In recent years, philosophers have debated whether the special authority that all subjects have on their mental states can be accommodated with the externalist theory of mental content. On the one hand, we have privileged access to our minds. It seems that we know the thoughts we have better than anyone else. For instance, while I know that at this moment I am thinking that I desire to have a holiday, it seems that another person can know my desire only if I provide him with some evidence. On the other hand, according to externalism, the contents of a subject's thoughts are determined by facts "external" to her, particularly facts about her social or physical environment.⁽¹⁾ Compatibilists think that they can provide a notion of privileged access that squares with their own preferred form of externalism.⁽²⁾ Conversely, anti-compatibilists argue against such reconciliation. ⁽³⁾

In this paper I discuss the rejection of compatibilism based on the argument that this doctrine implies the implausible conclusion that a subject should have knowledge of empirical facts without any observation of the external world.⁽⁴⁾ This criticism focuses on the epistemic status of externalism. The basic idea is that externalism is a philosophical theory that can be known independently from any knowledge of empirical facts. From this it is possible to infer that the subject who embraces compatibilism provides an example of an individual who has an implausible knowledge of the external world. Thus, what the compatibilist can know constitutes a reductio ad absurdum of her own doctrine.

I will proceed as follow. Firstly, I shall introduce a certain notion of privileged access and the form of externalism proposed by Hilary Putnam. Then, I will show how the anticompatibilist argument works in the case of these notions of privileged access and externalism.

The different notions that are implicit in the general concept of privileged access can be individuated using the classical analysis of knowledge as justified true belief.⁽⁵⁾ Often, philosophers have said that, while a person can be wrong about another subject's mental states, she cannot commit this error about her own mental states. This infallible access can be expressed by saying that if a certain subject believes that she has a certain mental state then she is justified in having that belief, and that she actually has the mental state that is the object of her belief.

Other thinkers have maintained that while we can ignore another subject's mental states, this is not the case with our own mental states. Usually, this epistemic feature is attributed to the fact that mental states are self-intimating. Using the analysis of knowledge given above, we can say that if a subject has a certain mental state, then she believes that she has it and this belief is justified.

There are reasons that can lead us to reject infallibility and self-intimacy. In particular, it seems that a decisive reply can be elicited in the case of propositional attitudes like beliefs and desires. The psychoanalytic tradition shows that we are prone to hide certain of our desires and beliefs. In addition, we attribute to one beliefs that one does not have.⁽⁶⁾ Infallibility and self-intimacy are not the only possible conceptions of privileged access however.

Another way to characterise privileged access is based on the nature of the justification we have in endorsing our beliefs about our mental states. A person is privileged in having immediate or direct knowledge of his own mental states.⁽⁷⁾ Another person can know what a certain subject is thinking only by means of something else, for example, observing his behaviour, using certain technical apparatus, or, if we describe the situation in a more abstract way, using some signs or criteria. It seems that the owner of these mental states, conversely, does not need to base her knowledge upon perceptual or empirical observation.

The notion of direct knowledge needs to be clarified in two fundamental respects. First of all, we have to make clear which mental states are directly knowable. Not all the mental states we have can be known directly. We have already seen that self-intimacy does not apply to all mental states. If I can have certain mental states without knowing them, then it seems that I cannot have a direct knowledge of them. In addition, it seems that we have to limit this kind of knowledge to those states that are currently present to the mind. It seems that if I want to justify my occurring beliefs about mental states that were present at an earlier time, I need to be certain that my memory is reliable. Second, not all justified beliefs I have about my mental states are direct. For example, I can realise what my real desires are after an elaborate process of self-analysis that I can perform only possessing the knowledge of a certain theory of the functioning of the mind.⁽⁸⁾

In the debate I consider, the notion of privileged access introduced above is said to provide an a priori knowledge.⁽⁹⁾ Although I follow this usage for reason of uniformity, it is important to indicate that, here, the expression a priori has a special meaning. Usually a certain subject is said to know a priori a certain proposition if and only if her knowledge of it derives from her understanding the concepts involved in that proposition and her knowledge of the rules of logic. Thus, mathematical and logical propositions are known a priori. However, in this context of the access to the content of our own thoughts, the notion of a priori is taken in a wider sense to indicate a type of knowledge that is independent of the observation of the world. ⁽¹⁰⁾ In Michael McKinsey's words:

A person knowledge's is a priori if and only if it would remain knowledge even if the person were radically deceived in his assumption and inferences concerning the existence and nature of the physical world that is external to his mind (McKinsey 1987: 3).

