Is Tyler Burge correct in claiming that the content of one’s mental states can only be determined in relation to one’s environment?

By: Urte Laukas, MSc Mind Language and Embodied Cognition 2015, University of Edinburgh

  1. Introduction

In a number of papers Burge (1976, 1982, 1986) put forward and defended what I will call the externalism about mental content thesis, which holds that mental states are individuated partly in connection with an organism’s environment. The initial push for the thesis came from Putnam’s (1975) Twin Earth thought experiment, which sought to establish semantic externalism, i.e. the view that the meaning of certain kinds of words is not determined by a speaker’s associations and internal state but also speaker-independent facts about the referent. In this essay I will (a) delineate the Twin Earth thought experiment in its original form, (b) present the way Burge extended the thesis of semantic externalism to mental content, (c) outline an objection concerning Burge’s implicit conception of the word-concept relation, (d) consider possible problems with the alternative picture and attempt to answer the challenge. Provided one accepts all the assumptions that Burge presupposes in order to make his case, it is a forceful argument indeed. However, I intend to demonstrate that they do not all seem to be unquestionably true. I will conclude that this may pose a problem for Burge’s position.

  1. Twin Earth

The setup of the thought experiment is the following: Putnam (1975: 139-143) pictures two planets identical to each other in all respects, except for one – whereas the liquid we call ‘water’ has the chemical composition H2O, the otherwise indistinguishable liquid on Twin Earth is composed of XYZ. He then imagines Oscar1 and his Twin Earthian counterpart Oscar2 in the year 1750 before the invention of modern chemistry. It is stipulated that the two Oscars are exact duplicates of each other and this includes them sharing all associations associated with the term ‘water’ and being in the same mental state (which is what Burge is going to counter). Nonetheless, since Oscar1 is only acquainted with H2O, when he talks about ‘water’, it refers to H2O, whereas for the same reason Oscar2’s uses of ‘water’ pick out XYZ and not H2O. This is what Putnam (1975) thinks most people’s intuitive judgments would be. The status of mental content is not discussed explicitly in relation to the original formulation. Yet a number of ideas expressed in Putnam (1975) potentially seem to have ramifications for the conceptualisation of psychological states, and I think Burge is right to press semantic externalists on that.

  1. Externalism about mental content

My exposition of Burge’s position cannot be exhaustive due to lack of space, some potentially relevant issues will have to be bypassed, yet the core of the argument should become clear. Burge (1982: 85-87) co-opts the Twin Earth thought experiment pretty much in its original form, although in his version the protagonist is called Adam. He also thinks there is no need to turn the time back to the year 1750, as long as Adam and his Twin Earth counterpart Adamte are not familiar with the chemical structure of the relevant substances. And most importantly, he denies that it is possible for Adam and Adamte to be in identical mental states in the scenario as presented.

Before I can proceed to explain the reasoning behind the claim, it is important to note two presuppositions[1] that seem to me to be required for Burge’s argument to take off as well as clarify some terminological issues. Firstly, the view appears to be assuming that thoughts are in essence not unlike propositional attitudes. Propositional attitudes serve to indicate what cognitive relations (such as belief, desire, hope, so on) people bear to propositions. For instance, ‘Adam believes that P’ (where P stands for the proposition ‘Water quenches thirst’ or ‘Water is a colourless odourless liquid’, etc.) would be an example of a propositional attitude. Burge (1979) also appears to use ‘think’ as a prepositional attitude verb, which may suggest that is indeed his conceptualisation of thought in general. Burge (1986: 8) prefers to use the term ‘intentional content’[2] rather than ‘propositional attitudes’, since he thinks that not all the relevant intentional mental states are explicitly propositional. However, whether thoughts as such are strictly speaking propositional or not may not matter all that much. It seems to me that in order for the argument to work, Burge has to assume that thoughts contain certain language-like elements, which refer in a language-like way. In this essay that is going to be all that is required for something to count as a propositional attitude. Sticking to using this term can be further justified by the fact that in Burge (1979) it is used freely in the relevant contexts.

