“The limit of language is showed by its being impossible to describe the fact which corresponds to (is the translation of) it) without repeating the sentence. (This has to do with the Kantian solution to the problem of philosophy)
(Culture and Value, p. 10)
The problem that I will address in this essay is the problem of intentionality, as seen by Wittgenstein. Since I cannot differentiate between Wittgenstein’s arguments and what I myself believe about this issue, my essay might be equally rightly be regarded as exegetical or claiming to originality. The assumption which I use is that Wittgenstein’s philosophy is – or at least might be regarded – as a far more unitary enterprise than it is customary to think of it. Here it is not the place to argue in favour of this approach – although arguments may be brought.
The commonest formulation of the problem of intentionality is “How is it possible for something to refer to another thing, to be about it?”. Put in this crude form, the explanandum (the “something”) of this question needs clarification: what is the something that is supposed to be about something else? What is supposed to refer? A common-sense answer is that it is only persons, or at least creatures endowed with minds, who can refer to objects: after all, in the absence of any mind the world would be inert, nothing would be about anything else. But mind obviously can refer to what is outside it, that is be intentional, via signs. Those things that refer are signs, they have meaning, and they have meaning to the extent at which they are used to refer to something. The problem becomes more complicated once we further ask: what is the property that things need to have in order to be signs? After all signs are things to be found in the world, they are a part of the world: any sign may be completely described in purely non-intentional terms, without describing it as a sign. What is responsible for the fact that some things are endowed with intentionality? For – we feel – the mind cannot use just any sign it likes in order to refer to anything else: if it did, then signs could not be used in communication. And the property of being intentional is bound with being the concept of being usable for communication – at least for the (transcendental) reason that some criterion is needed to legitimate the status of a fact as a sign, as an “intentional entity”: that is, if there were no communicable property that should characterize signs as intentional entities, then how could I legitimately speak of other people’s communicating their intentions via signs at all?
There can be, or at least so it seems, no such a thing as an intention that cannot be communicated, for if it could be, then no one could recognize it as being an intention (leave aside the question whether one could recognize it to be the intention of that thing about which it is an intention). Since intentions are intentions that can be communicated and recognized as such, there needs to be – it seems – a property that characterizes those parts of the world that express intentions; there should be, we feel, a property of signs in virtue of which they are signs. On the other hand, if there is such a property of signs that is responsible for their being signs, the place of the mind in the picture of intentionality appears unnecessary: for the mind could do nothing but discover that some signs are signs, being unable of turning natural things into signs.
I hope to have sketched in the above paragraph the difficulties with which the problem of intentionality, as discussed by Wittgenstein, is laden. Wittgenstein provided, at different times, at least three answers to the problem “what accounts for a sign’s being a signs?” or, as I would put it, “what is the source of intentionality”. These three answers are:
(i) The source of intentionality is the mind. The mind turns a sign into a sign
(ii The source of intentionality is the resemblance (or the isomorphism) between the sign and what is signified
Theses (i) and (ii) are present in the Tractatus. Thesis (ii) is harshly criticized in the post-Tractatus period. Thesis (iii) will appear – as a substitute for (ii) – in Wittgenstein's later philosophy. My purpose is to analyze these answers and signal the tension between them. I will formulate in italics the theses which I hold. I will start my considerations by noticing a puzzle in Tractatus:
A PUZZLE IN THE TRACTATUS
The reader of Tractatus may notice the following paradox: In 2.141 Wittgenstein writes that the image is a fact. It is natural for us to assume that not all facts are images, although (see 2.141) all images are facts. Then, sentence 3 reads that “das logische Bild der wahren Gedanken sind ein Bild der Welt”: The logical picture of the true thoughts is a picture of the world. Therefore, thoughts too are images, as we can learn from the whole section until 3.05. And – since all images are facts – it follows syllogistically that thoughts are facts. From 4.01 we can learn that der Satz ist ein Bild der Wirklichkeit: sentences are themselves images so – via 2.141 – they are facts. From here it very nicely follows that pictures (thoughts and sentences) are part of the world.
However, 5.61 tells us that the limits of the language mean the limits of the world: language appears as a mirror, which is isomorphous with the world, that is (via 1.1.) with the totality of facts. And language is the totality of meaningful sentences. In 5.61 we read that wir koennen nicht sagen, was wir nicht denken koennen: We cannot say what we cannot think. Thus, the limits of language coincide with the limits of thought. But how can thought (or language) have the same limits as the world, if thought (and language) are – as we saw in the previous paragraph - included in the world?
There is textual evidence in the Tractatus that the totality of images (thoughts, sentences) both is and is not part of the world. I take this to be a genuine puzzle, and one that needs to be elucidated. The alternative is between considering pictures as “natural” facts, as parts of the world, and considering pictures as being about the world. If pictures are facts, then intentionality has a substantial support; if pictures are outside the realm of facts, intentionality appears ethereal, mysterious, irreducible. As I shall argue, if pictures are facts their intentionality is unintelligible; and if pictures are not facts, they are not the source of intentionality: pictures do not mean anything, rather the mind uses pictures to mean facts, it means facts through pictures. We need pictures to be facts of the world since we need – this is the purpose of the Tractatus - to describe the conditions for meaningfulness and we cannot describe the conditions of meaningfulness unless we can describe what is meaningful. But only facts (i.e. entities belonging to the world) can be described. Wittgenstein’s celebrated answer to the problem “what turns mere facts into pictures” is the theory of isomorphism. I am not going to present the details of this theory, since the only idea that I think to be relevant for the purposes of this essay is that of the resemblance between pictures and facts.
The status of intentionality of, or through, pictures, in Tractatus becomes more complicated once we pay attention to the truth of solipsism that Wittgenstein admits. If isomorphism is the intentional mark of signs, is what turns a fact, a propositional sign into a picture, no place could be needed for the subject. Yet, as it is known, the Tractatus considered the problem of solipsism to be a genuine and important one. We do not exaggerate very much if we affirm that the subject of the Tractatus is identical with his own language, with the only language that he can understand: by this I mean that the subject cannot be conceived without his language and – presumably – vice versa. After all, 2.1. says that “We picture facts to ourselves”.
It can be replied that I am mistaken in ascribing to Wittgenstein of the Tractarian period the thesis of solipsism and that he mentions solipsism only in order to prove its absurdity via its identity with realism. Without entering into exegetical details, I will mention that, in my opinion, Wittgenstein does hold a solipsistic view, namely a transcendental solipsism and that he connects it with the problem of intentionality. Moreover, the theme of solipsism is mentioned in NB, where its Schopenhauerian roots are explicit: e.g. 2.9.16: “Here we can see that solipsism coincides with pure realism, if it is strictly thought out. The I of solipsism shrinks to an extensionless point and what remains is the reality coordinate with it. The philosophical I is not the human being, not the human body or the human soul with the psychological properties, but the metaphysical subject, the boundary (not a part) of the world. The human body, however, my body in particular, is a part of the world among others, among animals, plants, stones etc.” My personal opinion is that solipsism is a very serious proposal once we can recognize the transcendental status of the subject; thus the solipsistic Ego could be imagined as “what is common to all persons” or “the mind that is common to everybody who thinks/uses language”. Put in this form, solipsism has nothing dramatic or exaggerated about it and I see nothing ridiculous in claiming that I am a solipsist and I wonder why there are no other solipsists.
If indeed the subject is what imparts meaning to sentences, then the isomorphism between signs and reality, and the thoughts as psychological phenomena, appear at best to be unnecessary conditions for meaning. Wittgenstein seems to have realized this tension: thought, if it imparts meaning to sentences, must be isomorphous with sentences and with reality; but if so, then either thought or language is dispensable, replaceable.
Yet, Wittgenstein continued to hold that thought is isomorphous with reality: see NB 12.9.16: “Now it is becoming clear why I thought that thinking and language were the same. For thinking is a kind of language. For a thought too is, of course, a logical picture of the proposition and therefore it just is a kind of proposition”.