Having stated the kind of privileged access and of a priori knowledge here adopted, let us move to the formulation of externalism. The externalist's statement that mind is essentially linked to the world has been advanced to solve a series of interrelated philosophical problems concerning the nature of linguistic meaning and mental content. Here we will consider the version of externalism proposed by Hilary Putnam, who was interested in giving an account of semantical properties of terms and concepts for natural kinds. ⁽¹¹⁾ In general, objects that have some relevant property in common that can be used by scientists to predict and explain their behaviour constitute a natural kind.

Putnam has developed his theory of the semantic properties of natural kinds terms in opposition to the traditional picture of meaning that originated with the work of Gottlob Frege. Notoriously, in the Fregean theory, general terms have two fundamental semantical features. The extension of a general term is the class of things to which that term applies, while the intension constitutes the way in which this class is presented. According to Putnam, this traditional picture of meaning leads to the erroneous conclusion that terms refer to their extensions in virtue of some internal psychological acts. In his reconstruction this result derives from two unwarranted assumptions.⁽¹²⁾ Firstly, the intension of a certain term depends on the occurring in the subject of a particular psychological state. Let us call this assumption the individualistic thesis.⁽¹³⁾ Secondly, intensions determine extensions.

Putnam introduces his alternative theory of the meaning of natural kind terms as an adequate response to a celebrated thought experiment. Twin-Earth is a planet differing from Earth only in the fact that the substance whose chemical structure is XYZ replaces all the occurrences of H₂O and is indistinguishable from it in all ordinary circumstances. Oscar lives in Earth; his molecular duplicate Toscar lives in Twin-Earth. ⁽¹⁴⁾ When Oscar uses the term 'water' he refers to the substance H₂0, Toscar to XYZ. Thus, we can say that the term 'water' has different extensions if used by Oscar or by Toscar. Let us examine why this situation represents a problem for the traditional theory of meaning.

From the Twin-Earth thought experiment it is possible to draw a reductio of the traditional theory of meaning in the case of natural kinds terms. Toscar and Oscar are exact molecular duplicates; therefore, we can also assume that they have the same psychological processes going inside their heads.⁽¹⁵⁾ Given the individualistic thesis, the intensions associated with their uses of the word 'water' are identical. On the other hand, from the converse of the principle that intensions fix referents, the fact that on Earth the referent of 'water' is H₂O while on Twin-Earth it is YXZ, we can derive that the intensions are different. Hence, we have a contradiction. Let us consider the way in which externalists can escape this conclusion.

Putnam denies the individualistic thesis and assumes that the meaning of natural kind terms is determined by the external conditions that constitute their referent.⁽¹⁶⁾ Thus, the concepts expressed by Oscar and Toscar when they use the term 'water' differ, because H₂0 and XYZ are different substances. This resolves the contradiction inferred from the Twin Earth thought experiment.

Externalism is a more general theory than Putnam's proposal on natural kind terms. First, externalism is usually construed as a thesis of the individuation of mental states. This means that the external conditions not only determine the identity conditions of mental states, but also guarantee their existence. Putnam shows how changes in the environment produce changes in the concepts we have. Other philosophers have added that it is not possible to have a certain concept without the existence of the external conditions that fix its content. ⁽¹⁷⁾ Second, Tyler Burge has developed Putnam's proposal in more detail, extending his theory by arguing that it is not only that semantical features of words, but also intentional psychological properties depend on external conditions.⁽¹⁸⁾ In addition, he has argued that facts involving the linguistic practices of members of the subject's community are part of the external conditions that determine mental content. Finally, he has maintained that externalism applies, not only to natural kinds terms and concepts, but also to all types of expressions and concepts.

