Karl-Otto Apel, "The Problem of Philosophical Foundations in Light of a Transcendental Pragmatics of Language"

The Problem of Philosophical Foundations in Light of a Transcendental Pragmatics of Language

3. Does the Principle of Fallibilism Contradict the Presupposition of Indubitable Evidence?

The principle of fallibilism, to my knowledge first put forward by C. S. Peirce, represents an indispensable presupposition in the methodology of empirical science.³⁷ It is this presupposition that distinguishes empirical science from the philosophical science of German Idealism, from philosophical science in the sense of Plato, or even Husserl, from a science justified by essential insight and seeking “episteme.” But does not this distinction—as well as Albert's claimed insight into the difference between the search for evidence and the search for truth (or between criticism and justificatory rationalism) presuppose in turn a certain evidential, essential insight in the sense of philosophical knowledge?

I do not want to claim that this rhetorical question contains a definitive counterargument against Bartley's pancritical rationalism; rather, I am willing to grant that the principle of fallibilism—in a sense to be explained later—is to be applied even with regard to the insights of the formal sciences (logic and mathematics) and transcendental philosophy. In compensation for this admission, I would like to claim—also in a sense to be explained later—that evidence in the sense of indubitable certainty is methodologically indispensable for the empirical sciences as well. I want to clarify the significance of my claims through a discussion of Albert's dictum that one “can fundamentally doubt everything.”³⁷ The difficulty in this sentence, a sentence often casually pronounced by philosophers, is indicated by the historically remarkable circumstance that the founder of “fallibilism,” Peirce, polemicized against Descartes with the argument that one could not doubt everything, if the doubt was not to amount to a contentless “paper doubt.”³⁹ In empirical science a meaningful doubt presupposes, according to Peirce, that one does not doubt everything but rather proceeds from convictions that are taken as certain and assumed to be the standard both of what is to be doubted and of the new evidence considered possible in principle.

Quite similar arguments concerning the meaningfulness of doubting can be found in the later Wittgenstein.⁴⁰ We find in On Certainty, section 115, “Anyone who wanted to doubt everything would not get even as far as doubting. The game of doubt itself presupposes certainty.” In other words, doubt and thereby also criticism in Popper's and Albert's sense—is not explicable as a meaningful language game without in principle presupposing at the same time indubitable certainty. Wittgenstein generalized and radicalized this insight still further in section 114: “Whoever is certain about no facts, also cannot be certain of the meaning of his words.”⁴¹ In other words, every functioning language game (all agreement on meaning) presupposes that the communication partners, who must have learned the language at the same time as they acquired a well-established orientation toward the world, take numerous facts to be certain. In a real sense, convictions (be they principles or contingent facts) that are neither to be doubted nor to be changed function as “models” or “paradigms” of meaningful use.⁴² Thus, the conviction that the earth is a sphere that rotates on its axis and revolves around the sun is a “paradigm” for possible meaningful questions in, say, the language game of aeronautics and meteorology. The conviction that there is a real external world, “outside” of consciousness, is a “paradigm” for the language of critically questioning whether something is real or is only based on delusion, illusion, hallucination, or something similar.

It seems to follow from this that argumentation in everyday life and science must have recourse to evidence that is presupposed in the appropriate language game. Thus, “appeal to evidence” cannot, at least in this sense, be equated with “appeal to dogma” or “appeal to an arbitrary decision,” since criticism itself—as meaningful criticism in the framework of a language game—must be justified, at least virtually; that is, it too must in principle be based on “evidence.” To put it differently, criticism cannot somehow—as it appeared to Bartley and Albert—be made into a self-sufficient, ultimate stage of rational argumentation; criticism presupposes a transcendental-pragmatic framework (a meaningful language game) in which various possible attempts at justification and various possible critical arguments correspond to each other, at least in principle, through common appeal to “paradigmatic” evidence. This analysis reveals the essential structure of the institution of argumentation. Wittgenstein had something like this in mind when he wrote, “All examination, all weakening and strengthening of any claim whatsoever, occurs within a system. And such a system is not a more or less arbitrary and doubtful starting point of all our arguments, rather it belongs to the very essence of what we call an argument. This system is not so much a point of departure as it is the element in which arguments live.”⁴³

In light of the kind of argument about the meaningfulness of doubt that Peirce and Wittgenstein advanced, numerous imprudent or exaggerated claims of Bartley's and Albert's “pancritical rationalism” prove to be untenable. The simple equation of sufficient grounding through a return to evidence with appeal to a dogma or an arbitrary decision, and the proposal to put “the idea of a critical examination in the place of the idea of justification,” are two such claims. In fact, the language of the critical rationalists suggests, not infrequently, the misunderstanding of an anarchic criticism for criticism's sake, a critical reason without standards of criticism.

