SciELO - Scientific Electronic Library Online

vol.48 número143Respuestas disposicionalistas al problema wittgensteiniano-kripkeano sobre el significado y el seguimiento de reglas índice de autoresíndice de materiabúsqueda de artículos
Home Pagelista alfabética de revistas  

Servicios Personalizados




Links relacionados

  • No hay artículos similaresSimilares en SciELO


Crítica (México, D.F.)

versión impresa ISSN 0011-1503

Crítica (Méx., D.F.) vol.48 no.143 Ciudad de México ago. 2016  Epub 30-Ene-2020



Propositions, The Tractatus, and “The Single Great Problem of Philosophy”

Scott Soames* 

*School of Philosophy, University of Southern California,


Wittgenstein’s project in the Tractatus was to replace Frege-Russell propositions with a new conception capturing the essence of representational thought and language. This, he believed, was philosophy’s only real task. I argue that his account of atomic propositions was an incomplete realization of valuable insights, which, had they been slightly revised, could have been extended to all tractarian propositions. Had Wittgenstein followed this path, he would have made discoveries in the study of language and mind that are only beginning to emerge today. However, doing so would have meant stripping the Tractatus of its pretensions of fundamentally remaking philosophy.

Key Words: uses of sentences; truth bearers; meaning; thought; representation


El proyecto de Wittgenstein en el Tractatus consistía en reemplazar las proposiciones de Frege y Russell con una nueva concepción que capturara la esencia del pensamiento y el lenguaje representacionales. Según creía, ésta era la única tarea real de la filosofía. Argumento aquí que su explicación de las proposiciones atómicas fue una implementación incompleta de intuiciones valiosas, que, si hubieran sido ligeramente revisadas, podrían haberse extendido a todas las proposiciones tractarianas. Si Wittgenstein hubiera seguido este camino, habría hecho descubrimientos en el estudio del lenguaje y la mente que apenas empiezan a emerger hoy en día. Sin embargo, hacerlos habría implicado quitarle al Tractatus sus pretensiones de reconstruir de manera fundamental la filosofía.

Palabras Clave: usos de oraciones; portadores de verdad; significado; pensamiento; representación

For Wittgenstein, the central task of the Tractatus was to replace Frege-Russell propositions with a fundamentally new conception. I will identify what I take to be the essence of that conception with the goal of refining and perfecting his insights, while avoiding the problems that confounded him, in a way that allows us to advance a key philosophical project we have inherited not only from Wittgenstein, but also from Frege and Russell. In the Notebooks Wittgenstein says:

My whole task consists in explaining the nature of the proposition.1

The problem of negation, of conjunction, of true and false, are only reflections of the one great problem in the variously placed great and small mirrors of philosophy.2

Don’t get involved in partial problems, but always take flight to where there is a free view over the whole of the single great problem.3

To solve the one great problem, explaining propositions, was to identify the essence of representational thought and language. This, Wittgenstein thought, was philosophy’s only real task.

It should be noted at the outset that his task was not just to explain what propositions are, but to inventory the full range of them in order to construct a criterion of intelligibility, and use it to demonstrate the impossibility of finding any propositions the truth of which it was the job of philosophy to discover. The audacity of this project -to survey and categorize the range of all possible thought- was stunning. Naturally it didn’t succeed. Nevertheless, there is much to be learned from the attempt.

Like Frege and the early Russell, Wittgenstein took sentences to be the primary units of meaning, but unlike them he didn’t take meanings of sentences to be propositions. Instead, he denied that any entities were sentence-meanings. He agreed that propositions are bearers of truth, but he took them to be akin to meaningful sentences, rather than imaginary sentence meanings. For him sentences are linguistic facts consisting of expressions standing in syntactic relations. For them to be meaningful is for them to be governed by conventions. E.g., the sentence “Costa Rica is south of California” consists in the two names standing in a certain syntactic relation R -which involves, among other things, the first name being followed by the phrase “is south of” which is followed by the second name. The sentence is the fact that the two names stand in that relation. The conventions governing the sentence stipulate (i) that the names are used to designate the central American country and the American state, and (ii) that structures in which two names stand in R are used to represent the referent of the first as being south of the referent of the second. One who uses the sentence in this way represents Costa Rica as being south of California. From this we derive the condition the world must satisfy if one’s use of the sentence is to be true.

