In a reply to John Perry, Hector-Neri Castañeda once wrote:
But for me to use S [a sentence with a non-relative occurrence of the first-person pronoun] to make a statement, I must identify myself with the denotation, in the context of the utterance, of that non-relative occurrence of the first person pronoun in S. Clearly this identification involves thinking of myself as myself. (Castañeda 1983: 324)
In this passage, Castañeda submits that I need to identify myself in order to refer to myself with the utterance of the first-person pronoun 'I'. He then goes on to argue that for this to be possible, I ought to be able to think of myself as myself. In order to avoid an obvious circle, this latter capacity ought to be independent of my capacity to refer to myself with the first-person pronoun.
These considerations suggest the following generalisations. Any usage of the first-person pronoun presupposes some sort of preliminary capacity to think of myself and no account specifying the semantics of the expression 'I' can suffice alone for a full understanding of self-consciousness. One needs a more primitive form of self-consciousness in order to account for one's capacity to refer to oneself with the first-person pronoun in utterances expressing first-person beliefs.
One might hope to get rid of this problem by rejecting Castañeda's first premise. It has been argued that usage of the first-person pronoun does not, at least not always, presuppose the identification of its referent. Indeed, a certain type of beliefs expressed by utterances in the first person are said to be identification free (Evans 1982: 1980) or immune to error through misidentification (Shoemaker 1984: 7-8). Although these two notions present a number of important differences, they both involve recognition of the fact that the subject is not always required to identify himself when expressing a first-person belief or simply an 'I'-thought.
Typically, utterances expressing introspective beliefs about one's own mental states are said to be immune to error through misidentification. I do not need to identify myself in order to have a belief I would express with 'I have a pain'. But if I do not need to identify myself, then, one might hope to infer, I do not need to have any preliminary way of thinking of myself as myself either. And so, it might be concluded, all one needs in order to entertain genuine first-person beliefs is the capacity to think 'I'-thoughts that are immune to error through misidentification. In Bermùdez terminology this would be a central claim of a deflationary account of self-consciousness (cf. Bermùdez 1998: 10).
Bermùdez submits, however, that even if it does not involve the identification premise, the deflationary account faces a number of problems which prove once again the need for a more primitive notion of self-consciousness.
The problem is not with first person beliefs which are subject to error through misidentification. Bermùdez explains that any genuine first person belief expressed by an utterance of the form 'I am F' is either itself immune to error through misidentification, or it is analysable in two components, the predication component 'the F is F' and the identification component 'I am the F', which is immune to error through misidentification (Bermùdez 1998: 10). 'Any first person-content subject to error through misidentification will ultimately be anchored in a first-person content that is immune to error through misidentification' (ibid.: 7). The capacity to entertain thoughts about oneself which are immune to error through misidentification thus remains a specific trait of self-consciousness.
The worries, following Bermùdez, run deeper. He asks for an account that explains what it is to possess the capacity to make self-ascriptions that are immune to error through misidentification? (8). One might thus wonder what is responsible for the fact that self-conscious beliefs are typically (anchored in) beliefs about oneself which are immune to error through misidentification.
On Bermùdez view, the deflationary account is committed to the claim that it is the semantics of the first person pronoun which will provide an answer to such a question (11, 14). So, it appears, the claim would be that what accounts for the fact that self-conscious beliefs are immune to error through misidentification is the fact that the expression 'I' occurring in utterances expressing them has a specific meaning. Now, the specificity in the semantics of the expression 'I' is certainly the rule governing its usage, i.e. the well-known token-reflexive rule. After having discussed several versions of it, Bermùdez settles for its forth version, which states: 'If a person employs a token of 'I', then he refers to himself in virtue of being the producer of that token' (15). But here Bermùdez finds the deflationary view entangled in an explanatory circle. For in order to apply that rule, one must already know self-consciously that one is the producer of the token under consideration (17).
In a nutshell, then, the argument which lands the deflationary account in the explanatory circle goes as follows: (i) Immunity to error through misidentification explains why first person beliefs are self-conscious; (ii) The token-reflexive rule governing the usage of the expression 'I' explains why first person beliefs are immune to error through misidentification; (iii) But the application of the token-reflexive rule involves self-conscious first person beliefs; (iv) So the explanandum is part of the explanans (16).
As a first reaction to this argument one might wish to reject premise (ii). One might readily admit that in order to express a first-person belief in language one needs to master the token-reflexive rule. It is open to debate, however, whether the capacity to entertain first-person thoughts depends on one's ability to express them in language. Bermùdez thinks that premise (ii) is motivated by 'a principle that has governed much of the development of analytic philosophy' (12), namely the Thought-Language Principle, which states: 'The only way to analyse the capacity to think a particular range of thoughts is by analysing the capacity for the canonical linguistic expression of those thoughts' (13). So, it appears, in order to weaken the motivation supporting premise (ii), we first need to reject the Thought-Language Principle.
