Varia

 

Did Ramsey Ever Endorse a Redundancy Theory of Truth?

 

María J. Frápolli

 

Department of Philosophy I University of Granada (Spain) mjfrapolli@gmail.com

 

Recibido: 25 – 02 – 2011.
Aceptado: 03 – 06 – 2011.

 

Abstract

This paper deals with Ramsey's theory of truth and its aim is twofold: on the one hand, it will explain what position about truth Ramsey actually defended, and, on the other hand, we will pursue Ramsey's insight in the XXth century. When the name of Frank Ramsey is mentioned, one of the things that comes to mind is the theory of truth as redundancy. In the following pages we will argue that Ramsey never supported such a theory, but rather an analysis of truth noticeably similar to the prosentential account. In fact, the very word "pro–sentence" appears for the first time in the XXth Century in Ramsey's unfinished work "The nature of truth", written around 1929. Besides, we will show that the prosentential account of truth is a neglected trend throughout the history of analytic philosophy, even though relevant analytic philosophers, such as Prior, Strawson, Williams, Grover and Brandom, have endorsed it.

Key Words: Ramsey, truth, redundancy, prosentence.

 

Resumen

Este artículo trata sobre la teoría de la verdad de Ramsey y su propósito es doble: por un lado, explicar cuál es la posición sobre la verdad que Ramsey realmente defiende; y, por otro lado, perseguir algunas intuiciones originarias de Ramsey en la filosofía del siglo XX. Cuando se menciona el nombre de Frank Ramsey, una de las ideas que vienen a la mente es la teoría de la verdad como redundancia. En las páginas siguientes, argumentaremos que Ramsey nunca defendió esta teoría, sino que realizó un análisis de la verdad muy similar a la propuesta prooracional contemporánea. De hecho, la palabra "prooración" aparece por primera vez en el siglo xx en el artículo de Ramsey "La naturaleza de la verdad", escrito alrededor de 1929. Además mostraremos que la propuesta prooracional acerca de la verdad ha sido pasada por alto a lo largo de la historia de la filosofía analítica a pesar de que algunos filósofos de gran relevancia, tales como Strawson, Prior, Williams, Grover y Brandom, la han defendido.

Palabras clave: Ramsey, verdad, redundancia, prooración.

 

1 Introduction

This paper aims to discuss Ramsey's theory of truth. When the name of Frank Ramsey is mentioned, one of the (few) things that come to mind is the theory of truth as redundancy. In the following pages we will argue that Ramsey never supported such a theory, but rather an analysis of truth noticeably similar to the prosentential account. In fact, the very word "pro–sentence" appears for the first time in the XXth Century in Ramsey's unfinished work "The nature of truth", written around 1929. Besides, we will show that the prosentential account of truth is a neglected trend throughout the history of analytic philosophy, even though relevant analytic philosophers have referred to it from time to time.

Until the publication of Ramsey's manuscript On Truth (Rescher and Majer, 1991), scholars considered "Facts and propositions" (1927) as containing everything Ramsey had to say on truth. And although the materials in (Rescher and Majer, 1991) discussed the topic at length, the inherited view on Ramsey's theory of truth remains unchanged, as can be seen in the following text: "In both On Truth and in "Facts and Propositions" Ramsey defends his redundancy theory of truth: 'It is true that Caesar was murdered' means no more than that Caesar was murdered, and 'It is false that Caesar was murdered' means that Caesar was not murdered' […]" (Rescher and Majer, 1991: xiii). Engel and Dokic's book (Engel and Dokic, 2003) is the only exception.

 

