1.1. A conceptual problem

Traditionally, the Privacy Thesis is a springboard to an epistemological problem: How can I know that someone else is in pain? Although Wittgenstein sometimes speaks of knowledge, for him the Thesis is primarily a springboard to a conceptual problem: What could I possibly mean in speaking of ‘someone else's pain’?

Roughly put, the problem is this. Given the Privacy Thesis, I feel my own pains. As such, I will claim to understand fully what I mean by saying ‘I am in pain’. But, precisely because my pains are mine, there is no way for me to ascribe them to someone else. And this curtails what I could possibly mean by saying ‘he is in pain’.

To get this problem rolling, consider the following passage from Wittgenstein:

To defend the attempted ‘simple’ explanation against this line of thought, one might reply as follows:

But, when the Privacy Thesis is read a certain way, it blocks any reply along these lines, for the Thesis precludes anyone else from having my pain (type).Footnote ⁴ What we need is a way to ‘map’ my pain (type) over to you. But, it is just unclear how we could even try to effect this mapping. As Wittgenstein splendidly put the problem:

At this point, then, I will want to ask: What can I mean by saying ‘he is in pain’, or ‘he hurts’? There are a few options, but none of them seems very good. For example, I could suggest that ‘he hurts’ means that he is behaving (physically) as I do when I am in pain. Now, no doubt this suggestion is far too crude to handle all the varied uses of ‘he hurts’; I would need to draw a more sophisticated link between ‘he hurts’ and (dispositions to) behaviour. But, given the Privacy Thesis, the problem is that anything I could hope to mean by ‘he hurts’ could be rooted only in what is physically observable.Footnote ⁶ And that is a far cry, it seems, from what I mean when I say ‘I hurt’.

1.2. Zeugmas and solipsism

So now suppose I consider the predicate ‘x hurts’, as it occurs in the two expressions ‘I hurt’ and ‘he hurts’. Given what has come before, I will want to insist that ‘it is no longer [the] same function just with different arguments.’Footnote ⁷ Rather, I will insist that it has a very different meaning in these two occurrences. So, I will advocate the following:

Indeed, I will regard ‘you and I hurt’ as a bad zeugma, on a par with ‘I threw up and down’.Footnote ⁹

Consequently, I will want to mark a distinction between the predicate which I use for me, and the predicate I apply to everyone else. Given the Privacy Thesis, I cannot doubt that my pains are real, so I will reserve ‘x really-hurts’ for me. By contrast, I have been led to the view that it is unintelligible to think that anyone else experiences (my) real pain. So, when it comes to other people, instead of saying ‘x hurts’, I will say ‘x kinda-hurts’. (As suggested above, the sense I attach to ‘x kinda-hurts’ might be in the ballpark of ‘x’s body behaves like mine, when I really-hurt’.) Then, using this new vocabulary, I will say the following:

But I have called this Thesis ‘Solipsistic’, since it is exactly how Wittgenstein characterises solipsism in The Blue Book:

Note that, as befits a ‘grammatical’ investigation of solipsism, this characterisation of solipsism is meaning-theoretic. But the short point is this: I will have been led from the Privacy Thesis, via the Dual-Meaning Thesis, to some version of solipsism.

1.3. From solipsism to selfless solipsism

In fact, we can characterise this version of solipsism a bit more precisely. Consider the following reasonable thesis, due to Evans:

Now consider what happens when the Solipsistic Thesis meets the Compositional Constraint. For reductio, suppose that I understand both ‘I really-hurt’ and ‘Ali kinda-hurts’ as having a subject/predicate structure. Then, by the Compositional Constraint, I will also be able to understand the sentence ‘Ali really-hurts’. But the Solipsistic Thesis states that this makes no sense. So, I must understand at least one of ‘I really-hurt’ or ‘Ali kinda-hurts’ as not having a subject/predicate structure. However, my understanding of ‘Ali kinda-hurts’ surely does have that structure: there is nothing objectionable about replacing ‘Ali’ with ‘he’, or ‘you’, or any other term which picks out a human being.Footnote ¹³ So, I must conclude that ‘I really-hurt’ does not have a subject/predicate structure after all. And, since it is hard to see how it could have any other structure, I must accept that my understanding of ‘I really-hurt’ is unstructured.

