In response to:

The Romantic Rationalist from the December 2, 1982 issue

To the Editors:

Despite Jonathan Lieberson’s unsubstantiated summary of my In Pursuit of Truth (a Festschrift in honor of Karl Popper’s 80th birthday) as a series of “sugary and obsequious expressions of praise” in your December 2 issue, I found this and the first part of Mr. Lieberson’s two-part essay on Karl Popper’s philosophy to be a generally fair and reasonable attempt to explicate Popper’s work. Lieberson’s achievement, however, is unfortunately marred and nearly nullified by a conclusion that seriously misunderstands one of the central aspects of Popper’s philosophy.

Lieberson begins with an essentially accurate description of how Popper’s method of “falsification” or conjectures and refutations seeks to improve upon the traditional Baconian scientific method of induction or absorption of knowledge from mere repeated experience. As Hume and even Sextus Empiricus before him had seen, no amount of induction or positive repeated experience can ever verify or even support a general theory (for all of our repeated observations may merely be at the tip of an iceberg that runs counter to our general theory); but even one negative or counter experience can, as Popper emphasizes, serve to logically falsify or refute a general theory. Thus, no amount of repeated observations of white polar bears can prove or strengthen a theory that all polar bears are white (for we may from then on encounter nothing but black polar bears), but observation of even one black polar bear—assuming it is indeed a black polar bear—means our theory that all polar bears are white cannot be right. Lieberson then correctly points out, however, that Popper’s fallibilism is so pervasive as to lead Popper to assert that even observations of black or white polar bears are theory-impregnated (we identify the black object that we see as a polar bear rather than, say, a crow, because of theories that we hold about what polar bears look like, the constancies of species, etc.), and thus conjectural, uncertain, and eminently unprovable. How, then, Lieberson asks, may conjectures-and-refutations and its uncertainty be considered an improvement over induction and its problems? And why, recognizing the inconclusiveness of both, should we reject induction and rejoice in falsification? Since conclusive knowledge is not possible through Popper’s method of conjectures and refutations, Lieberson concludes that Popper’s hope for a non-inductive growth of knowledge is an impossible and thus misleading and dangerous ideal, a romantic “wild-goose chase.”

The problem that Lieberson raises—the conjectural nature of falsifying observations—is indeed profound, and one that most intelligent people almost always bring up on their first reading of Popper. Indeed, had Lieberson come upon his knowledge of Popper a priori, or from some casual discussion in a classroom, then the conclusions that Lieberson draws from the fallibility of falsifications would be entirely understandable. But the fact of the matter is that Popper himself has continuously raised, addressed, and dealt with this problem throughout his writings, going back to his first published work on scientific method, Logik der Forschung of 1935; and, I am obliged to add, this problem is similarly raised and dispatched with in at least four of the “sugary” essays in my volume. The situation is actually quite simple. We indeed must begin, as Popper does, with the recognition that all observations—whether used to falsify or “verify”—are themselves conjectural, and of no firmer epistemic import than the wildest, concocted abstract theory. We are then faced with a choice: do we use these uncertain, problematic observations to build knowledge inductively, or via a process of conjectures and refutations as suggested by Popper? Our decision might take into account the fact that induction is, quite independently of the uncertainty of all observations, logically untenable (as Hume had shown, there is no logical warrant that allows us to jump from even a huge number of specifics to a general theory), but that falsification, or the negation of generalities by specifics, is (as Popper and others have shown) quite logically acceptable as a process, even though the contents of that process (the observations) may be forever uncertain. Our choice would thus seem to amount to this: use conjectural, uncertain tools in an illogical process (induction), or use conjectural, uncertain tools in a logical process (falsification). Granting the obvious fact that neither choice can yield perfect or certain knowledge, which one would you choose, Mr. Lieberson?

But if we opt for conjectures and refutations as at least being logically possible, does not the uncertainty of the observations used as refutations condemn us to stagnate in our knowledge, to wallow in a perpetual state of conjecture? Is Lieberson’s characterization of Popperian method as a wild-goose chase appropriate after all? It is not—as a careful reading of Popper and, again, any one of a number of the contributions to my own In Pursuit of Truth makes clear. Indeed, discussions of how knowledge can progress and even flourish despite the endemic uncertainty of our cognition predate Popper by many years, and in Peirce we even find an implication that knowledge grows precisely because it is uncertain (see, for example, the Collected Papers of Charles Sanders Peirce, vol. 1, paragraphs 135-149; for extended discussions of Peirce on certainty and fallibilism, see any of Peter Skagestad’s recent writings).

From among the many arguments for the growth of knowledge in an uncertain world that Popper provides, let us look at but one—the biological or evolutionary analogy central to the field of “evolutionary epistemology,” which is where my own interest in Popper most lies. Assuming the general accuracy of the Darwinian model (but of course alert to its inevitable flaws), we notice three aspects of evolution that have pertinence to the possible growth of uncertain knowledge: (a) all organisms and organic adaptations are imperfect relative to their environments (i.e., they don’t always survive or succeed); (b) all organisms and adaptations appear to develop via a series of trial and error encounters with the environment, with organic characteristics initially generated or “proposed” independently of the environment, and then either eliminated or not by the environment; (c) on the basis of the first two processes, evolution or progressive change does indeed seem to occur, e.g., organisms seem to have developed from simple to complex, from non-intelligent to intelligent, etc., across time.

