Content

Department of Philosophy   Northern Illinois University
Tomis Kapitan

FIRST-PERSON REFERENCE

A self is what is referred to in the first person: what the first person pronoun denotes in its proper and correct use. Hence, the structure of the mechanisms of self-reference reveal at least part of the structure of the self.
                                 Hector-Neri Castañeda,   "Attributions of Self-Reference" (Nachlass)
 

Castañeda's writings on self-consciousness and first-person reference span nearly three decades, and until his final days these topics continued to be at the forefront of his philosophical attention. Building upon his early work on practical thinking and the private language arguments, he was among the first to locate an indispensible element of self-reference in the genesis of action and to investigate the logic of attributions of self-regarding attitudes. Convinced that philosophical questions concerning the self are rooted in the use of first-person pronouns, he went on to develop his initial insights within a philosophical system designed to resolve a wide array of problems in metaphysics, epistemology, and ethics. Yet many of his proposals concerning self-reference and self-consciousness are adaptable to other frameworks and deserve separate attention in their own right.
 

The Irreducibility of First-Person Reference

One of Castañeda's best-known observations was that replacements of first-person pronouns by coreferring expressions such as names, definite descriptions, and demonstratives, commonly fail to preserve truth and meaning. A woman might use the sentence, `The one who wrote that letter lacks subtlety,' to express a belief that she would not express with `I lack subtlety' even though, unbeknownst to her, she is the one who wrote the letter. Conversely, she might assert `I am a clever writer' yet reject `The one who wrote that letter is a clever writer.' Other substitutions disrupt modal status, for instance, `I am uttering nothing' is contingently false, but `The person uttering this token is uttering nothing' is self-contradictory. This substitution also imputes conceptual abilities--e.g., concepts of person, token or utterance--that cannot be expected of all users such as small children. As such, the meaning of `I' is not that of `the person uttering this token.'

Related difficulties arise in the attempt to accurately report or ascribe first-person usage. For example, if I want to report what the editor of Soul said with the words,

(1) I am a millionaire.

it would be inappropriate to use,

(2) The Editor of Soul believes that I am a millionaire.

since `I' invariably expresses the speaker's self-reference--in this case, my own--not the attributee's, unless I know that I am the attributee. Instead, I must indicate that it is the the Editor of Soul who is believed to be a millionaire. By whom? Well, by himself, the very man I refer to with `the Editor of Soul,' and his belief is true just in case this referent is a millionaire. However, the report,

(3) The Editor of Soul believes that the editor of Soul is a millionaire.

is misleading if it ascribes to the Editor the belief,

(4) The Editor of Soul is a millionaire.

The Editor might not realize that he is the Editor of Soul and, consequently, might not believe what he would report with (4) despite his affirmation of (1). Similar difficulties arise by substituting other coreferring descriptions, names, or demonstratives for the occurrence of `the Editor of Soul' within attitudindal scope in (3). Even if the Editor of Soul takes `the Editor of Soul' to designate himself, it is doubtful that (3) captures what is affirmed with (1). If it did, then

(5) I am an Editor.

as uttered by the Editor of Soul, and

(6) The Editor of Soul is an Editor.

should have the same significance. Yet (5) seems contingent and informative in a way that (6) does not.

A more promising candidate for recording the Editor's belief is,

(7) The Editor of Soul believes that he himself is a millionaire.

The reflexive `he himself' serves as a proxy for the first-person pronoun by depicting a first-person reference that someone makes or is disposed to make. It too cannot be replaced by coreferring descriptions, names, or demonstratives, even when these are its antecedents. Nor does it behave as a typical bound variable as in,

(8)  Something x is such that (x = The Editor of Soul and the Editor of Soul believes that x is a millionaire).

This statement is true if (3) is, but since (3) can be true without the Editor believing that he himself is a millionaire, (7) cannot be parsed as (8). More generally, insofar as `he himself' is used to attribute belief in what a person would normally express by using the first-person pronoun, it cannot be replaced by the standard variables of quantification.

Although `he himself' represents or depicts someone's first-person reference, it is not itself an indexical. It is a quasi-indicator--a linguistic device for attributing indexical reference--whose proper abode is within oratio obliqua constructions. Like the first-person pronoun, it cannot be eliminated from the language of propositional attitudes without loss of content and change of truth-value. This fact constitutes further evidence for the irreducibility of first-person reference, though the order of explanation is exactly the reverse: quasi-indicators like `he himself' and `she herself' are ineliminable because they represent what is essentially first-person usage.1

At first sight, the claim that first-person pronouns constitute irreplaceable, ineliminable referring expressions is odd. It seems obvious that my `I' tokens designate the very same person picked out by the name `Tomis Kapitan,' or by various descriptions or demonstratives available to others as well as to me. So what does the irreducibility of first-person reference come to? What is its source? What is its significance?

These questions motivated Castañeda's subsequent studies of self-consciousness. He addressed the source of irreducibility from the outset when he spoke of first-person pronouns as unique mechanisms of reference or ways of referring through which one refers to oneself. But this raises only further questions: what are these referential "mechanisms" or "ways," and in what sense are they irreducible? To answer them we must examine what reference is for Castañeda and how indexical reference is distinguished from other forms.
 

