From the WAB archives: A selection from the Bergen Wittgenstein archives' Working Papers and audio-visual materials

Philosophical Investigations 244:"Physiognomic language-games["]? ¹

In chapter 10, "Wittgenstein on Private Experience" [pp. 241ff.], of their book Investigating Wittgenstein Merrill and Jaakko Hintikka try to elaborate a view according to which the late Wittgenstein adopts a Cartesian point of view with respect to metaphysics. The remarks on "physiognomic language-games", which are also in some respect a topic of chapter 11, "Differences and Interrelations among Language-games in Wittgenstein" [pp. 272f.], play an important constructive part in the argumentation in favour of this thesis. In this paper I shall discuss these remarks on "physiognomic language-games" and attempt both an internal (I) and an external critique (II).

I

In order to demonstrate the fallacies of the "received interpretation", the object of the Hintikkas' critique, it is useful to remind ourselves of section 293. It goes as follows:

"293. If I say of myself that it is only from my own case that I know what the word "pain" means -- must I not say the same of other people too? And how can I generalize the one case so irresponsibly?
     Now someone tells me that he knows what pain is only from his own case!-----Suppose everyone had a box with something in it: we call it a "beetle". No one can look into anyone else's box, and everyone says he knows what a beetle is only by looking at his beetle.---Here it would be quite possible for everyone to have something different in his box. One might even imagine such a thing constantly changing.---But suppose the word "beetle" had a use in the people's language?---If so it would not be used as the name of a thing. The thing in the box has no place in the language-game at all; not even as a something: for the box might even be empty.---No, one can 'divide through' by the thing in the box; it cancels out, whatever it is.
     That is to say: if we construe the grammar of the expression of sensation on the model of 'object and designation' the object drops out of consideration as irrelevant."

The Hintikkas quote next the two last passages (b, c) of this section and write "that the private object (the beetle, the sensation) drops out of semantical consideration according to Wittgenstein only when the semantics of sensation-talk is construed on an incorrect model. ... It is the privacy of these semantical relations, not the privacy of what is represented by their means, that Wittgenstein is critizing. (Sensations are private; sensation-language cannot be.)" [p. 250.]²

With the "model of 'object and Bezeichnung'" the Hintikkas mean "the model of reference unmediated by any language-game." [p. 254] Now, the physiognomic language-games are - by definition - language-games and not an aspect of the "model of 'object and Bezeichnung'". Therefore, the object belongs as relevant to the discussion. The beetle-game's need for some "foundation" other than that of the "model of 'object and Bezeichnung'" is at the same time a demand for the self-sufficiency of this "foundation", that is, a demand for the independence of the situation described in section 293. [Cf. also pp. 273ff and 278f..] If one could find such a "foundation", it seems, one could also compare private beetles in a public way.

The most important textual evidence upon which the Hintikkas base their argument for a constructive aspect, is section 244. Wittgenstein writes:

"244. How do words refer to sensations?---there doesn't seem to be any problem here; don't we talk about sensations every day, and give them names? But how is the connexion between the name and the thing named set up? This question is the same as: how does a human being learn the meaning of the names of sensations?--of the word "pain" for example. Here is one possibility: words are connected with the primitive, the natural, expression of the sensation and used in their place. A child has hurt himself and he cries; and then the adults talk to him and teach him exclamations and, later, sentences. They teach the child new pain-behavior.
     "So you are saying that the word 'pain' really means crying?"---On the contrary: the verbal expression of pain replaces crying and does not describe it."

Here the Hintikkas only quote the first passage (a), where they find "Wittgenstein's first and foremost example of 'how words refer to (more accurately: 'are related to', in German beziehen sich auf) sensations'" [p. 257]. It is decisive for the Hintikkas that here we deal with a conceptual or "grammatical" relation and not with a contingent one between the sensation and the spontaneous pain-behavior, such as the sensation's "natural physiognomic correlate" (p. 258). "Such a framework of spontaneous expressive behavior (including facial expressions, gestures, and other bodily movements), will be called physiognomic framework and a psychological language (or language fragment) based on it will be called a physiognomic language." [Ibid.]

Now we can look at the characterization of pain the Hintikkas find in section 244a. (I take it as a definition and call it "DEF-H".)

DEF-H

"..., a sensation is an instance of pain, if it is of the kind that typically goes together with what is normally taken to be, and is responded to, as pain-behavior." [p. 257)

To say that x typically goes together with y, which is normally taken to be, and is responded to, as z, presupposes - as the words "typically" and "normally" typically and normally are used - that there can be atypical and abnormal cases. If we exclude that x cannot go together with y and/or with non-y, and/or if we exclude that y cannot be taken to be, ..., z and/or non-z, then we would typically and normally have no reason to use the words "typically" and "normally". But how does this fit with the so-called logical, conceptual, or "grammatical" relation between x and y/z that the Hintikkas are speaking about? Would it not be more prudent and advisable to speak only about something like a statistical relation?

Since we want to be a bit more exact here, let us transfer the parameters explicit in section 244 onto DEF-H (and call the result, as a variation of DEF-H, "VAR-1").