It is tempting to think that the bearer of truth is the sentence, or propositional sign, in which the two names are united by relation R.

After all, this linguistic structure is governed by the conventions I mentioned. But it didn’t have to be. Had it been governed by other conventions, it would have meant something different, and had different truth conditions. So there is something, the syntactic structure, that is used to represent Costa Rica as being south of California, but could have been used differently, and so had different truth conditions. However, there is also something that has these truth conditions essentially. We say: “Necessarily the proposition that Costa Rica is south of California is true iff Costa Rica is south of California.” This wouldn’t be true, if propositions were mere syntactic structures, or tractarian propositional signs. Hence propositions aren’t such signs. Wittgenstein agrees; he takes it for granted that propositions have their truth conditions essentially.

Can we accommodate this by taking propositions to incorporate both propositional signs and the conventions governing them? Let the conventions be those I stated. Perhaps the tractarian proposition that Costa Rica is south of California is a use of the propositional sign in accord with those conventions. What is this entity -a use of a sentence S in accord with conventions? Since to use S is to do something, a use of S is a cognitive doing, an act or operation of some sort. It is the act of using the two names to designate the country and state, while using the relation R to represent the referent of the first name as being south of the referent of the second. This repeatable act type represents Costa Rica as being south of California, in the sense that for an agent to perform it is for the agent to represent them that way. Since Costa Rica is south of California, this use of the sentence is true. On this picture, uses of sentences are representational cognitive act types or operations. For them to be true is for agents who perform them to represent things accurately, as they really are. The fact that they are repeatable act types means that they are a certain kind of abstract object, distinguished from the concrete events that occur when agents perform them.

This reconstruction preserves several tractarian themes. (i) It explains the meaningfulness of the sentence without positing an independent entity as its meaning. (ii) It identifies the truth-bearer, the meaningful use, as an entity the truth of which is defined in terms of its representational accuracy. (iii) It preserves the idea that the constituents of the sentence are isomorphic to the constituents of the atomic fact that makes a use of it true. (iv) Since the conventions governing use are those governing the sentence’s constituents, no extra convention governing the sentence as a whole is needed. (v) The proposition has its truth conditions essentially because any possible agent using the sentence in this way represents Costa Rica as being south of California.

Although this is as it should be, it isn’t exactly what Wittgenstein had in mind. Uses of sentences do represent, or, as he like to say, picture, reality. So, on my story, these uses could, defensibly, be called “pictures”. But they are not exactly Wittgenstein’s pictures. His pictures are supposed to be facts, not acts.

  • 2.14 The picture consists in the fact that its elements are combined with one another in a definite way.

  • 2.141 The picture is a fact.

  • 2.21 The picture agrees with reality or not; it is right or wrong, true or false.

This tells us that truth bearers are facts. We know that propositional signs are facts. Could they be tractarian propositions? At one point, Wittgenstein says that propositions are perceptible, which may seem to suggest that they are.

  • 3.1 In the proposition the thought is expressed perceptibly through the senses.

But he also distinguishes propositions from propositional signs.4

  • 3.11 We use the perceptible sign of a proposition (spoken or written, etc.) as a projection of a possible situation.

  • The method of projection is the thinking of the sense of the proposition.

  • 3.12 The sign through which we express the thought I call the propositional sign. And the proposition is the propositional sign in its projective relation to the world.

  • 3.13 A proposition includes all that the projection includes, but not what is projected.

  • Therefore, though what is projected is not itself included, its possibility is.

  • A proposition, therefore, does not actually contain its sense, but does contain the possibility of expressing it.

  • 3.14 The propositional sign consists in the fact that its elements, the words, are combined in it in a definite way. The propositional sign is a fact.

Although propositional signs are facts, they seem not to be identical with propositions. Propositional signs are syntactic combinations of words which, though meaningful, aren’t individuated by what they mean. It is tempting to say that the sense of a proposition is a possible fact that consists of the objects designated by its names being combined in the way they are represented as combining. If that were so, then the sense of the proposition would the possible fact that would make it true, were that fact actual. But that isn’t Wittgenstein’s view. For him, no fact is merely possible. He registers this obliquely by saying that propositions don’t contain their senses. They can’t because there are no facts for false propositions to contain, and because we must grasp the sense of a proposition before we know whether it is true or false.