2 Ramsey's Theory

In "Facts and Propositions" (1927) Ramsey places his treatment of truth in the context of the analysis of the notions of belief and judgment, but he only unfolds it in "The nature of truth" (1929), the first chapter of On Truth. Let us begin with (1927). Its aim is to offer a logical analysis of belief, judgment and assertion. Ramsey considers belief to be a relation between two "factors". There is a mental factor, my present mental state, and an objective factor, facts or events in the world. And to say that I believe that Caesar was murdered is to say that a particular kind of relation holds between my mental state and the objective factor related to it. In this context, Ramsey says: "There is no separate problem of truth but merely a linguistic muddle" (1927/1990: 38). The emphasis here is laid on "separate", since he considers that the serious philosophical questions at issue concern the notions of belief, judgment and assertion. Ramsey seemed to assume that once these notions are correctly understood, the notion of truth will fall smoothly into place. Truth is profusely put to work in epistemic contexts, even though it is not an epistemic notion, and epistemic contexts are not the only ones in which truth earns its living. What Ramsey upholds in this paper is that the logical analysis of truth is independent of the analysis of belief, and that most of the traditional difficulties associated to the analysis of truth are in fact misplaced difficulties related with epistemic notions. To say, as Ramsey does, that the (separate) problem of truth is no more than "a linguistic muddle" does not commit one with a redundancy theory of truth. Different authors with substantive views on truth have expressed more or less the same feeling. Austin, a champion of correspondence, maintains that "the theory of truth is a series of truisms" (1950:152), and at the same time rejects vigorously that the truth predicate is logically superfluous. This is also Ramsey's case.

Ramsey explicitly says in (1927) and suggests in (1929) that propositions are the primary truth bearers. Indicative sentences are removed from the list of candidates as "not a serious rival" (1929/1991: 7). In (1929) he favors beliefs as the items that bear truth, but beliefs are truth bearers insofar as they possess what Ramsey calls "propositional reference". In other words, beliefs point to propositions, and it is this feature that makes them suitable for truth bearing. Using an updated terminology, we would say that propositions are the contents of propositional attitudes, beliefs among them, and these contents are the items to which agents attribute truth and falsity. Thus the analyses of truth bearers in (1927) and in (1991) are arguably equivalent. From now on, we will refer to truth bearers as "propositions", keeping the notion as neutral as possible.

The act of ascribing truth to a proposition is a truth ascription. Derivatively, we can also call "truth ascription" to the sentence by means of which the ascription is performed. There are different types of sentences that act as truth ascriptions. For our purposes, the most relevant are the following: (i) ascriptions that give the clues to recover the proposition to which truth is ascribed, and (ii) ascriptions that offer no clue about the proposition which is its content. We can call the ascriptions in (i) exhibitive truth ascriptions, and ascriptions in (ii) blind truth ascriptions. In this second case, the proposition referred to has to be contextually identified. Examples of truth ascriptions are the following:

(1) It is true that Caesar was murdered (exhibitive)

(2) What Victoria says is true (blind).

In (1927) Ramsey acknowledges these two types, and offers his famous "redundancy" analysis only related to the first one. He says:

"Truth and falsity are ascribed primarily to propositions. The propositions to which they are ascribed may be either explicitly given or described. Suppose first that it is explicitly given; then it is evident that 'It is true that Caesar was murdered' means no more than that Caesar was murdered, and 'it is false that Caesar was murdered' means that Caesar is not murdered. They are phrases that we sometimes use for emphasis or for stylistic reasons, or to indicate the position occupied by the statement in our argument. So we can also say 'it is a fact that he was murdered' or 'that he was murdered is contrary to fact'" (1927/1990: 38).

This passage contains two intuitions that deserve attention. The first one is the idea that exhibitive truth ascriptions not only have stylistic utility, but they mark a proposition's place in an argument. Ramsey does not pursue this latter point further, but a plausible explanation of the role that truth performs in exhibitive truth ascriptions is that it stresses that the content is safe as a premise, for instance. The second intuition is the identification Ramsey makes between the role of the sentential operators "it is true" and "it is a fact", an identification stressed by other analytic philosophers such as Strawson (1950), and that lies at the core of the intuition that gives rise to correspondence theories of truth.

Ramsey does not stop here, though; he continues: "In the second case in which the proposition is described and not given explicitly we have perhaps more of a problem, for we get statements from which we cannot in ordinary language eliminate the words 'true' and 'false'. Thus if I say 'he is always right', I mean that the propositions he asserts are always true, and there does not seem to be any way of expressing this without using the word 'true'" (op.cit. 38–39).

In this passage, there is also an identification between the role performed by the blind truth ascriptions "what he asserts is true" and "he is always right", and an explicit rejection of the redundancy view. Taking the two passages together, Ramsey's view is that, although truth has redundant uses, not all its uses are of this kind. Focusing on uses instead of concept–types is characteristic of pragmatist approaches to language.

