II--NATURALISM AND THE FATE OF THE M-WORLDS[*]
We make a huge variety of claims framed in vocabularies drawn from physics and chemistry, everyday talk, neuroscience, ethics, mathematics, semantics, folk and professional psychology, and so on and so forth. We say, for example, that Jones feels cold, that Carlton might win, that there are quarks, that murder is wrong, that there are four fundamental forces, and that a certain level of neurological activity is necessary for thought. If we follow Huw Price's Carnapian lead, we can put this by saying that we make many claims in many different frameworks.
How should we view the interconnections between claims made in different frameworks? As Price observes, an increasingly popular view is that well-founded claims framed in one or more of the languages of the physical sciences give a complete, as near as is now possible, account of what our world is like, and that what is said in the languages of psychology and morality, for example, is either talk about the very same entities and properties (in the wide sense that includes relations) in different words (reductionism, on one understanding of that contested term), or else is false talk (eliminativism), or else is not talk at all in the sense of claims about how things are (noncognitivism). Price seeks to resist this kind of naturalism by outlining a new way of looking at the interconnections, or lack of them, between claims in the different frameworks. I will explain why I am unconvinced by his `new way'.
The paper divides into three parts. The first outlines the orthodoxy concerning language that Price is rebelling against, emphasising how natural and appealing it is. The second explains, in terms largely independent of what I say in the first part, why I find Price's general approach unhelpful when we come to consider two familiar problems concerning the connection between certain things we say in the language of psychology and certain things we say in the languages of the physical sciences. The final part is a brief advertisement for the approach I favour to one of these problems.
1. Orthodoxy about representation and language
Why do foreign-language phrase books sell so well? Because they help us find food, shelter, museums and airports when we travel outside our own language communities. Our need for them highlights what is anyway obvious: much of language is a convention-generated system of representation.
Although it is obvious that much of language is representational, it is occasionally denied. I have attended conference papers attacking the representational view of language given by speakers who have in their pockets pieces of paper with writing on them that tell them where the conference dinner is and when the taxis leave for the airport. How could this happen? I surmise that it is through conflating the obviously correct view that much of language is representational with various controversial views. So let me clarify what I am, and what I am not, taking to be obvious.
Robert Brandom describes himself as an opponent of the representational view. Here is a typical passage
The absence of a nonregressive account of what it is to take, treat, or use something as a representation of something else is the source of another traditional sort of dissatisfaction with the representationalist paradigm of contentfulness. It lies behind Rebecca West's irritated response to the "mind as mirror of nature" model that it is hard to see why one would want a copy of the universe: "One of the damn things is enough." Progress in understanding intentional contentfulness is made by invoking representational relations only in the context of an explanation of what it is that makes representings graspable or intelligible as representings in a way in which what is represented is not.
However, the representational orthodoxy I am espousing does not think of representings as somehow more intelligible that what is represented. If I could have been present at the Chicago Bulls' final game, I would have found their NBA victory both intelligible and welcome. However, as I was in Canberra at the time, I needed something located in Canberra to give me the desired information. I didn't want another "of the damn things", only something nearby to tell me about the genuine article, and a set of sentences in the local paper filled the bill.
Further, representational orthodoxy is not the view that we understand `intentional contentfulness' simply `by invoking representational relations'. The orthodoxy holds that certain bits of language, and likewise certain mental states, represent how things are (which, if either, is the more fundamental is not part of the orthodoxy, though orthodox orthodoxy holds, correctly in my view, that the mental states are the more fundamental). It is a separate, hard question how exactly some bit of language or mental state gets to have the representational content it does have. In the case of language, it has a lot to do with conventions of usage that settle what thoughts the words are to be used to express, but exactly which usages are crucial is controversial. Although everyone here knows how `There is a circle drawn on the board behind me' represents things as being--everyone here knows what to do to the board to make this sentence true--there are problems about saying which usages are crucial for its representing what we agree it does represent. Is this the correspondence theory of truth? It depends what you mean by that theory. `Tibet is larger than England' is true just if things are as it represents them, and we might well count this a version of the correspondence theory. But, unlike some versions of that theory, it is an account of truth for sentences, not propositions, and invokes no mysterious, specific-purpose `fact' to be what makes the sentence true by virtue of the sentence corresponding to it. What makes the sentence true is the nature of the world we live in. And, finally, it is democratic: it allows that some may mean by `true' something like, say, `part of the best theory in the limit of enquiry', and leaves as a matter for further investigation the relationship between this conception and the one that falls out of representational orthodoxy.