The Tractatus strongly suggests that the solipsistic subject instills propositional signs with meanings. But if the solipsistic subject is what is responsible for signs’ being meaningful, isomorphism may play no decisive role. Rather than saying that pictures are the bearers of intentionality, we should say that the solipsistic subject uses pictures in order to refer to facts. Pictures thus become not the bearer of intentionality: their intentionality is derived from the subject. (To use an simile: if intentionality is the light, then pictures are not the Sun, but the Moon). Thus, the Tractarian account of intentionality becomes the scene for a competition for the title of being the source of intentionality: the competitors are the solipsistic subject and the relation of isomorphism between pictures and facts. A facile solution would be to say that intentionality is over-determined, that both the solipsistic subject and isomorphism account for intentionality – they are both sufficient “causes” of the agreement between pictures and reality. But this solution cannot work if we try to extend it to a modal level: since overdetermination does not rule out the possibility that isomorphism could turn (some) facts into pictures even in the absence of the subject, or that, vice versa, the subject could use non-isomorphic pictures to mean the same facts. Thus, I can formulate these theses:
THESIS 1. Wittgenstein’s early philosophy witnesses a tension over the status of signs being the bearers of intentionality and their being the medium through which intentionality is manifested.
THESIS 1.1.The conflict between signs as the bearers of intentionality and signs as the medium through which intentionality is manifested is the conflict between the SIGNS as being the source of intentionality and the SUBJECT being the source of intentionality.
One cannot forget that the Tractatus is committed to an isomorphist account not only of the relation between (verbal) language (i.e. the totality of sentences) and reality, but also between thoughts and the previous two kinds. However, the fact that thought (the totality of thoughts) is another isomorphous level with language and reality does not solve, but only moves the problem one more level. It is true that the early Wittgenstein believed that thoughts consisted of psychical elements and that they had the same logical structure as sentences and facts:
“A thought [Gedanke] consists of several psychical constituents that have the same sort of relation to reality as words. What these constituents are I don’t know” (NB, 1914-1916, translation by Gem Anscombe, p. 130, Oxford, Blackwell 1961); “Now it is becoming clear why I thought that thinking and language were the same. For a thought is, of course, a logical picture of a proposition and therefore it just is a kind of proposition” (NB, p. 82)
But to say so makes the role of the subject in intentionality no less mysterious, once we note that the subject of the Tractatus is not an empirical subject, not a collection of thoughts or psychical experiences: rather, the subject is a transcendental subject, the limit of the world and not a part of the world, equally close or distant from any part of the world, including thoughts. In Tractatus’ ontology there is no room for a person such that the thoughts are his thoughts. While being psychical, thoughts are not the thoughts of the subject, any more or less than any fact in the world is a fact of the subject. That’s why the puzzle about language and reality can equally well be formulated about thought and reality: that is, either thoughts are a part of the world, or they are a mirror of the whole world. Saying that the sentences derive their intentionality from thoughts is not informative, in this context, since thoughts are not closer to the subject than sentences or than any other fact.
(A way of escape is to say that isomorphism is created by the solipsistic self, rather than being discovered by him). Wittgenstein’s later philosophy assumed this consequence and denied the role of isomorphism in elucidating the status of signs-qua-signs. The early Wittgenstein did not believe in a psychical subject, but I think (as an exegetical hypothesis) that he later on criticized his former self for suggesting the illusion that considering thoughts to be the bearers of intentionality can do justice to the subject.
In this context I will discuss the theses of Marc A Joseph (Mental Representation and the Metaphysics of Meaning in Wittgenstein’s Tractatus, Philosophical Investigations 2000). He contends that Wittgenstein, both in his first and in his second philosophy, endorsed an anti-representationalist view of meaning (which I hold to be correct) while simultaneously holding that in the TLP it is the isomorphism (and not the subject) that is responsible for a picture’s having meaning” “Logical form underwrites the intentionality of language, we can use the one (the sentence) in place of the other (the state of affairs) and if we can use the one to mean the other in its absence, even if the other does not (as a matter of fact) exist. We can use a sentence to say something because it is a logical picture”. (Joseph, p. 135).
Marc Joseph notes that sentences (and pictures in general) are parts of the world, seeing no problem with this: as he writes, “a sentence can correspond to the state of the world because it is a part of the world, too. This is one of the deep insights of the TLP. The logical form of language is a distillate of the logical formal organization of reality, and the vertical connection between a language and the world is a selection of some subpattern of horizontal connections to mirror the whole pattern. Thus the projection of the world onto language is a form-preserving mapping of language onto itself. Linguistic representation is possible because reality is, so to speak, in harmony with itself; no tertium quid has to link language with things in the world beyond language because language is part of the system of the world”. (p. 136).
On my interpretation, I take this to be wrong: it is precisely to the extent at which language is a part of reality that a tertium quid is necessary. If language is a part of the world, then semantic relations are mundane relations; and all mundane relations are contingent, external. (If any mundane relation were internal, it could be depicted in an a priori picture. But – as we know – there are no a priori pictures). If language is a part of the world, then both language and world can be seen, so to speak, through a super-psychological faculty. This super-faculty should give reasons for the agreement (or lack thereof) between language and reality; it is at its level that questions such as “Why this picture is a picture of this fact and not of another fact?” could be asked.
But the reason why no such super-psychological faculty is needed is that the Tractatus contains, besides the isomorphist account, the solipsistic account, according to which there is no psychological faculty involved in signs' acquiring meaning. In my interpretation, the solipsism comes to "save" language from standing in an external relation with reality.
Moreover, Joseph does not see that if language were indeed a part of reality and if the projection of the world into a language were “a form-preserving mapping of reality onto itself” (p. 136), then nothing could prevent us, in principle, from depicting language itself. After all, language is a part of the world, and as such it should be able to be depicted like any other part of the world. But if so, then metalanguage would raise its ugly head again.
In my account, the real tension is between a representationalist account of meaning and a non-representationalist one. The non-representationalist account I call subjective because what it says is that a sign has the meaning it has because the subject meant it this way, and not other way. Joseph holds that the continuity between the first and the second Wittgenstein lies in the fact that both avoid the postulation of intermediate entities that account for the agreement between language and reality: TLP avoids the pitfall of the intermediate entity via the doctrine of the logical isomorphism, while Wittgenstein’s later philosophy will use a contextualist weapon: ("Umgebungen” are what accounts for a word’s meaning )
“In the Tractatus, the immediate connection between a sentence and the state of affairs it represents is found in their formal identity, while on the view we are now surveying, the immediate connection between, e.g., “Slab!” and the bringing of a slab consists in the fact that the utterance and the action lie side by side in the activity of the builder and the assistant”. (p. 141). “To be sure, there is a long way from the Tractatus’ correspondence of meaningfulness, but it shares with that theory the underlying idea that nothing mediates the immediate connection between a sign and its meaning… The practice “binds” the call and the action, however, not as something that interposes itself between the call and the action, but as something that surrounds them” (p. 141).
On the contrary, I believe that what unifies Wittgenstein’s thoughts on intentionality is the thesis of the irreducibility of intentionality and – associated with it – the thesis of the irreplaceability of language. I fully agree with Marc Joseph that, as the human body is the best picture of the human soul, so our words are the best picture of their meanings. (Joseph, p. 142), and that “when we hear an utterance or read an inscription, we do not infer its meaning from the objects its writer or speaker has put before us” (ibidem). -I take the emphasis here to be on the word “infer”. But I believe that it is precisely if we conceive of words (and language) as parts of the world that we have no choice but to infer, i.e. to try to discover and interpret, the meanings from the signs.