Let us now consider (1929). In contrast with (1927), Ramsey's purpose in this paper is to offer an analysis of the meaning of "true", and what he does here is basically to disclose his previous intuitions as they were presented in (1927). Ramsey declares that everybody knows what 'true' means, and that the philosophical difficulty lies in putting it into words, a view that Tarski, for instance, expresses exactly in the same way in (1935:152). Thus, Tarski's famous diagnosis that truth cannot be defined in natural languages admits a charitable explanation as an unelaborate formulation of one of the tenets of the prosentential account: that the repertoire of expressions standardly possessed by natural languages makes it difficult to give an exact counterpart of truth–terms. Ramsey's position is in fact that natural languages are short of simple propositional pro–forms, and for this reason the role performed by truth–terms in natural languages cannot be made explicit without the addition of propositional variables and quantifiers.

The philosophical problem then is not to understand what "is true" means, something which every competent user of language knows, but rather to say what it means, and this because natural languages lack the appropriate expressive tools to explain the meaning of truth without using the very term or a closely related one. Let us go to Ramsey's definition. In (1929/1991: 9), we read: "We can say that a belief is true if it is a belief that p, and p." And he explains that p is a sentential variable that can represent any propositional structure whatsoever. We might predicate truth ofa disjunctive proposition, and then say: the belief that either p or q is a true belief if either p or q; or we can predicate truth of a general proposition and say: the belief that every A is B is a true belief if every A is B, and so on. The simplest way of generalizing the pattern to cover any propositional form is using a variable such as p. And p, as a sentential variable, involves a verb already, and thus there is no need to add "is true" to the definition, which would render it circular, as Ramsey explicitly acknowledges. A similar treatment of the role of the sentential variable p in a definition of truth is given in (1927/1990: 39).

Ramsey's definition, "a belief is true if it is a belief that p, and p", as it has happened with Tarski's T–convention, has been interpreted as a formulation of the redundancy of truth. But this interpretation rests on a misunderstanding. Ramsey's point, and probably Tarski's intuition too, is that in natural languages we need truth terms a truth predicate, a truth operator, truth adverbs, etc. because natural languages do not possess enough pro–sentences, i.e. expressions that can act like the propositional variables of artificial languages. If we enriched natural languages with this kind of variable, as it happens in the semi–formalized formulations that Ramsey, and also Tarski, offer, truth terms would be dispensable. But this fact, far from supporting the redundancy of truth, explains exactly which role it performs. In Ramsey's own words:

As we claim to have defined truth we ought to be able to substitute our definition for the word "true" wherever it occurs. But the difficulty we have mentioned renders this impossible in ordinary language which treats what should really be called pro–sentences as if they were pronouns. The only pro–sentences admitted by ordinary language are "yes" and "no", which are regarded as by themselves expressing a complete sentence, whereas "that" and "what" even when functioning as short for sentences always require to be supplied with a verb: this verb is often "is true" and this peculiarity of language gives rise to artificial problems as to the nature of truth, which disappear at once when they are expressed in logical symbolism, in which we can render "what he believed is true" by "if p was what he believed, p. (1929/1991:10, Ramsey's emphasis)

As Ramsey says, words such as "yes" and "no" work as prosentences from a logical point of view but have the grammatical category of adverbs. If I ask "Is Victoria coming for dinner?" and you answer "yes", the grammatical adverb has a whole proposition as its content: Victoria is coming for dinner. If I ask "Is Victoria taller than Joan?" and you answer "yes", the grammatical adverb has as its content the proposition Victoria is taller than Joan. In both cases the adverb maintains the same linguistic meaning, i.e. it is not an ambiguous expression, but acquires a different content contextually determined. And its content is a proposition in each case: this fact can be expressed by saying that "yes" (and also "no") works as a propositional variable, a sentential pro–form or a pro–sentence. But consider the following examples (3) and (4),

