The Private Language Argument Isn't as Difficult, Nor as Dubious as Some Make Out
Roger Harris
Introduction
The sections of Wittgenstein's Philosophical Investigations which contain the Private Language (PL) Argument are dense, cryptic and wide ranging. I argue that a specific argument against a private language can be distilled from the text that is less involved and obscure than is often supposed in the immense secondary literature. It is also far less self-contained and isolated from the mainstream of philosophy than many make out, including Brian GarrettFoot note 1 and Michael Ming YangFoot note 2 in recent papers in this journal. It can be distinguished from arguments about rule-following, pain and the problem of other minds, forms of life, etc. and, as I have framed it, avoids Garrett's objections. Moreover, a number of would-be conclusions Garrett takes its proponents to draw from the argument, far from depending on it, are long standingFoot note 3 positions, on some of which the PL Argument itself depends.
§1. Privacy
Although, for Garrett, «It is crucial to be clear about the sense of `private'» invoked by Wittgenstein, privacy remains virtually unexamined in his paper. These days, privacy is best discussed using some terms which have become current since Wittgenstein wrote. The realm of necessary privacy within which, Wittgenstein maintains,Foot note 4 it is not possible to follow a rule, and so not possible to have a language, is the realm of immediate first person mental life. I.e. immediate phenomenal episodes of sensory experience, memory, and the other aspects of one's mind which one owns exclusively. These are available only through introspection, judgements about them are wholly subjective, and their properties are the qualia that have subsequently been presented as posing the `hard problem' of the mind/body relation.
Wittgenstein has been taken by some to deny the reality of subjective inner mental life. I think he does not, and must not, if his PL Argument is to succeed, so I will devote a few paragraphs to this issue. His discussion of inner mental life turns in part on the truism that you cannot feel my pain. This exemplifies the necessarily private content of that realm of exclusivity which is the inescapable corollary of my experiences and thoughts being mine and not yours, and vice versa. This sphere of necessary (rather than `logical') privacy is delineated by that proprietary possibility, namely that experiences, thought and whatever else comprises an inner mental life, can
This determines some key features of privacy and subjectivity. Let's look at privacy and introspection first (I discuss subjectivity in §4. below).
If qualia are the properties e.g. of immediate phenomenal episodes, then, just in so far as they can be owned, they must possess aspects which are not shared, and which therefore cannot enter any arena of public comparison. This does not make qualia ineffable, however. We can refer to them, and describe them in detail. So you cannot feel my pain, nor I yours, but, as a diagnostician, for example, I can perfectly well appreciate the finer points of your pain as pointers towards the pathology I investigate. Physicians undoubtedly enjoy a huge advantage over veterinarians in virtue of our well developed phenomenological vocabulary for pain. Wittgenstein takes the fact of public talk about private pains to show that pain descriptors cannot get their senses from episodes accessible only to introspection.
§2. Public Sense but Private Reference
Garrett maintains that the PL Argument:
In reality the Private Language Argument is not needed to undermine this model. It suffices merely to cite an example -- `tongle'. This expression applies to a series of my mental episodes with which I am acquainted and you are not, and necessarily cannot be. I am sure you do not understand the expression `tongle'. It has been customary since Frege to distinguish what an expression conveys -- its `sense' -- from what it picks out -- its `reference', because `meaning' is ambiguous in this regard. There is a flat contradiction in supposing that what an expression conveys, when it is used by you and I, and speakers of English at large, is something which is not equally accessible to all who use it. Since I cannot feel your pain and vice versa, what you feel when you have a pain cannot qualify as what I convey by describing my pain, but no more can what I feel, since that is not accessible to you. But even if it is only the reference of `tongle' that it inaccessible, as I «define it with reference to my» tongles, this `definition' on my part is not a success, for you are none the wiser, any more than Alice understood Humpty DumptyFoot note 5 when he reassured her that
«When I use a word,» Humpty Dumpty said in a rather a scornful tone, «it means just what I choose it to mean -- neither more nor less.»
«The question is,» said Alice, «whether you can make words mean different things.»
«The question is,» said Humpty Dumpty, «which is to be master -- that's all.»