Also, representational orthodoxy is neutral on the extent to which we are able to say how things are. Even though I do not know how much you weigh, I have the resources to say how much you weigh: my problem is that I do not know which of `weighs 70 kilos', `weighs 71 kilos', etc. are the right words to use, not that I do not have the appropriate words (or thoughts or concepts) at my disposal. But it may be that there are some ways things are that are in principle beyond our representational reach, in some reasonably strong sense of `in principle'. Further, representational orthodoxy per se is silent on the extent to which, when we say how things are, we are saying something about how we representers are. Some words say how things are in part in terms of how these things relate to us: plausible examples are `red' and `comic'. A laudable endeavour is to seek to say how things are independently of our perspective on them; as Price himself puts it, `one of the great projects in the history of thought has been the attempt to achieve the untainted perspective, the Archimedean view of reality.' But representational orthodoxy is not committed per se to this project: it may be that all words and sentences, to some extent or another, say how things are in relation to how we are. Words would still serve to say how things are, but these claims would concern how we are much more than we might have thought. We would always, to one extent or another, be saying how we are. This view need not be any kind of idealism. It might grant that there are ways things are quite independent of how sentient creatures are, but insist that we cannot capture these ways in language. (I do not, though, find this position attractive. Although the sentence `Gravity is ubiquitous' in our mouths and from our pens has the representational content it in fact has, at least in large part, because of how we are, how it represents things as being is correct from the Archimedean viewpoint, or so it seems to me.) My hope has been to recruit to representational orthodoxy by noting some controversial matters that it is not committed to. I need though to mention a commitment that might be thought controversial. Much of language not only represents how things are, it does so in a way which we users grasp or understand: competent English speakers typically know what they are saying about how things are when they utter various English sentences about bus timetables, psychological states and so on--therein lies the utility of mastering English (and likewise for French etc.). But, of course, some argue that content is a matter of evolutionary history in the sense of being what some structure is selected for. This is, no doubt, true for some notion of content, but is implausible for representational or informational content. The information we give out when we utter sentences like `Jones hopes that there is not a tiger behind her' or `The bus leaves at six' is not information about evolutionary history.
Also, orthodoxy insists that we do not in general make or create what we report on in language: when the doctor produces the word `broken' after looking at an X ray, what worries you is what she is reporting on, what she has discovered, and not anything she has created, and not anything the system of categories embodied in our language has created. We take the categories to be apt for capturing how things are, not for creating how they are. There is, of course, a view that when we classify objects together in language or thought--as all having, say, the possibly partially disjunctive features we use words like `deadly', `square', `broken', and so on, for--we are not capturing anything about the objects or their relations (including possibly their relations to us) which obtains independently of our classification. I hope that the words I have selected make clear how hard it is to believe this view. Moreover, the view can be internally incoherent. We are told that there is no sense to categories waiting to be captured in some set of objects independently of those we impose or create by the divisions we make (actually or potentially) in language and thought. But it turns out that we impose these categories by attaching the same word or thought to each member of the set classified as similar in the respect in question. But unless this attachment is to be magical, there must be a difference between the things that fall under a given word or thought and those that do not--as it might be, and very roughly, the things that fall under the word or thought normally have a special link to tokens of the word or thought. But, on the sceptical view in question, there is no sense to these tokens belonging together independently of how we group them together--what is sauce for the labelled is sauce for the labels. The idea that there is nothing more to objects falling under a classification than their so doing--falling under a classification is never a matter of how they are independently of actual and potential acts of classification--gets any plausibility it has from thinking, inconsistently, that the words and thoughts that do the classifying belong together independently of some further classification. You might, of course, appeal to a meta-classification to tie the classifying tokens together, but this would lead to a vicious infinite regress--it would be like `tying' a ship to the wharf with an infinitely long chain that never reaches the wharf. Michael Devitt calls the view that we impose the classifications, the cookie cutter theory of reality. But, it seems to me, matters are even worse: the view is inconsistent with there being a cookie cutter able to do the cutting.
