Induction, Popper, and Kuhn

December 2008

Among the responses to Hume's problem of induction have been efforts to demonstrate its nonexistence. These responses, as summarized by Brian Skyrms, attempt to show that the kind of solution sought by Hume is not actually necessary. The two arguably most famous efforts during the 20th century were essayed by Karl Popper and Thomas Kuhn. Neither of them addresses the more fundamental question of what constitutes rational justification, though an evaluation of their arguments depends on an answer to that question. This essay will briefly discuss their arguments, comment on the difficulties of defining rationality, and propose a solution.

Hume's challenge is to justify our belief that the proper use of inductive reasoning will generate true conclusions from true premises most of the time. He acknowledges that reasonable people do, inescapably, have such a belief. In the essay where he presents the problem, he observes that "none but a fool or madman will ever pretend to dispute the authority of experience, or to reject that great guide of human life" ("Essays" 32). He remarks elsewhere in a similar context that "Nature, by an absolute and uncontrollable necessity has determined us to judge as well as to breathe and feel" ("Treatise"). Even so, he says, there seems to be no logical reasoning on which we can base our inductive judgments ("Essays" 32).

Skyrms outlines three responses intended to show that induction needs no such basis. They are: "(1) It asks us to turn induction into deduction. (2) Someone who doubts the rationality of accepting scientific inductive logic simply does not understand the words he is using. (3) It asks for a justification beyond the limits where justification makes sense." (43). Elaborating on the second, he says,

Examples could be multiplied to show that the use of scientific inductive logic is a standard of rationality, that part of what we mean by being rational is accepting scientific inductive logic. Thus the question "Why is it rational to accept scientific inductive logic?" is as silly as the question "Why was Genghis Khan's father male?" (44).

But, according to Skyrms, this defense of scientific induction simply turns it into a dogma. He also notes that we can still ask why rationality, even so defined, is superior to any alternative way of thinking, however the alternative be defined (47-48). "Thus," he claims, "we are back at the traditional problem of induction" (47). Furthermore, he says, even if we accept that induction is rational by definition in some sense,

a. A stronger type of justification would be valuable if we could find one.

b. No one has demonstrated that a stronger type of justification is impossible (49).

It seems uncontroversial that we should have the strongest justification we can find, but it is not obvious that Skyrms has correctly rated this particular one. The Genghis Khan analogy bears further scrutiny, and so does his reference to dogma.

The label father was invented for attachment to a brute fact, that each of us has two parents, one and only one of which is male. No one disputes that we have two parents or that they are one each, male and female. To assert that father means male parent is simply to report the observation that all English speakers have reached that particular agreement about what father means. But no similar agreement exists about the meaning of rational. We may stipulate that some kind of consensus exists, as reported by a standard dictionary such as American Heritage: "1. Having or exercising the ability to reason. 2. Of sound mind; sane. 3. Consistent with or based on reason; logical: rational behavior. See synonyms at logical" ("Rational"). This implies on its face that a rational person would, if nothing else, not believe a contradiction, but then what are we to think when a philosopher of some renown disagrees (Plantinga 146)?

One thing we might think is that the collective task of defining rational is not yet finished. If the question be whether induction has any rational justification, any proffered response will simply provoke further debate until "What do you mean by rational?" is considered by everyone to be as vacuous a question as "What do you mean by father?" A consensus on the essence of rationality might or might not be sufficient to entail a solution to Hume's problem, but it does seem at least necessary.

Even supposing it were incontrovertible that all rational people must, by definition, endorse inductive reasoning, there remains the charge of dogmatism to address. At this point, though, one might as well claim that it is dogmatic to say that no woman can be anybody's father. If (as may or may not be the case), as a matter of brute fact, what every English speaker means by rational includes something equivalent to endorsing inductive reasoning, then either it is not a dogma or else there is nothing necessarily wrong with being dogmatic. Which is the case will depend on how dogma itself is defined and just why we must avoid it at all costs, if that be so. Skryms tells us neither, but in common usage the philosophical toxicity of dogma is taken to lie in its epistemological immobility. Dogmas are understood to be incorrigible: Whoever believes a dogma qua dogma must have been convinced that it is impossible for the dogma to be false.

