Comments on Popperian Falsificationism

The group of philosophers united under what we have called the "Empiricist Consensus" were most strongly influenced by the development of logical positivism, but there was at least one rather large group who could be fairly said to lie within the consensus but who presented themselves as opponents to the logical positivists; they were (and still are) followers of the philosophy of science of Karl R. Popper, known generally as "falsificationism."

Popper began his career in Vienna at about the same time as the Vienna Circle positivists, and he originally published his magnum opus, The Logic of Scientific Discovery, in German in the 1930's. Popper spent the years of WWII in New Zealand, after which he emigrated to England where he became established at the London School of Economics (LSE). His Logic of Scientific Discovery was translated into English only after WWII and it was only then that Popper became a major voice in philosophy of science in the English speaking world. After the death of Bertrand Russell, Popper became England's best known philosopher, and he enjoyed considerable fame not only for his work in philosophy of science, but also for his socio-political views which vigorously attacked the Marxist claim to provide a scientific economics and social theory. He built up around himself a large institute at LSE devoted not only to the Popperian philosophy of science, but also to demonstrating how the history of science actually embodied, it was claimed, Popperian methodology.

At a first glance Popper seems to share so much in common with positivist philosophers within the empiricist consensus that we tend to view him as only adding a bit different emphasis to basically positivistic "confirmationist" views, a sort of variant of the main empiricist theme. In fact, this is how philosophers looking at science outside of the consensus did view him. However, Popper himself saw his position as the very polar opposite of positivism and spent a great deal of his energies savagely criticizing the positivists and distancing himself from them. So it is important to understand not only how much he shares with the positivists, but also why (in his opinion) his view was so very different.

Like the positivists Popper accepts basically the deductive-nomological model of explanation and the hypothetico-deductivist approach to justification. Thus he regards the business of science as deducing from the conjunction of statements of laws and theories and initial conditions certain observational or "basic" statements which are in effect singular statements about the empirically ascertained "facts." He agrees with the consensus that the "context of discovery," which is concerned with the origin of hypotheses, is properly a matter for psychology and that it answers to no particular "logic" but is the product of the inspiration of genius. Like the positivists, he regards only the "context of justification" as properly of concern to the epistemology of science. In other words, he shares with positivist defenders of the consensus the view that philosophy of science strives to explicate the logical relation between hypotheses and the evidential support in experience which leads to their acceptance as scientific knowledge.

The positivists originally upheld a verificationist theory of meaning that was intended to distinguish "meaningful" scientific discourse, which could be empirically verified, from the empirically unverifiable, and hence "meaningless," discourse of what they pejoratively called "metaphysics." Eventually positivists came to recognize that only the foundational basic truths could conceivably be so verified (and even that was problematic). Laws and theories erected on this foundation could not be reduced to purely particular observational statements, and so the best one could get for such statements was "confirmation," not "verification." Nevertheless they retained the basic idea that meaningful non-analytic statements about the world must rest on empirical support and that statements which lie beyond the reach of any empirical evidence are meaningless "metaphysical" nonsense.

Popper was also anxious to draw a line between science and non-science, but he rejects the positivists' concern with separating meaningful scientific discourse from meaningless "metaphysical" discourse as an absurd and pointless strategy. Thus he regards positivistic concerns with language, with the "logical syntax" and the semantics of an ideal perfect "scientific language," as artificial problems irrelevant to justifying the rational authority of scientific knowledge. His concern is with what he called the problem of "demarcation" or of distinguishing science from pseudoscience, and it is from that task that his theory follows. The way in which Popper demarcates pseudo-science from genuine science is that the pseudo-scientist does everything to make his "theory" immune to any potential refutation, whereas the genuine scientists is willing, even eager, to take bold risks in advocating theories that have the potential to be refuted by observational evidence. It is this willingness to take risks that accounts for the growth of genuine scientific knowledge; its absence in the pseudo-sciences makes these bodies of belief stagnate,

Popper's main point is the extremely elementary logical point that if one takes the business of science as deducing observational consequences from statements of laws and theories and initial conditions, no amount of particular positive observational outcomes will ever prove (or verify) the truth of universal hypotheses or laws, for all such attempted inferences commit the well-known fallacy of affirming the consequent. However, even a single negative observational consequence allows us to validly infer that the conjunction of laws and initial conditions from which it is deduced cannot all be true.

Thus theories can be "refuted" or "falsified," by the well known valid principle of inference known as modus tollens. In short, observational evidence can never prove any general theories are true, but it can falsify them.  For this reason Popper's model of justification is known as "Falsificationism."
This logical point is indisputable and of course was not denied by positivists, who early on gave up talk of "verification" in favor of "confirmation." But this seems to be a rather slight difference in emphasis rather than a wholly different philosophy as Popper regarded it. Why did he so regard it?

