question is: if
Popper’s proposition about conditions when ad
hoc hypotheses are permissible always advisable?
Suppose, there is a respectable
scientific theory T with a perfect
performance before the moment M. Now
at M a new experiment E is held, and its observed results
contradict prediction derived from T
and assumed initial conditions C.
The result is reproducible and intersubjectively checkable. All checks of
initial conditions have failed to detect anything wrong with them. All proposed
ad hoc hypotheses which were empirically
testable (given current possibilities of testing) have failed too.
There may be some other
proposed ad hoc hypothesis H which, if true, would succeed, but given
current possibilities of testing, this hipothesis is not empirically testable,
or scientists didn’t contrive, for a while, an experiment which could test it. In
a future, such a test can become possible, and H will become empirically testable, and can happen successful.
Or there may be no such
hypotheses at present moment. In a
future, such a hypothesis H can be
contrived, and it can be empirically testable, and can happen successful.
Now, if T had a perfect performance before M, and continues to have a perfect
performance except in some special condiction peculiar to E, and there is no known alternative to T which would do better, it may be unadvisable to renounce T. It can be advisable still to consider
T as true on promissory notes that an adequate testable explanation
(compatible with T) for results of E will
be found. Or we can consider T
as the best approximation to truth we have at present moment, and to hope that in
future some better theoty T1 will be
invented to account successfully for both successess and failures of T. This two options are different as beliefs about T, but they have in common that T is retained (as either true or the
best known approximation to truth) in expectation of future developments. And
scientists can differ in their belief-options (some believe the first option
rather then second, while others vice
versa), which makes a little practical difference for science. There still may be a practical difference in
that some scientists will be more inclined to search for ad hoc hypotheses which could restore the stability of T and for means of their testing, while
other scientists will be more inclined to search for a new theory T1 which could successfully supercede T. Both activities are equally
important; there is no saying beforehead which will be successfull in any
particular case; such a competition between "conservatives" and
"revolutionaries" is highly profitable, and both sides can appeal to high
critical standards. The balance depends on how good is performance of T; how wide is domain of its success
and how wide is domain of its failure. It changes with time: the longer T fails to pay its promissory notes,
the less it is believed to be true.
Now if it fits with
Popper’s account? The answer is not quite clear, for Popper in different
contexts made different emphasizes which tend to different interpretations. I
think that Popper’s proposition about ad
hoc hypotheses in LSD is not to
be taken too literally, in the sense that in situations like obove described T is to be immediately renounced. At
least, Popper’s later explanations suggest more flexible approach. Popper
always (LSD included) emphasized that
empirical falsifications are never conclusive. And in "Realism and Ains of
Science" Popper admitted that in scientific practice falsification is not
– and is not to be – one-moment event; that it may need a long time for examination
of different possibilities; and that often the theory (T) is considered as refuted only after more successful alternative
(T1) is invented.
Now, as Kuhn asked:
"What is falsification if it is not conclusive disproof?" (Kuhn 1970,
15) I think the adequate answer should consist of two parts – logical and
practical.
The logical part of an
answer is that falsification is a detection of contradiction between
observational statement S and a set
of premises P which include a theory
under examination T and statements
of initial conditions C. Or it may
be described as conditional logical truth that if S is true then either T
is false or some of C is false. Or
as logical truth that either S is
false, or T is false, or some of C is false.
In the practical part of
an answer I will follow M.Notturno. Notturno points out that no "proof that a theory is false" (and, so, refutation in this strong sense) is
logically possible (except if a theory is self-contradictory); what is possible
is criticism.
"...criticism is not the refutation of theories. It is the setting of
problems for them. We set problems for a theory by showing that it contradicts
other statements that we believe to be true. Since contradictory statements
cannot both be true, we can force ourselves, in this way, to choose between a
theory and our other beliefs. If our criticism is effective, then we may, at
the minimum, have to refine our theory in some way so as to remove the
contradiction. But the criticism itself never proves that our theories are
false. And if this is what is meant by ‘refutation’, then it never refutes
them. The problems that we set may sometimes seem overwhelming, and they may
even lead us to think that our theories have been shown to be false. But if
they do, then we should remember that we are fallible human beings, and that
our fallibility means that what seems overwhelming at one moment may seem easy
to deal with the next. We may, for example, learn something new that makes what
once seemed obviously true seem obviously false." (Notturno 1999, 118-119)
Now, I think practically
empirical falsification need not be interpreted as anything stronger than criticism
of a special (empirical) kind which poses
a serious problem for a theory T.
This problem demands solution and, if satisfactory solution is not found,
weighs against a theory, gradually undermines it, corrodes beliefs that T is true, pushes scientists to search
for alternative theory to supercede T.
On the other hand,
observational statements have epistemological priority in the sense that theories
are to be estimated – and adopted or rejected – depending on how well they explain
and predict observable phenomena and on if there are alternative theories which
do it better. Empirical science wouldn’t exist if we rejected observational
statements which contradict our theories, just because they contradict our
theories, rather than reject theories contradicting those observational
statements which we judge to be correct descriptions of what is observed.
Empirical theories are invented to explain and predict observables, and are to
be judged by how well they cope with this task, and if there are other competitors
which would to the job better.
Literature
Kuhn, T. 1970. Logic of Discovery or Psychology of Research? In Criticism and the Growth of Knowledge,
ed. by I.Lacatos and A.Musgrave. Cambridge: Cambridge University Press.
Musgrave, A. 2009. Secular Sermons. Otago University Press.
Notturno, M. 1999. Science and the Open
Society.
Popper, K. 2002a. The Logic of Scientific Discovery.
Popper, K. 2002b. The Poverty of Hystoricism.
Routledge
Popper, K.
2005. Realism and the Aim of Science.
[Return to Dmitry Sepety’s
Personal Page]
[1] Sometimes
Popper goes so far as to state that “every natural law can be expressed by
asserting that such and such a thing cannot happen… For example, the law of
conservation of energy can be expressed by: ‘You cannot build a perpetual
motion machine’, and that of entropy by:
‘You cannot build a machine which is a hundred percent efficient’.
(Popper 2002b, 56) But I think it is obvious that such ‘expressions’ aren’t
equivalent with what they are meant to express (statements of natural laws).
For example, the law of conservation of energy is much richer in its contents
than the statement that perpetuum mobile
is impossible. It logically follows from the law of conservation of energy that
perpetuum mobile is impossible. But
it does not logically follows from the statement that perpetuum mobile is impossible that the law of conservation of
energy is true. There are many other statements (including empirically testable
predictions) which follow from the law of conservation of energy, but does not
follow from the statement that perpetuum
mobile is impossible.