Panentheism.  Revisionism.  Anarchocapitalism.


Karl-Otto Apel


Essays by Me

Essays by Others

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

Karl-Otto Apel

2. A Critical Reconstruction of the Münchhausen Trilemma

What does philosophical tradition really say about the problem of foundations?  The problem has arisen repeatedly, ever since antiquity, in connection with the impossibility of a logico-mathematical (apodictic-deductive) derivation of the fundamental principles, or “axioms,” of logico-mathematical thought and, thereby, of all the demonstrative sciences.13  Put bluntly: ever since the time of Aristotle, the problem of foundations has been made a problem of philosophical significance precisely through the fact that logico-mathematical arguments can justify neither the truth of their own premises nor the validity of their rules of proof, but rather can only check “the transfer of the positive truth value, truth, from the set of premises to the conclusion and, in the opposite direction, the transfer of the negative truth value, falsehood, from the conclusion to the set of premises.”14  Since Descartes, the Aristotelian comprehension of axioms as intuited fundamental principles that are neither provable nor in need of proof15 has been radicalized by taking evidence (or evidentness) as the requirement of philosophical foundations.16  It is already clear that as long as the problem of philosophical foundations is conceived in traditional terms, it cannot be a matter of formal logic.

Albert at first seems to recognize this.  For he does not understand Leibniz's “principle of sufficient reason” as it was understood in older logic textbooks, that is, as the most fundamental principle of thought, as an “axiom of logic.”  Rather, it becomes a “general postulate of the classical methodology of rational thought”: it is understood as a “methodological principle” which presupposes that “the intelligibility of reality is connected with the determinateness of truth.”17  (As a matter of fact, the foundationalism of classical modern rationalism corresponds, in my opinion, to a subordination of logic—and of the ontological correspondence theory of truth—to the quest for evidence; epistemology is given the status of prima philosophia.  This subordination of logic and ontology to evidence as the basic principle of the theory of knowledge is expressed most radically in the phenomenology of consciousness developed by Brentano and Husserl.)

In his treatment of the Münchhausen trilemma, however, Albert starts from the point of view of modern logic, invoking the authority of Popper and Carnap.18  He gives the impression that he could explain the aporias of the rationalist postulate of philosophical foundations by a trilemma derived by formal logic alone, that is, by a trilemma that is in fact derived only on the condition that philosophical foundations be purely deductive.  This condition leads to the alternatives: (1) infinite regress, (2) logical circle, and (3) ungrounded breaking off of the process of giving reasons.19

Now, whatever Albert's intention may have been, a critical reconstruction of his argument against classical rationalism must, in my opinion, make the following clear: no argument against the evidence postulate of classical rationalism is directly connected with the third alternative of the trilemma as derived by formal logical means.  Rather, the trilemma can itself be understood as an explication of the problematic of axioms that Aristotle pointed out and that raised the problem of philosophical foundations in the first place.  (If, with David Hilbert, we reduce the problem of the truth of the axioms of logic and mathematics to the problem of the absence of contradictions in “axiomatic systems,” there results—corresponding to the Münchhausen trilemma—a metalogical or metamathematical aporia of the philosophical foundations of deduction itself, as Cadel, Church, and others have shown.)20  Already this much is clear: unlike the logico-mathematical (and metalogical and metamathematical) problem of philosophical foundations, the modern principle of sufficient reason, as far as it requires an appeal to evidence, is from the start an epistemological principle—a principle that, to put it in a modern idiom, involves the pragmatic dimension of evidence for a knowing subject.

In our context this means that it would be legitimate to trace the aporia of philosophical foundations back to the third horn of the Münchhausen trilemma only if it could be proven that making evidence into a postulate is completely meaningless that it, in effect, implies the replacement of the search for truth by an arbitrary decision.  However, the required demonstration of the pointlessness of the evidence postulate cannot, in principle, be accomplished by formal logical means alone.  How then can the demonstration be accomplished?  Must not such a demonstration itself assume paradoxically that appeal to “evidence” is not an arbitrary decision, but rather indispensable to philosophical argumentation?