At the risk of seeming glib, I say that these expressions, attributing pain and the like, have a public sense but a private reference.Foot note 6 What it is (privately) to have them, in such a way that they are your pains and no one else's, cannot provide the public senses of the expressions employed in the phenomenological vocabulary for pain. Humpty Dumpty said words mean what he chooses them to mean. No they don't. If they didn't already mean what they do mean, there would be no meanings for him to choose between. `Headache' is a word in English, and I learn from others that it has a sense which allows me to refer to my private sensations.
In support of his defence of the Cartesian model Garrett cites Craig's claim that an expression's being used to refer to what can only be known to the speaker doesn't stop its being understood by others. That is no support: Garrett can refer to his own headache without his own headache being the sense of the expression `headache' as he uses it, because he speaks English, and `headache' already has a sense and a definition. If you understand the English word you are able to refer to the private pains of whomsoever you please, including yourself. Craig's claim is equally compatible with what I have claimed.
What makes this possible? The answer can only be «the public sense of the expression `headache' which we all use.» This cannot be an item known only to Garrett by introspection. An expression could not be generally intelligible if it were generalisable over nothing but a plurality of private mental episode tokens accessible solely to the introspection of a single individual. Applicability both to mental episode tokens to which you do and to those to which you don't have immediate access is a necessary condition for the intersubjective intelligibility of any mental content descriptor we might employ. So the PL Argument is not needed to dispense with private senses. These can quite independently be shown to be impossible.
§3. Private Facts
I wrote above of `what it is (privately) to have' pains because I think their necessary privacy crucially undermines the more usual locution `what it is like...' to have a pain, be a bat, or whatever. Suppose you grant that the realm of necessary privacy is defined by ownership and accessibility to nothing but introspection, so its contents, qualia are consequently excluded from any arena of public comparison. If so, then the only relations of `likeness' into which such qualia can possibly enter must be entirely confined to that necessarily private realm, unless these are not simple likenesses between monadic properties of qualia at all, but involve relations they bear (or are disposed to bear) to other qualia and to properties of items outside the private realm. I argue that this is just what they do involve.
If the individual qualitative character of qualia, disclosed by introspection, cannot belong in an arena of public comparison, then discussions of what they are like, and of possible undetectable `inverted spectra' of qualia must surely be beside the point. Your qualia are not open to inspection by me, nor mine by you. The respects in which your qualia can resemble mine therefore cannot be those qualitative respects available only to you as the `owner' of those qualia. So no question of qualitative resemblance between your individual qualia and mine can arise, because the individual qualitative character of qualia cannot enter the arena of intersubjective comparison. By the same token, then, no question of qualitative difference can arise either. So there can be no `fact of the matter' about inverted spectrum claims. This is why Wittgenstein departs from a line of discussion that began with LockeFoot note 7 when he says, of his `beetle in a box' that «It is not a something, but not a nothing either» (PI § 304). Your `beetle' is not open to inspection by me, nor mine by you.
That is not to say that your qualia do not resemble mine in any respects -- indeed, there must be respects in which they do. In order for us to speak of us both seeing crimson, yellow or turquoise, your qualia must resemble mine in respect of the relations they bear (or are disposed to bear) to other qualia and to properties of things and events in the public realm. Your perceptions and mine could not otherwise coincide when it comes to reacting to, acting on, or assenting to descriptions of what we perceive. So we can say that each one of my colour perception qualia and of yours (if you and I have accurate colour vision) belong to relational resemblance classes, which contain the relevant qualia of everyone with normal colour vision who lives in a polychromatic world. Those people whose discriminations are faulty -- (can't tell red from green, say) lack qualia in the relevant relational resemblance classes.
«Aha!» the Cartesian will say «then those who can't tell red from green don't know what `red' and `green' mean, so the meanings of `red' and `green' must be the inner states possessed by those who can tell them apart.» Indeed, Garrett writes:
There is no `so' about the identification of the senses of `red' and `green' with inner states. Let me take the suggestions in the previous paragraph a little further. Delineating the realm of the private turns on the distinctness of my inner life from yours, and this consists precisely in your not being privy to the private aspects of my qualia, and vice versa, since no properties of a private episode are open to inspection by any other. So what your experience of seeing the colour red has in common with mine cannot be a monadic property of that private phenomenal episode, (what it is for me to see something red).