This, however, has a surprising upshot. When I say ‘I really-hurt’ – which, given the Dual-Meaning Thesis, was just a way for me to say ‘I hurt’ more clearly – I do not attribute anything to anyone. The surface syntax is altogether misleading. It will therefore be clearer if, instead of saying ‘I (really-)hurt’, I were to say ‘it hurts’ – as one says ‘it snows’ – or, better yet, I were simply to say ‘hurt’.

In short, I will both have become a solipsist, and given up speaking of ‘me’. And this is precisely why Wittgenstein tells us that:

Admittedly, in charting the route from solipsism to selfless solipsism, Wittgenstein does not formulate anything quite as crisp as the Compositional Constraint. However, something like that Thesis is implicit in his remark that ‘if another cannot have my toothache, then also I – in this sense – cannot have it.’Footnote ¹⁵

The upshot, though, is that we have been led to what Canfield calls ‘the thesis of selfless solipsism’, which ‘lies at the heart of the Tractatus’.Footnote ¹⁶ It is worth pausing to consider just how Tractarian this selfless solipsism is. When I say ‘he (kinda-)hurts’, I say something about a particular object in the world. However, I say nothing about any particular object when I say ‘I (really-)hurt’. As such, I draw no boundary between myself and the world. Indeed, if ‘my’ hurt – or, perhaps better, if hurt – belongs to anything, then it belongs to the entire world. Aphoristically: when ‘I’ am happy, the world waxes; when ‘I’ am sad, the world wanes; so that ‘I’ do not feature in the world at all, but appear as its limit.Footnote ¹⁷

1.4. The grammar of ‘my pain’

So far, I have outlined a line of thought which begins with the Privacy Thesis, runs through the Dual-Meaning Thesis into the Solipsistic Thesis, and ends up with selfless solipsism. I now want to explain where this line of thought goes wrong. Wittgenstein gives us exactly the right answer: the mistake involves ‘wavering between logical and physical impossibility’.Footnote ¹⁸ But this needs some unpacking.

Consider, again, the Privacy Thesis. On one good reading, it is just false to say ‘I, and only I, can feel my pains’. Indeed, you have probably experienced at least some of my pains: like me, you may have suffered through dislocated shoulders and disappointments.

This reading, however, treats the phrase ‘my pains’ quite generically. One could reasonably read the Thesis more specifically. So: whilst you may also have suffered the pain of dislocated shoulders, your pain was not the specific pain that I felt.

That is no doubt true. But it also seems contingent, whereas the Privacy Thesis makes a claim about what is necessary. And that necessary claim remains in trouble. To see why, here is a thought-experiment, essentially due to Wittgenstein, which I will use frequently in what follows:Footnote ¹⁹

In the Sci-Fi Example, there is a sense in which you feel all of my specific pains. And, since the Sci-Fi Example is possible – given a permissive enough notion of possibility – there is a sense in which you can feel all of my (specific) pains. In that sense, the Privacy Thesis is false.

To defend the Privacy Thesis, I might draw a type/token distinction. I might insist that, even in the Sci-Fi Example, I have my pain-token, and you have yours, and although these are of the same type, they are distinct pain-tokens.Footnote ²⁰ But to say this, I must introduce a new count-noun, ‘pain-token’; and introducing this noun does not seem compulsory.Footnote ²¹ The mere fact that our bodies remain distinct, in the Sci-Fi example, only requires that we speak of distinct people in pain, not of ‘distinct pain-tokens’.

