5.1 Área económica
5.1.3 Iniciativas de formación
Regulative Ideas
1Translated by William Egginton* 1
The classical definition of truth that has largely determined the understanding of the concept in the history of European philos- ophy comes, as is well known, from Aristotle. Aristotle says: “To say of what is that it is not, or of what is not that it is, is false, while to say of what is that it is, and of what is not that it is not,
is true.”2This formulation of Aristotle’s has been understood, to
* The translator would like to thank Eric Little for a first version of this translation. My gratitude as well to Bernadette Wegenstein for her invaluable help with the final version.
This article was published in The Pragmatic Turn in Philosophy: Contemporary
Engagements between Analytic and Continental Thought, edited by William Egginton
and Mike Sandbothe, The State University of New York Press, 2004, all rights reserved, and appeared in Critical Horizons, vol. 4, no. 1, 2003.
1 This is the revised and greatly expanded version of a lecture I first held
in 1995 at a conference on “Pragmatism Without Regulative Ideas?” at the Institute for Cultural Science in Essen, Germany. This occasion explains the strong and originally central reference to Karl-Otto Apel—who was present at the conference and whose 75th Birthday marked the occasion of this event— as well as the title of the essay. Later workings over have somewhat shifted the weight, but the conclusion of the essay, in which I summarise my critique of Apel, remains unchanged. Since the German publication of Mike Sandbothe’s volume Die Renaissance des Pragmatismus (Weilerswist, Velbrück, 2000), I have once more revised my article, prompted in large part by a criticism of Richard Rorty’s (see notes 15 and 18). Several notes, and section 11, are newly written.
2 Aristotle, Metaphysics, in The Basic Works of Aristotle, ed. Richard McKeon,
a large extent, in the history of European philosophy in the sense of an ‘agreement’, ‘adequation’, or finally also a ‘correspondence theory’ of truth. The medieval formulation according to which truth is an ‘adaequatio rei et intellectus’, is only one version of this basic idea. Even Kant, in The Critique of Pure Reason, plainly pre- supposed a corresponding understanding of the concept of truth: “The old and famous question with which the logicians were to be driven into a corner . . . is this: What is truth? The nominal definition of truth namely, that it is the agreement of cognition with its object, is here granted and presupposed.”3
2
If one wanted to reformulate what Aristotle and Kant said in an easier and somewhat schematic format, one could do so as follows: (T) the assertion (statement or conviction) that ‘p’ is true if and
only if p.
And this biconditional is then in formal semantics reduced to the following schema:
(TI) ‘p’ is true if and only if p.
The problem that will concern us from here on emerges in both formulations: on the one hand, in both cases, the ‘truth condi- tion’—be it of the assertion that p, or of the sentence ‘p’—is for- mulated with the help of the same sentence for which or for whose assertion the necessary and sufficient condition for being true is sought—such that we, so it seems, don’t at all go beyond the sentence ‘p’, or rather the assertion that p. But on the other hand, both biconditionals should explain truth as a relation of agreement between a statement (a conviction) and reality, such
3 Immanuel Kant, Critique of Pure Reason, trans. Paul Guyer and Allen W.
that the expression ‘p’ would thus have to emerge on the left and right sides as differentiated functions: on the left side it is a question of the statement (assertion, conviction) that p; on the right side, juxtaposed to this, is the ‘state of affairs’ that p. Nevertheless, it is of course no coincidence that we can specify the ‘fact’ with which the statement (conviction) that p should ‘agree’, again, only with the help of the same sentence ‘p’, with whose help the statement was also made (or the conviction formulated).
Next we ask: how can we ascertain whether such an agreement exists between an assertion and a ‘thing’—a state of affairs, reality? Let us assume that someone claims: both doors are closed. I turn around and determine that: no, both doors are not closed. It is not the case, as it was asserted. Or in another situation, I deter- mine: yes, both doors are closed. It is the case, as was asserted. In both cases, therefore, I determine whether the assertion agrees with reality, by determining whether it is thus the case, as was asserted. The precondition for this is that I understand the sen- tence ‘both doors are closed’, that I know to what the expres- sion ‘both doors’ refers in this case, and that I can correctly use the predicate ‘is closed’. If these preconditions are given, I can determine, as a rule—if I find myself in an adequate position— whether the doors are closed. My ability to determine whether the two doors are closed is my ability to determine whether the assertion that both doors are closed agrees with reality (whether it is the case as asserted)—and this means: whether the asser- tion is true.