Price says that his views should be seen as part of a semantic minimalism. One thing you might mean by semantic minimalism is what I have just denied. A second is the view that it is automatically the case that any meaningful, syntactically declarative sentence is representational. Suppose we ask, How does `If it rains, the match will be cancelled' represent things as being? The second view is that we can give the answer by using the sentence itself, as in
`If it rains, the match will be cancelled' is true iff if it rains, the match will be cancelled.
We have, it seems, a method that delivers how a sentence represents matters, for every meaningful, declarative sentence. However, representational orthodoxy insists that matters are not so simple. Perhaps every meaningful, declarative sentence is representational, but there is nothing automatic about it.
How a sentence represents things as being is an a posteriori, contingent matter. It is a matter of the thoughts about how things are that the words and sentences are, under the contingently adopted conventions of the language, used to express. Roughly, `square' stands for being square in English because the conventions of English imply that the use of the word `square' expresses thoughts about things being square. Equally, `carré' stands for being square in French, because `carré' is a word to use in French to express the thought that things are square.
This means that some class of meaningful, declarative sentences might fail to be representational by virtue of not having the right connection with thought to enable us to identify how they represent things as being. This is exactly what many hold concerning indicative conditionals. They ask, How should we settle how `If P then Q' represents things as being? And answer, By reference to when competent speakers produce it when seeking to express how they take things to be. But competent speakers produce `If P then Q' just when the conditional credence of Q given P is high enough for assertion. But, those who insist that indicative conditionals are not representational go on to observe, the conditional credence of Q given P is not the credence of anything: it is rather a quotient of credences. Hence, they conclude, there is no answer to how indicative conditionals represent matters, for the way we use indicative conditionals does not serve to determine truth conditions for them.
There is, though, one regard in which representational orthodoxy is minimalist. If a sentence represents how things are, the sentence itself gives the conditions under which it is true: it is true just if things are as it represents things, and one good way to give that is with the sentence itself. It is, accordingly, a consequence of orthodoxy that a complete answer to when the sentence `Snow is white' is true is given by the sentence itself: `Snow is white' is true iff snow is white.
Although orthodoxy insists that the combination of meaningfulness with being declarative does not per se show that some sentence is representational, it insists that a great deal of language is representational--representing being, after all, a prime function of language--and I am going to assume that the discourses we will be concerned with in our discussion of inter-framework negotiations are representational. This assumption is controversial in the case of ethical sentences but not, I think, in the case of semantic and psychological sentences. If much of language is representational, discussions of this fact had better be representational, but they are discussions in semantics and psychology. To say that some sentence represents how things are is to say something about its meaning to the effect that it serves the role of expressing how a user of the sentence takes things to be. And it would be very strange to hold that many sentences serve to say how things are, but the sentences we produce to say that they do so themselves fail to say how things are--you can say it, but you cannot say that you can say it!?
Finally, I should make explicit what has been implicit: orthodoxy insists that these various representational discourses all seek to represent how the one, huge world is: there is just one huge, complex world which makes, or parts of which make, our successful representations true (or at least make the contingent ones that will be our concern here, true). Price has an argument against this kind of ontological monism, or, as it sometimes seems to be, an argument that there is no sense to asking, One world or a plurality? He sources the argument to Carnap, and rests it on the point that we cannot step outside any and every framework, that of necessity we must work within some framework or other: as he puts it, `there is no absolute, theory-independent ontological viewpoint available to metaphysics'. I agree completely: you can't have a view without having a view, and having a view requires enough by way of belief, and so conceptual sophistication, to count as a framework; any amount of belief requires a lot of belief. But ontological monists insist that `one-worldism' is part of any and every coherent framework; we follow Quine in finding the idea of different kinds of existence utterly mysterious. But then no stepping outside is called for: monism is there in each and every framework. Obviously, this is no argument for ontological monism. Indeed, there is a sense in which we cannot argue for it, because we, or I anyway, cannot think of anything more obvious from which we might derive it. My point is simply that orthodox representationalism does not make the mistake of assessing a framework from outside any framework.
You now have the alternative, or one alternative, to Price's view of language before you. It agrees with Price that language can play many roles--warming up the audience before a rock concert is very different from giving a lecture in theoretical physics--but insists, against Price, that much of language is unified in being part of the enterprise of representing how things are; it is, that is, not a pluralist view in Price's special sense. And it takes ontological monism for granted--provided that this is understood as the view that there is one kind of existence and one huge world that makes our true representations true, not as the false view that everything is of the same kind. At a number of points Price notes a pervasive view that makes his own `practically invisible'. He says, for instance
When philosophers think of description, or representation, they often have in mind a familiar picture: on one side, the World, on the other side Mind, or Language (the medium of representation). Once this picture is in place, the path I want to take is practically invisible.