In order to avoid misunderstandings, I shall at this point make clear the strategy of my argumentation.  In what follows I by no means wish to defend the position of classical rationalism that—in the sense of the Cartesian primacy of the theory of knowledge as a theory of consciousness—reduces the search for truth to the search for evidence.  I do not, therefore, want to defend an empiricist or rationalist “philosophy of ultimate origins,”21 a theory of knowledge that is supposed to be “a solution of the problem of origins and of validity all at once.”22  Such a strategy seems to me unpromising because epistemic evidence as such, however indispensable, is restricted to the evidential consciousness that has it.  Traditional theory of knowledge, qua theory of consciousness, cannot show from its own conceptual resources how epistemic evidence—that is, the evidence accompanying judgments regarding conceptual syntheses of the ideas in some individual consciousness—can be carried over to the intersubjective validity of linguistically formulated statements.  The goal of Popper and his followers, viz., intersubjectively valid statements, seems to me to be the proper methodological aim of the scientific-philosophical search for truth.23  I completely agree with Popper and Albert that the “evidence” of convictions for a particular consciousness is not sufficient for the truth of statements.  Beyond this, however, and quite contrary to Popper and his school, I shall argue that the fact that only the critical discourse of scientists can decide the intersubjective validity of scientific results has consequences for the theory of truth.  In my opinion, the customary move in logical empiricism by which the linguistically mediated problematic of the intersubjective validity of statements is reduced to a (syntactic-semantical) logic of science, and the problems of traditional theory of knowledge are banished into psychology, just misunderstands the problem.

Albert seems to be of the same opinion, since in his discussion of the character of critical methodology he rightly rejects the reduction of the theory of science to an “application (or even a part) of formal logic, including the relevant elements of mathematics, even, in the best case, including elements of the semantics of artificial languages.”24  In light of “the contemporary distinction between syntax, semantics, and pragmatics,” Albert calls for a consideration of “the epistemological relevance of pragmatics”25—of the linguistic and extralinguistic states of affairs that constitute the context of problematic statements.  The pragmatic context includes, according to Albert, “those states of affairs which are the referents of the statements about them” and also “those states of affairs which make up the context of human epistemic activities—that is, not only the isolated activities of reflection and observation by single individuals, but also critical discussion as a model of social interaction and those institutions that support or weaken, encourage or discourage, critical discussion.”26  With good reason Albert draws the conclusion that his “criticism of the classical theory of knowledge” and the necessity, which he derives from this criticism, for a “choice between the principle of sufficient reason and the principle of critical examination” are matters that are to be dealt with “under the rubric of pragmatics.”27

I would like not only to support this evaluation of the problem but moreover to take it seriously, since I understand the pragmatic conditions of the possibility of scientific knowledge, at least in part, as Kant did: as conditions of the possibility of intersubjectively valid knowledge and (scientific and philosophical) critique of knowledge—quite unlike Carnap and Hempel, who understood pragmatic conditions merely as the empirical sociological or psychological contexts irrelevant to the validity of knowledge.  My assessment of pragmatics must be correct at least to the extent that the conflict, which occurs in “pragmatic realism,” “between the principle of sufficient reason and the principle of critical examination”—whether or not it implies a decision between alternatives—is concerned with the conditions of the validity of scientific knowledge.  I would like therefore to propose, as the philosophical extension of logical syntax and the semantics of ideal scientific languages, a transcendental pragmatics of language concerned with reflection on the subjective-intersubjective conditions of the possibility of linguistically formulated and, as such, intersubjectively valid knowledge.  Here I shall attempt briefly to summarize the main lines of a transcendental semiotics—a transcendental-pragmatic reconstruction and extension of the foundations of the logic of language and science—that I have developed elsewhere.28

The possibility and necessity of a transcendental-pragmatic approach or method of inquiry is, in my opinion, demonstrable in a radical way by reflecting on the conditions of the possibility and the intersubjective validity of logical syntax and semantics themselves.  As C. S. Peirce recognized, it is a logical implication of the three-dimensionality of the sign function, and thereby of sign-mediated knowledge and argumentation, that the intralinguistic (syntactic) sign functions and the reality-related (referential semantic) sign functions presuppose a (pragmatic) interpretation of the signs by a community of interpretation.29  This presupposition obviously also applies to the corresponding semiotic disciplines; logical syntax and semantics are, as abstractive subdisciplines of semiotics, only a means of “indirect” (that is, mediated through the construction of ideal systems of rules) elucidation of scientific-theoretical argumentation.30  Hence, they are in principle dependent upon their extension and integration in a pragmatics of argumentation.  This, however, means that pragmatics is the philosophical discipline that deals with the subjective-intersubjective conditions of understanding meaning and of the formation of consensus in the ideal, unlimited, community of inquirers.  Peirce already essentially conceived of such a semiotic transformation of the Critique of Pure Reason, in the sense of a “normative” semiotic logic of inquiry.31