I discuss the anticompatibilist argument only in the case of externalism in relation to natural kind concepts. It is commonly assumed that the content of our beliefs depends on the content of the concepts that occur in them. Thus, if we can defend the incompatibility between the requirement of externalist criteria of individuation of our concepts and our privileged access to them, then this result can be extended to the knowledge of the content of beliefs where these concepts occur.

The goal of the argument of the anticompatibilist is to show that a compatibilist has to accept that knowledge of empirical fact is obtainable without observation. The argument is developed in two steps. The first premise is that the compatibilist, defending a form of privileged access, has to admit that a certain subject could know a priori that she has a thought involving a certain concept. For instance, Oscar knows without any recourse to experience that he has the concept water. The second premise states that for the compatibilist, accepting that externalism is true, the subject knows a priori that if she has a certain concept then the external conditions that individuate this concept hold. In our example, Oscar knows a priori the conditional that if he has the concept of water, then water exists. From these two premises it follows that Oscar knows a priori that water exists.⁽¹⁹⁾ Given that the latter is an unpalatable conclusion, we have a refutation of compatibilism.

The first criticism that can be moved to this argument is that it is invalid because it is based on a disputable logical principle. According to many philosophers, if someone knows a priori that a certain proposition p implies the proposition q and she knows a priori the proposition p then she should not know a priori the proposition q. Using a logical terminology, we can say that a priori knowledge is not close under logical implication. A plausible counterexample to substantiate this non-closure can be the following inference: I know a priori that I if have a toothache, then teeth exist, I know a priori that if I have a toothache, therefore I know a priori that teeth exist.⁽²⁰⁾ But it is seems clear that the conclusion is not a priori. Let us consider the replies the anticompatibilist can produce.

It appears that there are at least two possible ways to meet the previous objection. It is possible to defend anticompatibilism showing that all the supposed examples of non-closure actually concern cases of knowledge that is not a priori. In the case of the example provided, Paul Boghossian argues that our beliefs concerning toothaches are grounded on the empirical belief that we have teeth. ⁽²¹⁾ On the other hand, some philosophers have argued that the anticompatibilist argument can be supported by adopting a principle weaker than the closure of a priori knowledge and that is more plausible. In particular, the modal version has been proposed that states that if is it possible that someone could know a priori both p and that p implies q, then it is possible that she could know a priori that q. ⁽²²⁾ These replies are far from conclusive, but show that this attack on the anticompatibilist's argument should be based on a more accurate analysis of the logic of a priori knowledge. Thus, leaving open the question whether the anticompatibilist's argument is invalid, let us move to the analysis of its premises.

The truth of the first premise follows from the fact the compatibilist accepts privileged access. Conversely, the second premise can be challenged. In particular, If the externalist is committed to the existence of a necessary metaphysical relation of dependence between concepts and external factors, it seems that she is not forced into admitting that we should know this correlation a priori. In fact, after the seminal investigations of Saul Kripke, we have compelling reasons to think that it is possible to separate the epistemic notion of a priori knowledge from the metaphysical notion of necessity.⁽²³⁾ Thus, in particular, if the content of a certain concept is necessarily individuated by some external conditions, it is possible to discover this necessary correlation a posteriori.

This seems to be the line of argument followed by Tyler Burge.⁽²⁴⁾ In order to defend the conclusion that: "one can know what one's mental events are and yet not know relevant general facts about the condition for individuating those events" (Burge 1988: 651) he takes as an instructive example Descartes' argument that the mind can exist independently of the body.⁽²⁵⁾ Descartes, from the fact that he could know directly the existence of himself and of his thoughts and doubt the existence of the physical world, deduced that mind was necessarily independent of the world. Burge notices that what Descartes shows is only that he cannot deduce, from the knowledge of his thoughts, the existence of the physical world. ⁽²⁶⁾ But this is compatible with the idea that this thought depends necessarily and essentially on the existence of a physical world. Similarly the knowledge of having certain thoughts can be separated from that of the external conditions that individuate them.⁽²⁷⁾