Garrett has a further argument to fulfil his undertaking:
This turns on the claim that a confusion is involved between qualitative and numerical identity when we say that you cannot feel my pain. You cannot feel my present pain token, because it is mine, but that does not show that you cannot have the same pain type as me. Indeed, it does not, but nor does it refute my earlier argument. Namely, «If qualia are the properties e.g. of immediate phenomenal episodes, then, just in so far as they can be owned, they must possess aspects which are not shared, and which therefore cannot enter any arena of public comparison..» If God could look into both our minds, He might see that our pains were of exactly the same type. However, since neither of us can do this, whether or not our pains are of the very same type must be irrelevant to the way senses are acquired by the terms used in our phenomenological descriptions of pain.
The resemblance your experience of seeing the colour red has to mine can, therefore, only consist in relations our experiences bear (or are disposed to bear) to other qualia and to properties of things and events in the public realm (e.g. to red and non-red things). The fact that our discriminations coincide is crucial (this is Wittgenstein's PI §242 «agreement in judgements»), because the agreement that something is red rests on this. This is a simple matter in the case of perceiving a colour. (It's much more complicated for other areas of inner mental life, which I won't discuss now.) There is a property of my immediate phenomenal episode with which, in so far as it is mine and not yours, you cannot be acquainted. What matters to us both are
If I am colour blind and don't have the relevant qualia then I don't have qualia that enter into the relations to one another and external things that are relevant to the correct use of `green' and `red', and my judgements do not coincide with yours.
This is not behaviourism, because our immediate phenomenal episodes have to exist for the key relations to have their relata. It is because those episodes are inscrutable to all but their owners that the question cannot arise for you of what it is for me to have them, however closely our judgements coincide, and their private qualitative character can be no part of their public sense. So, without the qualia with the necessary relations to red and green things, our colour blind subject knows that there are colours others perceive, but does not know how to use the expressions `red' and `green' of red and green things. This is not because the qualia are the senses of these expressions, but because it is on the relations those qualia have to external things that the agreement in judgements rests that supports the use of (provides the senses for) `red' and `green'. The colour blind subject lacks the qualia to which `red' and `green' refer. «So he doesn't know what we mean.» Well, yes, but bear in mind that `mean' is ambiguous as between `intend to convey' and `denote', and in this case it is the latter which is lacking an object for the colour blind subject. This account, notwithstanding its sketchiness, has the virtue that, unlike the Cartesian model, it is not impossible.
Moreover, it is only because our judgements coincide, and we agree, that we can overcome subjectivity and distinguish judgements from fancies, illusions dreams and hallucinations which are not necessarily inwardly phenomenologically distinct from veridical perception. Clarification of the conditions for the possibility of this agreement is the outcome of the Private Language Argument.
This agreement, however, falls far short of epistemic warrant, since speakers notoriously agree on superstitions of every kind. Collective use of a linguistic idiom rests can objectively rest on agreement independently of the objectivity of the subject matter that idiom is used to describe. `Language game relativism' arises from failure to mark this distinction. The possibility of agreement revealed by the PL Argument provides the justification for the general strategy involved in describing `language games', rather than the PL Argument requiring this strategy for its vindication. On this I part company with Ming Yang over what depends on what in Wittgenstein's skein of argument.
Now, given all that, one might ask, if there is already an argument against private meanings, why do we need an argument against a private language? The arguments in §2 & §3 above, however, show that the senses of expressions in a public language cannot be mental items with which people are exclusively acquainted by introspection. The PL Argument takes aim at the idea that such mental items might nonetheless be the meanings in play in a private language. To show that this is impossible, it is necessary to show that rules cannot be followed in the necessarily private sphere accessible exclusively to introspection.
§4. Subjectivity and The `is right / seems right' Distinction
There is more to this distinction than «the unobjectionable [idea] that meaning is normative» which Garrett's concedes. Indeed, it is independent of and necessarily precedes considerations of normativity. It turns on contrasting a realm only available to introspection with a sphere that embodies multiple perspectives, where the former is characteristically subjective. It is because it is necessarily solitary that introspection can only disclose what seems to be. Solitary animals and persons can check how things seem to one of their senses by bringing others into play -- e.g. by touching, smelling, tasting or listening to what they see, but, once they have done that, all they can know is how things seem to them.
To contrast how things seem with how they really are requires subjective and objective (or at least inter-subjective) standpoints to interact. So long as there is only necessarily solitary (i.e. private) knowledge, disclosed solely by introspection there can be no distinction between how things seem and how they are. For you, if the taste of liquorice seems nice, it is nice -- a paradigm of subjectivity since Hume (if not Hobbes).