On my view, what unifies the Tractarian picture of meaning with the later Wittgenstein’s one is the centrality of the subject and its non-psychological status. Assigning responsibility for meaning to the subject is not a psychological move – it can be seen as a psychological move only if we do not notice the transcendental, i.e. non-psychological status of the subject. This is the mistake made, in my opinion, made by Joseph, when he writes that “Wittgenstein’s early philosophy of language differs radically from traditional mental representationalism. Rather than founding the possibility of meaning on the intervention of a mind, he instead finds the scaffolding of language in the “essentially necessary” nature of signs themselves”. (p. 137). In my view, it is true that mental representation is not the source of meaning – both in TLP and in Wittgenstein’s second philosophy, at least if by “mental representation” we understand a psychological entity, a “mental state”, as it is now fashionable to put it. But it is wrong to say that the intervention of a mind is not necessary for conferring meaning. On my view, the mind is indeed what confers meaning on “dead” signs – but the mind that is here acting is not a psychological, but rather a transcendental mind.
In the following section of my essay, I will explore the paths that are opened by both of the alternative accounts to intentionality. The first account, which sees pictures as facts (parts of the world) I will call a metalinguistic or substantialist: account. The second one sees pictures as deriving intentionality from the solipsistic subject; I will call this account a transcendental or solipsistic account.
In the light of the above discussion, the following theses emerge:
THESIS 2. If signs are the bearers of intentionality, then intentionality has a “substantial” component. There is a something which is the support of the intentionality
THESIS 2.1. If (2) , then signs can be described as something belonging to the world; they belong to the same category as what is signified does
THESIS 2.2. But the job of describing them can only be effected in metalanguage.
If signs are the bearers of intentionality, then they should, hopefully, be described in such a way as to account for their being intentional: a picture and the fact to which it correspond given the isomorphism could be presented, one alongside the other, in such a way as to asses their isomorphism. But it order to accomplish such a task, we would need to take a standpoint exterior both to language and to the world. (The same exterior point of view would be needed if we wanted to assess the isomorphism between thought and world, or between thought and language). This outside point of view is, obviously, a metalinguistic one: it is metalinguistic to the extent at which it is meta-real and “meta-psychical”. The temptation of this metalinguistic vantage point is manifest in the notion that the intentionality of linguistic signs is expressed in (pre-linguistic) thoughts. Wittgenstein has always aware of this temptation:
“The phrase ‘to express an idea which is before our mind ‘ suggests that what we are trying to express in words is already expressed, only in a different language; that this expression is before our mind’s eye, and what we do is to translate from the mental into the verbal language. (Blue Book, p. 41) – this quotation anticipates the opening passages of the Philosophical Investigations about learning a language as translating from a foreign language into our own.
However, this temptation is utterly rejected in Wittgenstein's post-Tractarian philosophy. I will present the reasons why.
2.3. If metalanguage were possible, then the question of the agreement between language and reality could legitimately be raised. We could not be satisfied by being told that “language agrees with reality”. We should ask: why, by virtue of what, does this agreement hold?
In PG, Wittgenstein continued to be preoccupied with the problem of intentionality. Meaning, or rather meaningfulness, was seen in analogy with “life”: see, for example:
. “If I try to describe the process of intention, I feel first and foremost that it can do what it is supposed to do only by containing an extremely faithful picture of what it intends. But further, that that too does not go far enough, because a picture too in its turn stands isolated. When one has the picture in view by itself it is suddenly dead, and it is as if something had been taken away from it, which had given it life before. It is not a thought, not an intention; whatever accompaniments we imagine for it, articulate or inarticulate processes, or any feeling whatsoever, it remains isolated it does not point outside itself to a reality beyond”. (PG, VII 100, p. 158)
His quest for the source of intentionality took the form of the obsessive question, that will recur in the Blue Book, “what gives live to mere dead things, by turning them into signs”? We saw that in the Tractarian period, his answer was ambivalent: both "the solipsistic will" and "isomorphism" provided the desired answer. Unlike the Tractarian period, his later answer does not involve any kind of isomorphism between sentences and facts; moreover, the thesis that thoughts are the ultimate source of intentionality and that they impart meaning to sentences (that is, turning dead signs into meaningful ones) is in its turn abandoned. One of the reasons is that Wittgenstein started to take seriously the possibility of multiple methods of projections between a picture and reality. (see PG p. 163) As I see the issue, it is important to note that Wittgenstein considered the plurality of methods of projection as a serious threat to the metalinguistic account of intentionality. Far from rejecting the Tractarian doctrine of the ineffability of semantic relations, the admission of the plurality of methods of projection is a reductio ad absurdum of the task to “say what can only be shown”, to express in language the relation between language and reality. This threat also confronts the abandonment of transcendental solipsism: if the subject is admitted as a relevant “cause” of signs' being signs and if this subject is considered an empirical one, then the relation between the (empirical) subject, its language and reality is an empirical relation itself; it could be said, expressed in language.
The distinction between “the picture as it is seen from the inside” (that is used as such and is “alive”, i.e. meaningful) and “the picture as seen from outside” (such that it appears dead, i.e. meaningless) continued to obsess Wittgenstein until the last period of his philosophical writing: See, for example: Zettel par. 233:
“’Only the intended picture reaches up to reality like a yardstick. Looked at from outside, there it is, lifeless and isolated’. It is as if at first we looked at a picture so as to enter into it and the objects in it surrounded us like real ones; and then we stepped back, and were now outside it, we saw the frame, and the picture was a painted surface.” In this way, we are surrounded by the intention’s pictures, and we are inside them.
But when we step outside the intention, they are mere patches on a canvas, without life and of no interest to us.
And also Zettel par 235: If I see the thought symbol ‘from outside’, I become conscious that it could be interpreted thus or thus; if it is a step in the course of my thoughts, then it is a stopping-place that is natural to me, and its further interpretability does not occupy (or trouble) me. As I have a time-table and use it without being concerned with the fact that a table is susceptible of various interpretations”.
Wittgenstein wrote in PG that the rejection of metaphilosophy should be a leading principle. His rejection of meta-disciplines is notorious, see for example “One might think: if philosophy speaks of the use of the word ‘philosophy’ , there must be a second-order philosophy. But it is not so: it is, rather, like the case of orthography, which deals with the word ‘ortography’ among others without there being second-order orthography”.
This echoes a remark in PG-VI-72, pp.115-116:
“I’m allowed to use the word ‘rule’ without first tabulating the rules for the use of the word. And then rules are not suprarules.
Philosophy is concerned with calculi in the same sense as it is concerned with thoughts, sentences and languages. But if it were really concerned with the concept of calculus, there would be such a thing as a metaphilosophy. (But there is not. We might so present all that this would appear as a leading principle).
I believe that Tractatus contains an explicit refusal of metalanguage and that this refusal is the basis of Wittgenstein’s distinction between showing and saying and of his criticism of Russell’s theory of types. However, the refusal of metalanguage in TLP is not unproblematic: it may be faced with serious obstacles that can be overcome by the postulation of the transcendental subject. This may be nicely illustrated by the puzzle that I presented in the previous section. The need for metalanguage arises with the need to explain the agreement between language and reality. But if metalanguage is possible, it violates the requirement that language and reality should be co-limited. Metalanguage thus violates the ineffability of semantic relations. But it is precisely this ineffability that is required by the colimitedness of language and reality: indeed, if relations between sentences and facts were describable in language, it could no longer be said that language depicts all the facts – and presumably only the facts, unless we are going to say that semantic relations are somewhat analoguous to facts. (That is: if metalanguage is possible, then there are some facts – namely the pictures – that are not mirrored in language. But this contradicts the status of pictures as facts).
Leaving Tractatus behind us, we are left with the following option: either we give up the co-limitedness of language and reality in order to account for the status of pictures as facts; or we abandon the notion of pictures’ being facts, worldly entities in order to preserve co-limitedness. If we embrace the latter option, we will most probably feel tempted to think that pictures are some meta-facts (or entities analoguous with facts) that do not belong to the world. Actually, Wittgenstein’s option was different: what turns a sign into a sign is not an entity. More generally: what turns signs into signs is not something that can be described in language.