A way to react to this line of argument is to deny that the relation assumed to hold between a content and its external conditions is metaphysically necessary. For example, it has been argued that this kind of relation renders externalism a trivial theory. ⁽²⁸⁾ In fact, there are many metaphysical necessary conditions for the existence of psychological states. For example, the existence of the parents of a person is necessary for the existence of her thought that "2 + 2 = 4". Thus, this biological fact will enter in the content of this mathematical thought. However this does not seem a strong criticism. Externalist can improve their theories of mental content, just specifying some particular conditions that are relevant for having a certain mental content. However, another line of attack is to show that the nature of the relation between content and external conditions depends on the nature of the theory that states it.

Some authors have noted that if the argument used by the externalist is based on a priori assumptions, the correlation between having a certain concept and its external conditions can be know a priori from the externalist himself. ⁽²⁹⁾ Paul Boghossian, (1997), points out that in the case of Putnam's proposal, the existence of this type of correlation between a natural concept, water, and the substance water can be inferred only if we know that the word "water" is one of these words that allows the Twin Earth experiment to be applied. According to Boghossian, the conditions that a word should satisfy in order to be appropriate for this thought experiment, or as he says to be TE-eligible, can be known independently from any empirical investigations.

Amongst the conditions of TE-eligibility for a certain word there are the assumptions that the user of this word aims to name a natural kind and that she does not know the chemical composition of the substance. ⁽³⁰⁾ These conditions depend on what the subject believes and intends. It seems that the fact that a certain subject uses a word in order to indicate a natural kind depends on her desire to use it in that way. In addition, her ignorance of the chemical composition of natural kinds depends on her beliefs. Thus, we can assume that, given privileged access, she knows a priori that these conditions hold. However, there is a further condition that seems to create difficulties with this argument.

It might be objected that in the Twin Earth argument, a central role is played by the fact that the word water in our world actually refers to a natural kind, namely H₂0. And it seems that this is a fact that the externalists know by using empirical evidence. Thus, the anticompatibilist has to show that it is possible to produce a Twin Earth situation and the relative externalist argument in a way independent from the assumption that, as a matter of fact, the natural kind term we are considering actually names a certain natural substance.

The reply proposed by Boghossian centres on the fact that the externalist position logically entails not only the fact that the content of a concept covaries with external conditions, but also that its existence depends on the existence of these conditions.⁽³¹⁾ In order to prove this, he introduces the case of Dry Earth. Let us assume a case of a possible world where there is no water at all, but it seem to its inhabitants that there is a clear, tasteless and colourless liquid flowing in their rivers and seas. What concept is expressed by the term "water" in this situation? According to Boghossian, there are only two replies.

The first possibility is that the concept water on Dry Earth is given by a cluster of properties which present the substance water in the "collective hallucination" of the Dry Earth inhabitants. For instance, the content expressed by the term water is given by a set of properties like clear, tasteless colourless liquid that can be used to have a bath. Thus, while a substance individuates the concept expressed by the word "water" on Earth, on Dry Earth a compound of properties individuates it. It seems that the variation of the external circumstances can vary the internal structure of a concept. However, the complexity of a concept depends on its syntactical properties, and it appears difficult to assume that syntax supervenes on external conditions. Thus, we have to assume that the concept of water on Dry Earth is an atomic concept. But in this case there can be no such atomic concept, because of the very nature of externalism. On Dry Earth there is no liquid to which the concept water can be applied. Thus, it follows that in the case of Dry Earth, externalism entails that we cannot have a certain natural kind concept. This is equivalent to saying that if we have a natural kind concept an external substance should exist. ⁽³²⁾

So far, we have considered that externalism is an a priori theory that entails, in a way completely independent from experience, that a subject's having a certain concept implies the existence of certain external conditions. The next step is to consider a subject, for example Oscar, who knows the externalist theory. Given privileged access, Oscar knows that he has a certain concept, say water, and this satisfies the first premise of the anticompatibilist argument. In addition, given that he knows externalism, this implies that he knows a priori the correlation between his concept of water and the existence of water. ⁽³³⁾ Hence, the conclusion of the anticompatibilist argument follows.