It has sometimes been charged that Wittgenstein relies on simple (trivial or avoidable) memory scepticism in order to press for this conclusion. In fact he doesn't cast any doubt at all on memory per se. The issue, rather, is that recognition and memory are not two separate faculties so that one could serve as a check on the other. How do I recognise some sensation S on 2nd and subsequent occasions when I have it? Only by remembering previous occurrences of S. So, were I to go back to `consult my memory' to check I have remembered S correctly, I would not find a second inner authority distinct from the faculty I had already employed to recognise S in the first place. Hence the remark (PI § 265) that proposing this is «As if someone were to buy several copies of the morning paper to assure himself that what it said was true.» The point is that there can be no `is right / seems right' distinction within what is disclosed by introspection alone.
Descartes notoriously bootstrapped his way out of this problem. He sought to institute an `is right / seems right' distinction by invoking the necessity of an alternative, infallible Divine standpoint as a source of `clear and distinct ideas' to contrast with the possibility that his own might have been wholly deceived by the demon. Wittgenstein is not concerned with being right tout court, but only with the possibility of following a rule, which requires an `is right / seems right' distinction to be applicable to putative instances of rule-following.
Nonetheless, there is nothing new about Wittgenstein's implicit assumption that, for knowledge necessarily restricted to introspection alone, there can be no `is right / seems right' distinction. What introspection discloses is subjective. Latter-day CartesiansFoot note 8 have sometimes assimilated the impossibility of being mistaken about what is immediately disclosed by introspection to the guarantee supplied to clear and distinct ideas by their divine origin. Hence the notion that sense data supplied incorrigible knowledge. However, the impossibility of being mistaken about sensations, or whatever else is immediately disclosed by introspection, far from being a form of infallibility, is a defect consequent upon the absence from that sphere of any `is right / seems right' distinction.
You can't be wrong about what is immediately before your mind, but no more can you be right, because the distinction has no application to the necessarily private sphere, any more than it has to clouds, birdsong or the stars. Where it is impossible to be wrong it does not follow that one must be right, it follows, rather, that in such a context no distinction between correct and incorrect performances can apply. So «what seems right is going to be right» and the necessary conditions are not met for following any normative rule which distinguishes correct from incorrect judgements or performances.
§5. Rule Following and Normativity
What are those necessary conditions and why are they not met? The absence of an `is right / seems right' distinction debars rule-following because the possibility that a rule be followed requires that a mistake be equally possible. This is another longstanding position on which the PL rests. From Kant's argument, now known by the motto `ought' implies `can', a more general feature of normativity can be derived:
`ought' implies `can', and so too does `ought not'.
If we think of any prescriptive sentence expressing a normative rule (including those of language) which distinguishes correct from incorrect performances, we find an implicit notion of its `normative content' (by analogy with the notion of the `cognitive content' of declarative sentences). For there to be a rule expressed by prescriptive sentence (for it to have any normative content), both what it enjoins and what it forbids must equally be possible. A sentence can have no normative content if it states a putative rule that enjoins what no one can do, or, equally, that forbids what cannot be done in any case. It would be entirely empty and irrelevant to any possible performance: i.e. equally irrelevant both to how the performance might be produced and to how it might be judged.Foot note 9
§6. The Private Language Argument Reconstructed
Now we have all the elements on which the PL Argument rests:
None of these claims, as I see them, is a novelty introduced in the Philosophical Investigations. What is novel, in relation to a private language, is to combine them so that if:
1. the privacy premise applies to our inner mental life, and
2. the subjectivity premise applies to introspective knowledge, then
3. the necessary condition for rule-following (and, hence, a language) set out in the normativity premise cannot be met within the private sphere of our inner mental life.
The key challenge that Garrett poses for this argument concerns contingently private languages, for it is not possible successfully to argue that all languages are necessarily public. A single person might invent a language like Esperanto, or the private code in which Samuel Pepys wrote his diary. Garrett also states that
These two sorts of contingent privacy are very different, and I doubt that a `socially-isolated-from-birth Crusoe' could devise a language all by himself. I won't press this point, however, so let's suppose he could. Garrett's key point is that the limitations of actual privacy upon language should be the same whether or not that privacy is necessary. He takes the key steps of the PL Argument to be:
and argues that (2) must be «indifferent to whether the private language is contingently private or logically private.»
