Karl Popper is the author of a striking treatise on scientific method, The Logic of Scientific Discovery, as well as the celebrated wartime tract against totalitarianism notorious for its irreverent denunciations of Plato and Hegel, The Open Society and Its Enemies. He is an independent, versatile, lucid, and eloquent philosopher, among the most distinguished of contemporary thinkers who have undertaken the task—once a commonplace aspiration among philosophers but currently regarded by most of them as unduly ambitious—of constructing a rational critical system that would illuminate the entire range of human experience—science, art, morality, politics.

Though he has been much honored, his reputation has always been uncertain. Some—and not only philosophers, but scientists, politicians, artists—have professed to find unsurpassable wisdom in his works, while others, no less acute, regard the work as too blunt, oversimplified, audacious, disfigured by blunders and ungenerous presentations of opponents’ positions and by poor scholarship. For many, Popper’s work is the last expression of the neo-Kantian critical rationalism which flourished over a century ago, before skeptical intellects rendered it obsolete.

As Anthony O’Hear explains clearly, Popper has been for many years engaged to show through argument and illustration that a pervasive way of thinking about human knowledge should be replaced by what he regards as a more rational and coherent one. What in fact goes on when we come to know a mathematical proof or the chemical composition of a substance or the name of a neighbor? What procedures do we follow, if any? And what, after all, is human knowledge? These questions have led Popper, like so many other philosophers, to the somewhat abstract concerns of the “theory of knowledge” and particularly to the analysis of the methods and aims of science and scientific knowledge, the most successful and reliable knowledge we possess.

A prominent tradition in modern philosophy has held that all of human knowledge is founded or “based” on “experience.” As Popper describes one version of this view, the mind at birth is like an empty box. The box has windows or openings—the senses—through which information passes in the form of “ideas,” “atomic data,” “molecular experiences.” These items are pure, and “directly” perceived by us; they form the building blocks or “foundations” of all our knowledge. They “associate” with one another, giving rise to concepts (“swan,” “whiteness”) and to expectations (“All swans are white”), which are strengthened by repetition of conforming instances; even the most complex, abstract theories of modern physics could be shown by a patient genetic analysis to be “built up” from these humble beginnings.

It is no longer widely held that we arrive at hypotheses like “All swans are white” by “abstracting” them from elementary experiences, but nearly every modern philosophy of science has agreed with the box theory that repetition of certain instances somehow supports or confirms our hypotheses and raises the degree of rational confidence we may have for claiming them to be true.

This entire view of science, according to Popper, is misconceived. To him it suggests that scientists are engaged in an impossible “quest for certainty.” For one thing, it relies on a primitive psychology that supposes that there could be an infallible foundation in experience for human knowledge. Like the pragmatist John Dewey and his own teacher, the psychologist and educational theorist Karl Bühler, Popper believes that in our quest for knowledge there are simply no “secure” starting points that do not have presuppositions; such starting points can be found neither in a priori dogma nor in sense experience: we are, he says, never in a situation prior to all theorizing. Whenever we see or smell or listen, or indeed think at all, we are in a determinate situation comprising prior interests, theories, needs, aims, expectations; the “pure” elementary data of the box theory are figments plucked out of thin air.

Secondly, Popper thinks that induction is mythical. No one has encountered or inspected all possible polar bears, but judging from the sample we have come across, can’t we rationally claim that most polar bears are white? And can’t this claim—while of course it cannot ever amount to a proof—be strengthened by seeing more white polar bears? Do we not know generalizations (such as “Most polar bears are white”) based on observed instances that are extrapolated to cover unobserved instances in the past and future? And is it not perverse to deny that we rely constantly, in science and in daily life, upon propositions for which the available evidence is not logically conclusive? And perverse to deny that such nondemonstrative evidence can be graded for its probative force? Indeed, don’t men’s lives depend upon such discrimination in the law?

We have already seen that Popper denies that we generalize from “sensations” or pure “elementary experiences” to the truth of hypotheses like “All swans are white.” If by “induction” he meant merely this, most contemporary philosophers would agree with him; for there are no “pure” sensations, and even if there were, they couldn’t entail the truth or falsity of a hypothesis. But Popper rejects “induction”in a second sense as well: he denies that we can rationally generalize from any number of statements such as “This is a white swan” to claims about the truth or probability of hypotheses such as “All swans are white.” Like David Hume, Popper argues that induction (in this latter sense) is not a logically reputable inference: a hundred or a million observed white polar bears provide no decisive reasons for thinking that all polar bears are white; good reasons could only be found by inspecting all polar bears that have existed, exist, or will exist and concluding that each of them is white. There is no “justification” of induction, for all attempted ones end up in an infinite regress or in some a priori dogma. We cannot be sure the future will resemble the past. No matter how many times you have witnessed the rising of the sun, you are not entitled to assume rationally that it will, or even probably will, rise tomorrow; it might explode tonight, or melt, or, for that matter, burst into song. This much can be found in Hume; but unlike Hume, Popper does not claim that induction must therefore be a non-rational habit or custom. He argues the startling and provocative position that induction does not exist and has never been used. We do follow something like a “method” in acquiring knowledge, he says, but it is quite different from induction.