To sum up, I have considered the tenability of compatibilism in the case of terms for natural kinds. I have shown that a serious problem for this doctrine is created when we realise that externalist reaches his conclusions a priori. In fact, it is possible to infer the implausible conclusion that at, least in the case of a subject who embraces compatibilism, there are persons who can know, a priori, things about matters of fact. It seems that in this case the challenge for the compatibilist is to show that externalism is an empirical theory.

1. Kripke (1980), Putnam (1975), and Burge (1979).

2. We can include among the compatibilists Tyler Burge (1988), Donald Davidson (1984 and 1987), John Heil (1988), Ernest LePore (1990), John Gibbons (1996), and Richard Miller (1997).

3. In any philosophical controversy there is a third possibility given by discussing the assumptions shared by both contenders. Glock and Preston (1995) show how this could be done in this particular debate using some intuitions from Wittgenstein.

4. This type of argument is employed in McKinsey (1991 and 1994), Brown (1995), and Boghossian (1997). However, it is important to note that in the current literature there is another strong challenge to compatibilism that I am not going to address. Some authors have expressed their worries about accommodating externalism and self knowledge using cases of "slow switching" of individuals between the actual world and counterfactual situations where the external conditions that fix the content of their beliefs change. According to this criticism, while the content of their belief changes, as stated by externalism, these individuals fail to recognise it. Thus, they do not have a privileged access to these mental states. This kind of approach has been adopted by Paul Boghossian (1989), Kevin Falvey and Joseph Owens (1994), pp. 111-15, and discussed by Peter Ludlow (1995) and Ted Warfield (1997).

5. This kind of analysis has been proposed by Armstrong (1968) p. 101, see also Alston (1971).

6. See for this kind of argument Alston (1971), pp. 230-31; and Heil (1988), p. 238.

7. Here the key distinction between immediate-mediated knowledge is drawn in a distinctively epistemic way. For different interpretations of the notion of immediacy and why the model presented here is preferable see Alston (1971), pp. 231-233.

8. Heil (1988), p. 242.

9. It seems that this use of the notion of a priori knowledge has been introduced by McKinsey (1991).

10. The fact that the usual notion of a priori cannot be applied to the knowledge of our own thought as been noted by Jessica Brown (1994), p. 149 n1.

11. Putnam (1975).

12. Putnam (1975), pp. 221-222.

13. With regard the first assumption Putnam argues that, despite the famous Fregean anti-psychologism about the nature of intensions (sinne), in his theory a commitment to the existence of a psychological act of grasping an intension is unavoidable. See on this Putnam (1975), p. 222.

14. The identity of these two individuals may be disputed on the basis that while the body of Oscar contains H₂0 the body of Toscar contains XYZ. However it seems difficult to think that this mere difference in the chemical composition of Oscar and Toscar could be sufficient to determine internal differences that justify the fact that Oscar and Toscar have, respectively, thoughts about H₂^{O and YXZ.}

15. From the point of view of the mind-body relation, this argument works at least assuming the thesis of supervenience of mental states over physical states.

16 An alternative approach is to assume the individualist thesis and abandon the idea that intensions are necessary and sufficient for fixing the reference of terms. This is the line followed by those who assume the so-called "narrow content" theory of meaning.

17. The assumption that the existence of external factors is necessary for the existence of mental states is made in McGinn (1989), p. 3. For arguments in support of it see Gallois and O'Leary-Hawthorne (1996), pp.2-3, Boghossian (1997), p. 166, and Burge (1982), p. 116.

18. Burge (1979).

19. This formulation can be found in Boghossian (1997).

20. This inference is discussed by Boghossian (1997), p. 175.

21. Boghossian (1997), p. 175.

22. Gallois and O'Leary-Hawthorne have argued in this way (1996), p. 3.

23. Kripke (1980).

24. McKinsey (1991 and 1994) has argued that Burge's compatibilism presupposes metaphysical dependence.

25. He considers the classical passages in Meditations VI, andPrinciples I, LX.

26. Burge (1988), p. 651.

27. Burge (1988), p. 651.

28. Michael McKinsey has put forward this argument in (1991), pp. 13-14, and in (1994), pp. 126-127.

29. On the other hand, Jessica Brown (1995) works out a similar strategy for Burge's anti-individualist argument. From her argument, it seems plausible that this anticompatibilist line of argument is viable also for concepts of non-natural kinds. See, p. 153.