Now to the extent that the trial-and-error evolution of organisms seems descriptive of the conjectures-and-refutations growth of human ideas—and despite some obvious differences (for example, the important role of intentional rationality in the development of human knowledge), the two processes do seem to have much in common—we have in biological evolution an example of how progress can occur in a world utterly pervaded by, indeed constituted of, imperfection or uncertainty. In other words, if we accept the biological evolution of imperfect organisms as real, the growth of uncertain human knowledge through non-inductive conjectures and refutations seems possible: the nihilism that Lieberson imputes to Popper’s thoroughly conjectural method is unwarranted.

Of course, Darwin’s theory of evolution and for that matter the living world itself may be a chimera; reality and all our perceptions of it may be false or even non-existent. Popper’s philosophy does hold open such disturbing possibilities. But Popper’s philosophy also allows, more, encourages us to choose an alternative to the despair of nihilism and the illogic of inductivism, an alternative which seeks to parlay our uncertainty into a genuine, hard-won, painfully groping growth of knowledge. Granted that such a choice is something less than rational—I elsewhere call it “pre-rational”—but a choice and possibility it nonetheless is. It is just this golden egg of opportunity that Lieberson’s banishment of Popper’s wild geese would destroy.

Paul Levinson

Bronx, New York

Jonathan Lieberson replies:

Paul Levinson claims that the conclusion of my pieces on Popper displays a serious “misunderstanding” of “one of the central aspects of Popper’s philosophy,” namely Popper’s views on the nature and status of “falsifying observations.” But he does not accurately report my thesis: I did not say that since falsifying observation statements (not “falsifying observations”) are “fallible” or “conjectural” Popper’s theory of science falls to the ground. Nor did I claim that “since conclusive knowledge is not possible through Popper’s method of conjectures and refutations, his views are unacceptable. My difficulty, as I explicitly stated [NYR, December 2] was that a combination of views held by Popper render his alternative to inductionism (as contrasted with Baconian inductivism, which nearly all contemporary philosophers disagree with) a self-defeating and incoherent account of scientific inquiry and the growth of scientific knowledge. As such, I went on, it does not constitute a serious alternative to inductionism.

Thus, although I certainly discussed it, the problem of falsifying observation statements was not my main concern. I was aware that Popper has repeatedly discussed this problem, which Mr. Levinson believes is “one that most intelligent people almost always bring up on their first reading of Popper.” I was not aware, however, until I read Mr. Levinson’s letter, that it has been “dispatched with” in his collection of essays. Mr. Levinson claims that the “situation” with regard to falsifying observation statements is “actually quite simple”: all observations are conjectural, “of no firmer epistemic import than the wildest, concocted abstract theory.” Granting this point, he continues, we should clearly prefer the process of falsification to that of induction, which is “illogical.” I do not agree. While it is true that observation statements are, in a sense, “theory soaked” (as Popper says), not all the theories in which such statements are soaked are of equal merit, and not all observations are “of no firmer epistemic import than the wildest, concocted abstract theory.”

I wonder whether Mr. Levinson actually believes what he says; for my part, I have no difficulty in concluding that the claim that I am now seated before a typewriter is of far greater “epistemic import” than the abstract theory that the world is entirely made up of butter. I also hold, for reasons I set forth in my articles, that we do upon occasion possess perfectly good reasons for accepting such observation statements as true, a view Popper does not hold. Secondly, while we await an accurate codification of inductive practice—a task to which many philosophers, statisticians, and others have devoted their labors—I do not think that we can responsibly and without qualification claim that induction is “illogical.” That induction does not conform to the standards of deductive logic is obvious, but as I took pains to point out in my essay, there are no good reasons for regarding deductive standards of inference as establishing the standard of rationality in science. In short, I think I can answer the portentous question Mr. Levinson poses: granting that observation statements are not infallible, and that neither the methods of induction or of falsification can yield perfect knowledge, I continue to hold that induction is an activity—a “method” if you will—that we can in some circumstances rely on. It turns out, accordingly, that my alleged “misunderstanding” of Popper is no such thing, only disagreement.

I must add that the force of the evolutionary tale Mr. Levinson tells toward the end of his letter eludes me. Presumably it is an argument that is intended to contribute toward showing that the “nihilism” I impute to Popper, the view that his account of science describes a wild-goose chase (with respect to the aim of discovering the truth), is unwarranted. But does it do so? First of all, the argument depends upon an analogy that is seriously imperfect: the example of “obvious differences” between the “growth” of conjectures and refutations and the trial-and-error evolution of organisms that Mr. Levinson mentions is only one of many that could be presented—another would be the lack of analogy between the truth of a scientific statement and the adaptation of an organism to an uncertain environment.

Moreover, it is not clear to me that, even if we grant the analogy, the claim that imperfect organisms can develop through trial-and-error encounters with the environment into increasingly complex entities damages any of the points I made. The key issue, it seems to me, concerns “progress,” which in the case of science means making some advance toward the aim of discovering the truth about the world. After all, the whole process of evolution might yet be a nonprogressive affair, displaying only a temporary “progressive” character, as indeed some celebrated and dismal evolutionary speculations have suggested. As such, the analogy does not seem to me to support Mr. Levinson’s thesis that he has presented a good argument for “the growth of knowledge in an uncertain world.” When scientists speak of the growth of knowledge, they do not mean, I take it, simply a gradual increase in the complexity of their guesswork, or the increasingly successful adaptation of guesses to still other guesses. A parlor game or the process of creating myths and fairy tales, spurred on by problems of internal consistency, might exhibit this character; but while science might be an uncertain affair, wouldn’t this be a grossly exaggerated and perverse description of this uncertainty?

This Issue

September 29, 1983