(3) Joan denied what Victoria claimed,

(4) Victoria didn't say that,

"what" and "that" refer to propositions too and thus they are prosentences from the point of view of their contents. From a grammatical point of view, nevertheless, they are singular terms. The class of pro–forms includes expressions that work in every grammatical category, natural languages possess prosentences, and also pro–adverbs, pro–nouns and also pro–adjectives. Generally speaking, proforms are not eliminable from a language without loss of expressive power. This does not mean that particular uses of proforms are not eliminable. Sometimes a pronoun can be substituted by a particular name, and a prosentence by a particular sentence. But the general category has a deeply specialized role to fulfill, which in artificial languages is performed by variables of different categories. The import of Ramsey's definition, "We can say that a belief is true if it is a belief that p, and p.", is easily misunderstood because we forget that it is not formulated in natural language terms. As soon as we use propositional variables, the truth predicate can be dispensed with. But then we have abandoned natural languages and entered the realm of languages enriched with propositional variables. In these hybrid languages, truth terms are no longer necessary; but only because their job has been taken over by expressions specifically added to perform it.

Ramsey rejected epistemic approaches to truth, such as pragmatism and coherentism. He thought that his view would be interpreted as a theory of correspondence, but he was aware that the correspondentist intuition was merely a rewording of standard truth ascriptions (1927/1990:39), and not an analysis of them. He also insisted that his approach neither includes the troublesome terms correspondence and fact nor the claim that truth is a relation (1929/1990: 11). Ramsey's early formulation ofa prosentential view will become clearer in what follows.

 

3 The prosentential account of truth

Several philosophers have used the term "prosentence" to characterize the status of truth ascriptions. Bolzano was the first one to use the expression "Fürsatz" with the meaning that we give to the term "pro–sentence" here and, as Ramsey did some years later, he applied it to the grammatical adverbs "yes" and "no" (Bolzano (1904/1930: 76, 2^nd edition). More than a century after Bolzano's use and almost fifty years after Ramsey's, D. Grover, J. Camp Jr. and N. D. Belnap Jr. (Grover, Camp and Belnap 1975), on the one hand, and C.J. F. Williams (Williams 1976), on the other, developed the prosentential account independently.

A prosentential view on truth has emerged from time to time in the analytic tradition during the XX century. The origin of this view is Aristotle, with his: "to say of what is that it is and ofwhat is not that it is not is true" (Metaphysics I0II^b27). Aristotle's sentence has been paradigmatically interpreted as a formulation of the correspondence theory. Although the intuitions under the correspondence theory are widely accepted, to embrace a developed and substantive correspondence approach is another story. The correspondence intuition allows a trivial and uncommitted implementation or a metaphysically burdened one. In the first case, it asserts that truth has to do with what is said by the users of language about the world together with how the world is, and it is a harmless position. In the second case, it asserts that truth is a relational concept between two poles, the pole of language, and the pole of world. Here an account of the two poles is called for and in doing so the room for philosophical disagreement emerges. Most proponents of theories of truth have vindicated Aristotle as a precursor. Ramsey, Grover, Prior and Williams, all defenders of prosentential accounts, are no exception. And it is significant that in (1929/1990: II–I2) Ramsey struggles to distinguish his view from a correspondence theory, while presenting at the same time his own proposal on truth as an attempt to clarify Aristotle's dictum.

The prosentential theory of truth aims basically to define the predicate "is true", i.e., to offer an account of how truth terms work and how they contribute to the overall meaning of the statements in which they occur. What we seek when we analyze the truth operator is to determine the logical form of truth ascriptions. A quick way into the topic is to ask what a speaker means by the use of a truth term or in which communicative situations does an ordinary speaker (as opposed to a philosopher) put the truth predicate to work. That which the sentence

(5) snow is white

is used to express (in a standard context of use) is true if, and only if, snow is white does not seem to be in need of explanation. But this use hardly allows truth terms to earn a respectful position in language. If truth ascriptions such as (6)

(6) "snow is white" is true

were the only (or the favored) contexts in which truth terms occurred, then the redundantist conclusion that everything that can be done using a truth term can be done without i would be hard to resist.