Orthodox representationalism is precisely this picture. The sentence that told me that the Bulls won, though a part of the world, is to be sharply distinguished from the world it is about.
2. Why I find Price's approach unhelpful
Price is offering us a new way to look at old problems. Obviously, one concern I have with it is that it rests on a view about language I reject. But independently of this, I cannot see how his new way helps us with these old problems. I will develop the point in terms of two familiar questions that arise when we discuss the interconnections between assertions in the framework of psychology, on the one hand, and, on the other, assertions in the framework of neuroscience and, more generally, of the physical sciences especially concerned with the genesis of behaviour.
The `too much causation' question. In the language of psychology we say that mental states enter into certain causal roles. We say that people's beliefs are caused by happenings around them, and cause certain bodily movements. As it might be, my seeing a certain goal being scored causes me to believe that Carlton won, which then causes my cap to soar into the sky. We also know that there is an account in the language of neuroscience that has neurostate N (or, more precisely, a whole series of neurostates, but I simplify) playing the distinctive, causally intermediate role between the sight of a goal being scored and the movement of the cap skywards. What should we say in response to this observation?
There are many, much discussed options, but our question is what might functional pluralism in particular have to contribute? Price's critical comments on ontological monism and his use of the word `pluralism' to describe his position might suggest that his position is that no response is especially called for. We simply have two accounts, both fine in their way, and the idea that, for example, the two accounts are in conflict (one says that it is a belief, the other that it is something else, a neurostate, that does the causing) is some kind of category mistake and is, in any case, deeply confused; equally, the idea that we should put the two accounts together and infer that the belief is the neurostate, in the style of Australian materialism, would also be deeply confused. This way of thinking of the pluralist approach might, I suppose, be described as a kind of quietism.
We know, however, that this kind of quietist response is wrong in lots of cases. Often the right response to two accounts of a causal role framed in the terms of two different discourses is precisely to think of them as in conflict. That is how vitalism in biology was overthrown: the discovery that it was cell division that explained the growth of plants was rightly interpreted as showing that the view that `vital forces' did the job was a mistake. On other occasions, the right response to two accounts of a causal role framed in the terms of two different discourses is precisely to put them together and infer an identity. To take a famous example, it was right to combine the fact that temperature in gases plays a certain causal role with the discovery that a certain molecular property plays that same role to arrive at the conclusion that temperature in gases is mean molecular kinetic energy. Price's view cannot be that it is always right to go quietist; but then, in the absence of an account of when it is right and when it is wrong to go quietist, we have no principled way of tackling the problem raised by the fact that a certain belief and a certain neurostate are said, in the languages of their respective frameworks, to play the same causal role.
Price might object that vitalism and cell biology, and thermodynamics and the kinetic theory of gases, are not examples of distinct frameworks in his sense. But this reply shifts the problem rather than solving it. It makes the question of how to individuate frameworks absolutely central--without an answer to it, we do not know when to apply his new way. I started by objecting that without an account of when it is right and when it is wrong to go quietist, his new way does not help us with what to say about the too much causation question, but it is equally true that without an account of when we have, and when we do not have, distinct frameworks, his new way does not help us with the too much causation question.
Price's passing comments on the `Armstrong-Lewis arguments for materialism about the mental' to the effect that they rest on `the principle that all causation is at base physical causation' suggest that he might argue that the right response to the too much causation question is to hold that belief causes behaviour in a different sense to the sense in which neurostates cause behaviour. I do not find this suggestion--to be distinguished, of course, from the view that mental states do not cause behaviour; instead they stand in a kind of explanatory relation to it that it is easy to confuse with causation--very attractive. It would seem no better than defending vitalism by arguing that vital forces cause plant growth in a different sense from the sense in which cell division does. But, be this as it may, this response is not an approach to the too much causation question distinctive of a new way via a functional pluralism about language, but instead is an instance of a rather bountiful traditional metaphysics.
