next up previous
Up: Holistic Reductionism. The Case Previous: Simple Things, General Facts

The Soft Machine

You are, by now, probably wondering about the relevance of Wittgensteinian hermeneutics to the problem at issue between Quine and Carnap. Some additional hermeneutics might help. The intricacies of Wittgenstein's constructions do have a fairly accessible front end, namely an intuition that touches directly upon the controversy between linguistic frameworks in a scientific mould and more pragmatic shiftings of language, exhibiting various degrees of precision. Here are two quotes expressing this intuition.

Sind also im Grunde die Sätze der Physik und die Sätze des gewöhnlichen Lebens gleich scharf, und besteht der Unterschied nur in der konsequenteren Anwendung der Zeichen in der Sprache der Wissenschaft? $\ldots$ Es scheint klar, daß das was wir MEINEN, immer <<scharf>> sein muß. ( Notebooks 20.6.1915)

Ich will nur die Vagheit der gewöhnlichen Sätze rechtfertigen, denn sie läßt sich rechtfertigen. ( Notebooks 22.6.1915)

These ideas are not compatible with standard reductionism. They do, in fact, anticipate the later Wittgenstein's anti-Tractarian self-criticism. But his declarations of intent do, on the other hand, not support floating boundaries within a continuously changing web of belief. To put it in terms of Wittgenstein's later philosophy his aim is to preserve ordinary language and, at the same time, reveal its grammar. In order to justify vagueness some theory is called for; this is what ``sentence variables'' are ultimately about.

Here is another statement to the same effect:

Wenn ein Satz uns etwas sagt, so muß er, wie er da steht, ein Bild der Wirklichkeit sein und zwar ein vollständiges. - Es wird natürlich auch etwas geben, was er nicht sagt - aber was er sagt, sagt er vollständig, und es muß sich SCHARF begrenzen lassen. ( Notebooks 16.6.1915)

Sentences cannot be regarded in isolation. Each sentence's cognitive content has to be seen as the role this sentence plays within the overall pattern of functional dependencies imposed upon language by its application to the world. Wittgenstein's Tractatus is - among other things - a holistic manifesto: There is no way to determine single pieces of its descriptive apparatus without committing oneself to the entire system underlying possible description. Every sentence, to put it somewhat melodramatically, has to bear the weight of language's obligations towards the world. Yet, Wittgenstein is refusing the Quinean move to blur the distinctions between a given level of analysis and its embedding in ongoing processes of inquiry. There is no single standard of precision to which reductionism can appeal. But we should not renounce our analytical demands either. Wittgenstein's motto still holds: ``Was sich überhaupt sagen läßt, läßt sich klar sagen, $\ldots$ ( Tractatus, Preface) It is just that there are many ways to follow this advice. Rejection of straightforward reductionism does not force us to embrace pragmatism. Hidden beneath the surface of logical atomism Wittgenstein's Tractatus offers a conceptual machinery sophisticated enough to mediate between ordinary language and the scientific enterprise, accommodating both the inevitable vagueness of our expressions and their indispensable rigour in well-defined contexts.

This is as far as hermeneutics will lead. We have discovered a Wittgensteinian device to establish an inter-dependency between analysis and holism. Is this more than tending to a museum piece? The metaphysical aura surrounding the Tractatus as well as its mostly non-standard treatment of logical issues prohibit any easy application of Wittgenstein's design to current theory of science. His intuitions might seem attractive, but - torn from their Tractarian environment - they lack technical support. I will conclude this paper by indicating how Wittgenstein's intuition can be put to work in a contemporary setting. Wittgenstein did not care much about the de facto content of the world. Very well, let us disregard obligations to implement one or the other kind of reductionism and focus on the constraints Wittgenstein's procedure imposes upon language. I have, at the beginning, mentioned Stephen Yablo's objections against Quine's ontology. According to Yablo Quine's scientism presupposes a distinction between literal and metaphorical speech. On pain of loosing any chance to control what counts as proper science and what is to be regarded as make-believe we have to be able to sort out genuine reference from merely rhetorical linguistic moves. Quine's rejoinder would, presumably, insist on fuzzy boundaries between metaphysics and science. Physical objects are, as he famously put it in ``Two Dogmas of Empiricism'', on a par with Homer's gods in point of epistemological footing.10 This is precisely the kind of methodological relativism Wittgenstein can teach us to avoid. He does not - even in the Tractatus - put forward one single, allegedly incontestable standard. His contribution consists in offering a synthetic picture, combining rigour and flexibility. Let me briefly indicate how this relates to metaphors.