A second presupposition appears to be that concepts refer in the same way as words do, which is in turn the subject matter of philosophy of language. The term ‘concept’ is here used in a relatively non-technical intuitive sense as a mental representation that has some kind of mapping onto words. The implication seems to be that if one accepts that the arguments for semantic externalism put forward by Kripke (1980) and especially Putnam (1975) are cogent for linguistic reference, then one is bound to accept Burge’s (1979, 1982) elaboration of them regarding the nature of mental states. The only way such a direct link might be possible is if concepts, which are mental entities, are taken to be language-like in their reference. In that case there would be no disagreement between the way propositional attitudes function in the mind and their being reported in natural language (such as English).

To come back to Burge’s (1982) development of the Twin Earth thought experiment, the argument is essentially the following: water and twin-water are two distinct natural kinds. Adam’s use of ‘water’ refers to water, whereas when Adamte uses exactly the same expression ‘water’, the word refers only to twin-water (up to this point it is all still in line with Putnam (1975)). If one is prepared to grant that the word ‘water’ refers to two different natural kinds in the two linguistic communities, Adam and Adamte’s concepts must differ as well, irrespective of the fact that all relevant experiences and associations are shared between them. As those concepts feature in the two speakers’ propositional attitudes, i.e. thoughts, their mental states must depend on what the relevant words refer to in the environment. Thus the conclusion is that mental content is individuated in relation to one’s natural and social environment. Importantly, Burge (1979) wants to insist that if sound[3], his line of argument would not be restricted to natural kind terms denoting scientific kinds in nature, but would be applicable to the more mutable social domain as well (e.g. terms such as ‘arthritis’, ‘contract’, so on).

  1. The word-concept relation

I want to suggest that there might be a potential problem with externalism about mental content, which arises from conflating the nature of two phenomena that should be kept apart: words and concepts. Hence my contention is going to concern the second assumption, which, as I have shown, seems necessary for the argument to proceed. Burge (1979: 116-117) can be read as quite explicitly acknowledging that the considerations propelling him to equate language and mental content in such a way are at least partly methodological. More specifically, Burge (1979: 116) writes: ‘I assume that a primary way of achieving theoretical understanding is to concentrate on our discourse about mentalistic notions’ and a few sentences later he says that ‘there is a methodological bias in favour of taking natural discourse literally, other things being equal.’ Since ambiguity, ellipsis and suchlike are given as examples of what should be avoided when theorising about discourse, I take it to be the case that these considerations are also meant to apply to word meaning, not just sentential structure. I think this may well be a useful principle in linguistic theorising about semantic meaning when language is viewed as something akin to a formal symbolic system. In order to construe language in such a way it is necessary to abstract away from its use by specific individuals. However, what I find rather less convincing is that that same global perspective should be kept when the locus of interest comes back to the individual and especially something like mental representation.

Importantly, I am not implying that I have a fully developed alternative account of what concepts are and how they operate and a decisive argument to prove it once and for all. What I wish to do in this paper is rather more modest. The aim is to challenge the unquestionability of the (near-)identity relation between words and concepts in terms of how they may refer. If my line of thought is at all plausible, independent grounds other than methodological simplicity should be sought to justify the assumption required for Burge’s position to be convincing.

My main proposal is that communication does not require absolute convergence between speakers’ concepts. As long as they bear sufficient similarity to each other or, in other words, approximate a highly idealised standard in a given linguistic community, communication should be possible. Admittedly, it would be rather difficult to specify what exactly would count as ‘sufficient similarity’ in this context, but I do not deem that to be a particularly detrimental objection. It is reasonable to assume that most, if not all, speakers want to be able to communicate successfully. It is thus likely that they would constantly strive for their concepts to converge with those of speakers around them. Such a process would tend towards averaging out conceptual discrepancies in a given linguistic community, although without ever quite achieving it[4]. The situation could be rather different for technical (legal, scientific, etc.) vocabulary, in that many people will have very little knowledge about the specifics of certain terms. Even that I do not regard as a major problem. It could be argued that being able to factor in people’s lack of precise understanding as part of their mental content is an advantage of the theory.