The discussion does not end here, since the point of critical rationalism does not yet seem to be fully grasped. It is now necessary to clarify the meaning of the principle of fallibilism, as formulated by Peirce.

In his discussion of Descartes, Peirce showed that one cannot doubt everything at once—say, the existence of a real outside world in toto;⁴⁴ rather, one can doubt virtually anything that is held to be certain—for instance, under the proper circumstances one might doubt the reality of each and every fact that is thought to pertain to the world outside of consciousness. This virtually universal doubt, for Peirce identical with the principle of fallibilism, also appears to be Albert's target when he writes, “A consistent criticism, which does not allow any dogma, necessarily involves a fallibilism with regard to every possible stage,” and “There is neither a solution of the problem nor an appropriate stage for the solution of certain problems, which must necessarily and from the start elude criticism.”⁴⁵

But how is this fallibilism reconcilable with Peirce's and Wittgenstein's critical arguments about meaningfulness, according to which every doubt and every criticism must in principle (that is, as a constituent of a meaningful scientific argumentation game) be justified through what is presupposed as indubitable evidence?

Peirce himself found it difficult to reconcile his fallibilism with his notion of certainty within pragmatism, in the sense of “critical common sense”; and he did not, I think, satisfactorily resolve this problem.⁴⁶ It seems to me that the two Peircian principles can be considered consistent if and only if a distinction is made between the level of reflection of prescientific and scientific language games, on the one hand, and the level of transcendental-pragmatic reflection on the structure of language games in general, on the other. (In my opinion, this is not a question of an arbitrarily repeatable distinction between levels of reflection in the sense of psychology, or even of the formal hierarchy of metalanguages in metalogic; it is rather a distinction that should unequivocally distinguish the implicitly self-referential claim to universality of philosophical statements from the individual or empirically general claims to validity of non philosophical statements.)⁴⁷

From the standpoint of philosophical reflection it may be said with regard to every language game, including the philosophical language game, that within its framework doubt and criticism are meaningful only under the presupposition that they can be sufficiently grounded by appeal to indubitable paradigmatic evidence. At the same time, it is also possible on this level of reflection to formulate a fallibilistic proviso, as a virtually universal doubt with regard to the paradigmatic evidence of all possible language games, except for the philosophical language game of doubt. Naturally, with this doubt all corresponding language games are made virtually nonfunctional (in the thought experiment to that effect). This is the case because every language game stands or falls (according to an insight of Wittgenstein later taken up by Thomas Kuhn in his analysis of “scientific revolutions”) with paradigmatic evidence. Nevertheless, this virtually universal, metascientific doubt is not a “paper doubt” in Peirce's sense. This is because the fallibilistic proviso does not claim to doubt a statement of empirical science for empirical reasons, but only opens, or holds open, the possibility of doing so. Merely opening or holding open the possibility of justified doubt—that is, of justified criticism—on the level of metascientific reflection is not contentless inasmuch as it justifies the methodological postulate of the virtually universal attempt at well-justified criticism.

It may well be said that this argument covers the relevant meaning of “fallibilism” in the sense of Peirce and Popper, who made it a principle of the philosophy of science. At the same time, however, my argument is consistent with the transcendental-pragmatic insight of Peirce and Wittgenstein to the effect that doubt and criticism within the framework of the game of argumentation always presuppose justification by actually indubitable evidence (and by the expectation of possible evidence!) as the condition of their possibility. We must, however, inquire into the reason why the principle of fallibilism, in the sense of a principle of virtually universal criticism, and the principle of sufficiently grounding doubt and criticism through appeal to evidence can be consistent. It is not at all self-evident, but rather, philosophically remarkable that on the one hand, any evidence at the basis of a scientific theory must in principle be open to doubt and criticism, while on the other hand, criticism must be sufficiently justified, in the sense that all doubt and criticism must end with appeal to indubitable evidence. A satisfactory answer to this question, in my opinion, requires no more and no less than an appropriate transcendental-pragmatic distinction and mediation between the epistemological philosophy of ultimate origins and the twentieth-century philosophy of linguistic analysis.