Recall his words. “The method of projection is the thinking of the sense of the proposition.” In thought, the proposition we entertain represents worldly items -the objects that are projections of the names in the propositional sign- as standing in the relation that is the projection of the relation R that unites the names in the propositional sign. We are told that the proposition “includes all that the projection includes, but not what is projected”. This last item, what is projected, is the sense of the proposition -the possible fact. It isn’t “included” in the proposition; nor are the objects and relations that are projections of the constituents of the propositional sign. But the rest of the projection is included. What are these remaining items? They must be whatever elements are responsible for determining what the names and the syntactic relation R project; they are the conventions governing the names plus the convention governing R. They are needed to determine what fact would have to exist if the proposition were true. These conventions, which aren’t included in the propositional sign, are somehow included in the proposition as what one must know in order to understand its representational content.

How are they included? The propositional sign is a purely syntactic structure in which symbols stand in a certain relation. Wittgenstein tries to identify the proposition using the phrase, the propositional sign in its projective relation to the world. Unfortunately, this language, the sentence S in its relation to the world, doesn’t pick out an entity other than S -any more than the phrases Scott-in-his-relation-to-USC, Scott-in-his-relation-to-his-wife, or Scott-in-his-relation-to-the-Latin-American-Association-for-Analytic-Philosophy pick out entities other than me of which I am, nevertheless, an essential part. There aren’t several Scotts, or Scott-complexes, here, just misleading ways of talking about the fact that I teach at USC, live with my wife, and lecture at the philosophy conference. The same is true of Wittgenstein’s talk of propositional signs in their projective relations to the world.

This confused terminology parallels all-too-familiar contemporary talk of interpreted versus uninterpreted sentences. When speaking of a language like English, these phrases don’t designate two kinds of sentences; they are two ways of talking about the same sentences.5 Any English sentence is a syntactic structure uses of which are governed by linguistic conventions. These uses have their truth conditions essentially. To say a sentence is meaningful is to say that the contingent conventions governing its use endow uses of it with representational content. These uses, i.e., these cognitive acts or operations, are propositions. Wittgenstein rightly denied that propositions are propositional signs, while wrongly attempting to identify them with sentences-as-used-in-accord-with-the-conventions. The remedy is to reject these pseudo entities and to slightly amend the Tractatus by taking propositions to be uses of sentences.

Insights and Errors in the Tractatus

In doing this we remain true to his idea in the Tractatus that although propositions aren’t sentences, talk of propositions is talk about sentences. However, the idea needs still further correction. It is essential to thought that agents represent things as being certain ways. It is not, or at least not obviously, essential to thought what, if any, artifacts they use in so representing them. Surely, one is inclined to think, any organism whose cognitions can be true or false represents things as being various ways. Sometimes it does so by using symbols. But there is no obvious reason to believe that an agent always uses symbols when thinking of something as dangerous, or when perceiving one thing as larger than another. When agents perform representational cognitive acts linguistically, the propositions they affirm may be uses of symbols. When they non-linguistically represent things as being certain ways, the propositions they affirm seem not to be symbols.

  • Thus, I am inclined to disagree with what Wittgenstein says at 4.0312.

  • 4.0312 The possibility of propositions is based upon the principle of the representation of objects by signs.

Max Black says, “It is essential to Wittgenstein’s conception that the proposition should be expressed in a sentence. . . A disembodied proposition would be an absurdity.”6 But why? If one kind of cognitive act, a use of a sentence, can represent things accurately or not, and so be true or false, why can’t the same be said of cognitive acts in which we non-linguistically perceive, imagine, or think of things as being certain ways? If these are possible thoughts, the tractarian insistence on symbolic representation misrepresents the essence of thought.