Besides co-limitedness, there is another reason why metalanguage – and the expressibility of semantic relations – is impossible. If semantic relations were expressible, they would be mundane relations. We could justifiably ask for a justification: why these semantic relations hold rather than others? Any answer that could be given to such a question would involve an infinite regress: it is the temptation of this kind of answers that Wittgenstein felt bound to address.
2.4. If we took seriously the question "In virtue of what does agreement between language and reality hold?", the most plausible answer to this question is “in virtue of a class of entities that is different from language and from reality”. These entities could be conceived of in metalanguage.
Wittgenstein’s attack against the metalinguistic account of intentionality takes the form of what could be called a version of the third man argument. If anything can account for the agreement between language and reality, the most tempting answer is to say “because of an entity that is interposed between the two”. This is the classical empiricist answer: the intermediary entity that is here found are the mental pictures, conceived as psychical elements.
It is, moreover, noteworthy that the postulation of signs would not be a progress towards explaining the semantic relations, but rather would double its complexity. Instead of asking "why is language in agreement with reality", we should ask "why is language in agreement with the intermediary level?" and "why is the intermediary level in agreement with reality?". The more general formulation of the same problem appears if we did not speak about intermediary entities, but only about metalanguage. If we had a metalanguage, then the problem “why are the true bearers of intentionality the meta-signs?" would be an additional problem, created by the desire to provide an answer to the problem "where do signs derive their intentionality from?". We would need to explain both the agreement between meta-signs and signs, and the agreement between signs and reality. We would thus need meta-metasigns in order to answer the question “why meta-signs are the ultimate bearers of intentionality?” and so on. (This is Wittgenstein’s version of Plato's third-man argument). As I presented the issue, the problem of "intermediary entities" is a particular case of the problem of metalanguage: for if we introduce intermediary entities (like mental pictures) between signs and reality, we are committed to mention signs, that's why we would need to use meta-signs.
It is easy to note that, starting with the PG, Wittgenstein attacks this account of meaning to the extent at which it recognizes that it is a consequence of the meta-linguistic view that signs are the bearers of intentionality.
It important, in my opinion, to note that Wittgenstein’s point was not to promote a skepticism about semantic relations. The plurality of methods of projection could lead to skepticism if we adopted the metalingustic account of intentionality. But this meta-linguistic account may be taken seriously only if we abandon the transcendental solipsism of the Tractatus and want to explain the semantic relations that hold between the language of an empirical individual and the world. Wittgenstein’s point was not that, given the plurality of methods of projection, we can be ignorant about the objects (or facts) that we mean: but rather that if we try to explain semantic relations by postulating entities, we will have no option but skepticism. This line of attack is pursued in the Blue Book:
“Someone say, he imagines King’s College on fire. We ask him ‘how do you know that it is King’s College that you imagine on fire? Couldn’t it be a different building, very much like it? In fact, is your imagination so absolutely exact that there might not be a dozen buildings whose representation your imagination could be? – And still you say ‘there is no doubt I imagine King’s College and no other building”. But can’t saying this be making the very connection we want? For saying it is like writing the words ‘Portrait of Mr. so-and-so’ under a picture. I might have been that while I imagined King’s College is on fire you said the words ‘King’s College is on fire’. But in very many cases you certainly don’t speak explanatory words in your mind while you have the image” (p. 39, Blue Book).
This nicely echoes a remark in PG:
“When I said that my image wouldn’t be a portrait unless it bore the name of its subject, I didn’t mean that I have to imagine it and his name at the same time. Suppose I say something like: ‘What I see in my mind isn’t just a picture which is like N (and perhaps like others too). No, I know that it’s him, that he is the person it portrays.’ I might then ask: when do I know that and what does knowing it amount to? There’s no need for anything to take place during the imagining that could be called ‘knowing’ in this way”. (PG par 99, p. 147)
If thought were a kind of language, then propositional language would describe thought and would thus be metalinguistic. I would continue Wittgenstein’s reasoning this way: even if while imagining King’s College I mentally articulated the words ‘King’s College’, then still the question could equally legitimately be asked: how do you know that the words you mentally uttered meant King’s College? How do you know you are not deceived by your words? After all, if my words had a meaning independent of what I mean by them, how could I be sure I correctly grasp their meaning?
If they secured, so to speak, the real King’s College being meant, then why need the translation into verbal language? If the verbal linguistic expression makes them clearer, why is the mental expression needed? One of the two – the mental picture of the verbal one – is superfluous. The entity that is superfluous is so because of the plurality of interpretations that accord with it. If meta-signs were possible, then signs would be superfluous; they would be multiply interpretable; no relation between signs and reality would be an internal (necessary) relation and the chain of intermediary links could be indefinitely continued:
“However many intermediate steps I insert between the thought and its application, each intermediate step always follows from the previous one without any intermediate link, and so too the application follows from the last intermediate step” (PG- VII 110, p. 160)
In the Blue Book, Wittgenstein’s inquiries into the problem of intentionality amount to a complete rejection of the pictorialist account: “I can express our trouble in a different form by saying ‘How can we know what the shadow is a shadow of? The shadow would be some sort of portrait; and therefore I can restate our problem by asking ‘what makes a portrait a portrait of Mr. N. ? ‘ The answer that might first suggest itself is: ‘The similarity between the portrait and Mr.N. “ This answer in fact shows what we had in mind when we talked of the shadow of a fact. It is quite clear, however, that similarity does not constitute our idea of a portrait; for it is in the essence of this idea that it should make sense to talk of a good or a bad portrait. In other words, it is essential that the shadow should be capable of representing things as in fact they are not…An obvious and correct answer to the question ‘What makes a portrait into the portrait of so-and-so? Is that it is the intention…But there are a great many combinations of actions and states of mind which we should call ‘intending’. .. (Blue Book, p. 32-33).
If we are not satisfied that intentionality, meaning something is an irreducible relation, then no other relation could explain why intentionality holds. As an exegetical remark: to the extent that Wittgenstein’s attacks are directed against the Tractarian account of intentionality, they are only partly motivated. They are motivated if they are directed against the isomorphist, that is meta-linguistic view of intentionality. But they are not motivated if they are directed against the transcendental-solipsist view of intentionality. As I argued above, both of these views are present in the Tractatus.
2.6. The recourse to intermediate entities is unsatisfactory, since it implies that the true bearer of intentionality are not linguistic signs, but , for example, thoughts. If so, then linguistic signs become dispensable. There can be no cohabitation between language and metalanguage. If there were a metalanguage, it would dislocate language: it would render it lifeless, devoid of intentionality.
We can expand the argument against meta-language, as it appears in Wittgenstein's post-Tractarian philosophy, if we start from purely Tractarian premises that – as I hope to show – Wittgenstein did not abandon. If metalanguage (and metathought) were possible, what would remain from the claim, as expressed in the Preface of the Tractatus, to set a limit to thought and to language “from the interior”? More seriously, what would remain from the transcendental status of the subject? Since the subject is the limit of his language (and of his thought), the subject would appear in a description of the world-and-language-and-thought, if such a description were available. The subject would be itself a part of the world; it would be an empirical subject, rather than a transcendental one. But – according to the Tractatus – there is no empirical subject.
If metalanguage is possible, then there is no place for the transcendental subject. The admission of metalanguage goes hand in hand with turning the transcendental subject into an empirical one: a collection of psychical experiences and (perhaps) of sentences. But – as the PG will go at great length to explain – this would amount to turning signs into meaningless, dead things. The refusal of metalanguage in the Tractatus is equivalent to the refusal of the empirical status of the subject. Likewise: once metalanguage can do its work, pictures become mere facts. But if the metalanguage is refused, then nothing may remain from Wittgenstein isomorphist account of intentionality. It seems that Wittgenstein realized this incompatibility and he saw himself obliged to renounce either transcendental solipsism or isomorphism. I willl argue that he abandoned isomorphism, and I think this turn of his thought is obvious. The less obvious point is that he did so while preserving a form of transcendental solipsism .