This much seems clear to me: If the modern epistemological version of the philosophy of ultimate origins (whether in the form of empiricism or of rationalism) is correct in its claim to have reduced the intersubjective validity of knowledge to (“my”) evidential consciousness, then it can hardly be understood how certain convictions may be doubted or criticized at all. If, on the other hand, the logic of science (oriented to the semantic analysis of sentences) is correct in its presupposition that sentences can only be justified by other sentences, while extralinguistic evidence of consciousness may only be considered as the source of external causal motives for the conventional formation of “basic statements,” then it is inexplicable that criticism always presupposes possible justification by evidence. A resolution of this dilemma is, in my opinion, possible with the (transcendental-pragmatic) presupposition that evidential consciousness and intersubjective validity of linguistically formulated arguments are, on the one hand, irreducible aspects of the idea of truth and, on the other hand, always, as such, peculiarly interwoven with each other in language games.

This argument has two consequences. First, contrary to the view of the modern theory of knowledge from Descartes to Husserl, evidential consciousness for me (be it evidence in the sense of empirical perception or in the sense of ideal or categorical intuition) cannot in principle be equated with the intersubjective validity of arguments. The reason for this lies in the mediating function of language, conceived as the transcendental condition of the possibility of an intersubjectively valid interpretation of the world, a function overlooked from Descartes to Husserl. It seems to be a consequence of this mediating function that, to the extent that perceptual judgments possess a communicable objective content, in the form of an assertion that interpretively transcends the subjective sense data supporting them, they underlie any possible criticism: criticism now means nothing other than a possible reinterpretation of the perceptual evidence, which is itself indubitable. Kant postulated prelinguistic forms of connection and schemata of “consciousness in general” to account for the objectivity and intersubjectivity of “experiential judgments” that a priori transcend merely subjective perceptual evidence; and the modern “genetic epistemology” of Piaget appears to confirm this postulate by means of empirical psychology. It must be pointed out, however, that the prelinguistic conditions of consciousness postulated by Kant as conditions of the possibility of the intersubjective validity of knowledge are not, as Kant himself knew, sufficient conditions for the intersubjective validity of the empirical knowledge of science; further conditions are necessary to account for the validity of the empirical propositions of science. In addition, from the viewpoint of a transcendental pragmatics of language, it must be supposed that even synthetic a priori statements, which for Kant and Husserl were also certain a priori (for instance, the axioms of Euclidian geometry or the Husserlian statements concerning the simultaneity of color and extension), can be given the status of intersubjectively valid principles of science only insofar as such statements, on the basis of tacit conventions, function as paradigmatic evidence for argumentation in specific language games.

By means of this distinction and connection between the epistemological and the linguistic-pragmatic viewpoints, it becomes possible to explain why the so-called crisis of modern physics could question the intersubjective validity of the theoretical principles of classical physics on the basis of a reinterpretation of experience through explanatorily more powerful theories—and could do so despite the recognition of certain a priori evident connections between representations, as subjective conditions of the possibility of primary experience (for instance, conceptual connections in the sense of Kant's “forms of intuition” and “schematized categories”). In my opinion, the transcendental pragmatics of language lead here to a conclusion that is contrary to the theory of evidence: the answer to the question concerning the intersubjective validity of knowledge cannot be given by appealing to epistemic evidence for the individual consciousness (or even a priori evidence for “consciousness in general”); rather, intersubjective validity requires postulating a consensus that is to be reached through argumentative discourse in the community of inquirers (Peirce, Royce).⁴⁸

