Popper, Falsification and Scientific Demarcation
Karl Popper
Falsification, not
induction
Sir Karl Popper:
1902-1994
The Problem of
Demarcation
Science was a game which
required that in some specifiable circumstances we reject a given scientific theory.
For Popper, other sorts
of theory could perfectly well be meaningful, but they are not scientific.
So Popper was not a
“commit it to the flames” sort of guy.
This was first offered as
a logical feature of the theory, that is, what is MEANT by “science, ad that
which serves to distinguish instances of science from, say, instances of
metaphysics or pseudo-science.
A scientific theory must
have implications for observation, and when our observations show those
implications are false, we can reject the theory.
But notice, if we were to
adopt the criterion of verification as is, then scientific statements would be
no less meaningless than metaphysical statements since neither statements are
verifiable.
Science is not
distinguished from non-science on basis of methodology. No unique methodology specific to science
Science consists mostly
of problem solving.
A demarcation between
science and pseudo-science is established by falsification. A theory is scientific only if it is
refutable by a the observation of a conceivable event.
Every genuine test of a
scientific theory is based on an asymmetry between verification and
falsification
Falsification is the idea
that science
1. advances by
unjustified, exaggerated guesses followed by
2. unrelenting criticism
Any "positive
support" for theories is both unobtainable and superfluous.
All we can and need do is
create theories and eliminate error.
Scientists never actually
use induction according to Popper. It is
impossible to verify propositions by reference to experience.
This is his “answer” to
the “problem of induction.”
Good scientific practice:
E.g. Einstein’s general relativity
Probably
the most important eclipse in the history of science occurred on 29 May 1919.
Just six months after the end of World War I, British astronomers used it to
test a new idea that came from Germany in 1915.
The
proposition was that gravity affected light, space and time itself, and as a
result the Sun would deflect starlight passing by it. Changes in the apparent direction
of stars in the sky, seen close to the Sun during a total eclipse, could
confirm the idea.
The
announcement of favourable results in London on 8 November 1919 signalled the
replacement of Newton's theory of gravity by the theory of general relativity.
Its originator, a 40-year old Berliner called Albert Einstein, at once became
the most famous scientist in the world.
Expeditions
of astronomers photographed the eclipse in difficult tropical conditions in
Brazil and on the African island of Principe. At the time, the Sun was in front
of a useful cluster of stars, the Hyades. The astronomers compared the relative
positions in the sky near the Sun with the positions of the same stars as
previously photographed in the night sky.
The
expeditions and analyses, organised by the Royal Society and the Royal
Astronomical Society, were masterminded by Arthur Eddington of Cambridge. He
was one of the very few experts who immediately understood Einstein's theory
and appreciated its revolutionary importance, if it was correct.
Even
for stars almost in line with the Sun, the shift in apparent position is less
than two seconds of arc, or a few ten-thousandths of a degree. The 1919
measurements confirmed that the Sun bent the light rays by roughly the right
extent - less than predicted in Principe, more than predicted in Brazil.
Conjecture: mass of the sun bends
the path of light
·
If the apparent location of the observed star doesn’t
shift, the theory is wrong.
·
It will have been refuted.
So the mark of a scientific
theory is whether it can be falsified by observation
“Falsificationists…
prefer an attempt to solve an interesting problem by a bold conjecture, even
(and especially) if it soon turns out to be false, to any recital of a sequence
of irrelevant truisms”(CR: 231)
This gives us:
(i)
a glimpse of scientific method
(ii)
a demarcation criterion for science
Scientific method:
·
Scientific theories have deductive consequences
·
They can be falsified but not confirmed.
·
The objective of scientific theorizing is to put forward
(bold) hypotheses and then test them in order to falsify them
·
Theories are falsified by basic statements
(what is a basic
statement?)
Confirmation and
Pseudoscience
Demarcation:
So scientific theories
are those that can be falsified by basic statements. (pace Quine)
Good scientific theories
do not make themselves immune from falsification by use of ad hoc
hypotheses.
Progress of Science:
Science progresses by
eliminating theories that have been falsified.
But.. does this in fact
constitute process? Falsifying one of
finitely many potentially true theories might be thought to constitute
progress, like find one losing lottery ticket my narrow the field of potential
winners, but eliminating one of infinitely many possibly true theories does not
seem to qualify are making progress to the “truth.”