On the one hand, Morris and Carnap accepted Peirce's foundations of semiotics, in the sense of the three-dimensionality of the sign function (“semiosis”) and of the science of signs (“semiotics”); but—apparently because of the alleged impossibility of a noncontradictory self-reflection on the actual subjective conditions of sign interpretations32—they declared the pragmatic sign dimension to be the object of an empirical (behavioristic) discipline for which, at best, we might supply semantic conceptual explications in the form of a constructive, “pure, theoretical pragmatics.”  Whatever one thinks of the possibility of such a treatment of the pragmatics of language,33 it is certain that the “conventions” that, according to Carnap, underlie the construction of formalizable syntactic-semantic systems of rules—and, to that extent, also underlie the construction of semantic explications of empirical-pragmatic concepts—cannot be thematized philosophically in this way.  The normatively relevant conventions, which alone make possible the conceptual explications in a formal language necessary for a theoretical pragmatics, cannot themselves be made the object of such a pragmatics.  Hence, the theoretical pragmatics that Carnap had in mind—and that he semanticized in a rather a priori fashion—cannot replace the methodological arguments that Popper and Albert find essential.  In view of the contemporary demand for a semiotic transformation of transcendental philosophy, and in view of the fact that the presuppositions of modern constructive theories of language have not been rationally reflected upon, one might characterize the function of transcendental pragmatics for the philosophy of science as having to reflect on the conditions of the possibility and validity of conventions.  A tacit substitute for such reflection within linguistic analysis can be found in Carnap's provisional, ordinary-language “introductions,” which are—because of their use of implicitly self-referential “universally quantified propositions”—strictly speaking, expressed in an officially unlegitimizable “paralanguage.”  Here we find, in my opinion, the heritage of Wittgenstein's image of the ladder in the Tractatus.  The problem for constructive semantics captured in this image cannot be overcome until we accept the transcendental pragmatics of language as an unformalizable fundamental metadiscipline.

In the framework of the present investigation, I would like to support this approach by examining the unavoidable question concerning the conditions of the possibility of intersubjectively valid criticism.  I shall attempt to reconstruct and critically examine Albert's criticism of the classical postulate of sufficient reason from the point of view of transcendental pragmatics.

In this context I would first of all point out that the so-called Münchhausen trilemma facing any philosophical foundations can be logically derived only for sentences of an axiomatized system of propositions, in the sense of the syntactic-semantical construction of a so-called formal language.  That is, such a logical derivation is only possible under prior abstraction from the pragmatic dimension of argumentative language use.  To put it another way, only when one abstracts from the situation of the perceiving and argumentatively engaged subject, who offers his doubts and convictions for discussion in performatively explicable statements, is it possible to characterize the (deductively mediated) appeal to evidence as breaking off the process of giving reasons and to consider this presumed suspension, along with infinite regress and logical circularity, as the third horn of the trilemma.  For only from the viewpoint of syntactic-semantic abstraction, which cannot anchor language and knowledge to the lifeworld through objective or subjective (personal) deixis, can the meaning of the process of giving reasons be understood as a deduction of sentences (about states of affairs) from sentences (about states of affairs) that in principle cannot be broken off.  From the point of view of transcendental pragmatics, the logical process by which sentences are deduced from sentences—indeed, all “axiomatics”—can only be considered as an objectifiable means within the context of the argumentative grounding of statements through epistemic evidence.  (In this sense Aristotle's “apodictic logic” is in fact an “organon” of argumentative discourse—no more, no less.)  That is, the logical deduction of sentences from sentences is not itself the justification of the validity of knowledge—such an absolutization of the logical organon would in fact lead the problem of justification back to the Münchhausen trilemma—but is merely a mediating moment in the argumentative process of giving reasons, a moment that is indeed marked by a priori intersubjective evidence.