This discussion of raising evidential consciousness to the level of paradigmatic evidence for language games shows, on the other hand, that the procedure of arriving at a consensus on the basis of argumentative discourse in the community of inquirers can in no way be understood without taking into account the appeal to epistemic evidence clarified by epistemology. Thus it is clear, for example, that the reinterpretations of our primary experience by means of explanatorily more powerful physical theories must in turn be sufficiently justified by appealing to evidence that is paradigmatic for these language games. As is the case with such scientific theories, this evidence need not have the character of direct, clear evidence of primary experience. Thus, for example, in the case of the Riemannian space presupposed by the general theory of relativity, one presupposes public paradigmatic evidence for a language game that is not evidence in the sense of ideal perceptual space. In this case, however, the empirical verification of the physical theory is carried out by means of measuring instruments, which for their part, in both their function and their manufacture, presuppose evidence in the sense of the perception of ideal space, which is paradigmatic in the “protophysical” language game of Euclidian geometry. This example, I believe, elucidates the a priori necessary connection between argumentation related to discourse and (sufficient justification by means of) appeal to epistemic evidence—a connection that is not considered in the semantically oriented logic of science. Although the evidential consciousness that is always mine does not guarantee the intersubjective validity of knowledge, still the argumentative redemption of claims to validity in a scientific language game must refer back ultimately to that evidence which can, in principle, ultimately be validated by every single member of the interpretation community in his or her (empirical or a priori) evidential consciousness.

Here, one should particularly note that the paradigmatic evidence, upon which Wittgenstein's criticism and doubt rest in the framework of a language game, is yet not identical with the originally experienced epistemic evidence, but rather, can and must directly refer back to conventions. Indeed, Wittgensteinian objections to Kant and Husserl are correct: without the mediation of such conventions, epistemic evidence could not function as paradigmatic evidence for language games. The conventions of paradigmatic evidence as such, however, can in no way be traced back to an arbitrary decision; rather, as evidence presented in argumentation, they must be justified, however indirectly—for example, in the empirical verification of theories that they support—by reference to that original (empirical or a priori) evidential consciousness, of which they attempt to give a convincing interpretation. From the point of view of a transcendental pragmatics of language, the fact that evidential consciousness achieves intersubjective validity only as publicly acknowledged language game paradigms shows that giving reasons in arguments necessarily leads back to appeals to epistemic evidence.

It is not yet clear, however, how the transcendental-pragmatic mediation between the philosophy of consciousness and analytic philosophy of language yields an argument in favor of philosophical foundations. Indeed, the metascientific grounding of the principle of fallibilism appears to have shown that all indubitable epistemic evidence must be looked upon as relative to certain language games that can in principle be transcended by means of critical reflection. Thus, it appears that at the philosophical level of a reflection upon validity, the principle of (progressive) criticism does have priority over the principle of sufficient justification through appeal to evidence. The evidence presupposed in special argumentative language games is to be considered in principle revisable, while permanent criticism, which may presuppose in every particular context an appeal to evidence, retains, it seems, the last word on the level of philosophical reflection, which transcends all particular language games.

At this point we should recall that the reason why criticism appears to retain the last word on the (metascientific) level of philosophical reflection is that there exists a philosophical language game in which the scope of all language games can from the outset be discussed, and with a claim to universal validity. (Wittgenstein sought to minimize this claim through the notion of simple family “resemblances” among [language] “games”;⁴⁹ and the main line of analytic philosophy of science, including Russell, Carnap, and Tarski, objected to the implicit self-referentiality of the universal validity claim of philosophical discourse—although since Russell's theory of types these objections could themselves be articulated with universal validity only at the cost of self-contradiction.)⁵⁰ With regard to the critical rationalism of Popper, however, it is indisputable that he can justify his claim to replace the postulate of sufficient reason by means of principle of criticism only by making a universal a priori validity claim in philosophical argumentation.

Here, however, we immediately face the prospect of a new problem of justification, involving an appeal to evidence that cannot be doubted and criticized—at least not in the same way as the paradigmatic evidence of those language games that could be seen by philosophy as revisable and to that extent could be transcended. Corresponding to the reasons that seemed to speak for the final priority of criticism—that is, to the fact that philosophical reflection can and must consider all paradigmatic evidence as in principle revisable—we now have the fact that the philosophical language game itself must be able to appeal to evidence that in principle is not identical with any of the empirically revisable language game paradigms. In this way we can argue for the priority of philosophical foundations over the principle of permanent criticism.