A scientific theory
cannot be shown to be true. But some scientific theories do have varying
degrees of success. They resist falsification.
The difficulties
connected with my criterion of demarcation (D) are important, but must not be
exaggerated. It is vague, since it is a methodological rule, and since the
demarcation between science and nonscience is vague. But it is more than sharp
enough to make a distinction between many physical theories on the one hand,
and metaphysical theories, such as psychoanalysis, or Marxism (in its present
form), on the other. This is, of course, one of my main theses; and nobody who
has not understood it can be said to have understood my theory.
The situation with Marxism
is, incidentally, very different from that with psychoanalysis. Marxism was
once a scientific theory: it predicted that capitalism would lead to increasing
misery and, through a more or less mild revolution, to socialism; it predicted
that this would happen first in the technically highest developed countries;
and it predicted that the technical evolution of the 'means of production'
would lead to social, political, and ideological developments, rather than the
other way round.
But the (so-called) socialist
revolution came first in one of the technically backward countries. And instead
of the means of production producing a new ideology, it was Lenin's and
Stalin's ideology that Russia must push forward with its industrialization
('Socialism is dictatorship of the proletariat plus electrification') which
promoted the new development of the means of production.
Thus one might say that
Marxism was once a science, but one which was refuted by some of the facts
which happened to clash with its predictions (I have here mentioned just a few
of these facts).
However, Marxism is no
longer a science; for it broke the methodological rule that we must accept
falsification, and it immunized itself against the most blatant refutations of
its predictions. Ever since then, it can be described only as nonscience—as a
metaphysical dream, if you like, married to a cruel reality.
Psychoanalysis is a
very different case. It is an interesting psychological metaphysics (and no
doubt there is some truth in it, as there is so often in metaphysical ideas),
but it never was a science. There may be lots of people who are Freudian or
Adlerian cases: Freud himself was clearly a Freudian case, and Adler an
Adlerian case. But what prevents their theories from being scientific in the
sense here described is, very simply, that they do not exclude any physically
possible human behaviour. Whatever anybody may do is, in principle, explicable
in Freudian or Adlerian terms. (Adler's break with Freud was more Adlerian than
Freudian, but Freud never looked on it as a refutation of his theory.)
The point is very
clear. Neither Freud nor Adler excludes any particular person's acting in any
particular way, whatever the outward circumstances. Whether a man sacrificed
his life to rescue a drowning, child (a case of sublimation) or whether he
murdered the child by drowning him (a case of repression) could not possibly be
predicted or excluded by Freud's theory; the theory was compatible with
everything that could happen—even without any special immunization treatment.
Thus while Marxism
became non-scientific by its adoption of an immunizing strategy, psychoanalysis
was immune to start with, and remained so. In contrast, most physical theories
are pretty free of immunizing tactics and highly falsifiable to start with. As
a rule, they exclude an infinity of conceivable possibilities.
“We must not look upon
science as a body of ‘knowledge’, but rather as a system of hypotheses, that I
say, as a system of guesses or anticipations which in principle cannot be
justified, but with which we work as long as they stand up to tests, and of
which we are never justified in saying that we know that they are ‘true’, or
‘more or less certain’ or even ‘probable.’
Contra Induction
Popper objected to
inductivist views (according to which a scientific theory had to be supported
by observations) for two reasons:
He argued that it was too
easy to find evidence ‘supporting’ a given theory. (Here he used examples from Freudian
psychology, Marxist theory and other places.)
He argued that Hume’s
critique of induction showed that we simply could not actually support (i.e.
give good evidence for) a general claim using evidence from particular cases.
Contra Induction
Modus Tollens and Modus
Moron J
If T then O
Not O
Therefore
Not T.
Vs.
1’. If
T then O
2’. O
Therefore
3’. T.
The second of course is
an invalid Inference: (The polite, official label for the second argument form
is ‘affirming the consequent’…)
Popper’s influence on
scientists’ own views of their work has been considerable– perhaps even greater
than T.S. Kuhn’s influence.
His presentation of
science as grounded in the ongoing testing of our theories and always open to
the possibility that they will be refuted at the next test, resonates strongly
with them.
Philosophically, his
rejection of induction allowed him to avoid the difficulties the logical
empiricists struggled with in their attempts to understand confirmation.