From Popper's point of view the positivists and the H-D theorists who speak in terms of "confirmation" have got science's aim altogether backwards. The goal should not be to try to find support in favor of theories, but rather to find evidence which would refute them. Since the positivist is still seeing the scientific enterprise as looking for evidence in the empirical facts which confirms theories, he is still trying to make the logical relation between theories and observation inductive; i.e. he is looking for justifying an inference from the particular, the observational facts, to the universal, the laws and theories. Thus the standard hypothetico-deductivist like Hempel is still in effect a covert inductivist.

Popper grants with Hume that the problem of induction is insoluble, but he insists -perhaps rather astonishingly- that induction is never used in science! There is no reasoning, he claims, from facts to theories. Genuine scientists are not looking for support from the facts for their theories, says Popper, so there is no need for any inductive inference from the observational to the theoretical.
If scientists are not seeking support for their theories from the observational evidence, what are they doing when they deduce observational statements from theories? Popper answered that scientists instead are looking for ways to test their theories against the observational evidence, where by "test" he means "attempting to refute" their theories. What makes a theory "scientific," and thus answers the problem of demarcation, is the fact that it is testable. What makes it "accepted" in science -at least provisionally (for Popper all acceptance is provisional)- is the fact that it has been thoroughly tested by the most severe tests which the scientist can think of, and it has not been falsified. When it attains this status (severely tested and not refuted) then Popper calls it "corroborated." But to say this, in Popper's vocabulary, does not mean that we have good reason to believe it is true, that it is supported by the facts, but rather that, so far, it has not yet been refuted by the facts.
Thus Popper claims to have "solved" Hume's problem by eliminating induction altogether from science.
Popper's view also differs from the positivists in that he thinks that what he describes is a theory of scientific growth and change over history. Recall that the positivists were rather indifferent to the question of whether their model of scientific knowledge resembled real, historical science because they saw their task as normative; they were describing an ideal of how a "rational" science ought to be, even while admitting that actual human science was often less. Unlike the positivists Popper makes no claim to describe an idealized perfected or "rationally reconstructed" science; what he claims to describe is real historical science. The story of the growth of knowledge, he thinks, is the story of putting forward "bold,"or "risky"conjectures, and then testing them by making empirical predictions from them, until they are refuted. Then a new hypothesis is conjectured (often a modification of the old refuted one) and the testing begins all over again. The growth of knowledge is not a matter of "construction" upon a firm foundation (as depicted in the "pyramid model") but of "criticism" of speculative, but testable, conjectures. New hypotheses come from criticisms made of discarded old hypotheses rather than the accumulation of empirical data.
Thus Popper speaks of the progress of knowledge as "criticism and the growth of knowledge" which is a process of "conjectures and refutations." Inevitably every conjecture ultimately meets its nemesis in some fact with which it cannot deal.  It is then considered "refuted" and is replaced with a better conjecture, where "better" here of course cannot mean more "confirmed" or supported by the evidence, but it means more testable.

Good theories, according to Popper, must be testable, but "testable" means potentially falsifiable, refutable. Therefore, in proposing theories, the more refutable, (i.e. the more "testable"), the better. Popper expresses this point by saying that "conjectures" must be "risky" or "bold." Of course it would do little to advance the growth of knowledge repeatedly to propose totally off-the-wall "risky" conjectures just to shoot them down. The occasions when science advances most are when attempts to refute risky conjectures fail, thus corroborating bold guesses, or occasions when safe "modest" conjectures relying mostly on accepted beliefs are, surprisingly, refuted, thus falsifying "established" wisdom.

As a consequence, over time, the progress of knowledge demands that theories become more and more testable, and hence more and more "risky." It follows that the rational scientist never defends any theory to be established "beyond refutation." Since the more precise and the more comprehensive a hypothesis is, the greater the "risk" it takes of being refuted, the historical direction of science will be towards ever more precise theories, comprehending ever greater domains of phenomena.
For these reasons, the Popperian can never believe it is rational to hold any theory to be true; to do so would be "irrational" because it would bring the critical process by which knowledge grows to a halt. Inasmuch as one can never know that the "last possible test" has been devised, Popper holds that any theory can, and ought to, be tested until it is ultimately shown to be false. No rational inquirer ever holds his theory to be true, nevertheless Popper holds his doctrine is realistic in that the process of criticism and the growth of knowledge is in effect a method of filtering out the false, thus narrowing the bounds within which truth can be found, even though it can never be known to be in fact found. Thus as knowledge grows we do indeed learn more and more about the nature of reality; it is the nature of reality which "selects" those theory's which in effect survive best the scientists' onslaught of tests.