I hope that my reconstruction of the PL Argument shows that it is not privacy per se, but the restriction to knowledge of one's inner life by introspection which is responsible for the absence of the `is right / seems right' distinction. So it is merely within the sphere of inner mental life known only by introspection that a private language is impossible. The nature of actual privacy is, therefore, not indifferent to whether it obtains necessarily or only merely contingently. It is rather like ownership: you can transfer the ownership of your house to me, but, while you live, neither your body nor your mind can similarly become mine if deeds of ownership are exchanged.
§7. Is the Characterisation of Normativity Inescapably Modal?
This, however, is not enough to dispose of the root of Garrett's objection, namely that we cannot require a modal criterion be met in order to establish an actual fact. He writes:
For my reconstruction of the PL Argument relies no less on a modal criterion -- the normativity premise above. There are several possible responses to this. I think that, despite its apparent reasonableness, we will have to abandon the presumptions in the quote above and, with them, the more general presumption that we can never require a modal criterion to be met in order to establish an actual fact.
The other possible responses are.
1. Sever the link I assumed when I wrote of `the necessary condition for rule-following (and, hence, a language)'. Perhaps a language can be spoken without any issue ever arising regarding correct or incorrect use of its expressions. (But Garrett himself concedes the normativity of language.)
2. Devise a non-modal characterisation of normativity so as to do away with anything resembling the notion of the `normative content' of a rule I sketched above. (This will presumably show that `ought' need not imply `can'.)
One or other, or both of these two possible philosophical challenges may well be met and overcome. I am not going to hold my breath. Instead I will briefly consider the implications of an inescapably modal criterion for normativity, and an inescapably normative characterisation of language.
Let's return to Garrett's claim above. If Crusoe's language can be contingently private, as Garrett claims, that can only be because it is possible to translate it into ours. I.e. if his language is only contingently private there is a possible world in which I, too, can speak `Crusoese'. Since I speak English, I must, in that possible world, be able to translate his language into English. There would be no grounds to believe that I spoke Crusoese if I could not and, indeed, no grounds to believe Crusoese was a bona fide language if its translation into English was not possible. It wouldn't be a necessarily private language if it couldn't possibly be translated, it wouldn't be a language, period. These considerations are quite independent of the PL Argument.
It is false, therefore, to claim that «His speaking a language cannot rest .... partly, on such modal considerations.» Consequently «intrinsic facts about him, his world, and his intentions» cannot suffice for it to be true of Crusoe that he speaks a language. Suppose, furthermore, that Crusoe has taught the parrots on his island to parrot words in his contingently private language. The difference between Crusoe and the parrots would be that, unlike the parrots, «we might understand and correct him».
In fact, there is also the possibility that he might understand us and correct himself. There is something profoundly wrong with the customaryFoot note 10 conventionalist `community' account of the source of correction in rule following -- i.e. that following a rule is only susceptible to external correction. Self-correction (reflexive imputation of a rule) must be no less important to rule-following than correction by others. Two people in dispute over how to follow a normative rule correctly would otherwise be compelled each to say to the other: «You can, and I cannot tell whether I am following this rule correctly, but equally, I can, but you cannot tell whether you are correctly following the rule you seek to follow in correcting me,» which I hope is plainly absurd. The issue here is not the epistemic question `How is a correct judgement made?' (This depends on the sort of rule and practice involved). It is, rather, what the necessary conditions are for the possibility of there being a distinction between correct and incorrect judgement.
The answer must be a form of agreement which two or more people can have reciprocally with one another. (The reciprocity is all that can debar the absurd scenario of the previous paragraph.) The sort of agreement at issue here is one which may be granted, refused, broken, honoured, entered into and withdrawn from. It is a normative agreement.Foot note 11 It is not a convention. A convention assigns two or more items (of a general kind) to some arbitrary relation, which could as easily be otherwise (drive on the left, use an expression or a sign in a certain way, etc.). In order for it to be possible to institute a convention there needs to be a prior, non-arbitrary agreement as to which items are to participate in the arbitrary, conventional relation.