What ‘agreement’ between a statement and reality means can thus only be clarified when we reflect on our ability to find out in many cases—for instance, through perception—whether things are as asserted. And, if we have learned a language, we can do this in many elementary cases.
Naturally, things are much more complicated when speaking of logically complex moral, aesthetic, mathematical, historical or philosophical judgements (or statements or convictions). Then we cannot persuade ourselves, as a rule, by merely ‘looking into it’, whether it is the case as asserted: here, as a rule, the idea of a direct comparison between a statement and reality makes no sense. In order to find out whether it is the case as asserted, we are, on the contrary, dependent on ‘indirect’ procedures, namely, on ways of reasoning that so and so is the case—whereas the possible ways of reasoning for statements pertaining to the past, for aesthetic or moral judgements, for mathematical or scientific assertions or convictions, are respectively completely different. We try to decide the question whether something is so, as one says it is, with reasons (where in many cases the reasons which are at our disposal are not sufficient for bringing about such a decision). At this point it becomes clear that the idea of a ‘cor- respondence’ between a statement and reality (or between a state- ment and a fact) suggests a misleading picture: namely, it suggests the picture of a relation of agreement—ascertainable from some standpoint (which cannot be ours, but perhaps that of God)— between statements or beliefs and a piece of reality, or the things themselves. If, however, one detaches the idea of agreement between what one says and what is (really) the case from the way we justify or deny assertions or beliefs with reasons—or by calling on perceptions—then this idea becomes something com- pletely incomprehensible. For how then should one think such an agreement between statements (thoughts, convictions, etc.) and reality—two totally incommensurable relata: how and what should be tested here as agreement; and who should carry out such a test? A correspondence concept of truth, such as seems to be suggested by Aristotle’s formulation, becomes either incom- prehensible or metaphysical, or both at the same time, if we seek to think of the idea of ‘just like’, that is, ‘agreement’, indepen- dently of our justificatory praxis.
3
Let us now return to our biconditional (T): the assertion that ‘p’ is true if and only if (really) p. The intuition that this sentence expresses could also be reformulated as such: an assertion is true if and only if it is the case as was asserted. We can now think what place such an explanation of the concept of truth can have in our practice of making assertions. This practice is of a nor- mative kind: assertions are moves in a language game that are ‘justified’ or ‘unjustified’. We are entitled to assertions if we have good reasons to assert that p, or if we have convinced ourselves through our perceptions that p—or also if Someone whom we have good reason to trust has said to us that p (that is, reason for the assumption that this Someone could provide good rea- sons). What we learn when we learn a language is—among other things—to judge in a reasoned way and to distinguish between justified and unjustified assertions (convictions). This suggests a new interpretation of the biconditional (T), which frames it no longer as an attempt to interpret truth as an agreement between statements and states of affairs, but rather as an attempt to deter- mine the place the word ‘true’ has in our assertive and justifi- catory praxis. Accordingly, we could now read the biconditional as such: someone is justified in asserting that p is true precisely when he or she is justified in asserting that p. And this could now be further interpreted as saying: to say that an assertion is true is nothing other than to say that the assertion is legitimate (grounded, justified). Truth would then become no more than ‘warranted assertability’ or ‘rational acceptability’. The concept of truth would consequently be drawn back onto justification. 4
Of course, upon closer examination it appears that something in this reduction—or equation—cannot be correct. Nevertheless,
