20th World Congress of Philosophy Logo

Theory of Knowledge

In Spite of Davidson's Arguments for "The Folly of Trying to Define Truth," Truth Can Be Defined

Dan Nesher
University of Haifa

bluered.gif (1041 bytes)

ABSTRACT: Davidson’s argument against the possibility of defining truth draws upon the work of Tarski. However, Tarski’s assumption that the semantic conception of truth holds only for formal languages which are not semantically closed is not as plausible as it seems to be since it can be shown that this would result in the impossibility of formulating a theory of truth, because the epistemological presuppositions of formal semantics undermine any theory of representation of reality in which our cognitions can be true or false representations. Yet Davidson concludes that "there cannot be a definition of ‘For all languages L, and all sentences s in L, s is true in L if and only if . . . s . . . L’." I am challenging Davidson by introducing into his above scheme my own definition of truth — "For all languages L, and all sentences s in L, s is true in L if and only if we prove s in L" — and then showing how to prove this definition philosophically.

bluered.gif (1041 bytes)

I. Introduction: Can we define truth?

Davidson argues for "the folly of trying to define truth" and claims that Tarski's "accomplishment was accompanied by a proof that truth cannot (given various plausible assumptions) be defined in general" (Davidson, 1996:269). Tarski's plausible assumptions are that his "semantic conception of truth" can be formulated only for formal languages which are not semantically closed. But these assumptions are not so plausible as they seem since it can be shown that if we accept them it is impossible to formulate a theory of truth because the epistemological presuppositions of formal semantics undermine any theory of representation of reality in which our cognitions can be true or false representations (Nesher, 1996). Yet Davidson concludes from Tarski's theory of truth that "there cannot be definition of `For all languages L, and all sentences s in L, s is true in L if and only if ... s ... L'."

I would like to start by challenging Davidson about his claim for the impossibility of defining truth and to introduce into his above scheme my own definition of truth; then I will show how to prove this definition philosophically:

[1] `For all languages L, and all sentences s in L, s is true in L if and only if we prove s in L'.

We can see immediately that the plausible assumptions of Tarski's "semantic conception of truth" for semantically formal languages do not hold in my definition of truth since I define truth in the same language in which it is used. Moreover, according to pragmaticist epistemological theory of truth the terms "languages" and "sentences" in the above definition are replaced by the conceptions of the cognitive languages of mind and cognitions respectively. Further on, the notion of proof should be extended to include Peirce's trio that involves not only the deductive formal procedures but also material logic procedures of abduction and induction and at the low grades of self-consciousness and self-control of perceptual process the proof is only "quasi-proof" (Nesher, 1996). In the pragmaticist philosophy the above definition of truth cannot be the formal starting point but the result of a completed inquiry into meaning and truth as aspects of human cognitive processes.

II. The folly of trying to start a theory of truth from the definition of truth

In the axiomatic-deductive method we start from undefined terms and therefore the definitions of concepts themselves must be somehow vague, and moreover the axioms which are formed with these concepts are themselves unjustified even though the "logical" operation justification is based upon them. The problem of the uncertainty of axiomatic-deductive formal systems is related to their epistemological foundations and our understanding of the justification of their formation including their definitional starting point.

The problem with the philosophical "analytic" tradition is seeing the problem of knowledge and truth as a logical problem of formal systems instead of an essentially philosophical-epistemological problem. If we attempt to build up a formal theory in this tradition, as Davidson has tried to do with his theory of truth, and to start philosophizing from definitions, as Socrates attempted unsuccessfully, we may find that this is impossible (Davidson, 1996:293ff.). Davidson himself found that he cannot define formally the most elementary concepts like truth, knowledge, and belief and that he has to accept them as undefinable (Davidson, 1990:313-314, 1996:264). Hence, all our system of knowledge starts with undefined concepts and we run the risk that the entire analytic-philosophical program will be "ending with a whimper of failure" (Davidson, 1996:263).

Davidson thinks about definitions in terms of axiomatic formal systems, and therefore as the beginning of any philosophical enterprise,and since he has been trying unsuccessfully to define the concept of truth he now feels quite strongly that definitions of important philosophical concepts are impossible (Davidson, 1996:275).

