SORITES, ISSN 1135-1349

Issue #12. May 2001. Pp. 60-65.

Meaning, Normativiy and Reductive Naturalism

Copyright © by SORITES and Deborah C. Smith

Meaning, normativity and reductive naturalism

by Deborah C. Smith

In `The Normativity of Meaning', Eric Gampel argues that the capacity to justify a linguistic usage is essential to meaning and suggests that this fact entails that naturalistic theories of meaning must take a non-reductive form if they are to be viable. I will argue that reductive and non-reductive naturalisms stand or fall together in the face of Gampel's argument that meaning plays an essential justificatory role. I will further argue that, if they fall, the lesson to be learned is not that we should avoid reductionism, but rather, that we should steer clear of physicalism in our meaning theory; if Gampel's argument is cogent, any theory of meaning will have to make reference to at least some abstract objects.

According to Gampel, the fact that constitutes the meaning of linguistic expression x in language game L sets a standard for correct and incorrect usage of x in L. That is, the meaning of a linguistic expression is like a rule in that it defines a difference between correct and incorrect linguistic usage of that expression, and thus provides a potential justification for a subject's linguistic usage of x. Given that a subject S intendsFoot note 3_1 to play language game L, he ought to use x in such and such a way. He is justified in using x in some ways, not in others. Gampel argues that this justification is hypothetical and neutral in that the meaning of x `does not tell us whether or why we ought to play the language game'(Gampel, 1997, 227-228). That is, the meaning of x in L cannot by itself give me a reason to play the language game (this is the sense in which the justification is hypothetical) and there is no specific requirement or restriction on the kind of reason (e. g. moral, epistemic, etc.) that would serve as my justificatory basis for playing the language game (this is the sense in which the justification is neutral). Gampel dubs the thesis that meaning plays an essential justificatory role the EJRM.

After formulating and explaining the EJRM, Gampel goes on to argue that such a condition puts pressure on the naturalist. This argument can be outlined as follows:

1. Any adequate theory of meaning must a) not conflict with the EJRM and b) provide an explanation of the essential justificatory role of meaning. (Gampel, 1997, 230)

2. Reductive forms of naturalism conflict with the EJRM

3. Thus, reductive forms of naturalism are inadequate theories of meaning.

4. Genuinely non-reductive forms of naturalism (which limit themselves to token/token identity claims) do not conflict with the EJRM.

5. Therefore, any adequate form of meaning naturalism must be genuinely non-reductive.

The pressure put on naturalistic theories of meaning by the EJRM is, according to this argument, that they must take a non-reductive form if they are to be viable theories. Naturalism is not ruled out entirely by the EJRM, according to Gampel; rather, the EJRM puts restrictions on what shape a naturalist theory can take.

The support provided by Gampel for premise 2 of the argument in some ways resembles some of G. E. Moore's arguments for ethical non-naturalism and in other ways resembles arguments against the identity theory (indeed, any materialist theory) of mind. This argument, which he calls the `normativity argument' can be outlined as follows:

1. Suppose that the EJRM is correct: having the capacity to (hypothetically) justify a linguistic use is essential to meaning.

2. No natural fact has as an essential property the capacity to (hypothetically) justify linguistic use. [Actually the claim Gampel makes is stronger. See e. g., pp. 231-2, `it is not essential to such facts to have any sort of normative role....']

3. So, meaning facts are not identical with (reducible to) natural facts (by the substitutivity of identity).

4. Thus, if the EJRM is correct, reductive forms of naturalism are false. Reductive forms of naturalism fall afoul of the EJRM.

The objections to this argument which he considers and the replies to those objections he offers similarly mirror objections and replies in the ethical debate over naturalism and over materialism in philosophy of mind. For this reason, I would like to focus instead on another aspect of Gampel's analysis of the pressure the EJRM allegedly puts on the naturalist.

Gampel makes it clear that he takes it to be the reductive aspect of reductive naturalism and not its commitment to naturalism that makes it run afoul of the EJRM.