This criticism takes us a step toward Wittgenstein’s later philosophy, when he rejected identifying the essence of thought with the referential essence of language. But my critique goes further, while also pointing in a different direction. I’m not making the later Wittgensteinian point that there are no apriori limits to the variety of uses of language, though that too may have merit. My point is that there is no apriori requirement that representational thought be symbolic. The tractarian account of atomic propositions is, for me, an incomplete realization of three genuine insights. (i) Declarative sentences are representational, not because they express primitively representational Fregean or Russellian propositions, or because they name possible facts, but because of how they are used. (ii) They are bearers of truth because they are used to represent things as bearing certain properties and standing in certain relations. (iii) The truth conditions of a use of an atomic sentence are read off its representational properties; a use is true at a world-state w iff were w actual things would be as the use always and everywhere represents them.

This approach allows different propositions to be true at the same world-states. Hence, it raises a question. What use of which sentence is identical with the proposition that Costa Rica is south of California? There is no more reason to identify it with a use of an English sentence than there is to identify it with a use of a Spanish sentence, or with a use of a sentence of another language. The proposition we are looking for is something all representationally identical uses of certain sentences have in common. With this in mind, consider the representational act of using some sentence or other to represent Costa Rica as being south of California. Anyone who uses a particular sentence S in this way, thereby also performs a general representational act that one can perform without using that sentence. If acts of using particular sentences are propositions, then this general representational act should also be. It is a proposition that everyone using any individual sentence to predicate being south of California of Costa Rica thereby entertains.

What about the act of predicating the property being south of California of Costa Rica -cognizing the two as so-related by any means whatsoever. Surely, it is the best candidate for being the proposition that Costa Rica is south of California. If it’s not possible to perform this most general act without using a sentence, then it’s identical with the act of using some sentence or other to so represent the country and the state. If, as I believe, it is possible to perform the general act without using any symbolic intermediary, then it alone is the proposition we seek. Of course, it’s not really the proposition that Costa Rica is south of California. There isn’t just one proposition the representational content of which is exhausted by its representing Costa Rica as being south of California. There are many such propositions. Having seen this, we must reject the assumption, which is the source of Kripke’s puzzle about belief, that sentential clauses, that A is B, are fine-grained enough to pick out all the propositions we need.7

They aren’t.8 Nevertheless, there will always be a single cognitively most general proposition p that is entertained whenever we entertain a cognitively more specific proposition q representationally identical with p. In our example, it is the act of predicating being south of California of Costa Rica, no matter what means one uses to perform the predication.

The lesson here is worth emphasizing. Once we identify propositions as representational cognitive acts or operations, we must individuate propositions in the same way we individuate other act types. Think of the relationship between the act of driving to work and the act of traveling to work. Anyone who performs the first, thereby performs the second as well. However, since there are many ways of getting to work, one can perform the second act without performing the first. Thus the acts are different. When we apply these individuation conditions to cognitive acts that are propositions, we generate pairs, or in some cases n-tuples, of representationally identical propositions that are nevertheless cognitively distinct, because the cognitive demands they place on agents who entertain them are different. These propositions are not merely equivalent, or true in the same possible world-states. They are representationally identical in predicating precisely the same properties of precisely the same things. Nevertheless, they are cognitively distinct, which (often) means that agents can bear a propositional attitude to one without bearing it to the other.9 All of this follows rather naturally once we take propositions to be cognitive acts of a certain type, including uses of sentences in accord with conventions, as the proper reconstruction Wittgenstein’s problematic characterization of propositions as sentences-in-their-projective-relation-to-reality.

Truth-Functionally Complex Propositions

Suppose then that atomic propositions are acts of representing objects as being certain ways, sometimes or always using sentences to do so. How should we understand truth-functional compounds of these propositions? Shouldn’t they also be acts of representing objects -tractarian metaphysical simples- as being various ways? It is natural to think they should. There are at least two ways of achieving this. One way is by associating any proposition that represents things as being so-and-so with the property being such that things are so-and-so. This association is achieved by a cognitive operation that converts a proposition p into a property, being such that p, that is guaranteed to apply either to everything or to nothing. Starting from the propositions that a is F and that b is G, we can derive the properties being such that a is F and being such that b is G, each of which is had by everything if it is had by anything. Next we disjoin the properties, generating the disjunctive property being such that a is F or being such that b is G. Since to predicate it of anything is to represent a as being F or b as being G, the disjunctive proposition that a is F or b is G can be identified with the proposition that predicates this property of a and b (or perhaps, of everything). Other truth functions can be treated similarly.