One of the arguments that occurs repeatedly in PG is that, if language were not the source of intentionality but thoughts, it would be altogether unintelligible why language is (meaningfully) used at all. (I call this argument “the argument of the dispensability of language”).
I.2.: “We regard understanding as the essential thing, and signs as something inessential. But in this case, why have the signs at all?...”
Correlation between the words-as-I-use them and their reference is ineffable, grammatical, necessary. It is precisely because I do not infer my reference, my meaning from my words. (If I did infer, then my meaning from any other entity, then I could conceive that entity – be it linguistic or psychologic – separated from the meaning I infer from it. This is how I understand Wittgenstein's using of examples of the type "Utter the words a b c d and mean by them 'The weather is fine' ". Thus, when I express my intention through some signs, those signs are indispensable from my intention; when I mean something by some signs, I cannot make sense of my using those signs without their meaning. If my meaning depended on a different level than the linguistic signs which I use (for example, on my thought as a psychological yet inherently meaningful process), then I could make sense of the separation between my linguistic signs and their meaning.
“Thoughts”, which Wittgenstein (unlike Frege) continues to regard as psychical elements, seem to embody in the post-Tractatus period a meta-linguistic role and they fail to play this role well for the reasons already presented. What we expect from metalanguage is to elucidate the nature of the agreement between language and reality: but it cannot do so, since it cannot do better than depicting sentences and facts as two parts of a more encompassing whole. This can only be done from a vantage point situated outside language and reality; from this vantage point sentences appear as mere propositional signs, devoid of meaning. While in the Tractatus Wittgenstein flirted with the notion that ascribing to thoughts the responsibility for conferring meaning on propositional signs could do more than mere signs as used (or enlivened) by the subject could do, he later realized this mistake. Thoughts cannot do more than sentences can: if they could, then they would acquire a metalinguistic status and would make language dispensable. Instead of fulfilling the job of explaining why signs refer to world as they do, thoughts would cause signs-qua-signs to vanish away. The reason is that an outer perspective of signs would expose the possibility of variously interpreting them so as to be in accord with reality. (Thus, a good paraphrase of the quotation from PG that I gave would be “If we had a good metalanguage, why have language at all? If we had a good vantage point from which we could contemplate our grasping of reality, why have grasp reality at all? After all, from this vantage point – and only from it - we could realize that our grasp of reality was arbitrary and replaceable”:
”It might be said: in every case what is meant by ‘thought’ is the living element in the sentence, without which it is dead, a mere succession of sounds or series of written shapes.
But if I talked in the same way about a something that gives meaning to an arrangement of chessmen, something that makes it different from an arbitrary collection of bits of wood, I might mean almost anything! I might mean the rules that make the arrangement of chessmen a position in a game, or the special experiences we connect with positions in the game or the use of the game.
It is the same if we speak of a something that makes the difference between paper money and mere printed bits of paper, something that gives it its meaning, its life”. (PG- V 65, p. 107).
Language would be dispensable and questions concerning its adequacy (or lack thereof) to reality could intelligibly be raised. This is the Aristotelian view of language: we use words because it is more convenient to do so than to bring the objects that words stand for). I call this view semantic reism, and I claim that semantic reism is essentially a metalinguistic view: questions concerning the adequacy between language and non-language cannot be raised – or answered – in language. Semantic reism, the doctrine of the separation between words and their meanings, is aroused by the temptation of metalanguage. Wittgenstein remained famous for the rejection of semantic reism. I understand the insistence of the late Wittgenstein against the reification of meaning as pleading against the dissociability of meaning from the bearer of meaning. The reason for this insistence is that if meanings were detachable, then we would be able to grasp meaning without the mediation of language. But we cannot replace language, or at least cannot express in language its replaceability. This theme continued to obsess Wittgenstein until his last philosophical writing: see, e.g. Zettel for the connexion between the autonomy of language and its irreplaceability:
".Suppose I wanted to replace all the words of my language at once by other ones; how could I tell the place where one of the new words belong? Is it ‘images’ that keep the places of the words?” (Zettel par. 10).
On my interpretation, it is direct realism, that is involved by the transcendental solipsistic account of intentionality, that provides for language’s role as mediating reference to reality; if one rejected the solipsistic account in favour of a metalinguistic one, then we would have an equally direct access to language and to reality. But this would involve turning language into the dead support of it
This is the origin of Wittgenstein’s criticism of the representationalist account of meaning: if thought could refer directly to reality insofar as it imparts meaning to the physical sentences, then why need the physical signs at all? See, e.g., PG IV 46 (p. 86): “One might wish to ask: So is it an accident that in order to define signs and complete the sign-system, I have to go outside the written and spoken signs? Where I do that don’t I go right into the realm where what is to be described occurs? But in that case isn’t it strange that I could do anything at all with the written signs?” (p. 89). Direct realism, as I understand it, is nothing but the claim that if a symbolism reaches reality directly, then any other symbolism would reach reality by the "filter" of the previous symbolism. By this, I do want to deny the obvious intuition that "we can sometimes think without words". (and I think Wittgenstein is not commited to deny it). What he would have denied, is that in the case where we think in mental pictures, we could safely later come on and describe in language our previous thinking in mental pictures.
I hope to have presented in the last two sections of my essay that there are three arguments against the metalinguistic view of intentionality:
a) the “third man’s argument”: if an intermediary entity is postulated, then why not a new intermediary entity and so on?
b) the plurality of the methods of projection: any entity that is supposed to explain the relation between words and their reference can be interpreted in several different ways;
c) the dispensability of language: if there could be a vantage point from which to grasp language and reality, why have language at all?
and that Wittgenstein saw these three arguments as closely linked. All these arguments against the metalinguistic view of language account for the general picture of the "autonomy of language". Its consequences are, as we shall see, anti-psychologist to the extent at which they are anti-metalinguistic:
THESIS 2.51 The arguments against the metalinguistic account of intentionality can be derived from the Tractarian transcendental solipsism
I take it to be unproblematic that the solipsist Ego in the TLP is a transcendental subject, and not an empirical one. By this I mean that the Ego is not a part of the world; Wittgenstein’s famous dictum that solipsism, consistently followed, amounts to realism, is grounded in the claim that the solipsist, purporting to say something meaningful, cannot do so if he isolated his private world from the outer, remaining world; rather, what the solipsist means is that the public world is identical with his private one. The thesis of my paper is that the “consistently followed realism” that Wittgenstein speaks of is nothing but direct realism, i.e. a realism that does not distinguish between appearance and reality. (An alternative definition of “direct realism” is “realism that does not require any intermediate entity between the subject and the reality”). Thus, the rejection of intermediary entities and rejection of metalanguage in PG signifies the preservation of the Tractarian solipsism under the guise of direct realism. For what solipsism and direct realism have in common is the denial of the possibility of discoveries. Wittgenstein’s thought was: if intermediary entities – such as mental pictures – were needed to account for my meaning something, then I would learn, discover that I meant something rather than something else. I could be proven wrong when trying to refer to something.
I assert that after the Tractatus Wittgenstein abandoned the isomorphist account of intentionality: when I mean something by a picture, the alleged resemblance between the picture and its intended reference plays no role. This is the rejection of the pictorialist (isomorphist,) account of intentionality. My claim is that this move goes hand in hand with the rejection of the pictorialist account of deriving a picture of: “for describe whatever process (activity) of projection we may, there is a way of reinterpreting this projection” (p. 33). What the post-Tractatus Wittgenstein rightly abandoned from his younger self is not transcendental solipsism, but rather the pictorialist account of intentionality; and the reason why he abandoned it is that, if isomorphism between language and reality were important, it would amount to a version of indirect realism, according to which whenever we mean something, there are some objective entities that are intermediary between us and what is meant.
