The points raised by Fuller & Slater's (2000) challenging and detailed commentary fall into three broad categories. One group concerns the circularity involved in the paradox of self- consciousness. A second group concerns what makes a nonconceptual "I"-thought a genuine "I"-thought. Finally they address my argument that concept possession requires language mastery. In this reply I address their worries in these three areas.
1. Fuller & Slater (2000) raise two questions concerning the circularity that generates what I term the paradox of self-consciousness (Bermudez 1998, 1999). It is impossible, in my view, to explain "I"-thoughts in terms of linguistic mastery of the first-person pronoun because linguistic mastery of the first-person pronoun presupposes the ability to think "I"-thoughts. This is so because successful employment of the first-person pronoun requires the intention to refer to oneself, where this intention is itself a first-person thought - one that would be reported directly using the first person pronoun or reported indirectly by means of the indirect reflexive pronoun (Castaneda's "I*").
2. The first question that Fuller & Slater pose is whether this requirement generalises. Do all linguistic rules need to refer to the psychological states of the speaker? It seems clear that they do not. This leads them to their second question. Why can we not offer an extensional understanding of the token-reflexive rule that any token of "I" refers to its utterer? They suggest that such an extensional understanding might take the form: Jane's token refers to Jane, John's token referred to John, and so forth.
3. The problem with this proposal is that it doesn't give the right account of what it is to understanding a given employment of the first-person pronoun. Suppose Jane is in front of me but I've never seen her before and don't know her name. Jane says "I need a drink". Of course, I have no difficulty whatsoever understanding what she has said. Yet on Fuller & Slater's proposal I would be unable to understand her utterance. Without knowing who Jane is, it would be impossible for me to apply the rule that Jane refers to Jane when she employs the first-person pronoun. Things would be even more complicated if I were under the misapprehension that Jane was Judith. In that situation following the Fuller & Slater proposal would lead me to the false conclusion that Judith needs a drink.
4. The correct way of construing what is going on, of course, is that the utterer of the sentence "I need a drink" needs a drink - irrespective of who it was that uttered the sentence or who I think it was that uttered the sentence. What leads me to understand the sentence in this way is my understanding that the utterer of the sentence employed the first-person pronoun with the intention to refer to herself*. Hence the circularity.
5. What makes it the case, ask Fuller & Slater, that the primitive "I"-thoughts associated with the self-specifying information in visual perception (what they helpfully call "I-ECO"- thoughts) really count as "I"-thoughts? They identify three special features of "I"-thoughts on my behalf. A genuine "I"-thought must have the Perry feature of picking out their thinker in a distinctive way that is not shared by other thoughts that might happen to pick out the same object. Second, they must be based on sources of information that are immune to error through misidentification. Thirdly, they must be self-reflexive. That is, they must be such that it is impossible to think them without knowing that they are about oneself. Which of these features are possessed by "I-ECO"-thoughts?
6. Two preliminary points. It is not clear that there really are three different features here, because the most obvious way of understanding the Perry feature is in terms of the self-reflexive feature. What makes the way in which "I"-thoughts pick out their thinkers so distinctive is precisely that they are such as not to allow the thinker to fail to realise that he himself (he*) is the subject of the thought.
7. Secondly, it is not right that all "I"-thoughts must be based on sources of information that are immune to error through misidentification relative to the first person pronoun. There seems to be some confusion as to what the property of immunity to error through misidentification actually amounts to. It is not the property of a thought itself (in the way that Fuller & Slater seem to suggest). If it were then there would be no difference between the immunity property and the self-reflexivity property.
8. Self-conscious thoughts can be based on a range of different sources of information. Some of these sources can provide information either about the self or about other people. Testimony is a case in point. I can learn facts about myself by being told them by others, in the same way as I learn facts about anything else. But there are other sources of information about the self that provide information purely about the self. These sources of information are such that, if we know from them that somebody has a particular property, we ipso facto know that we ourselves have that property. Introspection is an example. If I know through introspection that someone is currently thinking about self-consciousness then I know that I myself am thinking about self-consciousness. Introspection cannot provide information about anybody other than me. This does not mean that introspection (and other comparable sources of information) cannot be mistaken. They certainly can, but they do not permit a certain type of error. Judgements made on the basis of them cannot be mistaken about who it is that has the property in question.
9. Clearly, not all "I"-thoughts are based on sources of information that are immune to error through misidentification. "I"-thoughts derived from testimony will not, for example. But it seems plausible that the immunity to error through misidentification feature will be a sufficient condition of something being a genuine "I"-thought (even if not a necessary condition).
10. So, we can rephrase the question of why "I-ECO"-thoughts should really count as "I"- thoughts in terms of the question whether "I-ECO"-thoughts are based on sources of information that are immune to error through misidentification. And it seems clear that the self- specifying information in visual perception IS immune to error through misidentification. I cannot, for example, perceive an affordance and then wonder whether it is an affordance for me.
11. In the case of ecological self-perception, for example, the immunity to error was explained in terms of certain features of the content of visual perception - the fact that the parts of the body appear in the visual field as what Gibson calls subjective objects (because they can vary in perceived size only within a limited range); the fact that the field of view is bounded by the body; the way in which affordances are directly perceived in the environment. And so forth. Further details will be found in the online summary and in the relevant chapters of the book.
12. Fuller & Slater take issue with my proposal that there is a constitutive link between concept possession and linguistic abilities. They attribute to me the following two claims:
(a) Concept possession requires both a capacity for justification and a grasp of (at least some adequate subset of) properties associated definitionally with category membership.
(b) Each of these requires a natural language.