On one view, for at least some definitions, it arguably is the case that they indeed cannot be false. On another view, however, absent certain essentialist presuppositions, a definition as such is not so much true or false as it is more or less useful. This follows from the observation that words are defined by usage, not by fiat. The proposition A father is a male parent can be confirmed or falsified only insofar as it accurately states whatever meaning people intend to convey when they call someone a father. It is a useful definition if it does so, and otherwise if it does not. There is not obviously any other fact of the matter about any necessary or sufficient conditions for being a father.

The manifest utility of scientific induction creates a powerful intuition that there must be some good reason for doing it. The failure of philosophers collectively to agree on what that reason is might say something about the problem's essential intractability, or it might say much more about prevalent thinking within the philosophical community. The modern scientific enterprise was itself a product of that community ("Science" 2), and it is not unthinkable that philosophers might learn something useful from their progeny.

Popper and Kuhn are among those who have tried to do so. Each attempts to show that Hume's challenge is misdirected. For Popper, even though a scientist may be committed to finding the truth, the procedure of science itself is not actually about a striving for truth but about a winnowing out of falsehoods. For Kuhn, the scientific enterprise is a kind of evolutionary process working on the consensus of those engaged in the enterprise.

According to Popper, "Induction . . . is a myth. It is neither a psychological fact, nor a fact of ordinary life, nor one of scientific procedure . . . . The actual procedure of science is to operate with conjectures" (III). But the conjectures must be so formulated that some potential observation could prove them false. Any theory, he says, is scientific if it is falsifiable, and not otherwise (I). To the question of why a theory ought to be believed simply because it has not been proved false, Popper gives no clear answer beyond the observation that as a matter of fact, scientists do believe theories on that basis: "So long as a theory stands up to the severest tests we can design, it is accepted; if it does not, it is rejected" (III).

According to Kuhn, science does not really work that way, nor could it. He notes that there is never a perfect fit between theory and observation. "There is no such thing," he says, "as research without counterinstances" (79) and therefore, "If any and every failure to fit were ground for theory rejection, all theories ought to be rejected at all times" (146). This, together with several other difficulties he discusses throughout his book, suggests to Kuhn that we probably should give up the notion of science as an inherently truth-seeking enterprise (170). He suggests instead that theories compete with and succeed one another in a process analogous to natural selection (146). His thinking is effectively encapsulated by Steven Weinberg:

Kuhn recognizes that Maxwell's and Einstein's theories are better than those that preceded them, in the same way that mammals turned out to be better than dinosaurs at surviving the effects of comet impacts, but when new problems arise they will be replaced by new theories that are better at solving those problems, and so on, with no overall improvement.

There is no necessary connection between truth and solutions to problems because, Kuhn says, the problems themselves, and the characteristics of acceptable solutions, are defined by the paradigms under which they are investigated (35-42). Even the key terms in which the problems are stated, he says, are defined by the paradigm (149-50). There are thus no paradigm-independent rules according to which one can gauge how closely a theory might approach objective reality (3-4).

Popper (apparently) and Kuhn (manifestly) thus agree that nothing in the methodology of science guarantees even the probable truth of any scientific belief. Popper says the methodology of science is intended rather only to falsify beliefs, while Kuhn denies that there is any methodology at all for distinguishing truth from falsehood. Their two essays, though, are overall too dissimilar to allow either to be ranked over the other as characterizations of the scientific enterprise. Popper is attempting to define the essential difference between genuine science and other intellectual enterprises that claim-falsely, in his judgment-to be scientific. Kuhn is attempting to show, without reference to competing enterprises, that genuine science is neither what Popper says it is nor what most lay people who have never heard of Popper think it is.

Beyond that observation, the constraints of this assignment preclude a substantial comparative critique of Popper's and Kuhn's arguments. It may be noted, however, that Hume's challenge may be rephrased thus: "Why should anyone believe what science tells us is so?" And it may be noted that neither Popper nor Kuhn actually answers that question. We do believe, and according to Hume we can hardly do otherwise, but is there a good reason why we are so compelled?

As Kuhn notes, the traditional Western epistemology of Cartesian rationalism has failed to provide a rational justification for scientific induction, despite its unquestioned utility. (121). Whatever other merits there are in Weinberg's counterargument against Kuhn, there is a telling omission. Weinberg declares that "scientists like myself . . . think the task of science is to bring us closer and closer to objective truth" and that he thinks they are succeeding. But Kuhn never says otherwise. He says scientists cannot prove even on probabilistic grounds that they are in fact getting any closer, and Weinberg offers no evidence to the contrary other than his personal conviction that it is so.

