20th WCP: Social Externalism and Non-Empirical Errors

1. Introduction

Ever since Putnam's and Burge's attacks against individualist accounts of meaning the individualist has felt squeezed for space. Very little manoeuvreing room, it seems, is left for the philosopher who wants to deny that meaning depends on the speaker's social environment.

One option, popular amongst individualists, is to grant that reference is socially determined but argue that there is nevertheless a notion of meaning or content that can be understood individualistically. That is, the individualist can opt for a bifurcation of content into "wide content" (connected with reference and truth) and "narrow content" (understood purely internalistically). (1) The trouble with this option for a committed individualist is that it entails an admission of partial defeat. By accepting bifurcation the individualist gives up on reference and is forced into an internal sphere of (rather obscure) non-intentional content.

This leaves the individualist a second option; she can deny that reference is socially determined and reject the conclusions of the anti-individualist thought experiments put forth by Putnam and Burge. For instance, she can deny that Burge's patient uttering "I have arthritis in my thigh" is to be interpreted standardly, as speaking of arthritis. And she can deny that the expert and the non-expert are to be ascribed the same "elm"-concept. This would leave individualism intact, with reference and content alike determined individualistically. But the trouble with this option is that it seems so thoroughly unattractive. Are we really to say that Burge's patient has his own "arthritis"-concept and expresses a true belief when uttering "I have arthritis in my thigh"? Or that the non-expert speaks of a different tree than the expert? Would that not imply a fragmentation of concepts and reference such that no common core of thoughts and, indeed, no common world of objects, would remain?

The individualist therefore seems faced with a dilemma: She can give up on reference, thereby compromising her individualism, or she can reject the communitarian conclusions and accept conceptual and referential fragmentation. Faced with these options even the strong headed individualist will be tempted to desert into the communitarian camp.

In this paper I will try to instill some courage into the individualist, by urging us to rethink the apparent dilemma. We have become accustomed to thinking of the anti-individualist thought experiments in a particular way which, I argue, is not compulsory but rests on problematic philosophical assumptions. Once these assumptions are brought into the daylight it is clear that the dilemma can be avoided and individualism will seem much more tenable.

2. Social Externalism

To focus the discussion let us consider Burge's well-known thought experiment. A brief summary of it will suffice.

The experiment proceeds in three steps. First, Burge hypothesizes a speaker who is "generally competent in English, rational and intelligent." (2) This speaker, Burge says, has a large number of attitudes commonly attributed with content clauses containing "arthritis". In addition the speaker believes falsely that he has developed arthritis in his thigh. The belief is false, Burge argues, since, unknown to the speaker, in his community "arthritis" does not apply to ailments outside the joints. Second, Burge considers a counterfactual world where all is as before except that in the counterfactual speech community "arthritis" applies not only to a rheumatoid disease of the joints but to various other rheumatoid ailments. This means, according to Burge, that in the counterfactual world "arthritis" expresses a different concept (the concept of a disease of the joints as well as the ligaments). Burge then concludes, in the final step, that a member of the counterfactual community, in all 'internal' aspects identical to the actual speaker, who utters "I have arthritis in my thigh" is not misusing the word "arthritis", but using it correctly. He is expressing a true belief. As a result, meaning and mental content vary with social environment.

Now, the standard assumption is that the first step is the crucial one: Once it is accepted that the speaker makes a mistake when uttering "I have arthritis in my thigh" the anti-individualist conclusions follow. (3) To say that the belief expressed is false, it is assumed, is to say that the speaker has the community concept. For this reason individualists have typically focused their attention on the first step, arguing that the speaker does not make a mistake but, instead, has his own concept of arthritis and therefore expresses a true belief.

Thus, consider Davidson's response to Burge. Davidson criticizes Burge and rejects the idea that, as Davidson puts it, "what we mean and think is determined by the linguistic habits of those around us". (4) To make his point Davidson considers a version of Burge's arthritis-example. Suppose, Davidson says, that he, who believes the word "arthritis" applies to inflammation of the joints only if caused by calcium deposits, and Arthur, who knows better, both sincerely utter "Carl has arthritis". According to Burge, Davidson says, the two speakers must then mean the same thing by their words, and we should report them as having the same belief. But, Davidson argues, it is not correct that his and Arthur's beliefs should be reported the same way. If Smith were to report that Davidson and Arthur both believe that Carl has arthritis, he would be misleading his audience, unless he added "But Davidson thinks that arthritis must be caused by calcium deposits only". This shows, Davidson argues, "that the simple attribution was not quite right; there was a relevant difference in the thoughts Arthur and I expressed when we said 'Carl has arthritis'". (5)