Bermùdez submits that the principle can actually be split into two subprinciples, the Conceptual-Requirement Principle (41) and the Priority Principle (42). The first principle makes content attribution in general depend on conceptual abilities. The second principle makes conceptual abilities depend on linguistic abilities. Bermùdez thinks that it is the first principle which ought to be abandoned, not the second one. Indeed, the first part of Bermùdez book argues in favour of the idea that it is possible to ascribe non-conceptual content independently of conceptual and linguistic abilities.
The central argument in the book is then to show that (i) there are mental states with non-conceptual first-person contents and (ii) that those contents can be used in order to explain why first person beliefs are self-conscious (see for instance p. 103).
In what follows I shall concentrate on the first claim, the claim that there are mental states with non-conceptual first-person content. This, I take it, is the claim that there is a sort of primitive non-conceptual self-consciousness. I want to determine the property in virtue of which a specific non-conceptual content qualifies as self-conscious on Bermùdez' account.
Let me first recall a point which has often been raised in connection with conceptual self-consciousness. Suppose somebody claims, following the line of argument discussed above, that what is specific about self-conscious first-person beliefs is the content determined by the token-reflexive rule. A suggestion along this line would immediately generate the question as to whether it is enough to grasp that content in order to have a self-conscious belief. If the content of my utterance 'I am F' is specified as <the producer of this token is F>, this won't do. For, as it has been argued, I might under certain conditions believe that the producer of this token is F without believing that I am F. It is thus possible to entertain the token-reflexive content in a way which fails to be self-conscious.
Examples of this sort, which I shall call 'Castañeda-examples' (in honour of the variety of them he has provided, starting with his famous Editor of Soul example) might be used to establish an adequacy condition for any account which aims at determining a specific content in virtue of which a mental state qualifies as self-conscious. We might say that a content, be it conceptual or not, is sufficient to yield self-consciousness only if it is not possible to entertain that content in a way which is not self-conscious.
Notice that this is not to deny that the content under consideration might yield self-consciousness when entertained, grasped, instantiated, etc. in a specific way. The point is that, in such a case, what remains to be determined is precisely that specific way, at least as much as the content which is supposed to lay at the source of self-consciousness.
In a first set of arguments (Chapter 5) Bermùdez appeals to Gibson's ecological approach to perception. He suggests using that approach against the Schopenhauer/Wittgenstein view that the self cannot be perceived and that there is no place for the self in the content of perceptual experience (105). Bermùdez submits that Gibson's theory of ecological optics shows 'beyond reasonable doubt that there is self-specifying information available in the field of vision' (109). Carrying self-specifying information, however, does not seem to be quite enough for non-conceptual self-consciousness (115). One also needs to take into consideration the intimate relation between perception and action. 'Perceptual contents are first-person contents in virtue of their immediate connections with behaviour' (118). So far the claim would be that what makes perceptual content self-conscious is the fact that it carries self-specifying information and that it stands in an immediate relation to action.
Let us look more closely at the first feature, self-specifying information in perception. One of the central claims in this respect is, as Gibson puts it in a passage quoted by Bermùdez, that 'vision obtains information about both the environment and the self' (110). Now, for information about the self to be relevant with respect to self-consciousness, it must be about the self as the self. So, what is it for visual information to be about the self as the self? Bermùdez quotes Gibson stating that 'vision is kinaesthetic in that it registers movements of the body' (110). But how does the addition of information about the body to information about the environment generate self-consciousness? Even if one thinks that this is how it should work, one would certainly need some further details in order to rule out circularity. So, for instance, it can't be simply the fact that one is obtaining information about one's own body: any person looking at the subject could have that sort of information. So maybe what is crucial is that it is information about one's own body as one's own body. But here self-consciousness would simply be presupposed and not explained.
In the passage quoted above, Bermùdez appears to be suggesting that what is crucial, in fact, for perceptual content to be self-conscious is its immediate connection with behaviour. As in Perry's famous example, what explains my behaviour is that I believe self-consciously that I am about to be attacked by a bear. But are we allowed to run the explanation in the other direction? Can we say that what explains the fact that my belief is self-conscious is its intimate connection with a certain type of behaviour?
Consider the following argument: mental state M (be it conceptual or not) caused behaviour B; behaviour B can only be explained by a state with first-person content; so M must have first person content. Even if one accepts the argument as such, it is hard to see how it could count as providing an explanation of the phenomenon under consideration. Just try the following parallel: the car caused the accident; the accident can only be explained by the fact that the car was moving; so the car must have been moving. This certainly does not count as an explanation of the fact that the car was moving.