If one worries about deriving these all-or-none properties from propositions, one can tell a different story. This time we let the disjunction of propositions that a is F and that b is G be the act of operating on them to produce a proposition, not a property, that represents the pair a,b as standing in a relation that consists of the first’s being F or the second’s being G. As before, other truth functions can be treated similarly.10

For our purposes it is not important which of these act-theoretic accounts of truth-functionally compound propositions we choose. The crucial point is that, whichever we choose, truth-functionally compound propositions turn out not to predicate truth or falsity of their propositional constituents. Far from being a defect, this is a tractarian desideratum. According to the Tractatus, nothing can be intelligibly stated about the representational relationship between propositions and the world. Since truth for propositions is defined as representational accuracy, predicating truth of a proposition violates this doctrine. Partly for this reason, Wittgenstein denied that the grammatical predicate “is true” expresses a genuine property. In the Notebooks he says there is no representational difference between p and the claim that p is true. In fact, he calls the latter a pseudo-proposition that attempts to say what can only be shown.11 Max Black makes a similar point. He says:

(a) [ ⌈“p” is true⌉ ] must be regarded as misleading and excluded from formulation in “a correct ideography” [the ideal object language of the Tractatus]. For there is no place in Wittgenstein’s conception of language for talk about propositions, as seems to occur in (a). All significant propositions refer to the world by having their components stand proxy for objects in the world, but a proposition is not an object, and any method of symbolization that suggests the contrary must be incorrect.12

Although it may seem astounding, Black is right. According to the Tractatus, no proposition predicates anything of propositions.13 The Tractatus, which says so much about propositions, says there are no propositions about propositions. Since some propositions are negations, conjunctions, and disjunctions, they don’t predicate truth or falsity of their constituent propositions. QED.

This presents an interpretive problem. The idea that one can’t intelligibly predicate truth of anything can hardly be taken seriously by anyone who wants to give a semantic theory of referential uses of language, or a philosophical theory of representational thought. Because Wittgenstein attempted both, this leaves us with two interpretative alternatives. One, suggested by Black, is to provide some of his talk of truth conditions with interpretations in which truth isn’t predicated of anything. The other is to avert our eyes from his incorrect doctrines about truth and reference until we are forced, in the final pages of the Tractatus, to include them in the scope of his conclusion that most of the Tractatus is unintelligible. My reading is a blend of these strategies.

Propositions that predicate truth of other propositions can’t be excluded from uses of sentences of what is, in effect, the tractarian metalanguage -i.e. the language in which the Tractatus is written. So, I will continue to say that negations are true whenever the negated propositions aren’t true, and so on. But we shouldn’t interpret sentences of the ideal object language of thought postulated by the Tractatus as predicating anything of propositions. Although this limits its expressive power, that is nothing new. We know that the hidden tractarian language of logical form doesn’t include reports of what agents believe, assert, or know.14 But it must include sentences expressing negative, conjunctive, and disjunctive propositions. Thus, we can’t take truth-functional compounds to be propositions that predicate truth or falsity of their constituent propositions.

My act-theoretic story is consistent with this prohibition. The only alternative I know of, which may well be what Wittgenstein had in mind, is mysterious. It says that the disjunction of p and q is the unique proposition that must be true iff p is true or q is true -without explaining what that proposition is, what it represents as being what ways, or how it can have truth conditions at all. Because of this, the mysterious analysis requires two theories of truth- one defining truth for atomic propositions as representational accuracy and one reducing truth for truth-functional compounds to the truth or falsity of atomic propositions. Two theories of meaning are also needed. To know the meaning of an atomic sentence is to know which things it represents as being which ways. To know the meaning of a truth-functional compound is know how its truth or falsity is determined by the truth or falsity of atomic sentences.