Davidson's suggestion, therefore, is that the fact that he believes that arthritis is caused by calcium deposits only, implies that his word "arthritis" must be reinterpreted and that his "arthritis"-thoughts are not shared with Arthur (who does not believe that arthritis is caused by calcium deposits only). Similarly, Davidson would presumably hold that the fact that the patient in Burge's thought experiment believes that one can have arthritis in one's thigh implies that he should not be attributed the concept of arthritis, but, rather, the concept of "tharthritis". Thus, the anti-individualist conclusions are prevented. Instead of ascribing the speaker the standard concept of arthritis, he is to be ascribed his own concept. (6)

The trouble with this individualist response, as suggested above, is that it implies a problematic fragmentation of reference and concepts. If two speakers cannot disagree about the causes of arthritis (or about whether arthritis afflicts the joints only) without it following that they must have "different thoughts", then it is hard to see how two speakers could ever share any thoughts at all. There are countless disagreements of this sort within a community and if every such disagreement implies a disagreement in meaning there could be no sharing of concepts and no such thing as a disagreement in belief. The idea of a common store of thoughts, as Frege liked to put it, vanishes and with it the notion of a common world of shared objects. You believe arthritis is caused by calcium deposits only, I don't, and so we have different "arthritis"-concepts and speak of different diseases. I believe that elms have dark-green oval leaves, you don't, and so we speak of different trees.

Thus, the individualist seems faced with the dilemma sketched above: She must either accept conceptual fragmentation or give up on reference and endorse a bifurcation of content. However, the dilemma is only apparent. It is quite possible for the individualist to grant step one of the thought experiment (that the speaker uttering "I have arthritis in my thigh" makes a mistake), thereby preventing conceptual fragmentation, and yet avoid the anti-individualist conclusions. To see this, let us have a closer examination of the presuppositions of Burge's thought experiment.

3. Non-Empirical Errors

A central component of Burge's reasoning is the notion of a non-empirical or conceptual error. The speaker uttering "I have arthritis in my thigh", Burge says, does not make an "ordinary empirical mistake", but a conceptual or a linguistic one. (7) This, Burge argues, is an essential presupposition of the thought experiment. If the thought experiment is to work, Burge says, "one must at some stage find the subject believing (or having some attitude characterized by) a content, despite an incomplete understanding or misapplication. An ordinary empirical error appears not to be sufficient." (8)

It is clear why this is so. Assume that Burge's arthritis patient uttering "I have arthritis in my thigh" only makes an empirical error. This means that "Arthritis afflicts the joints only" is an empirical claim. Consequently, the fact that the counterfactual community rejects this claim does not mean that their word "arthritis" must have a different meaning. Rather, the disagreement between the two communities would simply be one of theory, (9) and there would be no reason to say that the speaker in the actual world expresses a different belief than the speaker in the counterfactual world.

For the thought experiment to go through, therefore, as Burge himself makes quite clear, it has to be assumed that "Arthritis afflicts the joints only" plays such a special role within our English speaking community (i.e. the actual community) that a counterfactual community which rejects this statement must be one where "arthritis" has a different meaning. That is, Burge has to assume that "Arthritis afflicts the joints only", in the actual community, has the status of a non-empirical truth of some sort.

How, then, are we to understand the notion of a "conceptual" or non-empirical error? Burge is rather unclear about this. The talk of such errors suggests that he dismisses Quine's criticisms of analyticity, but Burge claims to be adhering to Quine's criticisms of this distinction. That the subject makes a conceptual error, Burge says, should not be taken to suggest that the error is "purely" a mistake about concepts, involving no error about the empirical world: "With Quine, I find such talk of purity and mixture devoid of illumination or explanatory power." (10) Even though there is no such thing as purely conceptual truths, Burge argues, there are "analytically" true and "analytically" false beliefs that are linguistic in the sense that "they are tested by consulting a dictionary or native linguistic intuitions, rather than by ordinary empirical investigation." (11) Thus, Burge says, while the line between conceptual and empirical error is "pretty fuzzy", in the case of the speaker uttering "I have arthritis in my thigh" we must say that his error is of the former kind, since in the actual community "arthritis" does not apply to ailments outside the joints "by a standard, non-technical dictionary definition". (12)

Burge's suggestion, therefore, is that we can endorse what might be called a "low-profile" notion of analytic or conceptual truths, one which is quite compatible with Quine's rejection of the analytic/synthetic distinction. "Arthritis afflicts the joints only" is a conceptual truth in the sense that this is what the dictionary or our intuitions tells us. To say that the speaker uttering "I have arthritis in my thigh" makes a conceptual error is just to say that he goes against the dictionary definition or our intuitions. It does not require a commitment to a high-profile notion of conceptual truths.