The elimination question. Sometimes something's internal workings tell us that it lacks psychological states. Once we know how a telephone answering machine works, we know that it lacks a psychology. And the same is true for things that look like us, and move through the environment much as we do. Humanoids controlled from Mars, as in a variant on Christopher Peacocke's case, and creatures that work by look-up tree, as in Ned Block's example, lack a psychology precisely because they work in the wrong way for having a psychology, no matter how good a job they do of passing as us `from the outside'. But it would be quite wrong to infer from this that anything for which there is a sufficiently complete account of its internal workings, and how they lead to behaviour, givable in physical terms lacks a psychology. We have a rich psychology, and yet we know that there is such a physical account of our internal workings. (Some say that the way we work inside may turn out to show that this rich psychology does not include the having of propositional attitudes. But that is another question, and one that need not concern us here.) By the elimination question, I mean the question of how to mark off the kinds of internal workings that debar something from having a psychology, from the kinds that do not debar it from having a psychology.
I cannot see how Price's new way helps us tackle this problem. It would clearly be wrong to invoke one or more of: the distinctness of frameworks, the denial of ontological monism or of monism about causation, the impossibility of assessing one framework from inside another, and the different functions of different areas of discourse, to argue that no account in the terms of some physical theory of how something works inside could show that it lacks a psychology. It would be wrong because we know that some accounts in physical terms of internal workings are inconsistent with having a psychology. The kinds of cases described by Peacocke and Block (and others) are compelling; and, more generally, it would be absurd to refuse to move from information framed in the terms of the various physical sciences to information framed in the terms of psychology: if we learn that the room next door is full of chlorine gas, we learn that very little thinking is going on in it. But now it is hard to see how Price's new way could make a distinctive contribution. The general thrust of his remarks is very much towards putting up barriers in the way of moving between accounts couched in the terms of different frameworks. Once we see that it is beyond question that sometimes the barriers have to come down, it is hard to see what he should say about when they should come down, and what to say when they do. And, of course, he could not object in this case to our regarding psychological discourse and physical discourse as different frameworks; it is an example he himself gives of different frameworks.
3. The approach I favour to the elimination question
The essential problem is that some accounts of how X works inside framed in the languages of the physical sciences are consistent with X having a psychology, and some are not. What makes the problem hard is that we have to, somehow or other, connect what is said in two quite different vocabularies. The nub of why I find Price's approach unhelpful to the elimination question is that it emphasises the differences between the vocabularies, whereas what we need for the purpose at hand is some way to interconnect them. We need to find links so that we can address when we have, and when we do not have, a clash between what is said in one vocabulary and what is said in another. Too much talk about how fundamentally different the discourses are invites a kind of incommensurability thesis we know must be wrong.
The beauty of the Ramsey-Carnap-Lewis approach is that it gives us a principled way to discern connections between different vocabularies. As applied to the elimination question, the approach gives us a way of discerning the structure essential to having a psychology, and of discerning the structure in one or another account of how our X works as given in physical terms. And in both cases we specify the structure in basically the same terms. We are then in a position to see whether the structure ruled to be essential to having a psychology is consistent with the structure discerned in the physical story of how X works, and because both structures are specified in the same terms, this part is easy. If the structure ruled to be essential to having a psychology is consistent with the structure discerned in the physical story of X, the physical story is consistent with X's having a psychology, and X may have a psychology (may, not will, as there may be other constraints to be met); if it is not, X does not have a psychology.
Thus, to give the barest outline of this much told story: we take the core claims about how various mental states interconnect and how they connect with the environment, as framed in a combination of the language of psychology and our everyday language for giving the nature of our environment, and write them out as a long conjunction. We then get rid of the mental state terms by replacing them with bound variables. We now have a sentence about structure. It says how any creature's states, and, in particular, our X's states, must be interconnected, one to another, and to the environment, in order for it to have a psychology. We now take the proffered account of how X works as told in physical language, and perform the same task of discerning this latter account's structure by, in this case, replacing the physical terms by bound variables. We end up with two accounts framed in essentially the same terms, namely, bound variables, expressions for relations between the states referred to by the bound variables, and expressions in the everyday language for the environment, and are able to inspect to see if the second structure is consistent with the first. If it isn't, X lacks a psychology.
There are many queries that might be raised against the Ramsey-Carnap-Lewis approach, most notably how to find the core claims concerning how various mental states interact with one another and with the environment, and whether it is threatened by the arguments against description theories of reference. I think it can give good replies to these queries, but I cannot argue that here. The point I want to close with is simply that at least we can see how, in broad outline, the approach addresses the elimination question.