VAR-1:

"..., a child's sensation is an instance of pain, if it is of the kind that typically goes together with what adults normally take to be, and what adults normally respond to, as a child's pain-behavior."

But here it seems as if that which is taken by the adults to be a child's pain-behavior might not really be pain-behavior. Of course, we know such things from adults: for example the pretence of pain. But it could also be that real pain occurs whithout there being any pain-behavior. Of course, we are also familiar with this: the concealment of pain. Finally it could happen that there are both pain and pain-behavior, yet the adults do not take them as such. And neither is this strange to us: sometimes we err in identifying the behavior of others. However, if only one of these things is possible we cannot say that the physiognomic language-games connect our psychological vocabulary with our private sensations, but only that they typically and normally do so. With respect to certainty there is no longer any conceptual difference (only, perhaps, a statistical one) between the physiognomic language-games and those language-games for which the physiognomic ones should function as a foundation. One question arises here: What, then is the criterion according to which we can distinguish between the normal and typical cases on the one hand and the abnormal and atypical cases on the other? It is clear that the physiognomic language-games cannot function as such a criterion if we take DEF-H or VAR-1 as in any relevant sense defining or characterizing them. Shouldn't we, then, better take

VAR-2:

".., a child's sensation is an instance of pain, if it is of the kind that goes together with what adults take to be, and what adults respond to, as a child's pain-behavior."?

Here it seems that none of the above doubts can arise. Indeed, this characterization fits much better with the Hintikkas' thesis about "the incorrigibility of primary language-games" [p. 279f.]. Can we then generalize VAR-2 into

VAR-3:

".., a sensation is an instance of pain, if it is of the kind that goes together with what is taken to be, and responded to, as pain-behavior."?

But also this formula cannot be valid, because it seems now that all pain-behavior - including that of the adults, who we know to be capable of pretending pain - forces us to accept a pain-sensation. Whatever nice results VAR-3 may possibly bring us, the costs incurred should not be so high as to require that we forget nearly all we know and nearly all we are certain about, when it was just these things which forced us to look for a "foundation" in the first place. Why do the remarks in section 293a,b seem to be so powerful? Simply because we do encounter things like the adults' simulation and concealment of pain-sensations, and because we know that adults can err while identifying the mental processes and events of other adults. We cannot check the adult's utterance "I am in pain!" as we can check his utterance "The cat is on the mat!". There is at least an epistemic difference between an adult's toothache and cats on mats. Of course, not everything that looks and sounds like a cat on a mat is really such. But there are at least some well-known and generally accepted procedures to find out whether something that looks and sounds like a cat on a mat is what it seems to be. Both cats and mats are visible things which we can take in our hands. Are pains visible entities? Of course not. We can have pain in our hands, but cannot take pain in our hands. And this is why it is so easy to simulate or to hide pain, if we want or are forced to do so. As candidates for simulation and concealment cats on mats are not very suitable. Much better are beetles in boxes. Both however, are more suited for these purposes than a child's pain!

The absence of the parameters explicit in section 244 in the Hintikkas' characterization of pain allows us to bring two different cases under one definition; but we have to pay the price of losing the differences which allowed us to take the one case as the "foundation" of the other.

If we cannot make the step from VAR-2 to VAR-3, does this mean that we have indeed two concepts of pain, one for children and another for adults? Don't we also talk about the child's milk teeth and the adult's (mature) teeth? Also the Hintikkas say that Wittgenstein on the one hand favors physiognomic language-games but, on the other, does not exclude other public frameworks for our talk about private sensations. In particular he does not exclude the possibility of physiological correlates of different sensations as a public semantical system of coordinates [cf. pp. 273ff.]. If different sensations can have different physiological correlates - and for them to be useful as a public system of coordinates they must be different, otherwise different sensations could not be publicly identified as such - shouldn't we then also say that physiologically different teeth are correlated with different sensations, a milk tooth with a milk tooth-pain and a (mature) tooth with a (mature) tooth-pain? Suppose that such a state of affairs were possible. Which concept of pain would the adults teach the child, or at least try teach it? Isn't it just a happy chance that the child learns from the adults' teaching precisely the concept which is most fitting for both its private sensation and the physiological correlate thereof. And isn't this just such a happy chance as the one whereby we find openings in the cat's skin precisely where the cat has eyes and ears?

No, I believe the physiological correlates cannot help us out of the difficulties which the physiognomic correlates led us into. They only create new ones. It is not only that the whole story sounds strange - the question arises whether we have any good pictures available in the cases of pain in the head, in the foot, in the neck, etc. Of course, one could say now that no such physiological correlates were meant but instead something in the brain. That may be so, but one thing is clear: this is not very public, at least not in the situation described in section 244. And finally: before we can use physiological correlates as a public system of coordinates we have to clarify whether sensation x belongs to the physiological state s (and/or process p) or not. This has to be done in a public way, otherwise we would have what the Hintikkas do not want: a private relation between the public sensation-vocabulary and the private sensation.