This is clearly unsatisfactory. One might of course stipulate that whatever is found in the dictionary is "weakly analytic" (i.e. analytic in a sense which does not invoke any principled distinction between the empirical and the non-empirical), but this notion of analyticity cannot do the work Burge wants it to do. While the dictionary may define "arthritis" as "a disease of the joints only", we cannot conclude from this that "arthritis" must have a different meaning in the counterfactual community. To draw that conclusion it would also have to be claimed that a change of the dictionary definition (so that arthritis can afflict the joints as well as the ligaments) must be a meaning change, rather than a theory change, and this presupposes a high-profile notion of conceptual truth, one which is not compatible with Quinean skepticism about the empirical/non-empirical distinction. (13)

Burge's thought experiment therefore presupposes that "Arthritis afflicts the joints only" is non-empirical in a strong sense. And the question is why we should accept this assumption. Why should we accept the claim that "arthritis" must have a different meaning in the counterfactual community? It may be that the belief that arthritis afflicts the joints only is central to our understanding of arthritis, but what gives Burge the confidence that it is so central that giving it up must imply a change in the meaning of the term "arthritis"?

This, however, might appear to be beside the point. Why can't we simply grant Burge that "Arthritis afflicts the joints only" is a conceptual truth, and then proceed to meet his arguments? After all, Burge is not doing descriptive linguistics but is putting forth a philosophical thought experiment. However, as so often in the case of thought experiments caution is called for. Burge explicitly makes a claim about English (the actual community is said to be our normal English speaking community) and much of his argumentation consists in appealing to our intuitions about the English language. If Burge were simply stipulating that, in a possible language, "Arthritis afflicts the joints only" expresses a conceptual truth, then he could not rest his argumentation, as he now does, on intuitions about our actual linguistic practices.

Let us now see what this implies for individualism.

4. Individualism Reconsidered

The individualist, it was suggested above, seems faced with a dilemma: Bifurcation of content or fragmentation of reference and concepts. It should now be clear that the dilemma can be escaped by questioning the assumption underlying the thought experiment, i.e. the assumption that the particular statements appealed to (such as "Arthritis afflicts the joints only" and "Arthritis is caused by calcium deposits", etc.) express conceptual or necessary truths.

Once this assumption is questioned, the individualist can grant Burge the first step of the thought experiment, thereby avoiding referential and conceptual fragmentation, and yet resist the anti-individualist conclusions. That is, she can grant Burge that the speaker uttering "I have arthritis in my thigh" expresses a belief about arthritis and thus makes an error, but by rejecting the assumption that the error is in any sense conceptual she prevents the conclusion that "arthritis" must have a different meaning in the counterfactual scenario. What justifies ascribing the standard concept to the patient is not his reliance on the community experts but the fact that he, by and large, uses the word "arthritis" correctly, i.e. he uses it to express a large number of true judgements. (14)

This is not to say that there can be no such thing as "linguistic" disagreements. One need not subscribe to the traditional empirical/non-empirical dichotomy in order to grant that some deviances from the community practice are so radical that the speaker must be said to have misunderstood the community concept. Burge himself provides a good example of such a case. He considers a speaker who thinks that "orangutan" applies to a kind of fruit drink. (15) Such a speaker would use the word "orangutan" in radically deviant ways (saying things like "Pass me the orangutan" at the breakfast table, and encouraging you to buy more "orangutan" at the store), and so it is justified to conclude that she has failed to grasp the concept of an orangutan.

However, allowing for these kinds of cases does nothing to support Burge's anti-individualism. Instead, the individualist should simply insist that deviances of this type show not that the individual has an incomplete understanding of the community concept, but that she has no understanding of it at all and, consequently, cannot be attributed the standard concept. Rather, the individual's word "orangutan" should be interpreted as expressing the concept orange fruit drink (say). In fact, Burge himself agrees that this is so. If a speaker thinks that "orangutan" applies to a kind of fruit drink then "we would be reluctant, and it would unquestionably be misleading, to take his words as revealing that he thinks he has been drinking orangutans for breakfast the last few weeks." (16)

The individualist's strategy should therefore take the following form: First, she should question the uncritical reliance on the empirical/non-empirical distinction and the idea that isolated statements such as "Arthritis afflicts the joints only" express conceptual or necessary truths. Second, she should insist than when indeed an individual's use reveals that she has misunderstood a word in the community language, she must be reinterpreted. Since this will only happen in the case of radical deviances, there is no risk that normal disagreements imply reinterpretation and so the threat of referential and conceptual fragmentation is prevented.