THESIS 2.7. The road to metalanguage and its dead-end are paralleled by the status of the subject. If there were such a thing as metalanguage, then the subject would be an empirical, not a transcendental, subject.
If my thesis that rejection of metalanguage goes hand in hand with transcendental solipsism, we should expect that, since Wittgenstein did not abandon the rejection of metalanguage in the post-Tractarian period but rather reinforced it, he preserved transcendental solipsism. I hold that this is the case and that there textual evidence in favour of this interpretation. First, Philosophical Remarks contain clear evidence of Wittgenstein’s solipsism:
PB p 100 VIII 71: The visual space has essentially no owner. Let us assume that, with all others, I can always see one object in visual space – viz my nose. Someone else really doesn’t see this object in the same way. Doesn’t this mean, then, that the visual space I’m talking about belongs to me? And so is subjective? No. It has only been construed subjectively here, and an objective space opposed to it, which is, however, only a construction with visual space as its basis. In the –secondary- language of the objective – physical – space, visual space is called subjective.
In the model described above, neither the two eyes that see the objects, nor their position, need be included. That’s only one technique of representation. It does just as well if, e.g., the part of the objects that is ‘seen’ is indicated by shading it in. of course, you can always work out the position of the two eyes from the boundaries of this shaded area, but that only corresponds to translation from one way of speaking into another.
The essential thing is that the representation of visual space is the representation of an object and contains no suggestion of a subject. (PB, p. 100)
“We might also put the question like this: What in my experience justifies the “my’ in “I feel my pain”? When is the multiplicity in the feeling that justifies that word? And it can only be justified if we could also replace it by another word.
The experience of feeling pain is not that a person ‘I’ has something.
I distinguish between an intensity, a location etc in the pain, but not an owner” (PR par. 66)
PB p96, VI 66: “Visual space and retina: It is as if you were to project a sphere orthogonally onto a plane, for instance in the way in which you represent the two hemispheres of the globe in an atlas, and surrounding the two projections of the sphere somehow corresponds to a possible extension of what is to be found on the sphere.
The point is that here a complete space is rejected onto a part of another space, and it is like this with the limits of language in a dictionary”.
That is: if we desire to depict the relation of agreement between language and reality, we are committed to abandoning pure solipsism and the pure subject. We need a metalanguage, since our language is only part of reality. But, by mentioning it, we do this in language. Metalanguage, like pure subject, is inescapable: but this only means that there cannot be another metalanguage that should describe it. This is equivalent to saying that a sign cannot appear as intentional if it is mentioned, rather than used. Signs are like the subject: if the subject is transcendental, there can be no question of his mental states or his language as being compared with reality: the subject cannot be mentioned, he is not in the world but the limit of his world. On the contrary, if the subject is conceived of as empirical, he loses his power to instill the life of intentionality on signs, since he will be a part of reality along with signs and facts. I hope that my interpretation makes clear why the truth of solipsism in Tractatus, like the semantic relations (the form of representation and form of depiction) are inexpressible. This line of thought is pursued in PG:
“Only the intended picture reaches up to reality, like a yardstick. Looked at from outside, there it is, lifeless and isolated. It is as if at first we looked at a picture so as to enter into it and the objects which it surrounded us like real ones; and then we stepped back, and were now outside it, we saw the frame, and the picture was a painted surframe” (PG-VII-9, p. 146)
Also, PG contains an almost explicit reiteration of Tractarian solipsism:
PG 156: The subject – we want to say – does not drop out of the experience but is so much involved in it that it cannot be described
and of the claim that the transcendental subject’s will is the source of meaning:
“If we said that the will is what is responsible for imparting meaning to the dead signs, we should pay attention to the fact that the will here involved is a transcendental, not a psychological, will”. (PG-VII-107, p. 157)
If the will is conceived as psychological, i.e. as a part of the world, then acts of will would happen to us like any other events. (Like Kant and Schopenhauer, Wittgenstein felt that the free will cannot be conceived of as a part of nature; the will , if it is to be genuine free will, needs, so to speak, its point of application to be outside the world).
It is well known that Wittgenstein held, during his intermediate period, the no-ownership theory of the mental, the thesis that the personal pronoun is eliminable in psychological contexts. (In the Philosophical Investigations period, this led to thesis that first-person psychological sentences are not statements, but Ausserungen, avowals, devoid of truth value). I cannot help but see in this development a continuation of the Tractarian transcendental solipsism. The redundancy theory of the subject in psychological contexts, developed in PR and preserved by Wittgenstein to the end of his life, is to me an obvious continuation of the Tractarian identification of solipsisim with realism:
Of course, the question might be raised “Why hold a no-ownership account of first-person psychological utterances, and not of all psychological utterances, as a consistent solipsism would dictate”? I am not going to explore in details this question: what I believe to be the true answer is that, in Wittgenstein’s later philosophy, mental states are not entities that are designated by language, but rather ways that language refers to the world: that is, a true assertion of pain is nothing but a painful assertion. I am aware that this answer might be criticized, but I am not going to defend it in this essay.
A metaphor may help to understand how the thesis of the autonomy of language and the refusal of metalanguage is related to transcendental (or pure) solipsism and to the refusal of empirical (or impure) solipsism. The impure solipsist commits the same fallacy as a photographer who wanted to convince us of his photographic skills by having himself pictured by other photographers and adding to each photography he takes a second photography, depicting himself taking the previous photography. He could do this in the hope of convincing us that his photographies are veridical (or capture aesthetically relevant aspects of the reality) and that it is he himself who has taken those pictures. To the extent at which he succeeded in such a task, he would be able to present, for each picture he takes, an additional picture showing himself taking the said picture (to each picture there would correspond a meta-picture). But then the meta-pictures would have to be taken from a different angle than the object-pictures and to this extent they would present reality from a different perspective than the original picture. Thus, if we want to supplement a picture with the meta-picture which depicts the photographer, we cannot simply depict more reality than the object-picture did; we will have to alter our perspective of the part of reality taken by the original (object) picture.
These general considerations about the relation between language and reality can be applied to various other aspects of Wittgenstein's account of intentionality. An example is the theory of waiting, as presented in Philosophical Remarks, where Wittgenstein criticizes Russell's doctrine that the relation between waiting and what is waited was an external one. I do not intend to explore this issue in details, since I think that it is a cliché that Wittgenstein attempted to substitute internal (i.e. necessary and inexpressible) relations wherever Russell saw external (i.e. contingent and expressible in language) relations. Another famous example is that of memory. Malcolm correctly ascribes to the early Wittgenstein the pictorialist, isomorphist account of intentionality, taking notice at the same time of the tension between this account and the “nonrepresentationalist”, or – how I call it – solipsist – account of intentionality. Indeed, “a picture [has] an ‘inner’ similarity with what it depicts. This similarity is due to the presence, in the picture, of the picturing (pictorial) relationship. (die abbildende Beziehung). The picturing relationship, what turns something into a picture, belongs to the picture (2.1513). this picturing relationship consists of the correlations of the picture’s elements with things” (2.1514). These correlations are not something we find ready made, but something that we create when we turn a group of elements into a picture (Malcolm, p. 135). Malcolm, however, does not realize that it is only from the standpoint of the transcendental subject, i.e. from a solipsistic perspective, that the relation between pictures and what is depicted may be an “inner” relation, one that is contained in the picture. My understanding is that in the absence of a transcendental subject, we would be left, so to speak, at the mercy of metalanguage and would end in external realism: the relation between the subject and reality, like that between language and reality, would be contingent. We would have to say what the solipsistic subject shows, and turning the transcendental subject into an empirical one. The relationship between pictures and what is depicted would not be contained in the picture – it would have to be explained, accounted for: it would be contingent.