Popper is especially concerned to point out the flaws of the box theory and the belief in induction in the case of science. No scientific theory, he claims, not even the greatest of them, Newton’s universal mechanics, has ever been “established” or “verified”: after all, if Newton’s theory was certain or “inductively proved,” how could it have been over-thrown and superseded by Einstein’s theory of relativity? The simple fact, says Popper, is that we cannot “justify” a claim to the truth of any empirical hypothesis. Even trite statements like “This is a glass of water” or “This is a polar bear” are unverifiable; indeed, according to Popper, no positive reasons can be given for their truth.

Why is this so? The reason is that the words that appear in them, such as “glass” or “water,” are in Popper’s view “universals”—they depend on definitions that apply to all glasses and all water—and “cannot be correlated with any specific sense-experience. (An ‘immediate experience’ is only once ‘immediately given’; it is unique).”1 Moreover, these universal terms denote “law-like behavior” on the part of the objects they describe, and so “transcend experience”: to call something a bear, for example, implies many things that go far beyond what we actually observe, e.g., that it will die, that it will not grow wings, that it was created through sexual reproduction. For this reason, ordinary “observation” statements like “This is a bear” are “soaked in theory” and possess no privileged status as “final” or “ultimate” or as “more certain” than other statements.

Indeed, according to Popper, their epistemic status is no different from that of universal statements like “All polar bears are white,” although they are logically simpler. Nevertheless, even though the problem of showing one single object to be a polar bear is logically as difficult as showing that all are, Popper asserts that the simpler statements of observation are “easier to test” and might be more likely to be agreed on than other statements. But this is not because experience proves them; as he frankly acknowledges, experience can cause, or “motivate,” us to “accept” or “reject” such statements, but this “acceptance” or “rejection” is logically a “free decision” or “convention.” Nothing in life or logic compels us to accept it. So, in effect, according to Popper’s view, all empirical hypotheses—whether that copper conducts electricity or that this is a pencil—are “conjectural” in the radical sense that we shall never be able to have rational confidence that they are true, let alone “know” them for certain or with probability. Whatever knowledge we have is permanently “fallible” and “conjectural,” although most of us would agree to accept some conjectures as more credible than others.

But if this is so, how can we go about trying to secure human knowledge? Popper thinks he has a more rational and coherent answer than “inductivism.” We cannot justify a claim that a hypothesis is true, but we can retain both rationality and the empiricist’s demand that our knowledge be supported by observation. For while no number of white polar bears could establish or verify the claims that all are white, nevertheless a single polar bear that is not white can falsify the hypothesis. This inference, he says, is rational (unlike induction) and, moreover, is “based” upon observation—for people might come to “agree” that a particular black polar bear exists.


Popper’s description of the “logic” of the natural and social sciences rests upon this logical insight. We need not be disturbed by the loss of induction and the foundations of the box theory, he says: if we wish to seek the truth, then the only rational way of doing so is to do it indirectly. We cannot give any positive reasons for holding our theories to be true or probable. But we can replace justification—the practice of producing valid reasons in favor of the truth of our theories—by rational criticism—the practice of producing valid reasons against them. In science, instead of trying to show how right we are, we should forget about starting points and try to improve our current stock of hypotheses, whatever its source. First of all, we should make our ideas logically falsifiable or capable of being criticized. This is something like an intellectual duty of the scientist—a duty, Popper claims, that was not fulfilled by those thinkers like Freud and some followers of Marx who claimed to be “scientific” and yet failed to lay down in advance the conditions under which they would give up such theories as that all dreams are “wish-fulfillments” or that the collapse of capitalism will occur. Secondly, we should systematically search for errors in our theories. We should jump to bold conclusions and then ruthlessly try to overthrow them by tests; this is the rational way by which we might, with luck, learn from our mistakes and draw nearer to the truth.