Still, even if we agree to say that you and I have ‘distinct pain-tokens’ in the Sci-Fi Example, the Privacy Thesis remains shaky. To show this, we can simply follow Wittgenstein in pushing the Sci-Fi Example further still. So, suppose that you and I are note merely wired together using a super-computer, but that we have ‘a part of our bodies in common, say a hand’; that ‘the nerves and tendons of my arm and [yours are] connected to this hand by an operation. Now imagine the hand stung by a wasp. Both of us cry, contort our faces, give the same description of the pain, etc.’Footnote ²²

As Wittgenstein notes, I could refuse to count even this as a case where you feel my pain(-token). However, by now, there is nothing I could say to justify my refusal, beyond doggedly insisting: ‘If you feel it, it isn't mine’.Footnote ²³ But that insistence would amount only to a grammatical proposal about how to use the phrase ‘my pain’ (or ‘the same pain’).Footnote ²⁴ Otherwise put: in an effort to defend the Privacy Thesis, I have turned it into a proposed rule of grammar.

At this point, though, the Privacy Thesis has become completely harmless. After all, any grammatical proposal is optional. So, I could speak this way if I wanted to. But I must not make the mistake of thinking that my choice to speak a certain way has taught me anything new about the world,Footnote ²⁵ or anything about what my words could possibly mean.

2.1. The two uses, in context

Wittgenstein moves immediately from a discussion of solipsism – along the lines sketched in §1 – into the passage where he distinguishes two uses of ‘I’:

A vast literature cites this passage as alerting us to an important phenomenon.Footnote ²⁷ However, that literature rarely considers why Wittgenstein was interested in the phenomenon. Indeed, when this passage is cited, the first sentence is usually omitted. But that sentence shows clearly that Wittgenstein wants to connect the temptation towards solipsism – which I discussed in §1 – with misunderstanding the grammar of the two uses of ‘I’. As such, the vast literature citing this passage has taken us to interesting places, but very different ones than Wittgenstein intended.

In the rest of this paper, my aim is to do what that literature does not: to revisit Wittgenstein's work, with the aim of showing both why the subject-use of ‘I’ threatens to lead us to solipsism, and how solipsism is to be avoided.

2.2. Subjectivality and objectivality

I will start by unpacking the passage a little. Wittgenstein states that the distinctive feature of the object-use of ‘I’ is that it ‘involve[s] the recognition of a particular person’, and that in such cases one can make a mistake about who the person is. To unpack this remark, I will flesh out each of Wittgenstein's four exemplar object-uses of ‘I’:

The final case is ‘the wind blows my hair about’. To deal with this, I could re-use the Mirror Example. But, both because it will both prove helpful later, and also because it provides a connection with the material from above, I will deal with this by using a variant of the Sci-Fi Example from §1.4:

To capture what is at work in these examples, and in Wittgenstein's general distinction, I will now offer a schematic definition:Footnote ²⁹

I hope it is immediately clear that Wittgenstein's four examples of subject-uses of ‘I’ are subjectival, and that his four examples of object-uses of ‘I’ are objectival. But, to make it completely explicit, consider the (variant) Sci-Fi Example. Here, (a) my tell-tale sensations give me grounds to believe that the wind blows my hair about; but (b) if I were to discover that I was wired up to Chip, then those tell-tale sensations would cease to give my any grounds for thinking that that the wind blows my hair about, and would instead give me grounds to believe that the wind blows Chip’s hair about.

2.3. Subjectival statements as descriptions of phenomenology

In fact, having worked through the (variant) Sci-Fi Example in detail once, it should be clear that Sci-Fi setups can be used to highlight the objectivality of any of Wittgenstein's exemplar object-uses of ‘I’. Recognising this will allow me to show why Wittgenstein claimed that subjectival statements are attempts to describe phenomenology.