According to Popper there is no sharp distinction between the allegedly meaningless metaphysics which the positivists so scorned and meaningful empirical science. Instead, in the Popperian view, most current conjectures about the world started out their histories as untestable "metaphysical" theories, but as we learned more about the world, we discovered ways to make these conjectures more and more testable, such that what we "accept" today is just the most recent, and so most testable version which we have been able to come up with. So science is hardly opposed to metaphysics (as typical positivists held); indeed, it begins in metaphysics. Or alternatively, metaphysical theories may just be regarded as immature scientific theories. But a so-called "theory" which is in principle untestable, i.e. which could never be tested, is not science at all, but "pseudoscience."

Any theory that is a good theory in effect says certain things cannot happen: it denies truth to certain statements. These statements are its potential falsifiers. If observation leads us to accept one of these statements as in fact true, then the hypothesis in question has been refuted.

Popper's view avoids the problems of foundationalism which confronted the confirmationist logic defended by the typical positivists. Because the positivists assumed a verifiability criterion of meaning, they had to explain how theoretical statements which could not be directly verified could be "reduced to" the observation statements which could be directly verified. In order to make such statements a secure foundation, positivists took the typical empiricist line that our awareness of our own sensations is "incorrigible" (one cannot be wrong about what one is sensing) and so incapable of being in error. Thus these directly verified statements were in effect about human sensations. Popper disdains this view as "psychologism" which is unhelpful in securing any firm foundation for scientific knowledge.

For Popper the "basic" statements which serve to refute conjectured hypotheses are simply statements the community of scientists have agreed to accept as intersubjectively reproducible. Whether these statements are regarded as "observations" of pointer readings or properties of measuring instruments, or the neural responses of human sense organs, is not to the point. Popper's view is that the decision of practicing scientists to accept certain statements as "basic" is -like the decision to accept any statement- a provisional decision reached by convention as agreed upon regarding what is an "observation." However, this does not imply that such basic statements are "incorrigible" in the empiricist notion of sense data reports. We can test so-called "basic statements" as much as any other by demanding they be reproducible. The "data" are not "fixed" or "given"; but are always eligible for revision in the light of further new information. To say it is "basic" is simply to say for the time being it is accepted and used to test some other higher level theoretical conjecture.

Popper's view also avoids the problem of theory-ladeness. Since the basic statements used to test any given theory are naturally going to be expressed in terms of the theory, Popper agrees that of course the observations are theory-laden, but he claims this fact does not impugn their ability to serve as tests of theory. For foundationalists who look to observational evidence for support of theories, the foundation must be regarded as fixed, unalterably "given" by sensory reports. In the foundational metaphor, it is "bedrock." Playing on that metaphor, Popper admits that the provisional acceptance of theory rests on what might be called a "foundation" of empirical evidence, but it isn't "bedrock," it is a "swamp." Our acceptance of theories which we erect upon this empirical evidence requires that we "sink piles" into this swamp, i.e., we must test empirical claims as severely as we test all claims in science. How far must we sink such piles? Popper answers, just far enough to support the structure we build on top. If we find the theoretical superstructure beginning to totter, we may well need to reconstruct the observational foundations. It is precisely because of the theory-ladeness of observational evidence that it is so "reconstructible." Thus unlike the positivists, Popper accords no fixity to the alleged empirical "foundation"; when it comes to the need for knowledge to grow no statement in science enjoys a "hands off" exemption from potentially being discarded.

In emphasizing falsification to the exclusion of confirmation, Popper often speaks as though the scientist must be his or her own theory's worst enemy. However, while his logical point that theories cannot be proved, but can be refuted is unassailable, his argument seems to fly in the face of a psychological fact about scientific theorizing. Inasmuch as he claims to be describing real historical science such psychological counterevidence cannot be ignored.

The psychological point here is that the creation and promotion of any significant scientific theory requires a considerable investment of time, money, genius, and often simply hard work. Given such a commitment, the natural psychological tendency on the part of those who advance a theory is to protect it as far as possible from critical attacks. No doubt part of this defense must include the ability to foresee at least some of the more obvious challenges that a theory is likely to meet. In order to avoid ridicule, the scientist must prepare to meet some such challenges before unveiling the theory to the professional community, and for this reason careful attention to tests which would possibly refute the theory must be taken into account. However, no reasonable scientists would expect a new theory to be entirely free from blemish and so would admit that any new theory will face problems which need to be worked on in the future development of that theory. Thus one would expect the judicious scientist to advance a theory by calling attention to its victories, potential refutations that did not occur, but the same time admit that there are problems and that these need time, talent, and money to be adequately addressed. Popper concedes that fledgling theories must be treated "leniently" in order to get off the ground, but once a theory has received its initial free "honeymoon," it becomes "open season" on that theory and the good scientist will be the one who attacks it with the most severe tests that can be devised.