I believe that it is folly to start a theory of truth from the definition of the essence of truth by looking for the initial simpler and clearer idea in the Platonic-Cartesian tradition. Philosophy is not a logical formal system and therefore in a philosophical theory we cannot define truth at the beginning of our inquiry; but this does not exclude the definition or explication of truth at the end of the inquiry. Moreover, the enterprise to form a formal theory of truth a'la Tarski, especially with regard to non-formal languages and empirical sciences, undercuts the possibility of this enterprise (Nesher, 1996; comp. Davidson, 1990:319f.). Nevertheless Davidson still envisioned a kind of Tarskian formal theory of truth for natural languages, even though he aims at an empirical theory of truth. However, he understands that such a theory cannot be a real formal theory "but in a far looser sense of `theory' than Tarski had in mind" (Davidson, 1990:312n55).

III. The difficulties with the logico-formal-semantics and with the conception of truth

Nevertheless, Davidson feels the importance of the concept of truth for the theory of meaning and knowledge, but caught by the formal semanticist conception of theory, the logical hierarchy of languages, and the terminology of Logical Positivism, he has difficulty formulating an empirical theory of truth (Davidson, 1996:273ff.). I believe that the impassible difficulties for some philosophers to form theories of truth is basically their being caught by the traditional conceptions of language and its meaning and truth.

The main characteristic of the accepted philosophical framework in which difficulties with the conception of truth arise is the explicit or implicit assumption that thoughts or linguistic expressions meaningful and true independently of human cognitive performances and knowledge (Frege, 1918; Tarski, 1944).

This metaphysical realist conception of language is the reason for the predicaments and paradoxes with our conceptions of meaning and truth (Nesher, 1996). Formal semanticists theorize human cognitive representation of external reality by replacing the everyday subject matter by suitable abstractly characterized idealizations, chosen to preserve those features of the original subject that are relevant to the study at hand. Here they need abstract substitutes for thought, for reality, and for their representational relations. For thought they substitute language, a formalized version of parts of everyday language, and for reality they substitute something called a structure, a collection of things suitable for being correlated, as meanings, to expressions in the language. For the representational relation of thought to reality they substitute interpretation, that is, a function assigning to certain expressions in the language, as their meanings under the interpretation, certain objects in the structure.

The question is whether, if at all, this kind of idealization preserves and can explain the representational relation between thought and reality. I argue that even with such Tarskian formal semantics, an enterprise which aims to create an objective scientific semantics, these abstract idealizations cannot preserve the essential relations of mind representing reality. The formal semantic idealization only creates an illusion of relevancy to the relation between thought and reality (Putnam, 1994:315). The reason for this illusion is that the interpretation relation between language and the entities of the abstract structure is actually assigned to them by formal semanticists from outside these idealized domains; they assume a God's eye view which in natural situation we cannot afford of getting outside our cognitive skin (Davidson, 1986a:312). This illusion leads eventually to a metaphysical realist position that seems as if we can assume without philosophical justification that the semanticist-observer (the "metaphysical subject") can compare linguistic pictures with real states of affairs, as in the illusory situation in Wittgenstein's Tractatus.

But language is human cognitive performance in which we cannot assume this illusory position since, as Hume argues, we have no external point of view independent of our cognitions. Hence, if we cannot solve Hume's problem about our knowledge of the existing external reality we must accept, as Kant did, a kind of phenomenalism (Nesher, 1997). In spite of entertaining the possibility of Tarski's style of formal semantic theory of truth, Davidson generally accepts the phenomenological position and suggests that there cannot be any confrontation between our thought and external reality, and therefore we must reject Quine's "third dogma" of empiricism (Davidson, 1984).

IV. Can we have true or false beliefs without a confrontation with reality?

Davidson wants to form a theory of truth for our beliefs without confrontation with external reality about which these beliefs are true or false. He believes that we do not have any access to this reality, and without such access there cannot be confrontation between mind and this reality.

What brings truth and knowledge together is meaning. If meaning is given by objective truth conditions there is a question how we can know that the conditions are satisfied, for this would appear to require a confrontation between what we believe and reality; and the idea of such confrontation is absurd. But if coherence is a test of truth, then coherence is a test for judging that objective truth conditions are satisfied, and we no longer need to explain meaning on the basis of possible confrontation (Davidson, 1986a:307).