Let us then put the physiological correlates aside and try to find the place where our problem arose. This place has to be found between section 244, section 293, and the Hintikkas' characterization of pain.

II

Remember what made the situation described in section 244 so interesting for the Hintikkas that they gave it an own name. It was the search for something that could play the role which the "model of 'object and Bezeichnung'" played in their eyes in section 293. The problem with the "model of 'object and Bezeichnung'" was its character as a "model of reference unmediated by any language-game." Was it then Wittgenstein's mistake in section 293 to speak about the "model of 'object and Bezeichnung'" instead of the "model of the private (making of the) relation between 'object and Bezeichnung'"? Of course, for the Hintikkas pain is still a (private) entity and "pain" a (public) Bezeichnung, a (public) name for these "private innards" (v. Savigny). Section 244 gives us a public language-game, which can, as it seems to the Hintikkas, function as a foundation for preventing the objects from dropping out of consideration as they do in section 293. In the words of the beetle-game the Hintikkas can say that we can talk about, describe, and compare the (private) beetles in our boxes because we have learnt in another public game how to describe, how to talk about, how to compare public beetles. Also we can now behave as if there weren't really a beetle in the box and can, if there really were none in it, behave as if there were. And we can behave as if we behave as if there were not really a beetle in the box, etc..

Suppose now that Wittgenstein's expression in section 293c is correct, that is, he means what he says when he speaks about the "model of 'object and Bezeichnung'". The first consequence is that "the object drops out of consideration as irrelevant." Whatever "pain" means, it does not stand for a private object. That means that pain-behavior can no longer be a physiognomic correlate of a private object. Does that mean that there is only pain-behavior left? Do the differences we were certain about - the differences between honest and dishonest pain-behavior, between spontaneous and reflected pain-behavior, between the child's and the adult's pain-behavior - disappear, drop out of consideration?

We do not doubt whether a child is in pain when it has hurt himself and is crying. But do we doubt in each case in which an adult has hurt himself and is crying? If reflection is possible and/or necessary in such situations, then this is normally and typically³ no reflection of whether the adult is really in pain, but it is reflection on our possibilities to help or to express our sympathy in an effective way.

When do we doubt? We doubt, for example, when the adult is crying like someone who has hurt himself, without having visibly hurt himself. We doubt only in special situations, not in general. In these special situations we interpret the behavior that closely resembles pain-behavior; and in this interpretation we sometimes err. In other situations we do not doubt, we do not interpret, we simply react, help, or express our sympathy. Sometimes we don't even know what it would be to doubt whether the other is in pain or not. Whoever doubts in these situations must have other concepts of pain than the ones we have, or must be in confusion.

The child's spontaneous pain-behavior and our reaction to it play indeed a special role in the variety of our uses of our pain-vocabulary. It is just one of the situations in which children are learning these uses. It is not the only situation to have this character. There are others in which children learn what it means to simulate or hide pain. There are lots of such situations, changing from time to time and place to place. But it is in no, or only a few, cases primarily the description of pain which the children are learning here. Much more important is to learn how to behave in the case of one's own pain and, in the case of another's pain, how to help, how to express sympathy, etc. That is the way people are. That is the point of section 244. To repeat the quotation of that passage from section 244 which is not quoted by the Hintikkas:

"'So you are saying that the word 'pain' really means crying?'---On the contrary: the verbal expression of pain replaces crying and does not describe it." (My emphasis - R.R.)

As far as knowing that another is in pain and as far as descriptions play a role, we often know whether the other is in pain or not and what pain it is. As far as knowledge plays a role at all in connection with being in pain it is the pain of the other we (can) have knowledge about. With respect to my being in pain it is false and senseless to say "it is only from my own case that I know what the word 'pain' means" [section 293]. In this sense - the sense in question here - pain is not private.

The Hintikkas are right, I believe, in criticizing a tendency in the literature according to which there is a wide gap between the Wittgenstein of the TLP and the Wittgenstein of the PI. I also agree with the Hintikkas when they write that what they call vertical relations between language and world do not drop out of Wittgenstein's interest. But in the Hintikkas' specific interpretation of the Tractatus and it's relation to the later philosophy it seems to be necessary to find an analog for what allowed in the early philosophy the analysis of the proposition's real logical form. If we take the picture of the vertical relations between language and world seriously, we could say that, according to the Hintikkas, the pain, the private sensation, is at the bottom. Superimposed upon it comes the physiognomic language-game, while at the top are the secondary language-games. Following this picture, the physiognomic language-game has to play the role of the Tractatus' Elementarsatz.

But if we take pain to be not a private, that is an inner and for the others hidden, entity, the physiological language-game can no longer play the role of a mediating language-game. It was my aim to show that they cannot play this role even if we let the question about the nature, or the essence, of pain remain open. At least they cannot play this role in the way the Hintikkas handle the problem. But if we take pain to be not a private entity then there is also no longer any reason, nor any attractiveness in my view, to take the situation described in section 244 to be a (self-sufficient) language-game. Then the view is open to see that this situation is no language-game at all, but belongs to the one game with the word "pain".