Criticisms of Popperian Falisificationism:

Unfortunately the Popperian falsificationist model of science runs into a big problem with what is known as "holism." This is the argument that what is in fact "tested" by observational evidence (if anything is) are not individual laws or theories, but rather large constellations of belief which include not only the one "theory" that is allegedly undergoing empirical "testing," but also a whole array of "auxiliary hypotheses" the truth of which is more or less tacitly taken for granted. According to the thesis of holism, by suitable modification or buttressing of the proper auxiliary hypotheses, any theory can always be "saved" from potential refutation. Furthermore, it is often claimed that historical research shows that scientists do frequently do precisely this. Hence it would seem that, contra Popper, theories cannot be definitively refuted any more than they can be verified or proved.

Popper acknowledges the holist argument, but he claims that it is possible to arrange "crucial experiments" in which two rival hypotheses make incompatible predictions but are made to share all the same "auxiliary" hypotheses or "background knowledge." In such a case, since the two hypotheses share the same background beliefs, the one which makes a prediction that is falsified by observation will be definitively refuted while the one whose prediction is consistent with observation will be corroborated. Popper holds that the history of science provides examples of just such crucial experiments.  Critics of Popper will admit that while this may occasionally occur, one may reasonably expect that there will be other situations where the rival hypotheses are simply too different to share much common background assumptions and no such clear cut crucial experiment is possible.

Because Popper admits that the observational evidence itself is never fixed and only adopted as a "convention" for the purpose of testing theories, when faced with a disconfirming observation, it would seem that the scientist has a choice of whether to reject the theoretical hypothesis or the observational evidence. To protect thereby a favored hypothesis from potential refutation, one might propose additional auxiliary hypotheses which have the effect of nullifying the apparently negative observational evidence. Such hypotheses are referred to as "ad hoc" hypotheses.

Popper maintains that he can distinguish between modifications of hypotheses which are "permissible" from those which are purely ad hoc, and according to his methodology, prohibited. Permissible moves are those which render the whole conjunction of hypotheses and auxiliary assumptions more testable; those which do not lead to new testable consequences but function only to "save" the hypothesis supposedly under test are ad hoc and outlawed by the logic of science. The epistemologist essentially prescribes methodology to the scientist in this respect.
Unfortunately the distinction over which particular moves make the collection of assumptions more or less "testable" turns out to be difficult to draw. To make matters worse, Popper claims his view describes real historical science. Yet real historical science reveals over and over again that knowledge advances by protecting hypotheses from refutation by ad hoc modifications that Popper would forbid. Thus, ironically, just like the positivists whose ahistorical formalism he so disdained, Popper also has to (despite his claims to the contrary) "reconstruct" science to fit his claim of how it ought to have been.
Other problems result from Popper's claim to have "solved Hume's problem." It seems positively perverse to try to deny that the accumulation of observational evidence ever leads to the formation of hypotheses. Yet to the question of where does a hypothesis come from (the question of the context of discovery), Popper replies from the refutation of a prior hypothesis, not from the collection of observational evidence. Has Popper really found an account of science which eliminates induction?
Hillary Putnam presents a nice argument to show why Popper has not really gotten around the problem of induction. The testing of theories is not done from a disinterested motive inspired by an interest in testing and testing alone. Testing theories is a relevant thing to do because we value well tested theories. Why should we value well-tested theories, if not because we have found that theories which are well tested in the past have continued to bear up well in the present, and we expect they will continue to do so in the future. But to reason in this way is just to make an inductive inference. If we had no reason to believe that well-tested theories could be relied upon in the future, then the motive for all the "testing" would evaporate.

Unfortunately Popper's response was basically to stonewall all such criticisms and to staunchly never budge. Although Popper exalted the virtues of open-mindedness and searching for evidence against one's views, in fact he held his own views unswervingly, brooking no criticism and demanding absolute allegiance to his views amongst his followers. This ironic disparity between the philosophy he promoted and his own personal response to criticism was so well-known as to have become something of a cliche, but Popper did little to dispel it and he persisted in making exorbitant claims in behalf of the virtues of his own philosophy. Many subsequently famous philosophers of science, including Lakatos, Feyerabend, and Laudan, at one time or another studied philosophy of science under Popper, although they broke with the master when their views departed from the orthodox Popperian line.