It should not be forgotten that the early Wittgenstein favoured a “mentalistic” view of meaning and intentionality, according to which psychical facts are responsible for imparting meaning on signs: this is the view that Malcolm ascribes to Wittgenstein: “When you think a thought, there is a ‘psychical’ structure to express – the thought in language’ the latter’s function is to project that same structure into the words of a physical sentence. The physical sentence becomes a picture by the virtues of the fact that the psychical picture is projected into in. If the picture is true, there are three things that have the same structure but different natures: the psychical picture, the physical sentence and the state of affairs in the world”. (Malcolm, p. 137).
Wittgenstein’s later thought seems to me to be this: if the physical sentence is isomorphous with the state of affairs, then there is no need for a third isomorphic entity (the thought): the reason why there is no need for the third entity is that either this entity is isomorphous with the previous two - and then why should it be privileged as the “bearer” of meaning, rather than physical sentence? – or it is the privileged bearer of meaning in virtue of another property than its isomorphism with the state of affairs – and in this case the entity that is the privileged bearer of meaning would make unintelligible the need for an intermediate entity that was the desired explanandum – in this case, the linguistic entity (the Satz).
I have presented so far the reasons of Wittgenstein’s opposition to the pictorialist account of intentionality, as it is expressed in the isomorphist theory of the Tractatus and in the temptation to explain the intentionality of language by appeal to thoughts. The pictorialist account that Wittgenstein rejects in PG and PR is a metalinguistic one: it is an account of intentionality that can only be allowed from a vantage point that is exterior to language and reality. Once this account is rejected, we remain with the Tractarian transcendental solipsism as the basis of semantic relations. When I mean something, there is no answer to the question “how can I be sure that it is it that I really meant” than “because my intention was to mean it’. That’s why I can formulate the following thesis:
THESIS 3. The alternative to pictorialism is to deny that intentionality has any substantial support. What this amounts to, is simply to say that when a word refers to something different from it (its non-linguistic reference), there is nothing that justifies why the word reaches its intended destination.
It could be claimed that this account makes semantic relations arbitrary. What I reply is that arbitrary connexion may be rightly diagnosed only from a standpoint that is exterior from the relata of the connexion. Thus, language – the medium through which I manifest my intentionality – can only be arbitrary if I can imagine a different way of reaching the same reality. But Wittgenstein's point is that insoluble problems would appear exactly once such a futile attempt is pursued. It is not the lack of a justification for the agreement between language and reality that Wittgenstein deplores, but rather the search for it. Wittgenstein's post-Tractarian philosophy was not a progress towards arbitrariness of semantic relations.
“I do not call a rule of representation a convention if it can be justified in propositions; propositions describing what is represented and showing that the representation is adequate. Grammatical conventions cannot be justified by describing what is represented. Any such description already presupposes the grammatical rules. (PB I-7 p. 54)
The famous remark that “to use a word without justification is not to use it without right” can be quoted in this context: of course, the ‘right’ to use words as we do is given in On Certainty not by the will of the solipsistic self, but by our form of life; thus, “our form of life” becomes in Wittgenstein’s later philosophy the inheritor of the Tractarian solipsistic subject. The dictum that "we cannot get outside language with the means of language" is equivalent with the dictum that "we cannot contemplate our grasp of the reality while simultaneously contemplating reality". But this is not a topic that I am going to explore here.
The idea that intentionality has no substantial support is present in Wittgenstein's last period of writing. See, for example, Zettel par. 6: “…and if I go on to tell someone the feelings, images etc which I had while I was making that remark (while I was making that allusion), these may fall out the typical picture of an allusion (or one such picture). But it doesn’t follow that the expression alluding to N., means: behaving like this, feeling this, imagining this etc."
Wittgenstein's insistence that meaning needs not hae any substantial support reminds Moore's arguments against the "naturalistic fallacy". Moore insisted that even if all good things had a certain property in common, it would still not follow that the meaning of "good" is that property. What Wittgenstein draws attention to could be called, given the lack of a better term, "natural intention fallacy": even in all cases where I rmean a certain object I display certain psychological or behavioural processes, it still does not follow that my meaning it consists of those processes.
To clarify the issue a bit further, two versions of the naturalistic fallacy argument may be imagined :
(1) (weak naturalistic fallacy argument) moral properties may not be necessarily coextensive with natural properties.
(2) (strong naturalistic fallacy argument) No natural property may be such that its presence could authorize us to conclude the presence of a moral property.
The naturalistic fallacy argument, as developed by Moore, points to (1). Thesis (2) is much less acceptable, since it implies that any inference of the type "He is good because he did so-and-so (where so-and-so is a description of non-moral acts) is not certain. (2) seems unacceptable because it leaves us with no possibility of correctly ascribing moral properties to other persons.
The "natural intention fallacy" argument may be formulate in two different versions
(1`): (weak natural intention fallacy argument): No intention may be necessarily coextensive with a state, event or property that is able to be described in purely descriptive (non-intentional) terms
(2`) (strong natural intention fallacy argument): No purely descriptive, i.e. non-intentional property, state or event may be such that its presence authorizes us to conclude the presence of an intention.
Again, thesis (2`) seems much less plausible than thesis (1`), for the reasons that I have already presented. I think it is undeniable that Wittgenstein entertained (1`) and that, at least in the case of the first person, he entertained even (2`). This leads to problematic conclusions, as we shall see below.
THESIS 3.1. If intentionality has no substantial support, then metalanguage is precluded and so are psychical phenomena as the source of intentionality
What metalanguage could do, is to present language and reality as two different parts of the same encompassing whole. (On the meta-linguistic account, language is a part of the world and semantic relations become worldly relations; on the contrary, according to the non-substantialist account of intentionality, what is responsible for intentionality is not a something. Signs-qua-signs are not entities, rather they are a medium through which reality is reached to).
The behavioristic trend in Wittgenstein’s later philosophy is traceable back to the main theme of the refusal of metalanguage and preservation of transcendental solipsism. I argued above that thoughts are dispensable for the same reason that we can succeed in referring to the world via linguistic signs. The thesis that “nothing is hidden”, that hidden processes are of no relevance, is a direct corrolary of the autonomy of language and impossibility of metalanguage. Postulating a realm of the inner in order to explain the relation between "outer" symbolic processes, such as behavior and language, and reality does nothing but double the problem of intentionality, since the inner, if it can be explanatorily relevant at all, needs to be itself symbolic:
“A description of what is psychological must be something which can itself be used as a symbol” (PG-p. 99, IV 58)
. Wittgenstein’s reasons for dismissing the relevance of the inner, of the hidden are the same as his reasons for dismissing metalanguage: indeed, if the inner could have any explanatory relevance, it would have the explanatory relevance that meta-signs have. Thus, the “behavioristic” trend present in Wittgenstein’s later philosophy can be traced back to his refusal of metalanguage and transcendental solipsism. For example, what distinguishes intentional from unintentional actions is not that a hidden process is present in the first case: “Even if my pencil doesn’t do justice to the model, my intention always does”. (IV-58). I take this remark to be ironical and I would continue it: If my intention does justice to the model, why do I need the pencil at all? Likewise, Wittgenstein’s anti-psychologist insistence, the denial of the empiricist tenet that mental pictures are involved whenever an act of meaning takes place is not meant (in my interpretation) to deny that we can sometimes refer to reality without words. Rather, Wittgenstein’s point, as I understand him, is that if we could refer to anything without language, then we could not use language in order to describe our non-linguistic act of intention.