Any statement ‘I am φ’ which says something physical about me is objectival. To see this, consider again the (variant) Sci-Fi Example from §1.4. There, I judge that the wind blows my hair about, when in fact Chip's hair is being blown about and I am receiving sensations from his scalp; so my claim is objectival. We can raise similar issues for any physical attribution that I could justifiably form on a sensory basis.Footnote ³⁰ For example: the statement ‘I am seated’ is objectival, since I may say this because I am receiving Chip's proprioceptive signals. Similarly, the statement ‘a tree is in front of me’ is objectival, since I may make this claim because I am receiving the signals from Chip's retinas. Generalising, enough Sci-Fi thinking will show that any statement which says something physical about me – any claim about my posture, my body, or my immediate environment – is objectival.

Consequently, if we are looking for subjectival statements, then we must turn our attention from physical attributions to mental attributions.Footnote ³¹ But this has a knock-on effect. Many attributions are mixed, and we must bracket the physical components of such attributions.Footnote ³² To illustrate: suppose I say ‘I am hungry’, but then discover that the situation is Sci-Fi, and that I am receiving the hunger pangs from Chip's body rather than my own. In this situation, I may reasonably want to say ‘it is Chip (not me) who is hungry’.Footnote ³³ This suggests a good sense in which the statement ‘I am hungry’ is objectival. Conversely, if my statement is supposed to be subjectival, then we must take the claim to involve only ‘narrow’ content; roughly, to concern only what is going on ‘in my head’. Then we can say: even if I am hungry in the narrow sense, only because Chip is hungry in the flesh-and-blood sense and I am being sent signals from Chip's body, still, this in no way threatens that I am hungry in the required (narrow) sense. That is: everything physical must be bracketed from any subjectival claims.

Similarly, any notion of duration must be bracketed away from subjectival claims. Consider ‘I was hungry’. Even after we have bracketed the question of exactly whose stomach rumbled, this past-tensed statement remains objectival. After all, I may say ‘I was hungry’ because I seem to remember being hungry, when in fact these are not my own memories of hunger but rather Chip's memories, which have been implanted in me by some (further) Sci-Fi mechanism. On discovering this, I may reasonably say ‘it was Chip (not me) who was hungry’.Footnote ³⁴

Subjectival claims must, therefore, be restricted to present-tensed, mental judgements.Footnote ³⁵ And only a little more generalising should convince you that subjectival statements concern nothing but sense data, or pure phenomenology, or some-such.Footnote ³⁶

Wittgenstein was aware of all of this. In a lecture on 6^th March 1933, Wittgenstein introduced the idea that ‘I’ has two uses. But he then immediately insisted that ‘I might have tooth-ache, even if there were no body – if my body was destroyed’, when the ‘I’ is used as subject.Footnote ³⁷ He repeated the same point about toothache later in that lecture,Footnote ³⁸ and also offered another example: ‘“This sound moves round my head clockwise” has meaning independently of the physical head.’Footnote ³⁹ In fact, the connection to phenomenology is completely explicitly in The Blue Book. Three pages after introducing his two uses of ‘I’, he wraps up his discussion of the subject-use, and says:

So, when Wittgenstein offers ‘I see so-and-so’ and ‘I hear so-and-so’ as exemplar subjectival statements, he (implicitly) takes it that everything physical is bracketed from such statements; that they concern only first-personal phenomenology.

2.4. From phenomenology to solipsism

Once we have recognised that subjectival claims amount to descriptions of phenomenology, however, there is a very quick route to solipsism.

We have seen that if I say (subjectivally) ‘I hurt’, then my claim amounts to a description of my phenomenology. By contrast, a claim like ‘he hurts’ does not concern my phenomenology.Footnote ⁴¹ As such, I must insist that the sense of the predicate ‘x hurts’ is totally different, when we consider ‘I hurt’ (understood subjectivally) rather than ‘he hurts’. That is, I will embrace the Dual-Meaning Thesis of §1.2. To mark the difference, I will want to use ‘x really-hurts’ for subjectival statements, and ‘x kinda-hurts’ for other people.Footnote ⁴² And so, exactly as in §1.2, I will arrive at the (meaning-theoretic) Solipsistic Thesis.