But then, even some mathematicians, in a field that for most of Western history was itself the paradigm of justifiable certainty, have come around to thinking that their truths are ultimately founded on nothing more substantial than personal conviction (Kline 318-19). Whether this is grounds for epistemological despair or provides a clue to something more hopeful might depend largely on one's philosophical temperament. If epistemologists cannot reach a consensus, the worst that can be said is that nothing will have changed.

What epistemologists might learn from the history of mathematics is that perfect certainty is not only unattainable but unnecessary.1 Mathematicians seem to be doing quite as well without it as they ever did when they thought they had it. The reality that scientists think they are studying is obviously very different in many respects from mathematical reality, whatever that is, and the differences are hardly trivial. For one thing, science seems to be the exemplar of a coherentist epistemology, while mathematics is the epitome of foundationalism. But mathematics is also integral to the scientific method. This could suggest that some kind of melding of coherentism and foundationalism would produce a consensus epistemology that would, incidentally, successfully address Hume's challenge. It would do this by, among other things, establishing a definition of rationality that would command a more nearly universal assent than is currently the case.

American Heritage records three senses of the word rational. The Oxford English Dictionary resolves them further into five senses relevant to this context:

These are all consistent with the word's etymology, according to which it is derived from a Latin root meaning "to think." But someone whose beliefs are contrary to mine is, I may suppose, thinking no less than I am thinking. If I say his beliefs are irrational, then, I imply that there is something wrong with the way he thinks. On what grounds, though, do I say that my way is the right way?

We may begin by depersonalizing the question. It seems uncontroversial, on any account of human origins, that we are a social species. Each of us belongs to some community, and in modern nations one may identify with several communities of more or less overlapping membership. One of those communities will comprise people who tend to think alike and who know, or claim to know, a substantial body of propositions in common. We may call this an epistemic community, and it has been argued (e.g. by Everett 2001) that it is prima facie rational to assent to whatever is believed by the acknowledged authorities of that community and, by the same reasoning, irrational to disagree with it.2 It seems like a good place to start, even if we do not accept all the inferences that Everett draws from it. That is to say, it seems useful to accept that a person is rational if they believe what they are told to believe by the authorities of their epistemic community, but not necessarily irrational if they dissent.

Such a position would start to address the concerns of Feyerabend (55) and others critical of what they think is the intellectual tyranny of science. It should also, if more widely accepted than it now is, do much to civilize some debates between advocates of any number of competing worldviews.3 We naturally suppose that because we have arrived at our own beliefs rationally, our beliefs must be at least probably true, and so contrary beliefs must be at least probably wrong. Such thinking is unobjectionable. But, despite a superficial analysis of its logic, it does not follow that anyone who holds contrary beliefs is ipso facto being irrational.

A common element of the dictionary definitions is that rationality entails the proper use of one's cognitive faculties. We need not decide who should adjudicate what constitutes proper use to note that no one should be faulted, either morally or intellectually, for failing to use those faculties in ways of which they are incapable. A rational worldview, whatever else may characterize it, must accommodate any brute fact; and it is such a fact that a substantial fraction of the world's people, whose minds are working just fine by any other criterion, cannot believe certain assertions made by the scientific community. That community, as described by Kuhn, can be seen as just an instance of an epistemic community, where the relevant authority is vested not in any particular individuals but in the consensus of the community's members.

The fear that such epistemic ecumenism would license some kind of postmodernist intellectual anarchy is not without evidential support (Daniel 1999). If astrologers are just as rational as their detractors, or if belief in young-earth creationism is no less rational than belief in evolution, then surely the very concept of rationality is practically vacuous? No. The difficulty of demarcating rationality from irrationality lies in the presumption that certain particular conclusions (e.g. that the Genesis story of human origins is a historically factual account of real events) cannot be reached by rational thought. The difficulty can be avoided by defining rationality in a way that minimizes the set of propositions to which it is applicable.

If the set were empty, then rationality would indeed be a vacuous concept. Let us stipulate that it is not empty. But it is universally agreed that not all propositions belong to it, either. That is, we all concede that rational people can disagree about many things, and not only about matters of taste. There either was or was not a historical King Arthur, and there seems to be a consensus that rational historians can disagree about which was the case in fact.4 It is generally supposed that in such cases, the disagreement is allowable because of a paucity of relevant evidence - the uncontroverted facts are insufficient to logically compel a choice among competing hypotheses. But as Kuhn argues, there is no paradigm-independent way to determine either what constitutes relevant evidence or how a given fact must be interpreted relative to a candidate hypothesis. The scientific consensus that Darwinian evolution is supported by an incontrovertible mass of evidence thus presents no logical challenge for anyone whose thinking is dominated by a nonscientific paradigm - even if that paradigm tries to co-opt the terminology of mainstream science.

