Is externalism about content compatible with our having privileged, first-person access to the contents of our own mental states?

I. Introduction

The question of whether externalism about content is compatible with the idea of privileged access to our own mental states has provoked much debate in the recent philosophical literature. The argument against the compatibility of these views is typified by the accounts offered by McKinsey (1991), Brown (1995) and Boghossian (1997), who claim that privileged access would appear to grant us a priori knowledge of the external world simply by reflecting upon the contents of our own thoughts. In this essay, I will show how the account of self-knowledge advocated by Tyler Burge (1988) may be used to overcome this type of incompatibilist objection and combine aspects of both views into a single coherent account. In particular, I will argue that the incompatibility arises from a misconception about the epistemic nature of self-knowledge, along with an incorrect model of how introspection works. Once these matters are resolved then there is no reason why the two doctrines should be considered incompatible. Before outlining Burge’s account of anti-individualism and privileged access, however, it will be useful to provide some background as to the nature of the problem.

II. Externalism About Content

The argument for externalism about content is largely based on a series of thought experiments proposed by Hilary Putnam (1975) in favour of semantic externalism, i.e. externalism about meaning. These were later developed by McGinn (1977) and Burge (1979) to include other forms of mental content and the importance of social and historical factors, respectively (Lau 2002: §2). Such experiments typically involve a protagonist who is in a particular internal physiological state and their doppelgänger, or twin, who has the same internal state, but to whom we would nevertheless attribute a different belief (thought, desire, etc.) based on the differing external conditions that obtain in her world. Thus, two physically identical individuals are said to be in different mental states, which are partly or wholly (depending upon the strength of externalism that is being endorsed) individuated by external factors, rather than those that are internal to the subject as had previously been assumed.1 This type of externalism causes a problem for the notion of privileged access — the idea that we have direct and certain knowledge of our own thoughts, beliefs, and so on — because we appear to have such knowledge on the basis of internal reflection, or introspection, alone. However, if our thoughts are partly individuated by what lies in the external world then how can we know what we are thinking without also having to carry out an empirical investigation of the world? It seems that the externalist must either deny that we have authoritative knowledge about the contents of our own thoughts, or give up the notion of externalism about mental content. On the surface, at least, it appears that externalism about mental content and privileged access are incompatible.

Burge (1988) seeks to overcome this conflict by offering an account of self-knowledge that purportedly accommodates externalism about mental content. His account revolves around the following four claims, each of which I shall consider throughout the rest of this essay.

1. 1.Our thoughts are at least partially individuated by external physical objects and/or the shared practices of a community of language users.
   

2. 2.Thought has a reflexive, ‘self-referential’ nature, which is directly accessible to consciousness.
   

3. 3.It is not necessary to know the enabling conditions of a thought in order to know the contents of that thought.
   

4. 4.Knowledge of one’s own thoughts inherits its empirical component from the concepts that these thoughts employ.
   

The first of these is the familiar externalist claim that in order to know what someone is thinking it is also necessary to know something about their environment. This may be because two speakers belong to linguistic communities that attach a different significance to the use of the same term. Alternatively, the physical environment of two speakers may differ in ways that they are unaware of, but that are nevertheless sufficient to alter the ‘content’ of their thoughts, causing them to refer to different things. The standard example of this is Putnam’s ‘Twin Earth’ scenario (op. cit.) where two individuals — call them P1 and P2 — inhabit physically identical environments with one important difference: the chemical composition of water. In P1’s world, water consists of the usual H2O, but in P2’s world, water has a different chemical formula, although is otherwise indistinguishable from ordinary water; i.e. it is colourless, odourless, potable, and so on. I shall call these two substances W1 and W2, respectively. Neither P1 nor P2 has any knowledge of chemistry or other means of identifying the kind of water that they come into contact with. To them it is just ‘water’ and they are unaware that there are two kinds, or of the existence of their twin. From a third-person perspective, we would say that whenever P1 is thinking about water then her thoughts are actually about W1 (H2O), whereas whenever P2 thinks about water then she is thinking about W2. As the two substances are not identical then the two individuals must be thinking different thoughts, despite them being in internally identical states. Consequently, their thoughts must be individuated by external factors.

Such examples are open to the following objection. If P1 and P2 were to switch places, then P1’s thoughts about water would now refer to the W2 in the world where she now finds herself, and vice versa for P2. It could be argued that rather than applying only to substances of a particular chemical composition, such as H2O, both individuals share the same concept of water, which applies equally to W1, W2 and any other water-like substance, irrespective of its chemical makeup. On this view, the fact that the term ‘water’ happens to pick out W1 or W2 in a particular world is a contingent feature of its usage rather than a necessary condition for its application. However, in the case of a substance or ‘natural kind’ concept like ‘water’ it may be assumed that the referent can be identified by some underlying structural or other unique property, even where this structure or property is unknown to the thinker (Putnam op. cit.). Thus, both P1 and P2 might agree that their terms for ‘water’ referred to two different substances if they later became aware of the differing chemical composition of W1 and W2, even though this difference was not apparent or even relevant to their earlier thoughts and actions. In the case of natural kind concepts, it seems at least plausible that P1 and P2 are having different thoughts when they think about water, and that the content of their thoughts is individuated by aspects of their environment, as Putnam and Burge suggest. In order to see how this could conflict with the notion of privileged access, we must turn to consider the epistemic nature of introspective thought.