The question is, how can coherence be a test of truth without confrontation with external reality? This is the key problem for Davidson and without explaining it we cannot be realists who accept "objective truth conditions" and have "a realist view of truth," and above all we cannot "insist that knowledge is of an objective world independent of our thought or language" (Davidson, 1986a:307). But Davidson rejects the possibility of such confrontation.

No such confrontation makes sense, for of course we cannot get outside our skins to find out what is causing the internal happenings of which we are aware (Davidson, 1986a:312).

Let us take Davidson's metaphor of "our skins" to mean our beliefs and other cognitive states or processes as sensations and imaginations. And yet the paradoxical situation is that if "we cannot get outside our skins to find out what is causing the internal happenings of which we are aware," how do we know that there is something outside our skins that is causing our internal happenings? If we accept the Humean argument that we cannot know anything outside our cognitive skins and assume with Kant that we cannot explain our internal happenings without assuming something outside that causes them, we have to explain how without going outside our skins we can know that there are things that cause our internal happenings. Hence we have to understand what can be the "objective world" and the "objective truth conditions" that determine our beliefs and whether this determination can work without confrontation.

What are the truth conditions of Davidson's empirical theory of truth?

A theory of truth, viewed as an empirical theory, is tested by its relevant consequences, and these are the T-sentences entailed by the theory. A T-sentence says of a particular speaker that, every time he utters a given sentence, the utterance will be true if and only if certain conditions are satisfied" (Davidson, 1990:313).

Davidson takes Tarski's convention (T) as the clue for his intuition of what are the truth conditions of linguistic sentences:

(T)`"Snow is white" is true if and only if snow is white'

However we cannot know from "Snow is white" as the name of a sentence whether it is true unless we know that the sentence itself Snow is white is true. Yet, what are the conditions in our perceptual experience that by knowing them we can indicate the truth of the perceptual judgment: Snow is white?

According to Davidson truth conditions are "what he [the speaker or interpreter] believes to be the case" (Davidson, 1990:318). But what is "the case" in which the agent of language believes and which "causes" him to hold his belief Snow is white to be true or false? It cannot be this perceptual judgment proposition itself but a kind of a indexical ("causal" or "dynamical") relation between this linguistic expression and something that is not a verbal expression (Davidson, 1990:320).

What the interpreter has to go on, then, is information about what episodes and situations in the world cause an agent to prefer that one rather than another sentence to be true (Davidson, 1990:322).

Yet the non-verbal perceptual "experience" that ended with this perceptual judgment is itself an element of our cognition and not "outside our skins," and the question is whether this experience can be the "objective truth conditions"?

Davidson feels very strongly that we need a concept of truth since we cannot have an empirical theory of knowledge without the formal theory of truth.

We are interested in the concept of truth only because there are actual objects and states of the world to which to apply it: utterances, states of belief, inscriptions. If we did not understand what it was for such entities to be true, we would not be able to characterize the contents of these states, objects and events. So in addition to the formal theory of truth, we must indicate how truth is to be predicated of these empirical phenomena (Davidson, 1996:276-7).

Yet it is not always clear and quite difficult to understand what these "actual objects and states of the world" are, whether "external objects and events" (Davidson, 1990:321) or Kantian cognitive states ("representations") that "the contents of these states" according to Kant, are the familiar objects which are "empirical phenomena" (Kant, 1781-87:A191\B236)

The conceptual problem for Davidson is to explain how the independent objective world causes from outside truth and falsity of our internal beliefs.

...and what ultimately ties language to the world is that the conditions that typically cause us to hold sentences true constitute the truth conditions, and hence the meanings, of our sentences (Davidson, 1996:275).

The difficulty with this formulation is whether the truth conditions are elements of the world or of our beliefs: whether they cause the truths of our beliefs or justify them. What therefore constitutes the truth conditions and how they affect the truth of our beliefs?

If Davidson abandoned the coherent theory of truth, as he concedes, and still retains the function of truth conditions in determining the truth and the meaning of our propositional beliefs, then the question is whether these truth conditions are themselves beliefs or not (Davidson, 1996:273, 1990:320). If truth conditions are elements of the external world they cause the truth of our beliefs but we cannot know how. But if they are beliefs why do they determine the truth of other beliefs and not vice versa? If truth conditions have a privileged position in our cognitive relations to the world we must explain what makes them privileged above other beliefs (Davidson, 1986b:332).