All this is understandably frustrating for those who would defend science against its detractors, but rationality simpliciter need not be our only intellectual desideratum. A major-league baseball superstar and a mediocre minor-leaguer are both playing baseball properly, but it does not follow that in every sense that matters, the minor-leaguer is just as good a ballplayer as the major-leaguer. Analogously, a person using his cognitive faculties properly can decide that creationism makes sense, but that alone does not put creationism on an intellectual par with evolution.

Another analogue suggests itself. The typical minor-leaguer agrees with everybody else that the typical major-leaguer plays baseball better than he does. Similarly, the deference given to science in at least a general way, even by adherents of competing worldviews, suggests a common awareness that science enjoys some kind of epistemological advantage over other worldviews. The reason for this deference is often said to be the manifest utilitarian success of the scientific method, but we can then ask why it has been so successful as to elicit the deference.5

This essay can barely begin to propose an answer, but it can do that much by citing Quine and Ullian's Web of Belief. In Chapter VI they propose five virtues that ought to characterize any of our beliefs: conservatism, modesty, simplicity, generality, and refutability. Although none is unique to science, it is commonly understood that they are indispensible to science and have something to do with the credibility of its pronouncements. To whatever extent other worldviews neglect them, to that extent they may be considered epistemologically suspect.

Other intellectual virtues may be seen as corollaries implied by Quine and Ullian's five, or else might be added to them. One such is a willingness to question authority, Everett to the contrary notwithstanding. What Everett's argument cogently establishes is not that it is irrational to dispute authority, but that in most instances a rational argument against an authority tends to be difficult for anyone but another authority in the same subject to construct. It is clearly not rational to argue, "Although X is an authority, no authority in fallible, so therefore I am justified in disagreeing with X." It is too often supposed that heretics are to be admired simply because they are heretics, never minding the quality of their arguments. Whenever heresy per se is virtuous, though, its virtue lies in its willingness to dispute authority, not in the disputing itself. A challenge to orthodoxy that is not well reasoned might provide its participants with some good intellectual calisthenics, but it otherwise it risks wasting everyone's time.

Challenging authorities is just a special instance of questioning assumptions. We tend to assume, not unreasonably, that authorities tend to know what they're talking about. We also assume, at all times, countless other things that we do not question and cannot question simply because, even if there were no other reason, life is too short. It does not follow that none of them should ever be questioned. Whatever the merit of "X is obviously true" as justification for believing X, "X is obviously true, and many vigorous attempts to falsify it have failed" has more merit.

Although pure foundationalism now seems unattainable, its methodology remains useful. A paradigm or worldview that is justified on coherentist grounds still uses assumptions. Foundationalism tried to find assumptions that could not be questioned. It turned out that there are none, in the sense that we can be absolutely certain of their truth. But if the denial of an assumption leads to intellectual chaos, then that is all the justification we need for accepting it. But it still needs to be shown that it will lead to chaos - and, as the history of Euclid's parallel axiom demonstrates, attempts to show that can be very productive. Occam's razor may advise us to discard unnecessary assumptions, but it doesn't tell us how to determine their necessity. Rational people can disagree, for just one instance, about how much can reasonably be assumed about the factual reliability of certain documents purporting to be records of Christianity's origins. This would have a bearing on any debate comparing the relative epistemological merits of Christian and secular worldviews.

The points of this essay may be summarized thus. Hume's challenge asked for a foundational grounding of science. The subsequent history of philosophy has given us reason to think that the challenge cannot be met. Popper tried to show that it did not have to be met, that the falsifiability of scientific hypotheses sufficed to justify belief in them pending actual falsification. Kuhn showed that Popper's model was itself falsified by the actual history of science - not that falsification has no role in science but that Popper greatly oversimplified its role. Science instead strives for a kind of coherence. It treats discrepancies between theory and observation, which would be falsifications under Popper's model, as puzzles to be solved within a current paradigm. Crises arise when the puzzles prove intractable, eventually leading to the creation of a new paradigm. Nothing about the process incorporates any epistemological tool for checking the congruence of a theory to actual truth. Scientists might be justified in assuming some congruence more or less, but they cannot prove it. Like members of any other epistemological community, they are rationally justified in believing what their worldview tells them to believe, and at some level, nobody needs any other justification.