Corresponding to this is the following important distinction, which has been characteristically overlooked not only by logical empiricists but also—at least in The Logic of Scientific Discovery—by Popper.  Only when one illegitimately abstracts, in the sense of an “abstractive fallacy,” from the transcendental-pragmatic interpretative function of the subject of knowledge and argumentation, thereby reducing it to an object for empirical psychology, is it possible to maintain that sentences can only by justified by other sentences and that the so-called observation sentences or basic sentences are merely motivated by the experiential evidence of the knowing subject, in the sense of causation.34  Against this the transcendental-pragmatic position takes the point of view of the argumentative knowing subject and attempts, not to explain (from the outside) his “behavior” in formulating sentences, but rather to understand it (from within): hence it must necessarily conceive of epistemic evidence as a reason for formulating observation sentences or basic sentences, although not as a reason from which these sentences might somehow be logically deduced.

It is by no means implied that epistemic evidence—for example, perceptions or ideal (categorical) intuitions—is to be thought of as an unquestionable and sufficient, linguistically independent (that is, prelinguistic and intuitive) basis for the meaning and truth of scientific statements or systems of statements (“theories”).  Such a view corresponds to the modern epistemological (empiricist or intellectualistic) philosophy of primordial origins, which I do not wish to defend, as I have already mentioned.  In my opinion, by virtue of the “propositional acts” (the identifying “acts of reference and predication”)35 upon which the formation of judgments depends, epistemic evidence is interwoven from the outset with language use and the capacities of the knowing subjects—in the sense of the interweaving of knowledge, language use, and activities in quasi-institutionalized “language games” or “forms of life,” as the later Wittgenstein analyzed them.  If knowledge, language use, and so forth were not so interwoven, a child could not learn language or acquire modes of behavior based on an interpretation of experience; that is, one cannot imagine a functioning language game without paradigmatic experiential evidence.  We could not communicate if we did not agree upon common experiential evidence, from which everything must proceed.

From this transcendental-pragmatic interweaving of possible epistemic evidence in language games, it follows, in my opinion, that the justification for the validity of knowledge can be equated neither with the logical deduction of sentences from sentences in axiomatized systems (as modern logic of language, or of science, does) nor with the appeal to nonlinguistic intuitive evidential consciousness (as Cartesian theory of knowledge urges).  Rather, justification, as giving reasons for the validity of knowledge, must always rest on the possible evidential consciousness of the particular knowing subjects (as autonomous representatives of the transcendental knowing subject as such) and on the a priori intersubjective rules of an argumentative discourse in the context of which the epistemic evidence, as subjective proof or objective validity, has to be brought to the level of intersubjective validity.  That this is necessary and also possible is guaranteed by the a priori transcendental-pragmatic “interweaving” of epistemic evidence, whose content is interpretable “as something,” with the rules of language use that Wittgenstein elucidated and that have been concretized and made precise, especially by Austin, Strawson, and Searle, as an interweaving of judgment, reference, and predication in speech acts.  According to this conception, it makes no sense to speak of “appeal to epistemic evidence” without presupposing linguistic discourse as a context for interpretation and logical coherence.  Likewise, it makes no sense to speak of substantial argumentative discourse without presupposing certain epistemic evidence, which the particular participants of discourse apply as their criteria of truth in the argumentative procedure of building a consensus.  This sort of interweaving of epistemic evidence in language games comprises, in my opinion, the transcendental-pragmatic explanation of the fact that all scientific discoveries are, as one says nowadays, “theory laden” and that the epistemic evidence that enters into “basic sentences” is more or less dependent upon those theories that are to be confirmed or falsified, or upon alternative theories.36

Now, one could perhaps take Albert's side and object that our treatment of the problem of philosophical foundations given by epistemic evidence begins with an inadequate, that is to say, already harmless explication of his concepts of “justification” and “evidence.”  One could say that the foundations, in the sense of evidence sought by classical rationalism, could only be absolutely certain or indubitable philosophical foundations.  The methodological search for truth, in the sense of the principle of fallibilism, seems then indeed to be incompatible with the search for evidence, because it could not recognize any final or indubitable certainty.  Let us examine this argument more closely, beginning with Albert's dictum that one “can fundamentally doubt everything.”