Nevertheless, examples such as (6) are also informative; they serve to illustrate the syntactic role of the truth predicate. Quotation marks are among the tools that speakers have at their disposal to convert sentences into singular terms. A sentence such as (5) is converted into a name of it by placing it between inverted commas, "snow is white", or prefixing the sentential operator "that", that snow is white. The truth operator has the converse function, i.e. to restore the status of sentence. It is a de–nominalizer, as Horwich (I990: 5) says. If we combine together in an expression the two kinds of device, quotation marks and the truth operator, the resulting expression is again the expression with which we begin. No wonder, since the two devices neutralize each other. A poor understanding of this process has been used as an argument for the redundancy of truth. The syntactic function of "is true" is also performed by other dummy predicates, such as "is a fact". Nevertheless, the contents of sentences such as (2), (7) and (8),

(2) What Victoria says is true,

(7) Everything the Pope says is true,

(8) The Theory of Relativity is true,

etc. (in standard uses) essentially require for their expression the notion of truth. (2) is structurally ambiguous between a reading in which the subject is a singular description and an alternative reading in which the whole truth ascription expresses a general thought. The two alternatives are (2A) b. and (2B)

(2A) a. México DF is a big city (said by Victoria in context C)

(2A) b. What Victoria says is true (said by Joan in context C)

(2B) Everything that Victoria says is true.

The content of (2A)b. could be displayed without making use of a truth term just by repeating Victoria's words. Sometimes, as we have seen that Ramsey acknowledges, there are stylistic or even logical reasons that advice the use of the truth ascription instead of a more transparent expression of the content. But there are also cases in which the truth ascription is unavoidable even in its singular reading, for we sometimes endorse contents that we do not understand or even do not know, as is the case in authority arguments. Be it as it may, the general reading (2B) does not allow the elimination of the truth term. For, strictly speaking, (2B) does not say anything. It is compatible with a potentially infinite list of propositions and can be seen as a rule that allows to deploy any proposition asserted by Victoria.

The basic claim of the prosentential theory of truth is that truth ascriptions are complex propositional variables. Then, they perform the roles reserved for variables in general. Natural language variables, i.e. proforms, perform typically three tasks: direct reference, anaphoric reference and generalization. Think of pronouns, the best known among proforms. We use them to refer to objects contextually salient, to give unity to a discourse linking the contents of referential expressions, and in the expression of general thoughts. In the same sense in which the pronoun "he" can be used to refer to John, to the President of Mexico, or to my son, depending on the context, a sentence such as (2) can be used to express the contents that Mexico DF is a big city, that it's heavily raining or that the cat is on the mat, depending on which one had been the content of Victoria's speech act. Nevertheless, the task in which proforms display their utility is in the expression of general thoughts.

General thoughts require variables. If a statement purports to cover a variety of items in a class, it has to include expressions that can range on the members of the class. The expression of general thoughts on numbers, such as (9),

(9) For any integer n > 2, there are no positive integers a, b and c that can satisfy the equation aⁿ + bⁿ = cⁿ (Last Fermat Theorem), need variables ranging on numbers.

The expression of general thoughts on sets and functions, such as (10),

(10) For each set A there exists a function f whose domain is the collection of non–empty subsets of A and, for every B ⊆ A with B≠ ∅f(B) ∈ B (Axiom of Choice)

need variables ranging on sets and variables ranging on function. The expression of general thoughts on concepts, such as (II),

(11) The extensions of any concepts, F and G, have the same size if and only if there is a one–to–one correspondence between the F–things and the G–things (Hume's Principle),

need variables ranging on concepts. The expression of general thoughts on propositions, such as (12),

(12) For every proposition p, p v ¬ p, need variables ranging on propositions.

Examples (9)–(12) are expressed in a semiformal language: English + ∅the appropriate kind of variables∅ + ∅symbols for particular operations∅. To express their contents in plain English, we need to include proforms, the natural language counterparts of variables in artificial languages. Truth terms are mechanisms for the construction of propositional variables, so the systematic use of truth terms in the formulation of general thoughts on propositions is exactly what we should expect. A reasonable translation of (12) is (13),

(I3) For all proposition, either it is true or its negation is.

In (I3) there are two nominal proforms, "it" and "its negation", and two sentential proforms, one explicit, "it is true", and another one elliptic, "its negation is". All of them are prosentences, the two former nominal prosentences, i.e. expressions that present the grammatical status of singular terms but that have whole propositions as their contents, and the latter two are sentential prosentences, expressions with the grammatical status of sentences which express propositions when used in appropriate contexts.