The question then arises of whether, at some other level, science or any other worldview happens to be more justified than the others. Besides coherence, there are some intellectual criteria by which any worldview may be judged. The continuing popularity of many competing worldviews demonstrates that nothing like a universal judgment has been reached, but it is reasonable for those who would like an answer to keep looking for one. It is also reasonable for those who think they have found one to try persuading others to accept it, provided only that their arguments not assume that proponents of competing worldviews are presumptively irrational.


1. Epistemologists in general are, of course, not unaware of what has been going on in mathematics for the past two centuries or of its philosophical implications. Nevertheless, after reading many more journal articles and other writings on the theory of knowledge than most laymen ever see, the author cannot shake the suspicion that many epistemologists continue to hope that some pure version of empiricist foundationalism may yet be vindicated.

2. Everett accepts the implication that scientists are thus being irrational when they challenge any orthodoxy. This is excusable, he thinks, as a kind of epistemic altruism (38-39). While this notion has its appeal, the author will argue that it is more useful to construe rationality so as to make it either unnecessary or redundant.

3. Kuhn's paradigms are highly reminiscent of what the author has seen referred to, in other contexts, as worldviews. The terms will be used interchangeably in this essay unless a distinction seems necessary.

4. The existence of an early Medieval British warlord named Arthur is apparently undisputed. It is the historical connection between that man and Malory's King Arthur that seems to remain a live issue.

5. For the time being, we are not assuming that utilitarian success has anything necessarily to do with the generation of true propositions. Even so, if people believe a scientific assertion, it follows by definition that they think the assertion is true. What remains to be addressed, but not in this essay, is the way and extent to which they are justified in so thinking.

(This is a slightly revised version of an essay written for a class on the philosophy of science taught by Dr. Susan Finsen at California State University, San Bernardino.)


Daniel, Barbiero. "In Defense of Science." Issues in Science and Technology (Summer 1999). Accessed June 10, 2008. is_199907/ai_n8828838/pg_2.

Everett, Theodore J. "The Rationality of Science and the Rationality of Faith." The Journal of Philosophy 98.1 (January 2001) 19-42. Downloaded April 22, 2007, from JSTOR.

Feyerabend, Paul. "How to Defend Society against Science." In Introductory Readings in the Philosophy of Science. Ed. E. D. Klemke, Robert Hollinger, and David Wyss Rudge, 54-65. Amherst, NY: Prometheus, 1998.

Hume, David. July 2000. "Essays and Treatises on Several Subjects Vol. II . . . ." In Essays and Treatises on Several Subjects Vol. II Containing an Enquiry Into Human Understanding. . . . Ed. L. A. Selby-Bigge. London, 1777. Leeds Electronic Text Centre. University of Leeds. Accessed April 7, 2008 hume/ehu/ehupbsb.htm.

Hume, David. 2003. A Treatise of Human Nature. Project Gutenberg. Accessed March 8, 2008

Kline, Morris. Mathematics: The Loss of Certainty. New York: Oxford University Press, 1980.

Kuhn, Thomas S. The Structure of Scientific Revolutions. Chicago: University of Chicago Press, 1996.

Plantinga, Alvin. Warrant: The Current Debate. Oxford University Press: Oxford, 1993. Oxford Scholarship Online. Accessed September 1, 2005 oso/public/index.html.

Popper, Karl. "Science: Conjectures and Refutations." Accessed May 20, 2008

Quine, W. V., and J. S. Ullian. The Web of Belief. 1978. Accessed June 10, 2008

"Rational." American Heritage Dictionary of the English Language. Houghton Mifflin: Boston, 2000. Accessed April 7, 2008

"Rational," Oxford English Dictionary, 2nd Ed. Oxford University Press: 1989. OED Online. Accessed June 3, 2008

"Science, philosophy of." Encyclopaedia Britannica. 2008. Encyclopaedia Britannica Online. Accessed May 20, 2008 eb/article-271805.

Skyrms, Brian. "The Traditional Problem of Induction." In Choice and Chance: An Introduction to Inductive Logic, 22-51. Wadsworth, 1999.

Weinberg, Steven. "The Revolution That Didn't Happen." The New York Review of Books 45.15 (8 Oct. 1998). Accessed May 21, 2008

Back to site home

This page last updated on September 10, 2011.