13. Aristotle's justification of the principle of noncontradiction can serve as an illustration of the classical problem of ultimate foundations.  After Aristotle first explains the nature of the so-called axioms of the mathematicians and then presents the principle of noncontradiction as an example of an axiom, he continues:

Some indeed demand that even this shall be demonstrated, but this they do through want of education, for not to know of what things one should demand demonstration, and of what one should not, argues want of education.  For it is impossible that there should be demonstration of absolutely everything (there would be an infinite regress, so that there would still be no demonstration); . . . We can, however, demonstrate negatively even that this view is impossible, if our opponent will only say something; and if he says nothing it is absurd to seek to give an account of our views to one who cannot give an account of anything, insofar as he cannot do so.  For such a man, as such, is from the start no better than a vegetable. Now negative demonstration I distinguish from demonstration proper, because in a demonstration one might be thought to be begging the question, but if another person is responsible for the assumption we shall have negative proof, not demonstration.

Aristotle, Metaphysics, trans. Arthur Platt, McKean ed. (New York, 1941), bk. 4, 10006a6-18.

14. Albert, p. 17.

15. Cf. Aristotle, Anal. Post. I, 2, 71b20ff.

16. To speak more precisely, Descartes ranks evidence (in the sense of “clara et distincta perceptio”) above truth (in the sense of the ontological correspondence between thoughts and states of affairs) and in this way raises self-consciousness, as certain of its own being, to the “first principle” of his philosophy.  Under the axioms that are grounded in clear and distinct ideas, Descartes first mentions the sentence, “All that is has a cause or a reason.” (Cf., for example, Principia I, 11.52, and Oeuvres, Adami Tannery ed., 7, 112, 135ff., and 164).

17. Albert, pp. 12-13ff.

18. Cf. Albert, p. 16.

19. Cf. Albert, p. 18.

20. Cf. Hans Lenk, “Philosophische Logikbegründung und rationaler Kritizismus,” in Metalogik und Sprachanalyse (Freiburg, 1973), pp. 88-109.

21. Under this title T. W. Adorno distances himself from the same type of modern theory of knowledge that Albert rejects.

22. Albert, p. 30.

23. The absolute necessity of linguistic argumentation is correctly emphasized by Popper, for instance, in his arguments against the intuitionist grounding of mathematics in a theory of evidence.  According to Popper, only argumentation can ultimately give rise to a decision concerning the validity of mathematical sentences.

As soon as the admissibility of a mathematical construction proposed by intuitionism can be called into question (and, naturally, it can be called into question), language proves to be more than a mere means of communication which would be in principle superfluous.  It proves to be rather an indispensable medium of discussion.

Popper, “Epistemology without a Knowing Subject,” in Proceedings of the Third International Congress for Logic, Methodology, and Philosophy of Science, Rootselaar-Stall ed. (Amsterdam, 1968), p. 360; reprinted in Objective Knowledge (Oxford, 1972), pp. 106-152.

24. Albert, p. 52.

25. Albert, p. 53. 26. Albert, p. .52.

27. Albert, p. 52.

28. Cf. K.-O. Apel, “Programmatische Bemerkungen zur Idee einer transzendentalen Sprachpragmatik,” in Studia Philosophica in Honorem Sven Krohn, ed. Timo Airaksinen et al. (Turku, 1973), pp. 11-36, and also in Semantics and Communication, ed. C. H. Heidrich (Amsterdam, London, New York, 1974), p. 79ff.; and cf. “Zur Idee einer transzendentalen Sprachpragmatik,” in Aspekte und Probleme der Sprachphilosophie, ed. J. Simon (Freiburg, 1974). See also my “Sprechakttheorie und transzendentale Pragmatik: zur Frage ethischer Normen,” in Sprachpragmatik und Philosophie, ed. K.-O. Apel (Frankfurt a.M., 1975), pp. 10-173.

29. Cf. my introduction to C. S. Peirce, Schriften II (Frankfurt, 1979). Eng. trans. Charles Sanders Peirce: From Pragmatism to Pragmaticism (Amherst, 1981).

30. Cf. Y. Bar-Hillel, “Argumentation in Pragmatic Languages,” in Aspects of Language, ed. Y. Bar-Hillel (Jerusalem, 1970), p. 208ff.