“For the purposes of our studies, it can never be essential that a symbolic phenomenon occurs in the mind and not paper, so that others can see it. One is constantly tempted to explain a symbolic process by a special psychological process; as if the mind could do more in these matters than signs can” (p. 99, PG IV 57)
:A consequence of the autonomy of language is that all relations that can be expressed in language are equally syntactic or semantic. (Rules that are truly semantic, that could justify the use of a word, are ineffable, like the Tractarian subject is ineffable): “One is inclined to make a distinction between rules of grammar that set up a connection between language and reality, and those that do no. A rule of the first kind is ‘This colour is called ‘red’’ - a rule of the second kind is “ ~~p <- -> p . With regard to this distinction, there is a common error: language is not something that is first given a structure and then fitten on to reality”. (PG-IV-46, p. 89)
3.2. If intentionality has no substantial support, then language cannot be compared with anything non-linguistic. Therefore, the question of the agreement between language and what is not-linguistic cannot be asked
On the background that I sketched above, it becomes easy to understand the autonomy of language that is associated with Wittgenstein’s philosophy. The thesis of the autonomy of language means, simply speaking, that language as it actually is is not accountable to any extra-linguistic instance. Autonomy of language derives from wittgenstein’s refusal of metalanguage, which is closely linked with his transcendental solipsism. As I showed, the progress of Wittgenstein’s thought from the Tractatus onwards was a continual commitment to the thesis of the autonomy of language. (The plurality of methods of projection, the “skeptical” possibility of various interpretations of a sign is not a consequence of this process of Wittgenstein’s philosophy, but rather a reduction ad absurdum of the meta-linguistic, pictorialist account). I hold that PG contains clear evidence of Wittgenstein’s refusal of metalanguage: see, e.g.: "What 1 cc of water weighs is called ‘1 gram’ – Well, what does it weigh? “ (PG II 23, p. 59).
What is new by comparison to the Tractatus is that Wittgenstein does no longer supplement the rejection of metalanguage with any doctrine of isomorphism. Thus, the autonomy of language is more consistently asserted. Asserting the autonomy of language is tantamount to asserting that one does not discover from any sign what one refers to:
"But if you say “How am I to know what he means, when I see nothing but the signs he gives?, then I say “How is he to know what he means, when he has nothing but the signs either?. What is spoken can only be explained in language, and so in this sense language itself cannot be explained. Language must speak for itself.” (PG: I.1)
This precludes both its being compared with reality in order to be tested or its being with an ideal language. (It is obvious that in order to assess language’s adequacy in respect of an ideal language we would need the meta-linguistic vantage point whose impossibility I argued for above). Language thus does not stand in need of improvement and is not amenable to improvement: indeed, any standards by reference to which we could judge its being better or worse could only be outside language. But if we could have such a vantage point outside language, we would not be able to express it in language
3.3. The consequence is that there can be no case of language falling to be in accord with reality. Whenever I want to refer to something, I succeed in doing so. There is no need for intermediary entities because there is no need to explain the agreement between language and not-language. Semantic relations are internal relations
If semantic relations are inexpressible and there is no tribunal in the face of which words could account for their meaning, it follows that we can never be wrong when trying to speak about something. So to speak, “referring to” is more than a success verb; even “trying to refer” is a success verb. (As a parenthetical remark: it makes Wittgenstein closer to a theory of “direct reference” than to a Fregean theory, according to the reference of a name is given by the descriptions that are synonymous with it; this remark is trivial and in need of improvement). This conclusion is difficult to digest – not only because it seems to violate the Quinean dictum that there is no sharp distinction between speaking about different matters and saying different things about the same matter. This conclusion is difficult to digest because of the so-called Donnellanian phenomena, that can occur not only in the case of definite descriptions but also in the case of proper nouns. Donnellanian phenomena are cases of failure of the language to refer to its intended meaning; are cases of disagreement between words’ intended reference and their actual reference (or between words’ reference and speaker’s reference). A defense of solipsism could consist in denying Donellanian phenomena and simply saying that there can be no situation of failed meaning.
But there is a deeper reason why the solipsist account of meaning that I presented above may rightly look suspicious. The reason is the strong intuition that we need some criterion to ascribe a meaning to another person's words. (The easiest way to conceive such a criterion is to say simply that another one's person's words mean the same thing as they would mean if used by me). We thus need a kind or another of connexion between some entity (behaviour, linguistic signs). The problem could be raised, in this context, whether this connexion is necessary, i.e. grammatical, or contingent, i.e. empirical.
I don't know how to answer these questions, and exploring them would most probably take the space of another essay. I will only sketch what I believe are some steps towards their solution: It is hard to deny the intuition that words are sometimes wrongly used. That someone who uses language can be criticized for failing to use the appropriate words. Moreover, if we denied that intentionality has any substantial support, we could not ascribe intentions to other persons. We could avoid both of the above problems if we adopted solipsism. (And we saw that solipsism is the consequence of denying the substantial support of intentions). But the price of Tractarian solipsism seems too high: the only way I can imagine to preserve Tractarian solipsism in Wittgenstein's later philosophy is to somehow equate the solipsistic subject with something like the transcendental We, with Our Form of Life. This solution is, I think, correct, but its disadvantage is that it does not explain the assimetry between the first-person and the third-person in the case of psychological utterances. (And it is certain that Wittgenstein maintained such an assimetry between personal subjects). Therefore, transcendental solipsism, in any form, cannot be ascribed to Wittgenstein's later philosophy without at least a word of caution. If one wants to avoid solipsism, then one needs to explain why language is (when it is) in agreement with reality. So we are back to 2.4 and we need a metalinguistic account. But we saw that postulating intermediary entities is of no help. So we need to look for another answer to the question “why is language in agreement with reality?”.
Wittgenstein considered this problem and he gave two answers: in TLP, the answer is “the isomorphism between language and reality”. In his later philosophy, he gave the answer “the circumstances (Umgebungen) in which an intention takes place.
However, these answers are unsatisfactory. Isomorphism is unsatisfactory because it can only be asserted by someone who stands outside language and reality. However, the Umgebungen solution is unsatisfactory because it leads to a question that Wittgenstein did not formulate but which, in my opinion, can legitimately be raised in the context of his later philosophy: If the circumstances are responsible for a sign’s having meaning, why have the signs at all? Why are not the circumstances alone sufficient for the intention to successfully “take place”? (It can be replied that the circumstances, like the act of intention itself, are the necessary but not the sufficient conditions for the intention; this is a serious objection that needs to be analyzed). Another important question that may affect the Umgebungen solution is whether the connexion between the Umgebungen and the intention is necessary or contingent. If it is necessary, then the intention itself is dispensable and the Umgebungen may be rightly considered the ultimate source of intentionality. If it is contingent, then it follows that I am never authorized to believe that I correctly identified other person's intention. This leads to a skepticism about other people's minds that is equally unacceptable. But this skepticism is the consequence of the thesis that I discover other people's intentions from their words or behaviour.
The dilemma is: I (my empirical person) is a person among the others, like many others. Therefore it is plausible that I express my intentions, I "mean" words the way other people do. However, in my own case I do not discover, I do not infer what I mean from my words or from my behaviour. But in the case of other people, I have no choice but to infer, to discover what they mean from their words, from their behaviour. Therefore, the connexion between their intentions and their behaviour (or their words) is empirical, contingent, while the connexion between my intention, what I mean and my words, my behaviour is necessary (grammatical). But this is unaccountable for on the premises of my being an empirical person among and like others.
This dilemma, which I think is a genuine one, shows the difficulties of the question "How can I ascribe the correct intention (meaning) to another person's words/behaviour". But answering it is outside the scope of this paper.
References:
Ludwig Wittgenstein- Tractatus Logico-Philosophicus, London : Routledge & Kegan Paul, 1974
Notebooks, 1914-1916, edited by G. H. von Wright and G. E. M. Anscombe. with an English translation by G. E. M. Anscombe
Philosophical Grammar, edited by Rush Rhees, Oxford, Basil Blackwell 1974
Philosophical Remarks, Philosophical remarks : edited from his posthumous writings by Rush Rhees, Oxford : B. Blackwell, 1975
Norman Malcolm – Memory and Mind, Ithaca, N.Y., Cornell University Press, 1977
Marc Joseph - Mental Representation and the Metaphysics of Meaning in Wittgenstein’s Tractatus, Philosophical Investigations 2000).