The Steps from Doing to Saying
Rudolf Carnap wrote that
we must distinguish two kinds of questions of existence: first, questions of the existence of certain entities of the new kind within the framework; we call them internal questions; and second, questions concerning the existence or reality of the system of entities as a whole, called external questions. Internal questions and possible answers to them are formulated with the help of the new forms of expressions. The answers may be found either by purely logical methods or by empirical methods, depending upon whether the framework is a logical or a factual one. An external question is of a problematic character which is in need of closer examination. (Carnap, 1950)
The exact nature of this ‘problematic character’ is left undetermined by Carnap. He is clearly drawn partly to dismissing external questions as metaphysical, and therefore requiring no attention, but he is also drawn to seeing them as questions of attitude or policy: practical questions, for which the answer would be given in terms of the benefits of the framework in question. This has suggested to me a template for thinking about pragmatism in general, which I distil into the following suggestion. You will be a pragmatist about an area of discourse if you pose a Carnapian external question: how does it come about that we go in for this kind of discourse and thought? What function does it serve, and what therefore is the explanation of this bit of our language game? And then:
(1) you offer an explanation of what we are up to in going in for this discourse, and
(2) the explanation eschews any use of the referring expressions of the discourse; any appeal to anything that a Quinean would identify as the values of the bound variables if the discourse is regimented; or any semantic or ontological attempt to ‘interpret’ the discourse in a domain, to find referents for its terms, or truth makers for its sentences. Instead:
(3) the explanation proceeds by talking in different terms of what is done by so talking, or by offering a revelatory psychology or genealogy or anthropology or even a just-so story about how this mode of talking and thinking and practising came about. and the functions it serves.
I do not offer this as a prescriptive, defining description of pragmatism old or new. Some thinkers who like the label may reject the whole enterprise of answering a Carnapian external question, rather than giving an answer of a certain shape to it. Quine, who thought of himself as a pragmatist, disliked the whole external/internal distinction. But I find that it helps me to draw up a rough map of some of our more important philosophical alternatives. It also makes close contact with the way things are seen by prominent neo-pragmatists such as Michael Williams, Huw Priee and Bob Brandom. While Brandom calls his latest work Between Doing and Saying the motto for my Wittgensteinian approach to things might be ‘from doing to saying’: place the discourse in amongst life’s activities, and you will gain a perspicuous representation of what is said when you use it.
To get a sense of the contrast we might imagine the difference between a Wittgensteinian approach to the philosophy of mathematics, and a more standard set-theoretic approach. The Wittgensteinian tries to give a ‘perspicuous representation’ of our activities with mathematical discourse, perhaps in terms of familiar doings such as measuring out bricks or planning floor spaces or keeping track of money. He starts with applied mathematics in other words, and hopes to see fully fledged systems of arithmetic as functioning as abstractions serving to book-keep for these homely activities. The contrast would be with an approach which takes ‘reference to numbers’ as a datum, and then puzzles over what numbers might be, how we might know about them, and why we should want to. Carnap’s fears that such thought will prove too ‘metaphysical’ obiously hover around such an enterprise, and continue to hover even if we propose the reduction of all mathematical entitites to sets. Clearly we get the same shape of approach with expressivism in modal and normative or evaluative contexts, so called ‘subjective’ approaches to probability, expressive or even secondary quality approaches to causal thought, and many others
Put this way the natural opposite to a pragmatist approach might be described as referentialism or representationalism or in some equivalent terms, for instance, as providing a truth-conditional semantics or an ‘account’ of what makes true sayings in the area. But this way of putting the opposition needs careful handling, for deflationism in semantics introduces an awkward guest, and some would say a cuckoo in the nest.
Initially deflationism is a valuable ally of pragmatism. One of the salient features of any assertoric discourse is that we are free with the idioms of truth and of ‘talking about’ things, or in other words reference, and representation. If I tell you that there is a chair in the kitchen, what I say is true under definite conditions, and we naturally say that I referred to the kitchen, represented it as containing a chair, and in a more general vein I was talking about how things stand. But the same is true when I talk about the distribution of prime numbers or the value of gratitude or the impossibility of perpetual motion. Abstract, modal, and normative vocabulary bedeck themselves just as naturally as any other with a propositional appearance and the associated semantic trimmings. But to the deflationist, this is no surprise and signifies no sinister flirtation with metaphysics. The vocabulary of truth is doing what it always does, and the vocabulary of representation the same.