III. What the Externalist Can Know A Priori

McKinsey (1991), Brown (1995) and Boghossian (1997) argue that the combination of externalism about mental states and privileged access would seem to grant us a priori knowledge of the external world simply by reflecting upon the contents of our own thoughts. One version of their argument goes like this:

1. 1. 1.If I have the concept water, then water exists.
      

   2. 2.I have the concept water.
      

   3. 3.Therefore water exists.
      

2. Boghossian: 165
   

If, as privileged access suggests, (1) and (2) are knowable a priori then the conclusion must also be knowable a priori, but this is not the case as (3) is an empirical fact and can only be known a posteriori. The compatibilist can respond to this in one of two ways. Firstly, by denying that thinking about water necessarily entails that water exists. For example, it could be that ‘water’ turns out not to be a natural kind concept after all, or that the thinker is a member of linguistic community that has a concept of water even though no actual water exists — Boghossian’s so-called ‘Dry Earth’ scenario (Boghossian: 170). In either case, one cannot know which of these conditions obtains without further empirical investigation, thus accounting for the a posteriori nature of the conclusion (McLaughlin & Tye 1998: 372; Brueckner 1992: 116). Of course, if water is a natural kind concept then my thoughts about it would constitute knowledge of the external world. However, I cannot know this with any certainty on the basis of a priori reasoning alone.

The second response to the above argument would be to deny that self-knowledge is a priori in the first place. Statements are normally classed as a priori in virtue of them being knowable in the absence of any material facts, or by logical analysis of the meaning of the terms involved. However, according to Burge’s second claim (p. 2), it is not from analysing our thoughts that we know what we are thinking, but from thinking them. On this view, in much the same way that perceptual evidence is considered to be empirical in nature, thinking itself constitutes a kind of experience. Therefore, rather than categorising everything we can know ‘just by thinking’ or in the absence of any external evidence as a priori, we should consider knowledge of our own thoughts and inner life to be a form of empirical evidence, and therefore a posteriori (cf. Brueckner 2000).2 This renders the above argument unproblematic, as there is no longer a discrepancy between the epistemic status of its premisses and conclusion, as both are only knowable a posteriori.

However, even if we accept that I cannot know whether water is really a natural kind concept, and that introspective knowledge is not a priori, it still seems that privileged access is able to grant me unwarranted knowledge of the world. If I can know what I am thinking, and what I am thinking is at least partly determined by my environment, then I must be able to know something about my environment just by reflecting upon the contents of my own thoughts ‘from the armchair’, so to speak. In order to see why this is not necessarily the case, we must examine Burge’s account of introspection and self-knowledge.

IV. Thought and Self-Knowledge

Burge’s account of self-knowledge is distinctive in that he rejects the Cartesian model of perception as a basis for how we come to know the contents of our own thoughts (Burge op. cit: 654). In perception, there is a clear separation between subject and object, with the properties of the latter being detected via the senses, as illustrated in Figure 1. This creates the possibility of error and misperception due to some intervening factor or poor environmental conditions, such as bad lighting, extraneous noise, and so on. Consequently, perceptual experiences may be non-veridical in nature.

Figure 1: The Cartesian model of perception and introspection

Although it is tempting to think of introspection as following the same pattern, with ‘thoughts’ as the objects that are detected by some sort of inner sense (namely introspection), this picture is, according to Burge, ‘deeply misconceived’ (ibid: 660). Whereas perception is concerned with objects that are independent from the observer and accessible to more than one subject (i.e. they are objective), thoughts exhibit neither of these characteristics (ibid: 657). In place of the perceptual model, Burge claims that thoughts have a dual nature that encompasses both content and evidence of the thought itself (Figure 2). This is most clearly illustrated in the case of so-called ‘Cogito thoughts’ (Burge: 1996), such as ‘I am now thinking about apples’. Here, the thought explicitly references both itself (‘I am thinking’) and its external content (apples). However, Burge extends this principle to apply to all thoughts, which are said to have a ‘self-referential’ quality (Burge 1988: 659–60) consisting of (at least) two components: (i) the thought itself, and (ii) the second-order awareness that one is thinking it. This second-order aspect of thought is what supposedly gives us direct access to the contents of our thoughts, and because each thought is already a part of consciousness there is no possibility of the kind of errors that can occur in the case of perception. Thoughts without this second-order component would simply be inaccessible to consciousness and therefore be unconscious thoughts to which we had no access. Thus, rather than being objects that we can observe either more or less veridically, thoughts are directly accessible to consciousness and, as such, constitute evidence of their own content. Consequently, one cannot be mistaken about what one is thinking.