But the rejection of induction
also leaves a major gap in his understanding of science– we don’t simply regard
science as a kind of intellectual game:
we use the results of science to guide our practice when we are engaged
in pursuing our ends.
So without some account
of induction, Popper’s position provides no account of how or why we should
prefer to act in ways our present well-tested but still-standing theories imply
will be successful.
Difficulties
Popper was know for being
extremely persistent and stubborn– as well as very demanding of the students
who worked with him. Nevertheless, his thought
certainly did evolve on these issues– but he never budged on any of his
fundamental views, including especially the complete rejection of induction.
He did attempt to capture
some sense of progress for science in terms of accounts of the truth-content
and question-content of theories, but these notions were always untenable, and
did nothing to resolve the lack of any commitment to induction.
He did some influential
work on probability theory, in which conditional probabilities (rather than
absolute probabilities) were taken as primitive.
A (nother) puzzle about
induction
There is something odd
about Hume’s (and Popper’s) critique of induction and its relation to the modest
attitudes of scientists towards even their most successful theories.
On one hand, it is clear
that (for example) despite its many brilliant successes, Newtonian physics has
turned out to be false.
The history science is
the history of such failures. This might
lead us to think that scientists should doubt that their present theories are
true as well.
But their modesty is not
a form of Humean skepticism. Quite
the opposite; the modesty arises and is justified by induction. If we were wrong despite failing to falsify a
theory in the past, they we may be wrong in the present and future.
In these cases, the theory
fits
a large collection of well-established facts, When this is the case, we don’t expect (nor
has it been the case that) similar observations of similar cases undermine the
theory later on.
The observations
scientists are ‘afraid of’ i.e. the ones they suspect may lead to trouble for
their presently held theories, are observations of new kinds of facts: observations of events at higher energies, or
on smaller scales, or of new phenomena.
But a true Humean has no
more reason to expect new phenomena to give trouble than she has to expect
entirely familiar sorts of cases to give trouble.
In either case, it’s only
induction that supports our present theory, and Hume and Popper both reject
induction as invalid (though Hume thought that it was also indispensable).
Further considerations
On the other hand, it may
well be induction that leads us to suspect theories are more likely to go wrong
when we apply them to new kinds of phenomena, or new extremes of parameter
values.
Still, pure inductive
skepticism like Popper’s draws no line here– induction is invalid, period—not
‘OK’ when applied to familiar cases and dubious when extended beyond them.
Fallibilism
Fallibilism: The position of accepting (some) of our
theories are true, while at the same time, accepting that any or all of them may
well turn out to be wrong.
In fact, scientists can
sometimes even predict, based on their theory, the sorts of circumstances in
which it’s likely to fail (cf. recent reviews of the LHC and its likely results
for the ‘standard model’ in particle physics).
But fallibilism does not
imply that we have no grounds for accepting our present theories– it only
implies that we suspect they may not turn out to be completely true.
Truth as Correspondence
There is something odd
about thinking in terms of truth here, as well.
The methods of science
(as Popper rightly saw) continually expose theories to further tests– even a
theory that is perfectly reliable in the range of applications we’ve already
tried can turn out to be false because it fails in some new application.
Perhaps, given the actual
record of science, it would be wiser to say that the goal of science is to
achieve greater and greater reliability and precision in our descriptions and
explanations.
Truth as Correspondence
In some sense this
actually brings us back to the idea of truth, but in a more reserved way:
Perfect reliability and
precision is both something we don’t ever expect to actually achieve, and
extremely close to what we would expect (even demand) from a theory we would
regard as true.
Of course we would also
expect maximal power (i.e. we would favor a theory that would constrain our
observations more tightly, assuming equal precision and reliability) before we
would be tempted to speak of ‘truth, the whole truth, and…
Truth as something else?
Searching for Truth in all the wrong places….
Popper’s skepticism about
induction leaves us with little grounds for confidence about making progress in
science—
Assuming everything works
without a hitch as we falsify theories, we may be right in rejecting one theory
after another
But…
We’re left with no sense
of progress towards the truth (no reason to suspect that our theories are
anything more than falsehoods not yet exposed).
Conjecture and Refutation
The pattern of scientific
work on this account is very simple and repetitive:
A scientific conjecture
is made.
Scientists try to refute
it.