37. Cf. my edition of C. S. Peirce, Schriften I und II (Frankfurt, 1967 and 1970), subject index.

39. C. S. Peirce, Collected Papers, 5 (Cambridge, Massachusetts: Harvard University Press, 1933), §§ 265 and 376.

40. For steering me to the following Wittgenstein references, I am particularly indebted to an unpublished paper by Dieter Mans.

42. Cf., for example, L. Wittgenstein, Philosophical Investigations (New York: Macmillan, 1958; trans. Anscombe), § 50:

There is one thing of which one can say neither that it is one meter long nor that it is not one meter long, and that is the standard meter in Paris. But this is, of course, not to ascribe any extraordinary propertv to it, but only to mark its peculiar role in the language game of measuring with a meter rule. Let us imagine samples of color being preserved in Paris like the standard meter. We define: “sepia” means the color of the standard sepia which is kept there hermetically sealed. . . . We can put it like this: This sample is an instrument of the language used in ascriptions of color. . . . What looks as if it had to exist is part of the language. It is a paradigm in our language game; something with which comparison is made.

It is—we should like to say—not merely the picture of the behavior that plays a part in the language game with the words “he is in pain,” but also the picture of the pain. Or, not merely the paradigm of the behavior, but also that of the pain.

With clear reference to a priori certain convictions, is the following in Remarks on the Foundations of Mathematics (Oxford, 1956, p. 30ff; trans. Anscombe):

Whence comes the feeling that “White is lighter than black” expresses something about the essence of the two colors? . . . Is it not like this: the picture of a black and a white patch . . . serves us simultaneously as a paradigm of what we understand by “lighter” and darker and as a paradigm for white and for black . . . . That connection, a connection of the paradigms and the names, is set up in our language. And our proposition is nontemporal because it only expresses the connection of the words “white,” “black,” and “lighter” with a paradigm.

44. That one cannot meaningfully doubt the real external world in toto can be shown also from the point of view of the later Wittgenstein. One cannot with Descartes meaningfully argue that all that is supposed to be real is finally merely my dream (or merely in consciousness), since the expression “merely my dream” (or, merely in consciousness) is meaningful only within the framework of a language game in which it is presupposed as paradigmatic that not all that is supposed to be real is merely my dream or merely in consciousness.

46. On this, see my introduction to C. S. Peirce, Schriften I (Frankfurt, 1967), p. 123ff.

47. There can be no theory of reflection that is formalizable in the sense of the analytic logic of science (no “symbolic model” of reflection), as G. Frey has shown in opposition to the call for a total objectification of human consciousness and its corresponding cybernetic simulation. Cf. G. Frey, Sprache—Ausdruck des Bewusstseins (Stuttgart, 1965), p. 37ff., and “Sind bewusstseinsanaloge Maschinen Möglich?” in Studium Generale, vol. 19 (1966), pp. 191-200. Just this insight shows that we have transcendental-philosophical knowledge concerning the theoretical distinction between any imaginable level of the metalanguage hierarchy and the level of reflection of philosophical sentences—and this knowledge can be philosophically explicated. See Theodor Litt's explication of the “self-stratification” of mind and language in Denken und Sein (Stuttgart, 1948).

48. Ct. my essay, “Szientismus oder transzendentale Hermeneutik? Zur Frage nach dem Subjekt der Zeichen-Interpretation in der Semiotik des Pragmatismus,” in Hermeneutik und Dialektik, Festschrift für H. G. Gadamer, vol. I, ed. R. Bubner et al. (Tübingen, 1970), now also in Transformation der Philosophie, K.-O. Apel, vol. 2, pp. 178-219. See Habermas's explication of the “discourse” theory of truth in “Wahrheitstheorien,” in Wirklichkeit und Reflexion, Festschrift für W. Schulz (Pfullinger, 1974), pp. 211-265. See also my “C. S. Peirce and Post-Tarskian Truth,” in The Relevance of C. S. Peirce (Lasalle, 1983), pp. 189-223.

50. Cf. Max Planck, “Russell's Philosophv of Language,” in The Philosophy of Bertrand Russell, ed. P. A. Schilpp (Evanston, Illinois, 1944), pp. 227-255, as well as my remarks above concerning “paralinguistic” introductions of a philosophy understood according to the paradigm of constructive semantics.

Next