In short, thinking about subjectivality has given fresh life to the Dual-Meaning Thesis, and thereby renewed the threat of solipsism. Moreover, the renewed threat is much harder to deal with than the original threat. Indeed, Wittgenstein's own response to this threat seems to emerge only gradually, over months or maybe years. Roughly, his thought progression is characterised by the following stages:

1. (1) The word ‘I’, as it occurs in subjectival statements, does not refer to anything.

This claim rebuts any version of solipsism which insists that some particular self is privileged. And this line of thought is clear from Wittgenstein's 1933 lectures onwards. However,

1. (2) Claim (1) is compatible with selfless solipsism.

Moreover, there are certain hints of selfless solipsism in the 1933 lectures. That said,

1. (3) We can avoid solipsism, of any form, by regarding subjectival utterances as more like moans than judgements.

This line of thought is floated in The Blue Book. But Wittgenstein's decisive answer to solipsism comes when he realises that:

1. (4) Considerations about private languages show that (almost) no truth-apt content can be attached to any subjectival utterance.

There are a few hints of this in The Blue Book, but the idea becomes much clearer in subsequent lecture notes.

In the rest of this paper, I will run through these four chronological stages of Wittgenstein's engagement with solipsism, considering stages (1)–(3) in the next section, and paving the way for the real answer to solipsism in §4.

3.1. ‘I’ as not referring

The main line of thought, in both Wittgenstein's 1933-lectures and The Blue Book, is this: the word ‘I’, as it occurs in a subjectival statement, does not refer to a particular self. Here, I will outline why Wittgenstein says this; in the next subsection, I will explain why this does not really answer the solipsist.

As explained in §2.3, subjectival statements are just attempts to describe phenomenology. As a matter of phenomenology, though, when I see a red patch, I do not observe myself seeing a red patch; I simply visually experience redness. Likewise, to use an example from Wittgenstein, ‘The experience of feeling a toothache isn't the experience that a person, I, has something.’Footnote ⁴³ More generally, according to Wittgenstein, the ‘truth is: No person necessarily enters into a sensory experience at all.’Footnote ⁴⁴ His point here is simply a rewording of Hume's observation that ‘when I enter most intimately into what I call myself, I always stumble on some particular perception or other[…]. I never can catch myself at any time without a perception, and never can observe any thing but the perception’.Footnote ⁴⁵

The upshot is this. Since subjectival statements are descriptions of phenomenology, and I am not an element of the phenomenology, the word ‘I’ in a subjectival statement does not refer to me. So, when I (subjectivally) claim ‘I hurt’, I do not attribute anything to anyone. As in §1.3, then, it would be less misleading for me to say ‘it hurts’, or simply ‘hurt’. In short: I disappear from subjectival statements.Footnote ⁴⁶

3.2. Selfless solipsism remains standing

Unfortunately, though, we cannot refute solipsism just by showing that the word ‘I’ does not refer when used in a subjectival statement.

In his Blue Book discussion of subjectivality, Wittgenstein clearly sets out the following line of thought. Before we investigate the ‘grammar’ of subjectival statements, we might think that the word ‘I’ must always refers to something. But, when we make a subjectival claim ‘I am φ’, it seems that we do not use the word ‘I’ to refer to some person that we have recognized by their ‘bodily characteristics; and this creates the illusion that we use this word to refer to something bodiless, which, however, has its seat in our body’.Footnote ⁴⁷ To defeat this illusion, then, it is sufficient to show that, in these cases, the word ‘I’ does not refer to anything after all. That is: by better understanding ‘the peculiar grammar of the word “I”’, we will rid ourselves of ‘the idea that the real I lives in my body’.Footnote ⁴⁸

Overcoming (this caricature of) Cartesian metaphysics is certainly some achievement.Footnote ⁴⁹ Moreover, the line of thought that tells against (caricatured) Cartesian metaphysics also tells against the specifically metaphysical version of solipsism which insists that one particular self – myself – is somehow at the centre of the universe.