A language deprived from the possibility of expressing general thoughts would be a very impoverished language from the point of view of its expressive power. Natural languages are not of this kind: they include means for the expression of general thoughts, without which we probably would not develop as the rational creatures we are. Truth terms help to construe prosentences, and thus they are essential to the expression of general thoughts that involve propositions.

And now let's go back to Ramsey. In both (1927) and (1929) Ramsey offered an analysis of the functioning of the predicate "is true" and drew its connections with the propositional variables of artificial calculi. He also introduced the word "pro–sentence" and analyzed the truth predicate by analogy with pronouns. He explained the intuitions that give support to the Redundancy theory of truth and also to the Correspondence theory, and showed that they can be accommodated within a developed pro–sentential theory. During the past century, some other philosophers develop a similar view independently, among them Strawson, Prior, Grover, Williams and, the most recent prosententialist, Robert Brandom. It is our contention that the prosentential theory is not only the theory that Ramsey embraced, but also the best informed proposal about the functioning of truth terms.

 

Bibliography

Austin, J. (1950), "Truth". In Blackburn, Simon and Simmons, Keith (eds.): Truth. Oxford, Oxford University Press, 1999, pp. 149– 162.         [ Links ]

Bolzano. B. (1904/1930), Wahrheit und Evidenz. Felix Meiner Verlag, 1974, 2^nd Edition        [ Links ]

Brandom, R. (1994), Making it Explicit. Reasoning, Representing, and Discursive Commitment. Harvard University Press        [ Links ]

Engel, P. and Dokic, J. (2003), Frank Ramsey. Truth and Success. London and New York, Routledge.         [ Links ]

Frápolli, M.J. (2011), "The Neutrality of Truth in the debate Realism vs. Anti–Realism". In Marion, M., Rahman, S., Primiero, G. (eds.), The Realism–Antirealism Debate in the Age of Alternative Logics, The Netherlands, Springer, Series: Logic, Epistemology & Unity of Science, pp. 85–99.         [ Links ]

Frápolli, M.J. (2012), On Truth. An updated approach to the meaning oftruth ascriptions. Springer, forthcoming        [ Links ]

Grover, D., Camp, J. Jr., and Belnap, N. Jr. (1975), "The prosentential theory of truth". Philosophical Studies: An International Journal for Philosophy in the Analytic Tradition, Vol. 27, No. 2, pp. 73–125        [ Links ]

Horwich, P. (1990), Truth. Oxford, Basil Blackwell (2^nd edition, Oxford, Clarendon Press, 1998)        [ Links ]

Mellor, D.H. (ed) (1990): Philosophical Papers. F. P. Ramsey. Cambridge, Cambridge University Press.         [ Links ]

Prior, A. (1971), Objects of Thought. Oxford: Clarendon Press        [ Links ]

Ramsey, F. (1927/1990): "Facts and Propositions". In Mellor (ed.) (1990), pp. 34– 51        [ Links ]

Ramsey (1929/1991): "The Nature of Truth". In Rescher and Majer (eds.) (1991), pp. 8–20        [ Links ]

Recanati, F. (2000), Oratio Obliqua, Oratio Recta. An Essay on Metarepresentation. Cambridge, Mass., London, England, The MIT Press.         [ Links ]

Rescher, N. and Majer, U. (eds.) (1991): On Truth. Original Manuscript Materials (1927–1929) from the Ramsey Collection at the University of Pittsburgh. Dordrech, Kluwer Academic Publishers.         [ Links ]

Sahling, N–E. (1990): The Philosophy of F. P. Ramsey. Cambridge, Cambridge University Press.         [ Links ]

Strawson, P. F. (1950), "Truth". In Simon Blackburn and Keith Simmons (eds.): Truth. Oxford Readings in Philosophy. Oxford University Press, 1999, pp. 162–182.         [ Links ]

Tarski, A. (1935), "The concept of truth in formalized languages". In Logic, Semantics, Metamathematics, Indianapolis, Hackett Publishing Company, 1983.         [ Links ]

Williams, C. (1995): "The Prosentential Theory of Truth". Reports on Philosophy, n° 15, pp.147–154.         [ Links ]