Following Horwich (1990), I take deflationism in the theory of truth to be a combination of three theses:
(A) That there is complete cognitive equivalence between Tp and p
(B) That conforming to that equivalence is all that is required to manifest complete understanding of the truth predicate
(C) That the utility of the predicate is purely logical: it is a device of indirect reference and generalization
The equivalence and the utility are the same whatever our subject matter. So if we are true to the folk, rather than seeking to debunk their sincere and intended sayings, or convict them of wholesale error in even deploying their favoured vocabularies, we will end up applying talk of truth and representation to whatever vehicles these sayings provide for the folk to travel in. For with truth comes a fully-fledged vocabulary of representation: when we speak truly we represent things as being thus-and-so, and the things we so represent are the things referred to or quantified over in our sayings.
In particular, notice that the word ‘description’ can go into the deflationist pot along with ‘representation’. We desribe how things stand with norms and values, possible worlds, or numbers and sets. We believe the results of our descriptions. Hence what Huw Price calls the ‘bifurcation theses’ between descriptive and non-descriptive uses of language, itself goes into the deflationist mix, and is apparently dissolved as effectively as truth, reference, and represention (Price 2008) . Other contrasts, such as that between belief and attitude, may go the same way. For there is nothing to prevent a theorist from allowing a promiscuous, catholic, universal notion of ‘belief’ —one that simply tags along with assertion, acceptance or commitment. But in that case, ask critics, what room is left in which to make pragmatism into something distinctive (Dreier 2002, Sinclair 2007)?
These points indeed show that it does not lie in where you end up. After deflationism, an expressivist, for instance, just as much as a Wittgensteinian in arithmetic or a Humean about causation, will be indistinguishable from anybody else in his everyday deployment of the relevant vocabulary. But we must not look only at the finishing line, after the race is run. Rather, whatever is distinctive comes in the route whereby you get to where you end up: the perspicuous representation that enables the pragmatist to put sufficient weight on the functions associated with the discourse to avoid putting any weight on the metaphysical imaginings that it might threaten to engender. It may be that any assertoric use of language associates itself with the same all-embracing semantic terms. Deflationism certifies as much. But the best functional story explaining how we ended up where we are may have a much finer grain. So, for instance, a perspicuous representation of how we have a descriptive- sounding evaluative laguage may itself eschew any truck with description, reference, facts or truth-makers but use as its only building blocks humdrum situations of choosing, preferring, recommending, or needing. Its promise is that with attention to these activities we come to see how our evaluative descriptions of things need no truck with the idea that we somehow respond to an autonomous realm of values: a metaphysical extra that we inexplicably care about on top of voicing and discussing our more humdrum concerns. It is here that a fine grained distinction between, say, describing and desiring will have its place, and similarly for the plurality of functions assiated with mathematics, modal assertions, normative assertions, and causal and dynamic assertions. All that the pluralisms of function would say is that we have to look below the surface, to see what belief amounts to in this area or that— Wittgenstein’s: ‘I’ll show you differences’. If you want to call the result ‘belief’ well and good – but it won’t necessarily be much like belief in other areas. And, I would say, if the content of these beliefs is sufficiently removed from explanation in the Carnapian external context, then these beliefs can equally properly, and metaphysically much more illuminatingly, be thought of in other terms.
For the rest of this address I want now to compare the story I have developed with pragmatism as it is presented in more ambitious writings, which see it as a global recipe for a particular approach to meaning and use. The idea derives from Wilfrid Sellars, and in the hands of Brandom and Price says that an account of use can be given across the board, and that itself explains, in general, the meanings we manage to give to our sayings (Brandom 2008, Price 2008). However within that broad camp, there are differences: Price, for instance is a self-confessed functional pluralist, just like my pragmatists, whereas Brandom shows less interest in differences of function, apparently holding instead that a blanket account of assertoric function will do all the explanatory work that is necessary to give a full acount of meaning adequate to any area whatsoever. This suggests not so much an alliance, as a wholesale takeover bid of the territory occupied by functional pluralists, albeit by a global company with a cognate set of interests.