We are dealing here with a perceptual process of observation and its causal relation with an extra-cognitive reality since we do not observe belief but we can believe in what we observe (Davidson, 1990:320). We have to explain why when we make a perceptual judgment we already know that it is true. We utter or accept `snow is white' if and only if the truth conditions for this utterance are satisfied, and it must be confirmed empirically for the truth of this utterance. But what these truth conditions are Davidson does not tell us (Davidson, 1990:309, 313ff.).

It seems that Davidson, following, Kant holds together without explanation the two conflicting philosophical perspectives: by rejecting our possibility of getting out of our cognitive skins, Davidson accepts phenomenalism and at the same time he assumes the God's eye view of metaphysical realism about the existing external world which causes our beliefs.

The way out of this paradoxical situation is to elaborate his own intended "argument that purports to show that coherence yields correspondence," to show how from inside our cognitive mind we can detect the relation between mind and world and explain our knowledge of the external world.

V. How do we know that truth conditions are satisfied? The relation of confrontation between mind and reality

Why does Davidson reject confrontation between mind and world, and what does he mean by confrontation?

...the causal relations between our beliefs and speech and the world also supply the interpretation of our language and out beliefs. In this rather special sense, 'experience' is the source of all knowledge. But this is a sense that does not encourage us to find a mental or inferential bridge between external events and ordinary beliefs. The bridge is there all right - a causal bridge that involves the sense organs. The error lies as Neurath saw in trying to turn this causal bridge into an epistemological one, with sense data, uninterpreted givens, or unwritable sentences constituting its impossible spans (Davidson, 1986b:332; italics added).

In other words, we justify our beliefs and actions by reasons and explain physical events by causes, but we cannot explain our beliefs by physical causes. Therefore we cannot explain it through confrontation of the world with our mind and the only way available to us is to explain beliefs "mentally" or "referentially" in relation to other beliefs. This is an analogue to Kant's argument about the unexplainable relation between the transcendental object and our experience (Kant, 1781-87:A104-110). I cannot suggest here the Spinozistic theory of representation and the explanation of how physical objects and events are connected to our experience of the truth conditions. It is enough to explain this through the feeling of duality in our experience between our soul and the stimulus when the latter is itself understood cognitively (Peirce, 1.322).

That shock which we experience when anything particularly unexpected forces itself upon our recognition ... is simply the sense of the volitional inertia of expectation, which strikes a blow... Low grades of this shock doubtless accompany all unexpected perceptions.... Its lower grades are ... that sense of externality, of the presence of a non-ego, which accompanies perception generally and helps to distinguish it from dreaming (Peirce, 1.332).

Thus, without going outside our cognition we can analyze "truth conditions" as components of "meaning conditions" of cognitive processes so that when we interpret our perceptual signs to understand the meaning of the perceptual judgment we already feel the elements of meaning that make it true.

Indeed, the conception of truth conditions is very elusive and usually philosophers accept it only intuitively as that which is somehow presented to our perception. Truth conditions cannot be elements of external reality to which we do not have direct access, but they cannot be just any cognitive states because then we would be in a vicious circle of holism. The answer to the dilemma of whether truth conditions are elements of our cognition or of external reality is that they consist of a specific stage in the perceptual process representing external reality (Nesher, 1997). The truth conditions of the perceptual judgments are the most rudimentary, they are feelings of duality of the indexical relation in the perceptual process and they are the indication of our confrontation with external reality. Thus truth conditions are components of the proof (or quasi-proof) procedure, and with them we verify (or falsify) the proof results owing to our confrontation with external reality (Nesher, 1997).

VI. The definition of truth as based on pragmaticist theory of cognitive truth

Elsewhere I analyzed the perceptual process and showed that structurally this process operates like a reasoning process in which the suggested abductive hypothesis is proved by a later inductive evaluation, though the perceptual process is at lower grades of self-consciousness and self-control (Nesher, 1997). But how can the proof make the suggested abductive hypothesis true? In opposition to Rorty's understanding of pragmatism I argue that a Peircean pragmatist is not only an ontological realist about the existence of independent reality but also an epistemological realist about our representations of this reality (Nesher, 1997).

The pragmaticist position, as I showed by analyzing Spinoza's and Peirce's conceptions of truth, is exactly what Davidson suggested "to show that coherence yields correspondence" (Davidson, 1986a:307). But because of his Kantian epistemology, Davidson cannot show how through our proofs of coherence and incoherence of our beliefs we make our cognitions true or false and therefore correspondent and non-correspondent with external reality.