First-person contents are needed in order to explain action. One might use the intimate connection between first-person contents and action in order to determine the content of a specific mental state. One might say that one's mental state must have a first person content if it is to explain one's action. But no mental state can be said to have a first-person content in virtue of its immediate connection with behaviour, if this is meant to provide a specification of what is responsible for the fact that a mental state is self-conscious.
Much of the argument concerning self-specifying information appears to depend on the fact that the information under consideration is given to the subject in a specific way. It is not the information itself, which is relevant, but the way it is given to the subject. This point appears even more in Bermùdez' discussion of somatic proprioception, which is also said to be 'a form of primitive self-consciousness' (131).
Bermùdez is well aware of the difficulty. Even if we grant, he writes, that 'somatic proprioception [I] involves perceiving certain bodily properties that are properties of one's self [I], this is not enough to make it a form of self-consciousness. Genuine self-consciousness requires the further feature that those bodily properties should be perceived as properties of one's self' (145). Let us see, then, how that further feature can be obtained.
Bermùdez starts off from the simple argument (135):
(1) The self is embodied.
(2) Somatic proprioception provides perception of bodily premises.
(3) Somatic proprioception is a form of self-perception.
(4) Therefore, somatic proprioception is a form of self-consciousness
Let us grant the first two premises (Bermùdez gives some quite convincing arguments in favour of the second). The third statement follows as long as 'self-perception' is not supposed to be self-conscious (as Bermùdez himself suggests, for the sake of the argument, on p. 135). The question thus is whether the fourth statement follows from the above premises.
Bermùdez discusses an important criticism against the simple argument which submits that there is an 'equivocation fallacy' (145). As far as I could understand, the equivocation would simply arise by passing, once again, from self-perception as perception of properties of the self, to genuine self-consciousness as perception of the properties of the self as properties of the self.
Bermùdez' first reaction to this argument is strategic. He thinks that the argument would have undesirable consequences concerning introspection. I shall return to this point later on. For the moment let me concentrate on Bermùdez central distinction between broad and narrow self-consciousness. He writes that the critic using the equivocation argument 'is pointing out that narrow self-consciousness is all that follows from the conjunction of (1) and (2)' (151). Bermùdez submits that this is 'fundamentally mistaken' (ibid.), although he recognises that narrow self-consciousness 'does not [I] exhaust the sense in which somatic proprioception counts as a form of self-consciousness' (149).
At one point narrow self-consciousness is defined as 'knowledge of one's bodily properties' (149). This alone, to be sure, is not enough for genuine self-knowledge. Indeed, Bermùdez formulates two requirements for a thought to be genuinely self-conscious. The first requirement is that such a thought 'should be about oneself in a way that is nonaccidental' (148) and the second is that those thoughts 'feed directly and immediately into action' (ibid.).
For a thought to be about oneself in a way that is not accidental means that it cannot ascribe properties to oneself without one's being aware that the individual in question actually is oneself. As Bermùdez notices, this appears to be a consequence of the immunity to error through misidentification of first-person thoughts. No doubt, mental states with perceptual content can have that sort of immunity just as much as beliefs. Immunity to error through misidentification, however, has been said to be insufficient for an explanation of the self-conscious character of first person beliefs. It cannot then be explaining the self-conscious character of perceptual states either.
I have stated above some of my qualms concerning Bermùdez usage of the action argument, the argument following which a mental state would be self-conscious by virtue of its immediate relation to action. Let me at this stage mention another problem which I take to be related to the 'functional characterisation' (148) of first person thoughts. Even if one were to accept the functionalist line of argument, following which a first person thought is defined by its essential relation to action, I doubt this would be of much help for Bermùdez defence of primitive self-consciousness. For the same line argument could have been taken with respect to first person beliefs. Self-conscious mental states with conceptual content certainly have an immediate and direct relation to action.
Narrow self-consciousness as applied to perceptual content appears to be genuinely self-conscious by virtue of the very same trait self-conscious beliefs can be said to be self-conscious. Therefore perceptual self-consciousness in the narrow sense cannot be used in an explanation of conceptual self-consciousness.
The burden must finally lie on broad perceptual self-consciousness. Recall that we are seeking a property by virtue of which a perceptual state would be genuinely self-conscious. Moreover, one ought to appeal to that property in order to explain why conceptual states based on those primitive forms of self-conscious representations are immune to error through misrepresentation, a property which is recognised as being essential to self-conscious beliefs.
Broad perceptual self-consciousness is said to be the recognition, through somatic proprioception, of the 'self-world dualism' (149), the fact that a self-conscious being registers the distinction between himself and the world. Now, somatic proprioception offers 'an awareness of the body as a spatially extended and bounded physical object that is distinctive in being responsive to the will' (150).