Even so, as with my Carnapian pragmatism, we have to start with some ingredients. The ingredients Brandom allows himself in order to characterise use in a way that fuels his project include activities of inference, and of criticism: a ‘social deontic’ world suggested, at least, by the idea of ‘score keeping in a language game’ where the game is one of making and rejecting inferences to and from individual sayings. The inferences in question are not purely logical inferences, but any of the wider class of material inferences, since these are a salient a part of our linguistic behaviour and as much a target for endorsement or criticism as any of they much rarer class of logically valid inferences.
In spite of its auspicious pedigree, however, the notion of a language game is not really appropriate. Our inferential moves do not belong to a self-contained game, existing only for the pleasure that can be given by conforming to the constitutive rules that make up the activity, which I regard as the essential characteristic of a game. Nor is a harvest in which we only gather firstly the syntax of peoples’ sayings, and then norms governing what they permit to be inferred from what else, sufficiently rich to bake the bread of semantics. The quick way of seeing that is to recognize that even if the social-deontic norms allow us to locate the logical constants and to conjecture their meanings, and allow us as well to distinguish out names, and predicates, it is still bound to be true that if they are interpretable at all in some domain, then they can be interpreted in any of innumerable domains of the same cardinality. You simply cannot conjure semantics out of syntax. Hence inferentialists are driven to add some recognition of the landscape in which peoples’ engagement with the world is more to the fore: fundamental use properties (Horwich 199?), or ‘indication relations’ ( Field, 1994).
In Brandom’s richer treatment this means adding explicit recognition of a version of Wilfrid Sellars’s entry and exit rules: starting with ‘the practical involvement with objects exhibited by a sentient creature dealing skillfully with its world’, progressing through a cycle of perception, performance, assessment of results and then further processes of ‘feedback-governed performances’, serving as the basis for the ‘special case’ of the practical intentionality exhibited by these transactions, which is semantic intentionality (Brandom 2008, pp. 178–9). Such practices, he insists, are ‘thick, in the sense of essentially involving objects, events, and worldly states of affairs. Bits of the world are incorporated in such practices, in the exercise of such abilities’ (p. 178). Such practices, we might add, are also far from games: they make up the serious business of life for any sentient creature.
Pragmatism is thus left only to fill in the space between any old sentient creature coping skillfully with its world, and specifically linguistic or semantic creatures, able to deploy the resources of language in order to aid that task. This may reasonably seem to be quite a diminution of its sphere of influence. For example the creature coping skillfully with its world may be supposed to exhibit phenomena of attention, memory, and recognition, each of which seem to be at least as important an ingredient in a burgeoning notion of reference as any phenomenon, such as anaphora, that emerges from thinking solely about permissible or mandatory patterns of inference. The referents of its terms are ready to be picked out by its patterns of action, including the patterns of feedback that identify what counts as its success in action, or achievement of its purposes.
Bringing in the subject’s worldly involvements in its world of desire and motivation, practice and fulfilment is bringing in a great deal. Indeed, it makes pragmatism of this kind, with this much enrichment, quite difficult to distinguish from, say, a Davidsonian account of the business of the radical interpreter. To identify a practice, in Brandom’s rich sense, is already to identify what counts as a perception or observation, as success in action, and as a modfication of strategy in the light of success or failure. It is to endow the subject with a psychology, leaving only the task of mopping up the interpretation of any signals that seem to aid the social regulation and coordination of the psychologies of conspecifics with which it cooperates or competes. The Davidsonian radical interpreter trigangulating holistically is using the same data in the same way. Yet Davidson is often offered as an example of someone advocating a theory based on truth conditions and therefore standing at some distance from modern pragmatism. It is desirable to understand that this opposition, at least, is much less substantial than it might have seemed.