Most importantly, however, this sort of agreement only has a point when it obtains between two or more individuals, since it is perfectly empty to suppose that anything can turn on your «agreeing» in this sense with yourself. This is why it cannot consist solely in some sort of relation of similarity or concomitance. The reason why not is to be found in PI §215 where Wittgenstein writes, of following a rule in the same way on more than one occasion:
In short, the relation between two instances, in virtue of which they amount to agreement in following some rule, cannot wholly consist in the same relation which one thing must necessarily have to itself -- namely, having properties in common. Following a rule must require that properties be shared by a plurality of instances, but enumerating those properties presupposes, and does not explain, what makes them the right properties to have picked out in order to identify successive instances of following that rule.
Intrinsic similarity, then, is a necessary, but cannot be a sufficient condition for agreement upon a rule. Thus, if speaking a language does depend on agreement upon some rules, then the intrinsic character of the speaker and his/her world and intentions cannot comprise a set of sufficient conditions for the existence of that agreement, and, therefore, the fact that a language is being spoken.
§8. Conclusion
I hope I have shown that the argument that there can be no private language rests on three unsurprising, far from novel premises concerning privacy, subjectivity and normativity, respectively. These premises stand:
I think this successfully untangles the PL Argument from the other issues Wittgenstein discusses in connection with privacy, rule-following, etc.: the considerations i.-iv. above are prior to the PL Argument; considerations regarding agreement (in judgements, language games and in forms of life) are consequent upon the PL Argument; and considerations regarding other minds scepticism and the indeterminacy of rule-following are independent of the PL Argument. So I hope I have indicated why Brian Garrett's conclusions do not follow, and, why, while I generally agree with Michael Ming Yang's substantive conclusions, I cannot agree with the route he takes to establish these.
Middlesex University
<r.harris [at] mdx [dot] ac [dot] uk>
[Foot Note 1]
`Cartesianism and the private language argument' Sorites 14 Pp. 57-62, 2002.
[Foot Note 2]
`Privacy, Individuation, and Recognition' Sorites 13 Pp. 90-98, 2001.
[Foot Note 3]
I would say that they were uncontroversial, were it not for what you might call the `Monty Python proviso' after the Python's brief sketch which went something like:
«`This is Mr. Smith, who contradicts everything anyone ever says.'
-- `Oh no I don't!'»
It is in the nature of philosophy that every philosophical thesis has been denied by someone, somewhere at some time.
[Foot Note 4]
I am not going to cite textual support for every remark I make about Wittgenstein's argument, but, where I do, I will refer to his Philosophical Investigations by the initials `PI' and the paragraph number preceded by `§'.
[Foot Note 5]
Lewis Carroll Alice Through the Looking Glass, http://sunsite.tus.ac.jp/coll/alice/alice_toc.html.
[Foot Note 6]
So, on this, I disagree with Michael Ming Yang. I think the impossibility of following a rule in a sphere known only by introspection does not result from the absence of identity criteria for e.g. pain episodes. In this context Ming Yang writes «What counts a kind, a type, an identity criterion, is hinged upon our way of life.» This perhaps confuses type identity with token identity (since it is to the latter that the slogan «no entity without identity» applies, which Ming Yang invoked). In any case, doctors can identify my pain tokens by extraneous criteria (like start and finish times), and my pain types from their diagnostic manuals.
[Foot Note 7]
Locke, J., 1689/1975, Essay Concerning Human Understanding, II, xxxii, 15, http://humanum.arts.cuhk.edu.hk/Philosophy/Locke/echu/
[Foot Note 8]
Adherents of a Cartesian theory of mind, that is, rather than of his rationalist epistemology.
[Foot Note 9]
What some call the `rule-following considerations' are often lumped in with the PL Argument, because Wittgenstein mentioned them in the same tract of the Philosophical Investigations. The impossibility of rule-following within the necessarily private sphere is quite distinct from the question whether public rule following can ever successfully be captured by determinate formulations of rules. Wittgenstein casts doubt on this because
(a) any explicit formulation always requires a further rule for its application, to create a regress of rules for following rules which must finish with the last rule followed «blindly» PI §§217 -- 219.
(b) there cannot help but be innumerably many such rules, and no end to the uncertainty as to which precisely is the unique rule being followed in the unfolding performances:
I will say no more about this, other than to reiterate that the impossibility of following a rule within a necessarily private sphere is a completely separate matter from the conditions surrounding following public rules.
[Foot Note 10]
E.g. in Kripke S. Wittgenstein on Rules and Private Language Oxford: Blackwell 1982 amongst many others.
[Foot Note 11]
A `contract', if you like, in the philosophical sense we derive from contractarianism.