If a theory of meaning is non-reductive it would escape the above argument. For instance, a token-identity theory such as Davidson's need not run afoul of essential normativity, since the token, in being a token of a meaning as well as a token of some naturalistic kind, is essentially normative. (Gampel, 1997, 232)

However, I am less than sanguine about the possibility of a non-reductive naturalism faring any better than reductive naturalism in the face of the EJRM, if the non-reductive naturalist is claiming that there are true genuine identity statements of the form `x=y' where `x' denotes a meaning token and `y' a naturalistic token. Just as it is difficult to see how any naturalistic type could have an essentially normative role, it is difficult to see how any naturalistic token could have such an essential normative role. Compare this point to the metaphysical paradox of the marble and the statue. It is an individual token of a statue form which is tentatively identified with an individual chunk of marble. The problem is that the statue seems to have certain essential properties (e. g. a particular shape) which the chunk of marble does not. And it does not seem to sufficiently resolve the metaphysical paradox to be told that, in being a token of the statue form the chunk of marble has its shape essentially. At any rate, if the non-reductive naturalist can make his view accord with the EJRM, I see no reason in principle why the reductive naturalist could not do so as well. If a naturalistic token can play an essentially normative role, why not an entire naturalistic type? After all, if some form of reductive naturalism about meaning is correct, then every token of the naturalistic type will be a token of meaning, and hence, essentially normative. The reductive and non-reductive forms of naturalism seem to stand or fall together given the EJRM.

But perhaps we should not construe the non-reductive naturalist's claim as the claim that every meaning token is numerically identical with some naturalistic token. Perhaps his claim is merely that meaning tokens supervene on or are in some other way correlated with but distinct from naturalistic tokens. Perhaps this is why non-reductive naturalism need not fall with reductive naturalism. In this case, I fail to see how the non-reductivist counts as a meaning naturalist at all. Not only is there no reduction of meaning to any set of naturalistic facts, but we still seem to have non-natural meaning facts in the account. If this wouldn't count as a form of meaning non-naturalism, I'm afraid that I don't see what would.

I think that a quick digression concerning the criteria for a naturalist view is in order here. Often, a view is called a form of naturalism if it (i) reduces some type A of (putative) entity, property or fact to another type B of entity, property or fact and (ii) type B is a naturalistic type (roughly, a type that would be recognized by the natural sciences). However, it is clear that Gampel, in allowing non-reductive forms of naturalism, is construing meaning naturalism in a very different way. What exactly is meant by the term `naturalism' as Gampel uses it? It is somewhat hard to tell. However, his reference to Davidson's token/token identity theoryFoot note 3_2 as a form of non-reductive naturalism makes it seem plausible to interpret Gampel as holding the view that a natural object or process will be a concrete, physical object or process.Foot note 3_3 It would appear that, on this view, what we might call meaning physicalism (the view that meaning tokens are concrete, physical tokens even if no nomic regularities correlate the two types and so no reduction is possible) is definitive of meaning naturalism.

What I want to claim is that, if the reductive naturalist's view does run into trouble with the EJRM, it is his physicalism and not his reductionism that is the cause of the trouble. If this is correct, the claim that the EJRM puts pressure on the naturalist is a bit of an understatement, to say the least. On this construal of naturalism, the truth of the EJRM and the cogency of the normativity argument entail that naturalism is false. The lesson to take away from Gampel's normativity argument is not to avoid reductive forms of naturalism, but to avoid naturalism altogether when constructing a meaning theory.

Does the EJRM give us any reason to shy away from reductionist accounts of meaning in general? I think not. And, indeed Gampel indicates as much in his discussion of functionalism. We are reminded that `functional accounts of speaker meaning, often called «non-reductive» because they allow the functional kinds to be realized in any of a number of ways, are still ruled out by the normativity argument, so long as the accounts attempt to define the relevant function in naturalistic terms'(Gampel, 1997, 233, my italics). It is not hard to see that a non-physicalist, and hence, non-naturalist functionalism would not run afoul of the EJRM. Suppose that I reduce meaning facts to facts about functions thought of as purely non-natural, abstract objects.Foot note 3_4 Let us suppose that `+' means what we ordinarily take it to mean -- addition. The meaning of `+' can be thought of as a certain abstract function (call it the addition function) which takes pairs of natural numbers as its argument and yields for every such pair a determinant natural number and which satisfies the recursion laws for `+': (x)(x+0=x) and (x)(y)(x+Sy=S(x+y)), where all this is understood in the usual way.Foot note 3_5 Such a function will be essentially normative in the way a rule is. If one is intending to compute the addition function, the function sets a standard for correct and incorrect computation. Given that I intend to compute the addition function, I am justified in answering `125' to `68+57' (to use Kripke's example). I am not justified in answering `x' where `x' denotes any natural number other than 125. Notice that this by no means guarantees that I will answer `125'. I might make an error in my computation, or, as Kripke suggests, suffer from some mental frenzy which would prevent my getting the correct answer. But, the relevant meaning fact, the abstract addition function, in conjunction with my intention to embody that function, is essentially normative; it tells me what I ought to answer. Such a theory, although reductive, accords with the EJRM.