In this paper, though, I have focussed on a meaning-theoretic version of solipsism, as characterised via the Solipsistic Thesis in §1.2. And this meaning-theoretic solipsism does not take any stance on whether ‘I’ refers in subjectival statements. Indeed, we saw in §1.3 that meaning-theoretic solipsism leads, via the Compositional Constraint, to selfless solipsism. And we obviously cannot criticise this version of solipsism, by insisting that subjectival uses of ‘I’ do not refer to any self.

Worse: insisting that subjectival uses of ‘I’ do not refer actually exacerbates the the threat of selfless solipsism. We can see this by rehashing some ideas from §1.3. When I say ‘Ali hurts’, I simply state a fact about Ali, an object in the world. But if ‘I’ does not refer when I (subjectivally) say ‘I hurt’, then I do not thereby attribute (real) pain to any object in the world. I draw no boundary between myself and the world, and so attribute hurt to the world itself. And that is selfless solipsism.

In short, many of Wittgenstein's remarks in 1933–4 actually seem to entail selfless solipsism. And it is not immediately obvious that this would be contrary to Wittgenstein's aims. After all, Wittgenstein had once claimed that ‘what solipsism means is quite correct; only it cannot be said, but shows itself’.Footnote ⁵⁰ Perhaps he still believed, in 1933–4, that ‘the I of solipsism shrinks to an extensionless point, and there remains the reality coordinated with it’.Footnote ⁵¹

In fact, the textual evidence is too thin to form a firm judgement on Wittgenstein's views in 1933. But it worth noting that Wittgenstein's 1933-lectures, preceding The Blue Book, contain nothing concrete about how to avoid this route to (selfless) solipsism. We have, for example:

Wittgenstein clearly says that it ‘is absurd’ to embrace the metaphysical version of solipsism, which insists that myself is the centre of the universe. But the rest of what he says is quite compatible with selfless, meaning-theoretic, solipsism.

If it were true that this ‘solipsism, strictly carried through, coincides with pure realism’Footnote ⁵³ – as Wittgenstein had claimed in the Tractatus – then perhaps we could live with it. However, if I arrive at selfless solipsism by the route outlined in §2, then my world will not coincide with the world of the realist, but will instead retain a detestably solipsistic residue. For, as things stand, I will have two languages: our ordinary language, and also a phenomenological language. Once I have recognised that a subjectival use of ‘I’ does not refer to anything, I will stop saying that the phenomenological language makes claims about me or about my experiences.Footnote ⁵⁴ Still, to describe the world in its entirety, I will need to use both languages. And there will be no possible comparison between ordinary pain, as described in ordinary language and attributed to ordinary human beings, and (real) hurt, as described in (my the?) phenomenological language, and attributed to (the limits of) the world. That, surely, is unacceptable.

3.4. Expressivism avoids solipsism

As mentioned: I am not sure about Wittgenstein's explicit views on selfless solipsism in his 1933-lectures. Fortunately, though, his Blue Book contains some remarks which commit him to denying that there is any truth in selfless solipsism.

In §3.3, I suggested that selfless solipsism is not just compatible with the idea that subjectival uses of ‘I’ do not refer, but is entailed by that idea. However, my argument implicitly depended on the premise that subjectival utterances express bona fide judgements. The problem, in a nutshell, is that any judgement I make would draw no boundary between myself and the world. But, if I make no judgement in saying (subjectivally) ‘I hurt’, then I neither ascribe anything to me nor to the world. (There is no hurt, and no unacceptably solipsistic residue.)

In principle, there are many ways to flesh out the idea that I make no judgement in saying ‘I hurt’. But a specific account, known as expressivism, holds this:

Expressivists have nothing to fear from selfless solipsism. Whilst my words ‘I hurt’ make no reference to me, nor do my moans, and nothing deep follows from either fact. So, by adopting expressivism, it is possible to maintain that subjectival uses of ‘I’ do not refer, whilst avoiding selfless solipsism. Perhaps, then, this is why Wittgenstein suggested that ‘[t]o say, “I have pain” is no more a statement about a particular person than moaning is’;Footnote ⁵⁵ it would allow him to avoid sliding into an unacceptable solipsism.