30. Here I omit the condition concerning the atomicity of the concept employed. See on this Boghossian (1997), p. 168.

31. Jessica Brown proposes a shorter reply to the same problem based on Burge's version of externalism. She states that Burge's argument for externalism does not require that a subject knows that water is H₂O. According to her, Burge should accept the following statements, where terms for specific natural kinds are replaced by variables. Necessarily, if x has a thought involving the concept of a natural kind k and x is agnostic about the application conditions of the concept k, then either x is in an environment which contains k, or x is part of a community with the concept of k. See Brown (1995), p. 152-153.

32. See Boghossian (1997), p. 174.

33. Of course, it is enough that there is just one subject who can make an a priori inference of the existence of external facts. The fact that not all the subjects are familiar with Putnam or Burge's argument cannot be used as a reply. See on this point Brown (1995), p. 153.
 
 

References

Alston, W.

1970 "Varieties of Privileged Access", American Philosophical Quarterly, 8, 223-41.

Armstrong, D. M.
1968 A Materialist Theory of Mind,Routledge & Kegan Paul, London.

Boghossian, P.

1989 "Content and self-knowledge",Philosophical Topics 17, 5-26.

1997 "What the Externalist Can Know A Priori", Proceedings of the Aristotelian Society, 97, 161-175.

Brown, J.

1995 "The incompatibility of anti-individualism and privileged access", Analysis 55, 149-56.

Burge, T.

1979 "Individualism and the mental",Midwest Studies in Philosophy 4, pp. 73-121.

1982 "Other Bodies", in Woodfield (ed.), Thought and Object, Oxford University Press, Oxford.

1986 "Individualism and Psychology",The Philosophical Review, 95, N. 1, 3-45.

1988 "Individualism and self-knowledge",Journal of Philosophy,85, 649-63.

Davidson, D.

1984 "First Person Authority",Dialectica, 38, N° 2-3, 101-111.

1987 "Knowing one's own mind",Proceedings and Addresses of the American Philosophical Association.

Falvey K. And J. Owens

1994 "Externalism, Self-Knowledge, and Skepticism", The Philosophcal Review 103, 107-137.

Gallois, A. & O'Leary-Hawthorne, J.

1996 "Externalism and scepticism",Philosophical Studies, 81, 1-26.

Gibbons, J.

1995 "Externalism and knowledge of content", Philosophical Review 105, 287-310.

Glock, H. J. & Preston, J. M.

1995 "Externalism and first-person authority", Monist 78, 515-33.

Heil, J.

1988 "Privileged access", Mind 98, 238-51.

Kripke, S.

1980 Naming and Necessity, Basil Blackwell, Oxford.

McGinn, C.

1989 Mental Content, Basil Blackwell, Oxford.

LePore, E.

1990 "Subjectivism and environmentalism",Inquiry 33, 197-214.

Ludlow, P.

1995 "Externalism, self-knowledge, and the prevalence of slow-switching",Analysis 55, 45-49.

1997 "On the relevance of slow switching", Analysis 57, 285-86.

McKinsey, M.

1987 "A priorism in the philosophy of language", Philosophical Studies 52, 1-32.

1991 "Anti-individualism and privileged access", Analysis 51, 9-16.

1994 "Accepting the consequences of anti-individualism", Analysis 54, 124-8.

Miller, R. W.

1995 "Externalist self-knowledge and the scope of the a priori",Analysis 57, 67-74.

Putnam, H.

1975 "The Meaning of 'Meaning'", in H. Putnam, Mind Language and Reality, Cambridge University Press.

Warfield, T. A.

1992 "Privileged self-knowledge and externalism are compatible",Analysis 52, 232-37.

1995 "Externalism, privileged self-knowledge, and the irrelevance of slow switching", Analysis 57, 282- 84.