To this duplication, I add three related worries. First, if truth-functionally compound propositions can be identified only by using an illegitimate truth predicate, then no agent can identify them without affirming pseudo-propositions, and thereby making a mistake. How can that be? Second, if understanding truth-functionally compound sentences requires knowing their truth conditions, which, in turn, requires knowing they are true iff various atomic sentences or propositions are true (or false), then mastery of the “ideal” language of the Tractatus requires knowing pseudo-propositions. But that’s impossible: pseudo-propositions can’t be known. Third, any theory that identifies understanding some sentences with knowing their truth conditions must invoke a notion of truth in which sentences S and ⌈“S” is true⌉ are not apriori consequences of one another.15 Wittgenstein had no such conception.

In short he had no defensible account of truth-functionally compound propositions. We have remedied this defect by providing an account of them that fits his treatment of atomic propositions. According to our account, truth-functional compounds are acts of using sentences to represent metaphysical simples as having properties derived from atomic propositions. This isn’t exactly what Wittgenstein had in mind. But it does preserve his most valuable insights.

General Propositions:

Finally, we turn to general propositions, which Wittgenstein expresses using a joint denial operator that takes indefinitely many propositions as arguments.16 Consider the proposition that all Fs are Gs. To express this we start with a formula containing free occurrences of the variable “x”. Each use of the formula predicates being both F and ~G of the object “x” is used to designate.17 The class of all such uses contains, for each object o, the proposition that o is both F and ~G. To jointly deny all these propositions is to predicate the property not being F unless it is G of everything. How does one predicate this, or any property, of everything? We can do this by generalizing what happens with Fregean definite descriptions. To predicate being G of the F is really to predicate determining something that is G to the individual concept associated with “the F”.18 Similarly to predicate not being F unless it is G of everything is really to predicate the property determining items that are not F unless they are G of a general concept -i.e. one that determines each thing.19 This idea isn’t explicitly tractarian, but it preserves the insights behind Wittgenstein’s rejection of propositions as abstract objects the representational natures of which are independent of cognitive agents.20


This completes my reconstruction of Wittgenstein’s revolutionary conception of propositions. His key insight was to turn the view of Frege and Russell on its head. Instead of attributing the ability of agents to represent things as being so and so to their bearing a mysterious entertainment relation to an equally mysterious abstract proposition that primitively represents things as being so and so, he took the representational features of propositions to be derived from the cognitions of agents. Focusing on pictures, models, and sentences, he saw that our use of them to represent objects as bearing properties and standing in relations was crucial to understanding propositions. Unfortunately, he failed to turn this insight into a real solution to his “single great problem” of philosophy. I have argued that his failure can be traced to his confusing uses of sentences to represent this or that with sentences-as-used-to-represent-this-or-that. The first is a cognitive act that represents the world because any possible performance of it does. The second is a pseudo-entity: a contingent artifact the truth conditions of which are essential to it. There is no such thing.

The two ideas, uses of sentences versus sentences-as-used, also generalize differently. The proposition that Costa Rica is south of California can’t be the use of a single sentence. But it can be the use of any sentence to predicate the property being south of California of Costa Rica. Even better, it can be the act of so predicating, with or without a linguistic intermediary. I have generalized this idea to quantified propositions and truth-functional compounds. The sentence-as-used idea is harder to generalize. One attempt to do that comes from Frank Ramsey’s insightful interpretation of the Tractatus, which can itself be expressed in the style of contemporary intensional seman-tics.21 The idea is to posit highly abstract artifact-types, instances of which are (imagined) sentences-as-used-at-a-context. These items are said to have truth conditions at possible world-states. Let S1 be any sentence and w i be any possible world-state at which linguistic rules govern its use. Given this, we generate a theorem: S1-as-used-at-w i is true-at world-state w* if and only if at w* x is so-and-so. We do the same thing for any sentence S2-as-used-at-w j . This gives us the set of world-states at which S1-as-used-at-w i is true and the set of world-states at which S2-as-used-at-w j is true. We then stipulate that the proposition of which S 1 -as-used-at-w i is an instance = the proposition of which is S 2 -as-used-at-w j is an instance iff those two sentences-as-used-at-their-respective-world-states are true at the same world-states. That’s what it is for them to be instances of the same artifact type.