Let me consider those two points separately. Could I not be aware of my body as a spatially extended and bounded physical object without being aware of the fact that it is my body? Again, there must be something in the way I am aware of all this that yields self-consciousness. It must be some specific trait of proprioception, as an access to that kind of information, that makes the trick. But what exactly is it?
The notion of one's own body as being responsive to the will would require a treatment on its own. Yet the emphasis in Bermùdez line of thought does not rest on this argument, so I shall allow myself to be brief. As phenomenologists from Husserl to Merleau-Ponty have remarked, there is an intimate relation between the will and the body. I do not normally experience my body as being moved by me, I move my body directly, without any further mediation. And so I can certainly make a distinction between self and nonself along the divide between objects I move through my body, as it were, and my body itself. For this to show that proprioception is self-conscious, however, we need to assume that the action carried out with the body is itself self-conscious. One might say that it is because I am aware of the fact that I myself am moving my hand that I perceive my hand as being mine. Yet, aren't then just removing the problem by one step. What is it for me to be aware of the fact that I myself am carrying out a action?
Bermùdez final answer to our question appears thus to rest on considerations concerning the content of somatic proprioception. On his view, 'the content of somatic proprioception is that certain events are taking place at particular bodily locations' (154). Content of proprioception thus has two aspects, a spatial one, determining the place in one's body where something is happening, and a descriptive one, determining the qualitative and quantitative aspects of the occurring bodily state (being bruised, being hot, moving in a particular direction, etc.). Now, which of those aspects is responsible for the proprioceptive state being self-conscious? Is it the qualitative aspect? Remember the traditional argument: I cannot feel pain and wonder whether it is really me who is in pain, whether the pain is really mine. Right, but that is immunity to error through misidentification once again. So it must be the spatial aspect.
Bermùdez explains that a bodily sensation can be located in one's body following two different criteria (cf. 154). One can simply locate a sensation in a limb (A location), or one can locate it with respect to other body parts (B location). We are then told that there are hinges in our body which provide the fixed points in terms of which we can determine A and B locations of our sensations and that B locations depends on A locations. All these considerations provide a better understanding of the phenomenology of proprioception. But I still cannot see what property is supposed to account for the self-conscious character of proprioception. The way Bermùdez specifies the correctness conditions of proprioceptive states certainly does not help any further. He thinks that their correctness conditions must be determined by looking at their functional relation to events causing them and actions being caused by them. He writes that 'the correctness conditions for explicitly body-directed actions, like scratching itches, are that the action should be appropriate to the disturbance that causes the proprioceptive state' (159-160). Unless it is supposed that the itch is perceived as belonging to the self, or that the scratching is felt as being carried out by the self, this correctness condition does not even require proprioception to be self-conscious.
Let us return to the strategic point mentioned above in connection with Bermùdez reply to the equivocation argument. The critic is supposed to claim that one can have a proprioceptive representation of one's own body without being perceptually aware that it is one's own body (cf. 146). Bermùdez thinks that one could just as well argue that one can be introspectively aware of one's mental state without being introspectively aware that they are one's own mental states. He submits that what militates against the second argument should be taken to militate against the first as well.
As far as introspection is concerned, the point seems to be clear. Introspection being immune to error through misidentification, one cannot imagine a subject being in pain and wondering whether it is really herself who is in pain. Bermùdez submits that the same applies to proprioception: 'One of the distinctive features of proprioception is that it is subserved by information channels that do not yield information about anybody's bodily properties except my own' (147). This is just to say that proprioception is immune to error through misidentification. Yet in that case the point against the equivocation argument, the point which is supposed to explain why self-perception in fact is a form of self-consciousness, relies entirely on the property of being immune to error through misidentification.
If it is the immunity to error through misidentification which accounts for the self-conscious character of proprioception, then one wonders what was wrong with the deflationary account from which we started. The point, notice, is not to deny that there is perceptual self-consciousness. What seems doubtful is rather its role as a primitive form of self-consciousness to which one ought to appeal in order to explain other forms of self-consciousness.
Castañeda, Hector-Neri 1983: "Meaning, Belief and Reference. Reply to John Perry". In: Tomberlin, J. E: Agent, Language, and World: Essays Presented to Hector-Neri Castañeda. Indianapolis: Hackett Publishing Company, 312-328.
Evans, Gareth 1982: The Varieties of Reference. Oxford: Oxford University Press.
Shoemaker, Sydney 1984: "Self-reference and Self-awareness". In: Shoemaker, Sydney: Identity, Cause and Mind. Cambridge: Cambridge University Press, 6-18. First appeared in: Journal of Philosophy (65), 1968, 555-67.