This is a step in the right direction. However, it does not provide much by way of an answer to an already-committed solipsist. As it stands, the selfless solipsist will just disagree with expressivism, and insist that his (subjectival) claim ‘I hurt’ really does express a full-fledged judgement. What we need is an argument that he is wrong. Hence, with Wittgenstein, we find ourselves saying:

In fact, the solipsist is to be brought to rest by developing a private language argument.

4.1. A quick connection to private languages

It is tempting to draw the following, straightforward connection between subjectivality and Wittgenstein's private language argument:Footnote ⁶⁰

1. (a) Subjectival statements are attempts to describe phenomenology.

2. (b) If phenomenology can be successfully described at all, then it must be described in a necessarily private language.

3. (c) If Wittgenstein's private language argument is correct, then there cannot be a necessarily private language.

4. (d) If Wittgenstein's private language argument is correct, then there can be no successful subjectival statements.

I established premise (a) in §2.3, and (c) is uncontentious. Moreover, (b) is certainly plausible. However, nothing I have said so far forces it upon us. For example, it is not immediately obvious that a narrowly phenomenological description of my pains could not help a doctor to diagnose my illness. So, in what follows, I will allow that (b) might be false, but will show how Wittgenstein's early discussion of private languages provides us with materials for showing that subjectival statements lack (almost) all public content.

4.2 Subjectival statements and testimony

In §2.3, I explained that everything physical must be bracketed from subjectival claims. In fact, we must also bracket anything relating to testimony. To see why, consider an example due to Coliva:Footnote ⁶¹

As such, the statement ‘I hate myself’ is objectival, at least as ordinarily understood. Contraposing, subjectival statements cannot involve any content which could be connected with the testimony of other people.

However, the impact of testimony extends beyond the purview of subconcious thought. In lecture notes dated shortly after The Blue Book, Wittgenstein makes this note: ‘Can a man doubt whether what he sees is red or green? (Elaborate this.)’Footnote ⁶² I will elaborate on this, on Wittgenstein's behalf, by using a variant of the Testimony Example:

As Wittgenstein required, this sort of Example generates a situation – admittedly, a rare one – in which a man – me – can legitimately doubt whether what he sees is red or green. But now I want to focus on what this Example teaches us about subjectival claims.

In the Relearning Example, we may take it that my colour-words pertain to the physical swatches. In that case, my claim ‘I see red’ is objectival; indeed, it is as objectival as ‘I have a red swatch in my hand’, and for the same reason. However, this is no surprise. In §2.3, we saw that my subjectival claims must be a description of my phenomenology, and not (for example) an attempt to describe some physical swatch. To make the Relearning Example interesting, then, we should assume that, in the Relearning Example, I am saying ‘I see red’ in an effort to describe my phenomenology .

Now, in §4.1, we considered the following:

1. (b) If phenomenology can be successfully described at all, then it must be described in a necessarily private language.

If we assume this, then I must offer ‘I see red’ in a necessarily private language. In that case, nothing the teacher might say in a public language could either give me any grounds for saying ‘I see red’, nor count against my saying that. At that point, the Relearning Example will become wholly irrelevant to the question of whether my claim ‘I see red’ is subjectival or objectival.

However, as explained in §4.1, I do not want to assume (b). Instead, I am assuming that we can (try to) describe phenomenology in a public language. And in that case, the Relearning Example remains relevant to my claim ‘I see red’.

Indeed, if we assume that we can (try to) describe phenomenology in a public language, then we must conceive of the Relearning Example as follows: by directly teaching us the names for the physical colours, the teacher is indirectly teaching us the names for the (our?) phenomenological colours.Footnote ⁶³ And on this understanding, the Relearning Example has this structure: I take it that the teacher was indirectly saying something about my phenomenology, when in fact she is indirectly saying something about someone else's phenomenology. If I were to find all this out, I would stop saying ‘I see red’, but would remain justified in saying ‘someone else sees red’.