It might seem that Brandom’s emphasis on material inference, with no initial basis from which to distinguish strictly logical inference, makes his project different from anything in Davidson. But it is hard to believe that in the end this makes such a difference. Any emphasis on material inference must eventually be tempered by the requirement that people who have even quite etensively different beliefs about things can nevertheless be interpreted as sharing a language, or meaning the same by their words (this is why their beliefs can clash). As the initial debates between realists such as the then Putnam, and radical incommensurabilists in the philosophy of science showed a long time ago, shared reference is an interpretative move that can properly be imposed precisely to curb the idea that a small divergence in inferential pattern immediately suggests a large collapse of shared meaning. Pragmatists must, therefore, acknowledge something like a Quinean division between centre and periphery, and it takes little more than that, coupled with the isolation of logical vocabulary applied in any area and in connection with patterns of inference that may recur with any subject matter, to resinstate a much more realistic conception of sharing a language. But again, it will not be one that separates such an approach to meaning from that other descendant of Quine, Davidson.
Be these things as they may, the element I wish to highlight is rather different. Suppose we ask exactly which bits of the world are incorporated in the exercise of these abilities that make up practical intentionality? Some, we might surmise, will be much harder to do without than others. Surrounding trees and rocks, prey and predators, will no doubt be parts of our practice, but what about the necessity of one event following another, or the duty to love our children, or the number 42? I have little idea how Brandom would answer such questions, but a natural line would bring him into closer alliance with Price, and with me. This natural line would recognize that a description of the sentientt situation will necessarily employ some environmental landmarks, and for human beings in particular it will see us as surrounded by middle size dry goods, distinguished by features that are irresistibly borne in on us in the normal deployment of our senses (the objects for which, as evolutionary adaptations, the senses are fitted). These are things which, in Price’s terminology give rise to what he calls one of the ‘attractors’ of the undifferentiated idea of representation that he describes as environmental representation or e-representation. Cleaving to deflationism, Price does not think that e-representation exhausts the idea of representation: the other attractor for the notion is the inferentialist notion, or in more general terms the promiscuous deflationist notion we have already saluted.
We now see room for a rapprochement between the Carnapian pragmatist and the post-Sellarsian camp. The route to metaphysics opens up precisely when the notion of the ‘environment’ is expanded sufficiently to include any old thing: properties, classes, numbers, propositions, values, norms, abstracta, necessities. But there is no reason, if we want to understand our cognitive functioning in terms that are remotely naturalistic, for doing that. It is confusing items which it may be reasonable to call ‘parts of our world’ —for after all we talk in terms of them—with items which are ‘just there anyway’: parts of the world in which we must inescapably see ourselves placed if we are to give any remotely realistic yet naturalistically economical account of human cognitive functioning, or the human situation.
I do not see this development as in any sense hostage to what are vaguely referred to as the ‘rule following considerations’. By stressing the contingencies underlying all our classifications Wittgenstein’s discussion certainly has the potential to alter the way we look at ourselves (even if, as some commentators suppose, only by warding off philosophical mistakes responsible for erroneous ways in which we might look at ourselves). But adding the equivalent of ‘that’s how we look at it’ or ‘that’s what we say’ after all our sayings is a wearisome game, and as Wittgenstein himself insisted, does nothing to obliterate working distinctions between what we find ourselves having to say, and what we can more or less easily imagine ourselves not saying, or not having said at some historical juncture, or not being about to say in the future (Wittgenstein 1969). And as I have argued at length elsewhere the rule following considerations most certainly did not, for Wittgenstein, operate as a ‘metaphysical wet blanket’ smothering distinctions within language, and muting all its rainbow hues into a uniform muddy brown of response dependency.