It is certainly useful to treat other speakers’ utterances as if they conveyed a fixed, detailed and definite content. And much of the time, for all intents and purposes, this is in fact accurate, since the differences between proficient English speakers’ understanding of ‘chair’, ‘table’ and many other words will be negligible. Several times in the paper Burge (1979: 107-111) makes it clear that he thinks words can be regarded as having fairly unambiguous well-established meaning that all competent speakers would know and agree on. Individual uses of certain words may diverge from the norm but there seems to be no doubt that such a norm exists. Burge (1979: 110) gives an example of a dispute regarding whether a particular object can be said to be red or not. But, he maintains, the people with an inadequate understanding of what ‘red’ means would give in ‘when other speakers confidently correct them in unison’ (Burge, 1979: 110). However, a little later Burge (1979: 112) writes: ‘What I have called ‘partial understanding’ is common or even normal in the case of a large number of expressions in our vocabularies.’ If that is true, it is not clear to me who could be taken to have the right conceptualisation of ‘red’ and on what basis. One could suggest the deciding factor is what view the majority holds, yet that might prove to be not an easy matter. Communities tend to split in their use of all kinds of words, minority interpretations shift to be majority interpretations, which is the prime driving force behind language change. Having a picture of different speakers’ concepts as non-fixed and non-identical would also provide a workable mechanism for phenomena like historical language change and sociolinguistic variation.

  1. Possible problems

It would be reasonable to suggest that, even methodological concerns to do with simplicity aside, the burden of proof when making such a claim still lies with me. I suppose part of it is going to be appealing to people’s intuitive judgments again, as is often the case in philosophy. One may find the conclusion of Burge’s argument unintuitive, i.e. the claim that one’s mental states could have different content purely as a result of a difference in the environment, irrespective of the fact one is not even aware of it. Yet reading through such an admittedly well-crafted case, there is seemingly no choice but to accept the conclusion. My suggestion is intended to point out what may have possibly gone amiss in the course of the argumentation. In the end adjudication is probably still going to have to rely on the relevant intuitions.

A different issue one may have is the following: one may think that the ‘idealised standard’ may be regarded as a mere terminological quibble not that different from there actually being a norm. It is true that in many everyday situations the speakers’ conceptualisations are not going to diverge significantly and hence could be treated as (nearly) identical. However, the cases that Burge (1979, 1982) is focusing on are precisely the ones where there are such divergences. Suppose one accepts that (private) concepts and (public) words are distinct non-identical entities that may refer in different ways. Then if Adam and Adamte share the same concept (which they would), one can grant that the word ‘water’ refers to two distinct natural kinds on Earth and Twin Earth, and yet still hold that their mental states are nonetheless the same.

  1. Conclusion

I hope to have demonstrated that presupposing (near-)identity of words and concepts regarding reference, as Burge seems to have done, may not be a wholly unproblematic move. This essay is not going to be the final word on the subject. However, what I hope to have achieved here is to highlight the need on Burge’s part for a more direct argument to warrant making the assumption, other than an appeal to methodological concerns regarding simplicity.



[1] The reason I use the term ‘presupposition’ or else ‘assumption’ instead of ‘premise’ in this case is that the relevant claims are not explicitly defended and yet the force of the argument rests on them being true.

[2] An intentional construction is one in which expressions occur obliquely, i.e. one cannot replace a given expression with a different one referring to the same object/person/etc. without affecting the truth value of the whole sentence.

[3] A sound argument is taken to be such that it is valid, i.e. the conclusion follows from the premises, and also the premises are all true.

[4] My proposal may be construed as Wittgensteinian (1953) in outline and I think justifiably so. Due to space restrictions I cannot elaborate on the possible similarities and points of divergence. However, one important difference is that his position was intended as an alternative picture in philosophy of language. I want to remain neutral on whether that is necessarily the best way to conceive of the abstract system that language is when taken on its own terms, i.e. divorced from its use by individual speakers. I deem the two to be separable subject matters.



Burge, T. (1979) Individualism and the mental. Midwest Studies in Philosophy 4 (1), pp. 73-122.

Burge, T. (1982) ‘Other bodies’. In Woodfield, A. (ed.). Thought and Object. Oxford: Oxford University Press, pp. 82-99.

Burge, T. (1986) Individualism and psychology. Philosophical Review 95, pp. 3-45.

Kripke, S. (1980). Naming and Necessity. Cambridge: Harvard University Press.

Putnam, H. (1975) The meaning of ‘meaning’. Minnesota Studies in the Philosophy of Science 7, pp. 131-193.

Wittgenstein, L. (1953) Philosophical Investigations. Anscombe, G. E. M., Rhees, R. (eds.), Anscombe, G. E. M. (trans.). Oxford: Blackwell Publishing.

Share your thoughts on this!

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s