the new interpretation of the biconditional (T) makes clear an internal relation between truth and justification—which in its correspondence theoretical interpretation is lost from view—and which I would like once again to clarify differently. In Ludwig Wittgenstein’s Tractatus Logico-Philosophicus, one finds the famous sentence: “to understand a sentence means to know what is the case if it is true.”4 This sentence denotes the fundamental idea of truth-conditional semantics. Wittgenstein’s sentence says: we understand a sentence if we know its truth conditions. For exam- ple, we understand the proposition ‘both doors are closed’ if we know under which conditions a corresponding assertion is true. Now it is clear that the knowledge of the truth conditions of a sentence p cannot be anything other than the knowledge of the conditions under which I am entitled to assert that p. If I have learned a language, and so long as I understand the sentences of the language, I know, as a rule, when the conditions are pre- sent under which I am entitled (or not entitled) to assert that p. This knowledge is only, in part, a propositional knowledge. To a considerable degree it is practical knowledge—a knowing how— as the later Wittgenstein always emphasised (I can ascertain, for example, through perception, whether the predicate ‘is closed’ applies to both doors). My knowledge of the truth conditions is, in practical terms, a knowledge of assertability conditions. Since the conditions under which I am entitled to assert that p are pre- cisely the conditions under which I am entitled to assert that p is true, and in that sense the knowledge of the conditions of assertability is the same as the knowledge of the conditions of truth, it would appear that truth is warranted assertability. Truth would become thus, as one says, an ‘epistemic’ concept, which should mean: a concept that can be traced back to ‘justification’.
4 Wittgenstein, Tractatus Logico-Philosophicus, trans. C. K. Ogden, London,
Upon closer inspection, however, something about this conclu- sion appears to be false.
5
Putnam has made clear that the words ‘true’ and ‘justifiable’ (referring to assertions or convictions) cannot be equated by ref- erence to the grammar of these words: one is entitled (has good reasons)—under certain conditions—to believe or to assert that p—and such reasons can later be revealed to be insufficient. ‘Justification can be lost’5—justification is relative to time or cir- cumstances and also to people, whereas truth ‘cannot be lost’— that is, a conviction or an assertion cannot today or for me be true and tomorrow or for you not be. Truth is trans-subjective and timeless. This points to an grammatical difference between the predicates ‘is true’ and ‘is justified’,6which would certainly
5 Hilary Putnam, “Reference and Truth,” in Realism and Reason: Philosophical Papers Vol. 3, Cambridge, Mass., Harvard University Press, 1983, p. 84.
6 Here I employ the concept of justification in a broad sense. An assertion
can be justified in a narrower sense by reasons that stand in an inferential con- nection with the propositional content of an assertion. In the broader sense, it can also be justified through reference to perceptions (or, indirectly, also to the reliability or believability of another speaker). Cf. the analogous differentiation by Robert Brandom in Making It Explicit, Cambridge, Mass., Harvard University Press, 1994, chps. 3 & 4. Incidentally, I concur with McDowell—differently than Brandom, Rorty, or Davidson—that convictions are also justified through recourse to perceptions (although I also have some reservations about the details McDowell’s argument). Rorty’s (Sellar’s, Davidson’s, and Brandom’s) causal interpretation (that is, elimination) of the concept of experience is, in my opinion, only convincing insofar as it is directed against the empirical ‘myth of the given’: namely, against the idea that ‘sense data’ or ‘sensory feelings’ could play an epistemic role in the formation of our empirical convictions. A remainder of this idea is still found in Quine’s concept of ‘stimulus meaning’. Now Rorty rightly points out, for instance, that Davidson—to whom, among others, he himself refers with regard to his own causalistic elimination of the concept of experience—has long since dismissed the Quinean version of empiri- cism. “Davidson substitutes a ‘distal’ theory of meaning formulated in terms of public external objects; he allows no intermediate terrain of philosophical
remain to be clarified. A corresponding attempt at clarification has been undertaken, in particular, by Putnam, Habermas, and Apel. The basic idea common to all three philosophers is as fol-
inquiry between linguistically formulated beliefs and physiology.” (Richard Rorty, “Dewey Between Hegel and Darwin,” in Truth and Progress. Philosophical
Papers Volume 3, Cambridge, Cambridge University Press, 1998). If the causal