However, closely allied to the rule following considerations is the thought that different parts of discourse blend into each other, or that there is no ‘disentangling’ leaving us an area of clean fact, and a distinct area of clean value, or in the empirical case a world of clean ‘is’ and a world of clean ‘must be’, or natural or causal law. This latter entanglement is saluted by Brandom as a major discovery of Kant, reinforced by Sellars: ‘the ability to use ordinary empirical descriptive terms such as ‘green’ ‘rigid, and ‘mass’ already presupposes grasp of the kinds of properties and relations made explicit by modal vocabulary’ (Brandom 2008, pp. 96–7). According to Brandom Sellars saw that this licensed the idea that modal vocabulary makes explicit what is already implicit in the inferential powers associated with more ordinary items of vocabulary. To say that a cat necessarily has weight, for instance, makes explicit an inferential license already possessed by anyone understanding the concept of a cat and of weight. Such an idea clearly has the potential to undermine any kind of simple empiricism, and any kind of simple foundationalism. On the other hand, it does not have the potential to undermine the distinctness of the contribution the modal element is making to the concept in question: this is why the Kantian thesis is worth stating (and the only reason that it can be stated). Similar remarks apply to the factual and the evaluative. Even if it could be established that, as it is put metaphorically by Hilary Putnam, values everywhere cast at least a little shade of pink over the grey of fact, this would not stop it being true that by so doing they do something distinctive and something that it is important to distinguish if we are to make explicit the architecture of cognition.
Environmental representation is essentially a matter of causal covariation. It can be thought of by comparing ourselves with the instruments we build to co-vary with environmental states: petrol gauges, voltmeters, windsocks and so forth. What the rule-following considerations can remind us that this is not a matter of wholesale opposition between the space of causes and the space of reasons. One of the functions of our cognitive machinery is to monitor our own capacities as input/output devices, and this is where Brandom’s notion of iterated episodes of feedback and adjustment is useful. I may be aware, for instance, that a verdict on the state of the world has not just popped into my head, but was the result of my having placed myself intelligently into a situation in which that verdict would not have resulted had the world not been thus-and-so. Or, I may entertain doubts whether I did indeed do this, and adjust my confidence accordingly, or go an look another time, or go over my notes again, or whatever is required. In this way I can monitor my own functioning: applied to myself the question ‘did I dream it up or am I remembering seeing it?’ is much the same as the question asked of a voltmeter ‘is it stuck at this reading, or is it covarying its output with the input in the way in which it is supposed to do?’. And in each case the empirical investigation of the question proceeds in much the same way. We look for evidence of causal receptivity, and if it is lacking we may take steps to improve the instrumentation operating on the subject matter. Kant was right that without such a grounding in our own reception of causal impacts from the immediate environment, the whole world of thought would be empty. The spaces of causation and reason cannot therefore be separated: by seeing ourselves as reasoning beings we do not stop ourselves from being at the same time natural animals.
Let us return to the pragmatic approach to particular metaphysical danger areas. We now decide to treat some part of what we say not in terms of environment—but how, then? What is the contrast, and how, having started with the contrast, do we regain entry into the world of ‘inferential representation’ or in other words aptness for assessment as true and false, and the resulting right to wear any semantic vocabulary that we care to deploy? Approaching this question takes us into the domain of what I christened ‘quasi-realism’, although I have long regarded the term with the same kind of embarrassment that ought to be felt by parents who call their babies things like ‘Honeymoon’ or ‘Sheetrock’. I no longer like the connotation of ‘as if’, which does endless amounts of trouble, nor the implication that ‘realism’ is a well developed ‘ism’ which can and should be imitated. Nevertheless, the task is properly identified. It is that of getting from some kind of doing, now thought of as the function of the sayings in question, to being comfortable with their assertoric status, or propositional clothing—their fit into the domains of truth and falsity. And this territory must be crossed without supposing that the doing in question is one of responding to an environment of an enriched kind: a modal or moral environment, for example. For that is precisely the kind of explanation that would explode the pretension to pragmatism and open the door to metaphysics once more.
It seems to me, as it has always done, that the right approach must be to see what happens if we do without the propositional surface. Suppose we did not have this piece of vocabulary: in what way would our practices be hampered? Provided the cost is identifiable and we therefore can be seen to have had a motivation to avoid it, then the way should be open to see ourselves, precisely, as having done just enough in order to avoid it. We can see ourselves as having enriched our inferential practices or our dealings with the world, without having licensed the philosopher to enrich our conception of the world with which we are dealing.