Compare this to a theory in which the meaning of `+' is reduced to a function thought of not as an abstract object but as a wholly physical state of affairs. To see how this would go, let's begin by thinking of the abstract addition function as a program or set of instructions (again, thought of as an abstract object and not a series of symbols in a particular machine language). Then, let us imagine some physical system which computes the abstract addition function whenever it encounters problems of the form `x+y' where x and y are natural numbers. We can say that the physical object embodies or instantiates the abstract addition function. To turn this picture into a purely physicalistic functionalism, we need to drop reference to the abstract function out of the analysis of the meaning of `+'. We need to identify the meaning of `+' not with the abstract function -- which (if there is any such thing) might not have been computed by any physical system at all, but rather with the function thought of as instantiated or embodied by the physical system. That is, the meaning of `+', on this view, is the function as computed by the physical system. Now our analysis makes reference to only physical objects and processes; meaning facts are identified with a type of physical fact.

Unfortunately for the physicalistic functionalist (if Gampel is correct), this fact will prove troublesome given the EJRM. For, it is hard to see how any fact about what a physical system does yield as answer to a given question could have any bearing on what it ought to yield without further appeal to an intention to compute a function thought of as an abstract object. It is at this point that Gampel's normativity argument ties up with certain of Wittgenstein's points concerning the machine as symbol for a function as discussed by Kripke in his development of Wittgenstein's skeptical paradox.

First, the machine [no matter what we take it to be so long as it is a natural object] is a finite object, accepting only finitely many numbers as input and yielding only finitely many as output -- others are simply too big. Indefinitely many programs extend the actual finite behavior of the machine.... Second, in practice it is hardly likely that I really intend to entrust the values of a function to the operation of a physical machine, even for that finite portion of the function for which the machine can operate. Actual machines can malfunction: through melting wires or slipping gears they may give the wrong answer. (Kripke, 1982, 34)

If indefinitely many functions extend the actual finite behavior of any physical system, then either the physical system is not justified in yielding any particular answer to a new problem involving `+' or it will be justified no matter what answer it yields. The former directly conflicts with the EJRM if we are to identify meanings with the workings of the physical system. The latter conflicts with the very plausible assumption that, if the notion of a physical system's correct functioning (its being justified in yielding the answers that it does) makes any sense at all, there must be something which would count as incorrect functioning.

So, again we see that, if Gampel is largely correct about the difficulties for some forms of meaning functionalism given the EJRM, it would appear that it is the physicalism of these forms and not the reductionism that is the problem. Both the physicalistic and the non-physicalistic functionalisms considered reduce meaning facts to facts about functions and get some explanatory juice out of doing so. But, it is only the physicalist whose view runs afoul of the EJRM.

To sum up, reductive naturalisms and genuine token/token identity naturalisms stand or fall together in the face of the EJRM. If they fall, the lesson to be learned is that we should steer clear of physicalism in our theory of meaning; any theory of meaning will have to make reference to at least some abstract objects (whether or not they constitute an autonomous set of irreducible meaning facts) in order to properly accord with the EJRM. Since naturalism seems to require what I have called physicalism, this has obvious implications for meaning naturalism; it entails that meaning naturalism is false (and not merely that the forms which it may take are restricted). However, the truth of the EJRM does not entail that we cannot have a theory of meaning that makes an interesting and explanatorily useful reduction of meaning facts to facts that involve other abstract objects such as sets or functions. After all, such a reductive non-physicalist might argue, if we need to postulate, e. g., abstract functions for other purposes, why postulate in addition to such functions, meaning facts? We achieve an advantage of theoretical simplicity by reducing the number of primitives we need to postulate in our total theory by reducing the meaning facts to facts about abstract functions. Of course we could reduce the number of primitives postulated by our total theory even more if we could reduce these abstract functions to the workings of various physical systems, thereby yielding a reductive physicalism. But that reduction can occur only if the EJRM is false or if there is something else wrong with the normativity argument.


Davidson, Donald. 1984. Inquires Into Truth and Meaning. Oxford: Clarendon Press.

Gampel, Eric. 1997. The Normativity of Meaning. Philosophical Studies Vol 86: 221-242.

Kripke, Saul. 1982. Wittgenstein on Rules and Private Language. Cambridge: Harvard University Press.

Deborah C. Smith

Kent State University