We will, of course, never be able to be sure we have found the truth. All that we will be able to do is to claim that our theories have so far not been falsified by severe tests, that they have withstood our efforts to overthrow them. In other words, the best we can hope for is that we find not positive but critical reasons for thinking that our ideas are not true but “preferable” to other theories in the search for truth. Popper believes that it is on the basis of such critical reasons that we are convinced, for example, that the Copernican model of the solar system has more truth to it than, say, Aristotle’s: our reasons for this conviction, he writes,

consist in the story of the critical discussion, including the critical evaluation of observations, of all the theories of the solar system since Anaximander, not overlooking Heraclitus’ hypothesis that a new sun was born every day, or the cosmologies of Democritus, Plato, Aristotle, Aristarchus, and Ptolemy. It was not so much the accumulation of observations by Tycho as the critical rejection of many conjectures by Kepler, Descartes, and others, culminating in Newton’s mechanics and its subsequent critical examination, which ultimately persuaded everybody that a great step had been made towards the truth. This persuasion, this belief, this preference, is reasonable because it is based upon the result of the present state of critical discussion; and a preference for a theory may be called “reasonable” if it is arguable, and if it withstands searching critical argument—ingenious attempts to show it is not true, or not nearer to the truth than its competitors.2

This picture of science as a matter of conjectures and refutations has not been without critics of its own. Falsification as a logical technique among others in science was known to some medieval writers and was stressed at the beginning of our century by the American pragmatist C.S. Peirce. But critics complain that Popper not only places too stringent demands on scientists in making their hypotheses falsifiable, but vastly exaggerates the role of falsification in scientific inquiry. He requires that scientists formulate their ideas so that potentially refuting instances be specified in advance of testing them; and further, that should the theory be contradicted by results of tests, it should not be rescued by ad hoc hypotheses that do nothing more than account for discrepancies and make no new predictions.

But critics of Popper, such as O’Hear and T.S. Kuhn, have suggested that many great theories in science, such as Newton’s theory of universal gravitation, were not falsifiable in this sense. For one thing, the interpreters of these theories frequently referred to such “ideal” or “limiting” conceptions as “frictionless surfaces,” and “perfectly rigid bodies,” which are only joined to the materials of actual experiments through long chains of reasoning and subsidiary hypotheses, so that the status of experimental evidence as confirming or refuting such theories is inherently vague. The adverse outcome of tests need not be taken as falsifying them, but rather as clarifying their scope of application.

Secondly, most important scientific theories are flexible systems of assumptions compatible with a variety of specific formulations: Newton’s theory, for example, does not predict observational results unless it is joined with a large number of other assumptions, themselves modifiable; we cannot therefore regard a false prediction as decisively overthrowing the theory—as opposed to one of these auxiliary assumptions. If the planet Jupiter were tomorrow to assume a square orbit, we most likely would not condemn Newtonian theory before postulating that some unusual interfering force created the deviation from Jupiter’s predicted orbit.

Finally, critics say that the historical behavior of scientists has rarely conformed to Popper’s prescribed method; when theories have gotten into trouble—as Newton’s “celestial mechanics” did on more than one occasion—scientists have often invoked ad hoc hypotheses to rescue them. As O’Hear notes, the physicist Pauli “postulated the existence of the neutrino simply to preserve energy conservation in the theory of radioactive nuclear disintegration, long before any test was envisaged or there was any theoretical basis for it.”

Was it necessarily “wrong” or “irrational” for scientists to behave this way? Such thinkers as Kuhn and Popper’s close colleague, the late Imre Lakatos, have claimed it was not: in their view, Popper incorrectly focuses upon the testing of individual theories, whereas in fact what is critical in science is the development of some fundamental insight or “paradigm”—that all things are composed of atoms, that light is a wave phenomenon. The improved expression of such an insight in a series of theories is what constitutes most scientific research. Such theories are rarely thought by scientists to be the whole truth in any case—they are called “promising” or “worthy of attention”—so that to try to refute them may be gratuitous.3

What testing accomplishes, these writers claim, is to assist us in locating the shortcomings of our ideas and to help us to improve them, so that it may be perfectly “rational” for scientists occasionally to dogmatically set aside apparent falsifications as “anomalies.” If they did not, the merits of their ideas might never be discovered. Popper’s exclusive stress on negative arguments, “counterexamples,” and destructive criticism would eliminate, in practice, both the leniency and the useful dogmatism that characterize science. Popper’s most ferocious critics, such as his one-time disciple Paul K. Feyerabend, go much further. They not only argue that most celebrated scientific theories have never been “falsifiable” or “falsified” in the sense Popper prescribes. They also insist that his “criterion” of falsifiability demarcating science from nonscience is just a logical toy, one more instance of the depressing attempt on the part of so-called philosophers of science to squeeze and warp the teeming variety of scientific attitudes and inquiries—for example, the manner in which they appraise evidence—into severe logical calculi and formal “rules.”4

What is Popper’s response to these criticisms? As we shall see in a second article in the next issue, he has altered few of the philosophical tenets he first formulated in his twenties and thirties, and has instead vigorously redefended his position and prodigiously extended and generalized his views in a metaphysical world picture unexpected from a philosopher widely thought to confine himself to the narrow ambit of epistemology and the logic of science.

(This is the first of two articles on Karl Popper.)

This Issue

November 18, 1982