As they point out, acceptance of these two claims is incompatible with the interpretation I put on Susan Carey's experiments in young children's biological understanding (Carey 1982). I suggested that Carey's 4-year olds were making use of inference patterns connecting the concepts "human being", "living animal" and "internal organs". But, as Slater and Fuller rightly note, one of Carey's principal conclusions was that the judgements the children accepted concerning these kinds differed systematically from those that are accepted as definitional by competent language-users.
13. I was careless in the interpretation I put on the Carey experiments. What I should have said is simply that the 4-year olds were operating at the conceptual level - that is to say, the level of justification and inference patterns. They displayed rational sensitivity to the kinds in question, rather than the simple perceptual sensitivity that I would locate at the nonconceptual level.
14. But there is a more important point here. Am I committed to holding some form of "necessary and sufficient conditions" account of concept possession, according to which possession of the concept C requires being able to offer some sort of definition of what all the things that fall under C have in common? It seems to me that I am not, which is fortunate, as such accounts of concepts are rightly held in low esteem among both philosophers and cognitive scientists.
15. Slater and Fuller are drawing the wrong conclusions from some of the things that I said about whether prelinguistic infants can properly be described as perceiving objects. The point I made was that there is a range of properties (object properties) that define what it is for something to be an object. These object properties are physical, kinematic and dynamical. I wanted to stress that unless young infants are perceptually sensitive to a reasonable subset of these properties they cannot properly be described as perceiving objects. Rather, we need to invent some technical term to characterise what they are perceiving (I suggested "object*"). But it by no means follows from this that possession of the concept of an object requires being able to list or otherwise spell out just what the object properties are. It is more plausible to construe the requirements on concept possession in terms of certain canonical inferences, in the manner of Peacocke 1992.
16. Slater and Fuller tentatively attribute to me the view that language is a requirement for concept possession because the definitional criteria for a concept are those associated with the natural language term that expresses it. It should be clear by now that this is not my view. But it might be helpful if I made some suggestions about how I understand the differences between linguistic cognition and non-linguistic cognition. As I mentioned in my PSYCOLOQUY reply to Robbins, the type of rational sensitivity appropriate at the conceptual level will involve making second-order judgements about the inferential relations between thoughts and other mental states (such as perceptions) as well as between thoughts tout court. That it is to say, it will involve intentional ascent. Intentional ascent, however, requires holding a thought in mind in a way that reveals its structure (otherwise it would be impossible to evaluate its inferential connections with other thoughts). This, it seems to me, is possible only if the thought is linguistically vehicled. At least, we have no grasp on any other way in which a thought can be held in mind in such a way that it can be the object of further thoughts. The argument, therefore, is that intentional ascent requires semantic ascent.
17. The role of language in cognition, therefore, is in making possible intentional ascent (or metarepresentation). Let me end by addressing a problem raised by this account of the differences between linguistic and non-linguistic cognition. The eventual solution I offered to the paradox of self-consciousness depended crucially upon the claim that creatures at the nonconceptual level are capable of representing intentions, both their own and those of others. But surely such representations involve intentional ascent and metarepresentation. How can this be reconciled with the argument that intentional ascent and metarepresentation are available only to linguistic creatures?
18. Let me start somewhat tangentially by distinguishing two types of desire. Goal-desires are desires for objects, while situation-desires are desires that particular states of affairs be the case. In standard philosophical terms, goal-desires are extensional while situation- desires are intensional. That is, in spelling out the content of a goal desire it is necessary to be sensitive to the mode of presentation under which the desired state of affairs is apprehended in a way that it is not when specifying the content of a goal-desire. Only in attributions of situation-desires, therefore, do we have intentional ascent and hence a need for language. An attribution of a situation- desire involves specifying that an individual stands in the desiring relation to a state of affairs characterised in a particular way (that is, by a particular sentence). In attributions of goal- desires, on the other hand, we have simply an individual standing in the desiring relation to an object. There is no need to mention how that goal is apprehended by the individual, and hence no need for a linguistic representation.
19. It seems to me that a comparable distinction can be made between two different types of intentions - extensional goal-intentions versus intensional situation-intentions. The strategy I would adopt, therefore, in defusing the tension between the idea that intentional ascent requires language and the claim that psychological self- awareness at the nonconceptual level requires being able to represent certain mental states is to deny that all representation of mental states involves intentional ascent. Intentional ascent comes in only when mental states have to be represented in an intensional manner - in a way sensitive to considerations of cognitive significance and mode of presentation.
Bermudez, J. L. (1998) The Paradox of Self-Consciousness. Cambridge MA. MIT Press.
Bermudez, J. L. (1999) Precis of "The Paradox of Self-Consciousness". PSYCOLOQUY 10(035) ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1999.volume.10/ psyc.99.10.035.self-consciousness.1.bermudez http://www.cogsci.soton.ac.uk/cgi/psyc/newpsy?10.035
Carey, S. 1982. Semantic development: The state of the art. In E. Wanner and L. R. Gleitman (Eds.), Language Acquisition: The State of the Art. Cambridge. Cambridge University Press.
Fuller, G. and Slater, C. W. (2000) "I"-Thoughts: CRITERIA, Constitution, and Concept Possession. ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/2000.volume.11/ psyc.00.11.059.self-consciousness.9.fuller http://www.cogsci.soton.ac.uk/cgi/psyc/newpsy?11.059
Gibson, J. J. 1979. The Ecological Approach to Visual Perception. Boston. Houghton Mifflin.
Peacocke, C. 1992. A Study of Concepts. Cambridge MA. MIT Press.