interpretation of the concept of experience refers to ‘public external objects’ as the causes of non-inferentially received convictions, then—against Davidson’s and Rorty’s opinion—there exists absolutely no further objection to an epis- temic role for (linguistically impregnated) experience (that is, perceptions). If I justify my assertion that the corner window on the second floor of the house across the street is open, with the indication that I see it (or have seen it), then here, in my opinion, it clearly concerns an epistemic relationship between my perception and my conviction. Granted I can also causally interpret the same relationship, as Davidson does in the context of his theory of interpretation. (It is the rabbit hopping by that causally brings about the conviction ‘Gavagai’ in my foreign-speaking interlocutor). But now it appears to me in this case that the epistemic interpretation is to have conceptual precedence over its inter- pretation as a causal relation, and this is for the following reason: to interpret a relation as causal means to assume that we could discover something empir- ical about the asserted causal relation. Of course, this is possible in the case of radical interpretation because the interpreter could discover that it is not the rabbit hopping past, but rather, something else that (the interpreter presumes) caused the foreign speaker’s conviction expressed by his uttering ‘Gavagai’. But what is here empirically clarified is nothing of the sort that under these or those circumstances certain objects of the public world cause this or that non- inferential conviction. What is clarified is, on the contrary, the question, which is the conviction that the foreign speaker expresses (that is, what ‘Gavagai’ means). But what about the interpreter’s conviction (that there is a rabbit hop- ping past), which is clearly already implied by his interpretation of the for- eign speaker? As to what concerns its cause, as a rule the interpreter has no choice (because most of his—non-inferential—convictions must also be true): it is the rabbit hopping past (that he sees hopping past). But the ‘principle of charity’ now says that, as a rule, the same objects will effect the same non- inferential convictions in speaker and interpreter. The interpreter, as noted, has no choice as to what affects him: the cause of his non-inferential conviction is that thing he—in an ordinary way of speaking—sees (that of which he sees that it is the case). And insofar as ‘seeing that’ is an ‘achievement’ word, he can also occasionally be mistaken. What I am driving at is that we can only
identify the causes of our non-inferential beliefs in Davidson’s scenario (which
Rorty makes his own), in that we say what we see. And that must be—still in Davidson’s scenario—as a rule the correct identification. Accordingly, another method of identification of causes of our non-inferential beliefs can in principle not be at our disposal (except if one would cross over again into ‘physiology’).
lows: if truth is internally connected with justification and never- theless not the same as justification (here and now), then further conditions must be stated in such a way that an assertion or con- viction justified under such conditions would be necessarily true. Truth would, therefore, be maintained as an ‘epistemic’ con- cept—internally connected with the concept of justification—and nevertheless the simple equation of the predicates ‘is true’ and ‘is justified’ would be avoided. The conditions in question would have to be ideal conditions; and the basic idea is that an asser- tion or conviction justified under such ideal conditions would be necessarily true—whereas the ‘necessarily’ should give expres- sion to a conceptual necessity. Putnam, Habermas, and Apel want to say that this is how we understand the concept of truth (or this is how we ought to understand it).
Let us first make clear the point of the basic argumentation strat- egy: if one eliminated the difference between truth and justifica- tion, there would be relativistic consequences, because it is easy to see that in the vertical dimension of historical time, as well as in the horizontal dimension of a plurality of cultures, situa- tions, and contexts, many mutually incompatible convictions are held to be true by different people, in different cultures, etc., and all with—prima facie—good reasons. This even goes for the history of science—it is thus not only a problem for cultural pluralism. But then there can be nothing disreputable in the (justifying) appeal to (lin- guistically impregnated) perceptions; on the contrary, the implication of a causal interpretation of non-inferential convictions lead us back to the possibility of justifying such convictions through recourse to perceptions. In reality, how- ever, the praxis of such justification has conceptual priority over a causal inter- pretation of non-inferential beliefs (as far as the possible identification of the corresponding causes is concerned). The causal interpretation is not false, but
first it is compatible throughout with an epistemic interpretation, and since we
must now give the epistemic interpretation an epistemic priority over the causal