I think it is not often recognized (or perhaps it is a question of myself having been slow to recognize) that far from threatening this project, deflationism itself offers an example of the very same strategy in action. We have already placed deflationism as giving a logical, i.e. inferential, role to the truth predicate. And this supercedes any idea that the predicate serves to introduce some arcane property or relation to which we are equally mysteriously sensitive. But this is exactly an example of the strategy about which I am talking. What would be the cost if we had no truth predicate? The cost would be that we could not generalize and indirectly refer. We could give lists along the lines of ‘George Bush said that p and not-p’ —but not sum them up in the generalization that everything Bush said was false. We could not give anyone the information that the witness said something true without spiling the beans over what it was, and neither could we generalize to tell students that if one member of a disjunction is true then the overall disjunction is true. So the truth predicate slots into place as a device of indirect reference or a device of generalization. That is its function, and it is enough to justify the use we make of it.
Now consider the ‘good’ predicate. What would be the cost of doing without it? The second clause in deflationism about truth gives the cognitive equivalence that establishes the conditions for understanding the truth predicate. The natural equivalent clause for the ‘good’ predicate would be that in a straightforward application to an identified subject, such as someone’s action or character, you assent to the assertion using it if and only if you are disposed to endorse, choose, recommend, or otherwise practically orientate yourself in favour of the action or the character. And then the story of function proceeds in parallel with the alethic case: we can now generalize and refer indirectly, talking of John’s good deed (without telling you what it was he did); saying that everything Bush did was good or that John’ character is spotlessly good and so forth. In other words we now bring the practical orientation into the sphere of the propositional, ready to take its place either as the conclusion or as a premise in inference. If used as a premise then its eventual output may be other approvals, or a modification of beliefs, or whatever else the addition of the proposition may do to affect inferences downstream of it. If we are ‘online’, or in other words asserting the proposition, then the output is itself primed to be online. If we are ‘offline’ or merely playing with the approval in our imagination, then the proposition is put into inferences as a supposition, which may or may not be discharged, exactly like any other.
Giving this kind of pragmatic story about the role of certain propositions—for we can now use the term in good faith—does not diminish the care they deserve nor our responsibility for verdicts cast in their terms, or in other words, our responsibility for doing our best to get them right. If I say that a proposition is necessary when it is not or that an act is obligatory when a better verdict would say that it is not, then my intellectual or practical life is set to go wrong. I would be insisting too much, since the verdict of necessity plays the role of ruling out alternatives, and this may be a very bad thing indeed to do. So self scrutiny is just as urgent in any such case as it is anywhere else.
There are things which little companies do better before they are taken over by large conglomerates. I hope I have done something to defend the idea that the piecemeal investigation of different uses of language, of the kind suggested so forcibly by Wittgenstein, may be more fruitful than acceptance of the prairie landscape insisted on by those global behemoths, realism and pragmatism.
Carnap, R. (1950) ‘Empiricism, Semantics and Ontology’, in Meaning and Necessity: A Study in Semantics and Modal Logic, enlarged edition (University of Chicago Press, 1956).
Horwich, P. (1990) Truth, Oxford: Blackwell.
Dreier, J. (2004) ‘Metaethics and the Problem of Creeping Minimalism’, Philosophical Perspectives 18: 23-44.
Sinclair, N. (2006) ‘The Moral Belief Problem’, Ratio 19: 249-260.
Sinclair, N. (2007) ‘Propositional Clothing and Belief’, The Philosophical Quarterly 57: 342-362.
Horwich, P. (1998) Meaning Oxford: Oxford University Press.
Field, H. (1994) ‘Deflationist Views of Meaning and Content’, repringted in Blackburn & Simmons, eds, Truth, Oxford: Oxford University Press, 1999.
Price, H. ‘Descartes Lectures’,
Brandom, R. (2008) Between Doing and Saying, Oxford: Oxford University Press.
Wittgenstein, L. (1969) On Certainty, Oxford: Blackwell.