The second step is
repeated until it succeeds.
While Popper acknowledges
that we can respond to a failed test by modifying our theory, or even by
modifying some assumptions that were made about the test, he is uncomfortable
with this sort of defensive strategy– to Popper this looks ad hoc.
Theory construction by
means of empirical guidance doesn’t take the sort of risk that Popper thinks a
good theory should.
Popper, Falsification and Scientific Demarcation
Conjecture and Refutation
The pattern of scientific
work on this account is very simple and repetitive:
A scientific conjecture
is made.
Scientists try to refute
it.
The second step is
repeated until it succeeds.
While Popper acknowledges
that we can respond to a failed test by modifying our theory, or even by
modifying some assumptions that were made about the test, he is uncomfortable
with this sort of defensive strategy– to Popper this looks ad hoc.
While Popper acknowledges
that we can respond to a failed test by modifying our theory, or even by
modifying some assumptions that were made about the test, he is uncomfortable
with this sort of defensive strategy– to Popper this looks ad hoc.
Theory construction by
means of empirical guidance doesn’t take the sort of risk that Popper thinks a
good theory should.
Norms and games
For Popper these are
rules that somehow constitute the game of science.
We can do other things
(make ad hoc adjustments to our theories or auxiliary hypotheses as new
evidence accumulates) but that is playing another game.
The surviving theories,
at any given time, are no more likely to be right than any other theory not yet
refuted by the evidence we have.
We have no reason even to
suppose that the most basic, reliable phenomena should continue to respect the
regularities we have observed in them (this is Hume’s skepticism about
induction, which Popper endorses).
But scientists are very
confident about their ability to reliably produce certain phenomena and the
success of their current theories at explaining them–even if they are
fallibilists about those theories when applied to new phenomena and
observations.
Our division of phenomena
into kinds is largely a practical matter- the use of certain kinds of
apparatus, making certain kinds of measurements, etc.
Competition and selection
It’s tempting to draw an
analogy between what Popper describes and evolution.
Falsification of a theory
would be something like extinction.
The scientific community
continuing to work on an unfalsified theory would be like the continuation of a
species over time.
But note this is perhaps
a telling metaphor. Evolution does not
proceed to ultimate perfection, just suitability to problem sets.
Further, survival s
secured so long as the entity can adapt.
Evolution is (as Darwin defined it) descent with modification.
On a similar view of
scientific “progress,” nothing rules out repeatedly modifying a theory to make
it fit the phenomena.
Extinction can still
occur, if no suitable modifications arise, but it might be forestalled
indefinitely by a creative mind.
In science, modifications
that are purely ad hoc rarely succeed over the longer term.
Further, in evolution
many species that are not optimally adapted to their environment can still
persist for some time.
In particular, in the
absence of superior competition work may well continue on a theory even after
scientists are convinced that it is wrong– after all, the work may well focus
further investigation on issues that will help with the development of a new
theory.
From logic to psychology
For Popper, as for the
positivists, the aim of philosophy of science was to lay out how science should
be done, and the criteria for this notion would be fundamentally logical– they
would turn on the logical relations between a scientific theory and
observations.
But the Duhem/Quine
problem led him to turn this logical line into a psychological one: the scientist is someone who should not
avoid falsification by blaming ‘auxiliaries’– she would should place
her theory in the harshest environment possible, and blame her theory if the
results don’t fit. Here the psychology
of decision making replaces the logic of theory and evidence…
This means that scientist
actually risks rejecting a true theory (if she retains a false auxiliary or
even a false observations and consequently regards the theory as
falsified).
The last shred of logical
progress (a steady accumulation of falsified theories) in Popper’s game of
science is lost here.
Further, the demarcation
Popper had sought becomes problematic too.
Popper can still draw a
logical line, if he can specify which theories would be falsified by some
possible observations– but the users of these theories could still defend them,
in principle, by shifting the blame elsewhere.
Godfrey-Smith also
emphasizes a problem for probabalistic theories (such as classical
thermodynamics)– if all they tell us is that certain observations are very improbable, then strictly speaking they
aren’t really falsifiable! Again, the
psychology rather than the logic of science becomes central here.
The problem of induction comes home to roost
Godfrey Smith suggests a
question Popper cannot credibly answer:
Suppose we are building a
bridge, and we have two theories we might employ to guide our work.