With pragmaticist epistemology we can show that the very process of perception is an instinctive quasi-proof of the perceptual judgments which as proved are "simple truths." In the perceptual process the coherent structure of cognitive states is formed and transformed into and embedded in the perceptual judgement propositions. Only after the perceptual processes are interpreted instinctively and evaluated positively are they generalized into perceptual judgments. These "simple truths" are distinct from "logical truths" of propositions and theories which are rationally proved or disproved to be true or false on the evidence of the proved perceptual judgments (Peirce, 5.570f.).

With the above analysis of the perceptual process, being instinctive and basically unconscious and uncontrolled by perceivers, we can explain why the foundationalists accept the perceptual judgment as the "given" bedrock of our knowledge (Nesher, 1997:IV). The feeling of duality inside the perceptual process is what Peirce calls "the natural instinct for truth," and this is "the source of truth" that Dummett is looking for (Peirce, 7.220).

...we often derive from observation strong intimations of truth, without being able to specify what were the circumstances we had observed which conveyed those intimations (Peirce, 7.46).

We have seen that these "strong intimations of truth" are a very primitive stage of our cognition of external reality. This should be the source of Ramsey's feeling for the redundancy of specifically emphasizing the truth of propositions if we already know them instinctively as such. This is also the reason that we are bewitched by Tarski's (T) convention and conclude through the disquatational appearance, the eliminative, or minimalist, or deflationist theories of truth. The source of this disquatational theory of truth is this instinctive process of quasi-proof that as perceivers we cannot analyze but feel it so strongly as a proof. We feel that we add nothing by predicating our perceptual judgment with "is true" and that we can do without it, and without being entangled in the problematic substantive theory of truth.

If we accept different kinds of proofs ranging from perceptual quasi-proofs to logical formal proofs, we might assume that all our cognitions should be true because otherwise we would not hold them. This is basically Davidson's intuition, and it could be right if human cognition were not a dynamic and continuous process of replacing relatively weaker true cognitions by relatively stronger ones through repeated proof procedures of refutation and confirmation. So thus we can find in the entire human dynamic cognitive processes true, false, and undecided doubtful cognitions as essential components of the evolution of human knowledge.

bluered.gif (1041 bytes)


Davidson, D. (1984) "On the Very Idea of a Conceptual Scheme." In Davidson, Inquiries into Truth and Interpretation:17-36. Oxford: Clarendon Press 1984:183-198.

____________ (1986a) "A Coherence Theory of Truth and Knowledge" in Truth and Interpretation: Perspectives on the Philosophy of Davidson, E. LePore, ed., Oxford: Blackwell 1986:307-319.

____________ (1986b) "Empirical Content" in Truth and Interpretation: Perspectives on the Philosophy of Davidson, E. LePore ed., Oxford Blackwell 1986:320-332.

___________ (1990) "The Structure and Content of Truth." The Journal of Philosophy Vol. 88, No. 6, June 1990:279-328.

___________ (1996) "The Folly of Trying to Define Truth." The Journal of Philosophy Vol. XCIII, No. 6, June 1996:263-278.

Frege, G. (1918) "The Thought: A Logical Inquiry." In E.D. Klemke, ed., Essays on Frege. University of Illinois Press.

Kant, I., (1781-1787) Critique of Pure Reason. Tr. by N. K. Smith, Macmillan & Co. Ltd. 1929.

Nesher, D. (1996) "The Pragmaticist Conception of Language and Truth and the Liar Paradox." (Manuscript).

_________ (1997) "Peircean Realism: Truth as Meaning of Cognitive Signs Representing External Reality." Transactions of the Charles S. Peirce Society. Vol. XXXIII, No. 1, 1997:201-257.

Peirce, C.S. (1931-1958) Collected Papers, Vols I-VIII. Harvard University Press.

Putnam, H. (1994) Words and Life. Harvard University Press.

Tarski, A. (1944) "The Semantic Conception of Truth." In H. Feigl & W. Sellars, eds., Readings in Philosophical Analysis. Appleton-Century-Crofts, Inc. 1949.

bluered.gif (1041 bytes)


Back to the Top

20th World Congress of Philosophy Logo

Paideia logo design by Janet L. Olson.
All Rights Reserved


Back to the WCP Homepage