Figure 2: Burge’s second-order model of self-knowledge

Burge’s third major claim (p. 3) is that it is not necessary to know the ‘enabling conditions’ of a thought in order to know its contents (ibid: 654). By this he means that the concepts and empirical knowledge that is referenced by each thought is presupposed by it rather than being part of the thought itself. I can therefore know that I am thinking about apples without knowing whether there really are such things (although I may remember seeing some and be inclined to believe in their continued existence). Instead, the empirical content of my concept of ‘apples’ is inherited by the thought, which is Burge’s fourth and final claim, illustrated in Figure 3.

Figure 3: Burge’s second-order model of self-knowledge

Even though the knowledge that we are thinking about apples cannot give us any certainty about our other beliefs about apples, Burge thinks that it is nevertheless sufficient to ascertain the content of the thought. This means that our knowledge of our own mental states relates primarily to the concepts that we employ, rather than any knowledge of the things to which these concepts apply, which is obviously empirical in nature. But if we only know that we are thinking about our concept of an apple and not about apples per se, does this not amount to a rejection of externalism — or at least a form of relativism — with respect to the individuation of thoughts? Burge’s response is that this concern is a result of confusing the first- and third-person points of view. It is only when we move from a first-person ‘privileged’ perspective to a third-person ‘explicative’ one that we need to take into account the external factors that individuate thought (ibid: 662). Thus, my direct experience of my thoughts is sufficient for me to know what I am thinking. I do not need to know how my thoughts are individuated, or to be able to explicate or defend this knowledge against all possible sceptical doubt  (ibid: 661, fn. 9).3 Consequently, Burge is effectively an internalist about the individuation of our own thoughts (i.e. from a first-person perspective), but an externalist about the individuation of the thoughts of others (from a third-person perspective). This allows him to maintain that privileged access is compatible with externalism, as the two doctrines no longer overlap.

V. Conclusion

The account of externalism and privileged access that Burge (1988) offers successfully combines aspects of both doctrines into a unified and coherent view of how we can know the contents of our own thoughts whilst also holding that mental content is — at least in the third-person case — externally individuated. Incompatibilist arguments of the type given by McKinsey, Brown and Boghossian may be dissolved either by denying that the subject has sufficient a priori knowledge to affirm their premisses, or by classifying self-knowledge as a posteriori or ‘introspective’ in nature, rather than a priori as is often assumed to be the case. Furthermore, such arguments fail to engage with the account of self-knowledge that Burge describes in which thought is directly presented to consciousness and derives its empirical component from our existing a posteriori knowledge and concepts, and not from our knowledge of thoughts themselves. By separating out these two aspects of self-knowledge and denying that it is necessary to know how one’s own thoughts are individuated in order to know what they are, Burge is able to overcome the main incompatibilist objection to these views. Consequently, there is no reason to doubt that externalism and privileged access are indeed compatible despite their appearance to the contrary.

——————

Bibliography

Boghossian, Paul A. 1997: ‘What The Externalist Can Know A Priori’, Proceedings of the Aristotelian Society vol. 97, pp. 161–75.

Brueckner, Anthony 1992: ‘What an Anti-Individualist Knows A Priori’, Analysis vol. 52, pp. 111–8.

—————  2000: ‘Externalism and the A Prioricity of Self-Knowledge’, Analysis vol. 60, pp. 132–6.

Burge, Tyler 1979: ‘Individualism and the Mental’. In French, Uehling and Wettstein (eds.), Midwest Studies in Philosophy IV. Minneapolis: University of Minnesota Press, pp. 73-121.

—————  1988: ‘Individualism and Self-Knowledge’, Proceedings of the Aristotelian Society vol. 85, no. 11, pp. 649–63.

—————  1996: ‘Our Entitlement to Self-Knowledge’, Proceedings of the Aristotelian Society vol. 96, pp. 91–116.

Brown, Jessica 1995: ‘The Incompatibility of Anti-Individualism and Privileged Access’, Analysis vol. 55, pp. 149–56.

Lau, Joe 2002: ‘Externalism About Mental Content’, The Stanford Encyclopedia of Philosophy (Fall 2004 Edition), Edward N. Zalta (ed.), http://plato.stanford.edu/archives/fall2004/entries/content-externalism.

McGinn, Colin 1977: ‘Charity, Interpretation, and Belief’, Journal of Philosophy, 74, pp. 521–535

McKinsey, Michael 1991: ‘Anti-Individualism and Privileged Access’, Analysis vol. 51, pp. 9–16.

McLaughlin, Brian P. & Tye, Michael 1998: ‘Is Content Externalism Compatible With Privileged Access?’, The Philosophical Review vol. 107, no. 3, pp. 349–80.

Putnam, Hilary 1975: ‘The Meaning of “Meaning”’, Philosophical Papers, Vol. II: Mind, Language, and Reality. Cambridge: Cambridge University Press.

—————  1982: Reason, Truth and History. Cambridge: Cambridge University Press.