What is this thing, the type of which S1 and S2 as-used-at-two-world-states are instances? On thing is certain; it isn’t anything resembling a sentence type or any other recognizable linguistic artifact. Thus we risk losing the initially promising idea that propositions are meaningful sentences, uses of sentences, or artifacts put to representational uses. But whatever these abstract types are supposed to be, we know they are identical iff their instances are true at the same world-states. With this we derive one of the most important, but also most problematic, doctrines of the Tractatus, namely, that necessarily equivalent propositions are identical. The derivation of this doctrine, which relies on truth conditions of sentences-as-used-at world-states, highlights what seems to me to be an underlying absurdity. What is really being said by those who speak of S-as-used-at-a-world-state @ being true at w? What they are really saying is that when S is used at given world-state, @, it expresses a proposi-tion that would have been true if w had been actual. But if that’s right, this development of Ramsey’s interpretation of the Tractatus presupposes propositions rather than explaining them.22

For these reasons, I take the cognitive act-type theory be the best reconstruction of tractarian propositions. With this in mind, let us return to the idea of a particular type of use of a sentence to predicate a property of objects. Such a use is true at a world-state w iff were the universe in state w things would be as the use represents them. Note, what a use of a sentence represents is not indexed to a world-state. It represents what any actual or possible agent who used the sentence in that way would thereby represent. Since this doesn’t change from world-state to world-state, uses of sentences have their representational properties, and hence their truth conditions, essentially.

This allows us to reconstruct an account applying to all propositions that vindicates rather than betrays the insights behind Wittgenstein analysis of atomic propositions. We proceed in stages. Stage 1 propositions are acts of using a specific sentence to predicate a property of objects. Stage 2 propositions are acts of using some sentence or other to perform the predication. Stage 3 propositions are acts of performing the predication whether or not one uses any sentence to do so. Each stage includes atomic and non-atomic propositions. At no stage is truth at the same world-states sufficient for propositions to be identical. At each stage, representing the same objects as bearing the same properties is necessary and sufficient for the propositions to be representationally identical. If all that mattered was representational identity, genuine propositions could be limited to stage 3. If, more plausibly, fine-grained propositions are needed to deal with the full range of Frege’s puzzle, then all three types should be propositions.23 This analysis takes us well beyond the Tractatus, while capturing the insights behind its account of atomic propositions and avoiding its problems with non-atomic propositions. It also avoids identifying necessarily equivalent propositions, which was a barrier to the breakthrough that Wittgenstein’s account of propositions might otherwise have been. However, he himself would not have agreed. Without the identification of necessarily equivalent propositions, the Tractatus would not have had the far-reaching consequences for philosophy, and its self-conception, that he passionately desired. These were among the consequences that led him to take the problem of the proposition to be “the single great problem” of philosophy. Had he correctly conceived, and then solved, that problem, he would have seen that its solution, though important to philosophy, linguistics, and psychology, would not have been the world-changing event he dreamed of.


Black, Max, 1964, A Companion to Wittgenstein’s Tractatus, Cambridge University Press, Cambridge. [ Links ]

Kripke, Saul, 1979, “A Puzzle about Belief”, in Avishi Margalit (ed.), Meaning and Use, Reidel, Dordrecht, pp. 239-283; reprinted in Peter Ludlow (ed.), Readings in the Philosophy of Language, MIT Press, Cambridge, 1997, pp. 875-920. [ Links ]

Ramsey, F.P., 1923, “Critical Notice of the Tractatus Logico-Philosophicus”, Mind, vol. 3, pp. 465-478. [ Links ]

Soames, Scott, forthcoming, The Analytic Tradition in Philosophy , Vol. 2, Princeton University Press, Princeton and Oxford. [ Links ]

______, 2016, “Yes, the Search for Explanation Is All We Have: Reply to Stephen Schiffer and Ben Caplan,” Philosophical Studies, vol. 173, pp. 2565-2573. [ Links ]

______, 2015, Rethinking Language, Mind, and Meaning, Princeton University Press, Princeton and Oxford. [ Links ]

______, 2013, “Cognitive Propositions,” Philosophical Perspectives, vol. 27, pp. 479-501. [ Links ]

Soames, Scott, 2003, Philosophical Analysis in the Twentieth Century , Vol. 1, Princeton University Press, Princeton and Oxford. [ Links ]