Back to Wittgenstein's Notebooks. There is talk about chairs and watches as well as about viruses or islands. A speaker's mastery of the language consists in her being able to deal with those expressions in such a way as to preserve linguistic agreement. One cannot, in ordinary circumstances, use the expression ``chair'' to refer to the square root of 10. As Wittgenstein might put it: there must be something we definitely mean in employing such terms - if we mean anything at all. Intuitive handwaving will not do. This claim, of course, provokes a further question. How can we come to know those presumably essential features? Wittgenstein's Notebooks invite us to reject this question. The important point is for us to be able to distinctively use a term. It suffices for this term to meet the requirements of precision any given situation imposes upon it. How should one, then, understand expressions like ``computer virus'' or ``Susan is an island''?

Regarding ``computer'' or ``virus'' as sentence variables we do, in fact, assume that these terms play some characteristic role in some particular set of sentences. Successful employment of those sentences in conversation is their source of meaning. Similar considerations hold for composite expressions like ``dangerous virus'' or ,,retro virus''. Now, if someone first comes up with a term like ``computer virus'' the underlying pattern of sentence variables is disturbed. There are many appropriate occasions to use ``computer'' and, on the other hand, just as many legitimate expressions including ``virus''. But (as long as a ``living metaphor'' is concerned) we are at least initially puzzled by sentences containing ``computer virus'' in their subject position. Two sets of sentences seem to be imposed upon each other; the realms of digital equipment and biology are simultaneously invoked - but we lack a well-established sentence variable. This metaphor works by suggesting that we employ a biological pattern to technological processes. Understanding metaphors, in this view, amounts to constructing a new kind of sentence variable built upon sets of sentences superimposed onto each other. If I may use a comparison taken from popular culture the procedure is similar to the common practice of using successful athletes to advertise energy drinks or leisure equipment. A pre-established, meaning-determinative scheme is, as it were, recycled to serve a different purpose.

These sketchy remark do not purport to contribute new insights into the theory of metaphor. Their aim is to indicate the philosophical potential of Wittgenstein's idea. Insofar as we do understand what ``computer virus'' refers to, we operate on the basis of some sentence variable. It functions as a dummy for a certain subset of sentences concerning computers, namely those sentences that can successfully be picked out by employing the biological prototype. This is, I readily admit, basically just common sense. It can, however, be applied to finally deal with the issues raised at the beginning of this talk. The problem was how to decide between the Carnapian attitude of working within well-defined contexts and Quine's more radical approach of permanent, universal revisability. Can dogs be turned into cats? Will machines ever be able to think? One typical reaction is to claim such feats to be impossible. Even imagining them seems to violate the meanings of the terms involved. The opposite position is, of course, to allow such events to happen after all, i.e. to concede that there might well be circumstances forcing us to drastically change our present meanings. My explorations have been heading towards a Wittgensteinian view regarding this stand-off.

Traditional logical empiricism wanted to base itself on hard facts. Science has to be distinguished from metaphysics and one disguise of metaphysics is science fiction. Epistemology has, consequently, to charter the terrain of legitimate research and to reject attempts to blur the line between scientific knowledge and private imagination. The pragmatists' reply subverts this kind of dualism. Hardness and softness are thoroughly mixed up. There is no a priori stance from which to rule out science fiction. This is Neurath's and Quine's position which is widely thought to have eclipsed Carnapian logical analysis of given scientific frameworks. Wittgenstein's method, put forward in his Notebooks, cuts across this antagonism.

Meine Methode ist es nicht, das Harte vom Weichen zu scheiden, sondern die Härte des Weichen zu sehen. ( Notebooks 1.5.1915)

We definitely know how to tell a cat from a dog - at one level. What about different uses, in different contexts? Well, it depends. We cannot expect that one level's use will simply project to a deeper (or more superficial) stratum. One established set of literal expressions might turn into a metaphor in different circumstances. Definite meaning is embedded within linguistic environments of changing granularity.

Still, this is not the end of the story. To discover the hardness of softness should not only be taken as a plea for some intermediate robustness of meaning. There is more to the difference separating cats from dogs. Someone seriously defending a distinction between natural kinds will not rest with the previous concessions. Such intuitions are often perceived to cover all reasonable levels of language use. ``Eliminate this difference and you are not left with any meaning to begin with! If your cats can turn into dogs I have no good reason to even enter into this conversation.'' There is no understanding without limits and this does not imply that actual understanding is incomplete or relative to a more comprehensive scheme. At certain occasions our capacity to make sense of given utterances hit rock bottom, the hardness of softness which should not be cushioned by soft talk.

The idea sounds somewhat dogmatic but it is just a corollary following from Wittgenstein's views in his Notebooks. Analysis is distinct from demolishion because it somehow saves its starting point. To pick an item apart is a different game. Traditionalists, it is true, frequently complain about destructive consequences of analysis, which forces analysts to defend their own work of decomposition. Yet, philosophical analysis has missed its purpose if it does not care for its analysandum. Is there a better way to care for concepts than to explore their limits, knowing when to stop?


next up previous
Up: Holistic Reductionism. The Case Previous: Simple Things, General Facts
h.h.
2000-05-05