31. Cf. my essay. “Von Kant zu Peirce: die semiotische Transformation der transzendentalen Logik,” in Transformation der Philosophie, K.-O. Apel (Frankfurt, 1972), vol. 2, p. 157ff. Eng. trans. “From Kant to Peirce: The Semiotical Transformaton of Transcendental Logic,” in Kant's Theory of Knowledge, ed. L. W. Beck (Dordrecht, Boston, 1974).

32. Cf., for example, C. W. Morris, Writings on the General Theory of Signs (The Hague, 1971), p. 46ff. and p. 56ff.

33. Cf. my critical introduction to C. Morris's Zeichen, Sprache und Verhalten (Dusseldorf, 1973).

34. As Popper wrote in The Logic of Scientific Discovery (London, 1959), p. 105, “Experiences can motivate a decision, and hence an acceptance or a rejection of a statement, but a basic statement cannot be justified by them—no more than by thumping on the table.”  Popper even speaks alternately of a motivational and a causal relation (cf. P. Bernays, “Reflections on Karl Popper's Epistemology,” in The Critical Approach to Science and Philosophy, Essays in Honor of Karl Popper [London, 1964], p. 38).  Albrecht Wellmer correctly remarks,

The method of linguistic analysis, which Popper holds in low esteem, is not necessary for demonstrating the untenability of the conception of a motivational relation between experience and its linguistic articulation. . . . Popper overlooks the fact that not only experientially based sentences but also experience itself transcends our momentary here and now.

(Methodologie als Erkenntnistheorie. Zur Wissenschaflslehre Karl R. Popper [Frankfurt, 1967], p. 156ff.)  

Like the logical empiricists, Popper is unable to think of a conceptual alternative to the disjunctive logical relation between sentences and empirical-psychological (external-causal) motivational contexts, or between linguistic universals and prelinguistic evidential experiences.  And under this (nominalistic) presupposition, Popper is correct in discarding as psychologism the “protocol sentences” qua “experience protocols” of the neopositivists.  (Cf. The Logic of Scientific Discovery, p. 95ff.)  He leaves himself no alternative but to trace back the validity of the “basic statement” to a “basic decision.”

Suppose, however, that our evident experiences are always linguistically interpreted experiences and as such transcend the momentary here and now.  Then two things follow: first, their evidence, as dependent on interpretation, can never be considered infallible; and second, such evidence can and must function as the internal justification of the meaning content of our linguistically articulated judgments of experience.  One will certainly not appeal to such experiential evidence in the way that a psychologist explains the convictions of a man by means of experiential evidence qua cause.  But one will appeal to it, in argumentation (and also in critical argumentation), as subjective testimony concerning objective evidence.  Popper is unacquainted with this concept of evidence, which is presupposed in transcendental phenomenology.  Rather, he equates (as does logical empiricism, only more consistently as regards the verdict of psychologism) “evidence” in the sense of the theory of knowledge with evidential experience or evidential feeling in the sense of empirical psychology (Popper, p. 46ff. and p. 99ff.)—as if evidence did not also belong to the necessary but not sufficient conditions of the validity of psychological knowledge.  If one reduces the criterion of truth (in the sense of a never infallible indicator) or objective evidence (which, to be sure, must be capable of being had by a knowing subject) to the psychological status of a subjective evidential feeling, then it certainly becomes necessary to replace the notion of something being objectively justified with the notion of unlimited testability or criticizability.  But without the presupposition of possible evidence, what meaning does the very idea of testing or criticizing really have?  Reference to the fact that an infinite regress can be avoided in practice by a decision can scarcely be a satisfactory answer to the question concerning the positive meaning of criticism.

35. Cf. John Searle, Speech Acts (Cambridge, ] 969), chap. 2.

36. I cannot here go into the consequences that, in the philosophy of science, result from the notion of the interweaving of evidence in language games.  Suffice it to say that experiential evidence can no more be seen as an interpretation-free basis for the intersubjective validity of knowledge than its being interwoven in language games can be understood as clear dependence on theoretically precise language use. Such a consequence—drawn by followers of Kuhn, particularly Feyerabend—leads to a relativism of language games or of theories, which Popper has correctly characterized as the “myth of the frameworks.”  Not only are there “language games,” but also, within all such language games, there is the transcendental language game of the unlimited communication community.

Posted August 29, 2007


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