Wittgenstein, Ludwig, 1922, Tractatus Logico-Philosophicus, trans. by C.K. Ogden, Routledge and Kegan Paul , London, reprinted in 1999 by Dover; retranslated by D. Pears and B. McGuinness, Routledge, London, 1961. [ Links ]

______, 1914-1916, Notebooks. 1914-1916, second edition, G.H. von Wright and G.E.M. Anscombe (eds.), trans. by G.E.M. Anscombe, Blackwell, Oxford, 1979. [ Links ]

1 Wittgenstein 1914-1916, p. 39.

2Ibid., p. 40.

3Ibid., p. 23.

4Throughout this essay, italicized quotations from the Tractatus employ the Pears and McGuinness translation. All other quotations from the Tractatus use the Ogden translation.

5On page 98 of his generally excellent commentary, Max Black (1964) seeems to succumb to the same error when he contrasts tractarian propositions, thought of a meaningful sentences, with “uninterpreted sentences”. This is followed on page 99 by the remark “The word Satz is used in German to stand for what we would call a ‘sentence’ as well as for what we would call a ‘proposition’ (or ‘statement’. . . ). Wittgenstein sometimes distinguishes the two senses by using ‘propositional sign’ (Satzzeichen, 3.12a) for the sentence. . . It is essential to Wittgenstein’s conception that the proposition should be expressed in a sentence. . . A disembodied proposition would be an absurdity. Thus it is natural for him to use Satz to cover both aspects - the perceptual sign and its sense. . . . [I]t is essential to a proposition that it makes an abstract truth-claim.” Essentially the same confusion occurs in his discussion on pp. 81-82 of a “picture-vehicle” and “a picture in the full sense when its elements have been coordinated in a determinate way with objects, upon the understanding that those objects are supposed to be connected as their proxies are in fact connected in the vehicle”.

6 Black 1964, p. 99.

7 Kripke 1979.

8 Soames 2015, chapter 4.

9For applications of this lesson to familiar problems in the philosophy of language and mind, see Soames 2013, 2015.

10See Soames 2016.

11 Wittgenstein 1914-1916, p. 9.

12Black 1964, p. 218.

13In the Tractatus no propositions predicate properties of anything other than metaphysical simples. This is discussed at some length in chapter 2 of Soames (forthcoming).

14On the hiddenness of logical form see Tractatus 4.002. For a discussion of the tractarian inexpressibility of propositional attitude reports see pp. of 240-244 of Soames 2003.

15Otherwise knowing the apriori truth that the earth is round iff the earth is round would provide a monolingual speaker of Spanish with apriori knowledge of the truth conditions of the English sentence “the earth is round”.

16Since the arguments can be given by complete sentences or by formulas containing free occurrences of variables, we need variable binders he didn’t provide. For details about how to construct an appropriate tractarian system, see chapters 2 and 3 of Soames (forthcoming).

17The relevant formula is N(NFx, Gx). One gets the effect of Tarskian assignments of values to variables by assigning truth conditions to uses of formulas in which the convention governing variables is that they can be used to designate any object. Chapter 3 of Soames (forthcoming) explains how to use this idea to get Tarski-like results.

18This is what is called “mediate predication” in Soames 2015, 2016.

19 Soames 2016.

20This way of explicating quantification exploits the fact that unrestricted universal quantification is the only quantification in the Tractatus. If the system included all generalized quantifiers —all Fs, some Fs, most Fs, etc.— it might be better to take quantificational statements to predicate higher-order properties —e.g., being true of all, some, or most Fs— of lower-order properties. See Soames (forthcoming).

21 Ramsey 1923.

22Pages 12-13 of Soames (2015) apply this argument to versions of contemporary possible-worlds semantic theories that identify propositions with sentences-as-used-at-world-states.

23See Soames 2013, 2015.

24This essay was delivered as the concluding keynote address to the Latin American Association for Analytic Philosophy on May 28, 2016 in San Jose, Costa Rica.

Received: May 31, 2016; Accepted: June 09, 2016

Creative Commons License This is an open-access article distributed under the terms of the Creative Commons Attribution License