One theory has been
tested in many cases, and passed all the tests.
Another has yet to be
tested, but (trivially) has passed every test we’ve ever put it to. (i.e. zero)
Which should we use? Both are, to date, unfalsified.
But…in Popper’s defense,
There is something odd
about this example, though: such a new theory, to be relevant, has to cover a
lot of the same ground as the old successful theory in terms of explanatory
power, etc. – and in order not to be falsified, it must succeed on those
tests.
(But there may be a worse example in the wings
for Popper.)
Upshot of Godfrey Smith’s
question:
If we discount inductive
success thus far entirely, we have no reason to suppose the (“confirmed”)
theory will succeed in a further application.
But also, without
induction, failure in some cases in the past gives us no reason to suppose the
theory will fail in a further application.
(Such a theory could not
be true, but it could still be a reliable guide to future results!)
If we discount inductive
success thus far entirely, we have no reason to suppose the (“confirmed”)
theory will succeed in a further application.
Of course if we can
specify a finite range of theories within which we can show the truth must lie,
then we can proceed by elimination in a quasi-Popperian way.
Taking risks
Even with all these
reservations, Godfrey-Smith thinks Popper was onto something worthwhile:
The idea that scientific
theories must take risks– they must have enough ‘empirical content’ (even as
qualified by Duhem and Quine) that they really could be shown (or at least
justifiably accepted to have been shown) false.
But this focuses more
attention on how a theory is deployed (pragmatic consequences), and how its
users respond to difficulties, than on the ‘logic’ of the theory itself.
We can proceed
scientifically, if we treat the theory as subject to empirical correction and
even refutation.
Or we can proceed
non-scientifically, by isolating and sheltering the theory from any and all
empirical threats.
These choices are even
open to Freudians and Marxists.
What could falsify the
Theory of Evolution?: Precambrian Rabbits
A standard one-line
answer to the question, what would refute evolution.
"Precambrian
rabbits" or "fossil rabbits in the Precambrian" are reported to
have been among responses given by the biologist J.B.S. Haldane when asked what
evidence could destroy his confidence in the theory of evolution and the field
of study. The answers became popular imagery in debates about evolution and the
scientific field of evolutionary biology in the 1990s.
Some accounts use this
response to rebut claims that the theory of evolution is not falsifiable by any
empirical evidence.
This was offered as a
justification for regarding evolution as a scientific theory in light of
Popper’s insistence that falsifiability is an essential feature of a scientific
theory.
Popper himself had
expressed doubts about the scientific status of evolutionary theory, although
he later concluded that the field of study was genuinely scientific.
Rabbits are mammals. From
the perspective of the philosophy of science, it is doubtful whether the
genuine discovery of mammalian fossils in Precambrian rocks would overthrow the
theory of evolution instantly, although, if authentic, such a discovery would
indicate serious errors in modern understanding about the evolutionary process.
Mammals are a class of
animals, whose emergence in the geologic timescale is dated to much later than
any found in Precambrian strata. Geological records indicate that although the
first true mammals appeared in the Triassic period, modern mammalian orders
appeared in the Palaeocene and Eocene epochs of the Palaeogene period. Hundreds
of millions of years separate this period from the Precambrian
Of course it would take
very careful work to establish such an astounding result.
But there are many other
tests evolution has passed and continues to pass (the in-principle independent
tree-structures of taxonomy, biochemistry, development and the fossil record
continue to match up, for example).
Applying new tests to
these trees has led to many new results and much refinement of our views on the
course of evolution—but they have always fit within the narrow range of results
acceptable for evolution.
Objection: Falsificationism is ‘No criterion of truth’:
There is no Theory/Observation distinction:
If this is true, then
accepting an observation statement will require accepting the theory in which
the observation statement occurs.
But if this
is true, then is it unclear if or how we could really falsify a theory via
observation, at least in a truly revolutionary way.
Popper, Falsification and Scientific Demarcation
All observations are
selective and theory laden
‘Here is a glass of water’ is theory laden
In accepting the
statement we must accept a significant amount of theory.
We have only as much
justification for accepting the observation statement as we do for the theory
of which is it a part and which makes it possible.
Upshot: we cannot use
observation to establish the truth of a theory
So how can we establish
the truth of scientific theories?
We Kan’t!