In short: if my statement ‘I see red’ describes my phenomenology in a public language, then my statement is objectival. Conversely, if I want to insist that my statement is subjectival, then I must bracket the intended meaning of my statement, so that it is narrower even than an attempt to describe phenomenology.Footnote ⁶⁴

4.3. ‘Hello, World!’

In fact, though, the amount of bracketing that is required would leave ‘I see red’ with (almost) no public content at all. To show this, I will consider various attempts to express the supposedly remaining content in a public language, and show why they fail.

I might start by suggesting the following:

However, ‘this colour’ is a demonstrative, and – in Wittgensteinian spirit – you should ask me what is demonstrated. I might respond by accompanying the word ‘this colour’ by a gesture, perhaps pointing at the colour-swatch I am looking at. But this will bring me back to the colour of swatches, and away from my phenomenology. So I will be tempted, instead, to insist that my words ‘this colour’ are accompanied by some inner, mental version of pointing. I want, as it were, to point to a location in my own visual field, or perhaps a spot in my inner colour-space. But such an ‘inner pointing’ is, of course, utterly inarticulable, as Wittgenstein brought out as early as The Blue Book.Footnote ⁶⁵ If my aim is to convey something public, this is not the way to go.

As such, I must give up on any attempt to say something specific about colour – whether by using a name or a demonstration – in a public language. So I must make my claim more generic. I might try suggesting:

The Relearning Example does not threaten this claim. However, we can bring it into jeopardy just by slightly tweaking the Example. Imagine that the group is being taught names, not just for colours, but also for textures. I am rubbing a swatch of hessian, and hear the teacher say ‘you're seeing red’. I take her to be addressing me, and so say, ‘I see red’. But in fact, the teacher is speaking to someone else. My eyes are closed, and I am seeing nothing, but feeling hessian. And this example directly connects with one of Wittgenstein's own remarks about private languages: if my aim is to provide my own labels for my phenomenology,

Summing up: I have been led to acknowledge that the public content I would express by saying (subjectivally) ‘I see red’ must be even thinner than what I would express by saying ‘I see [something]’. And at this point, I must simply concede that there is no public content left to my claim that ‘I see red’, if it is supposed to be subjectival. Or rather: the only content which could possibly remain is the sort of content that is usually conveyed with a claim like ‘I am conscious’. And, since the content is to be public, we should not hear either claim as uttered in hushed, reverential tones – as flagging some deep riddle of consciousness – but as if they had been said by someone who has woken up from their anaesthetic and wishes to indicate that they are once again receptive to stimuli. It is a version of the ‘Hello, World!’ that one uses to show that a system is (barely) functioning.

4.4. Solipsism avoided

I have established that subjectival claims must lack (almost) any public content. It only remains to show how this allows us to avoid solipsism.

The route to solipsism outlined in §2 ran as follows: in considering subjectivality, I realise that my aim is to describe pure phenomenology, and so I embrace the Dual-Meaning Thesis. However, once I have seen what I would have to bracket in the face of the Relearning Example, I am no longer compelled towards the Dual-Meaning Thesis. Certainly, the sense of ‘x hurts’ would have to be different, when on the one hand I say ‘he (kinda-)hurts’, and on the other hand I try to make a subjectival claim ‘I (really-)hurt’. But this is not because my ‘I (really-)hurt’ describes real pain, as it were; rather, it is because it fails to describe anything. It is the difference between saying something of an object, and saying ‘Hello, World!’

The crucial point in the refutation of solipsism, then, is as follows. If ‘I hurt’ expresses more than an utterly minimal public content, then it must be objectival. Now, this point is quite compatible with expressivism, as discussed in §3.4. But it is equally compatible with rejecting expressivism. Expressivism is just orthogonal to the real refutation of solipsism.