Thinking Reference

The term `refers' and its cognates are multifaceted. We are familiar with talk about the reference of singular terms, but, for Castañeda, referring is first and foremost something we do intentionally, for example, when we direct someone's attention to a thing. Since we cannot purposely direct attention upon what we ourselves do not attend to, there must be a psychological process of consciously "picking out" an item to be talked about. This, too, is a kind of reference--thinking reference--wherein one focuses upon one item to the exclusion of others for the purposes of thinking or saying something about it.2 Linguistic "mechanisms of reference" play a causal role in producing and communicating references, and the irreducibility of any such mechanism is a function of the peculiar semantic or conceptual resources that guide its use. Indexicals, in particular, are irreducible to non-indexical mechanisms since they embody a unique access to referents. Let us expand upon this more carefully.

To thinkingly refer to something requires some means of distinguishing it from other things of which one is aware. These means--modes of presentation or identifying properties of the referent--constitute the conceptual mechanisms underlying reference. If I report that the Louvre is in Paris, for example, I must have some way of picking out the Louvre, say, as the largest art gallery in France or qua that huge building over there across the Seine. Each description corresponds to a mode through which the referent is distinguished, even if this mode is not itself articulated in my report. Several modes of presentation might be operative within a single act, as when one says; "Look at that one, the statue to the left, the only Michaelangelo in the room," and while some modes may be generic at least one must permit the user to identify or individuate the referent.

It is tempting to say that a mode of presentation cannot guide reference unless it corresponds to a genuine status or property of the referent, but this is not quite right. If the one I pick out with `the man drinking a martini in the corner' is neither a man nor drinking a martini then the expressed mode is not satisfied. Since I have referred, however, I have utilized some identifying mechanism even if best conveyed through demonstratives, e.g., `that person,' or, simply, `that.' If one of these does not hit the mark, then I have not succeeded in referring to anything. We must recognize the following satisfaction principle:

One thinkingly refers to an item only by distinguishing it through a mode of presentation uniquely satisfied by it.

While it is another matter to characterize precisely the act of distinguishing, we can allow that a mode is operative in reference only if it is either satisfied by the referent or implies a mode that is so satisfied.3
 

Indexical Execution is not Indexical Interpretation

Indexicals offer a simple means of making, expressing, and communicating thinking references. They are particularly useful when proper names or descriptions are either cumbersome or unavailable, say, when demonstratives like that or that thing over there represent the only way of picking out what suddenly looms into vision. Or, a kidnapped heiress locked in a trunk of a car might think: It is quiet here now, without having any other means of locating herself. In such cases indexicals provide autonomous mechanisms for executing a reference that do not depend upon the user's possession of other ways of referring to the same items.

How is indexical reference explained? The context-sensitivity of indexicals precludes both a causal chain account and a solution in terms of the reference-fixing meanings, senses, or characters of linguistic types.4 One popular solution takes character in combination with context of usage to do the trick. That is, by (i) grasping rules specifying the relations between tokens of a given indexical type to their referents, and (ii) understanding relevant features of context, one is able to understand what indexically expressed thoughts are about. For instance, the relation of a `now' token to a duration is that of being the time of the token's utterance, while that of an `I' token to its referent is given in a rule such as:

(I) A token of the first-personal indexical pronoun `I' refers to the speaker of the utterance in which it occurs.

Hearing you say, `I am now going to throw the ball over there,' for example, then my grasping the characters of `I,' `now,' and `there'--as displayed in rules like (I)--and knowing that you uttered the sentence, when you uttered it, and what region you demonstrated, enable me to determine what your referents are.

To so account for my interpretation of your indexical utterance, however, is not to explain your indexical reference. In fact, if all indexical references were guided by token-reflexive rules then none would be autonomous.5 Obviously, to interpret what another person is referring to indexically one must have a separate access to the referent--for which reason quasi-indicators have antecedents--but this does not explain the speaker's production of first-person pronouns. One who uses `I am hungry' to demand food is referring with `I', but not by first apprehending that very `I' token and then determining its referent by recourse to context. Otherwise we would be at a loss to explain how `I' references are initiated.6 The speaker's production of a first-person pronoun is not guided by a rule suitable only for its interpretation and, therefore, the mechanisms for executing a first-person reference must be distinguised from those used in its interpretation.7
 

How Indexical Identification is Achieved

If token-reflexive accounts cannot explain the execution of a first-person reference, what will? Castañeda's solution exploits the idea that reference is indexical because it originates in and reflects a thinker's experience of objects from a particular point of view or perspective: "indexicality is simply the general involvement with experience that indicators denote" (Castañeda 1990c, 68). Accordingly, each indexical referent is singled out through a unique perspectival mode of presentation or, as Castañeda preferred to say, a particularized perspectival property, a relational attribute that the referent has in virtue of being experienced by the user. There are two central features of such a property; an indexical form of experience and a position within a perspective. Let us look at each in turn.

I. The Indexical Form. Part of the productive meaning of any indexical token is the manner in which the speaker encounters the referent, as evidenced by the information values of indexical types. If I say, You will be more comfortable in this chair, then I am addressing someone I take to be within my present communicational reach. Similarly, my saying, It was hot here yesterday expresses something about my perspective on a place and time designated in a way that It was hot in Chicago on July 13, 1997 does not. No indexical is a mere label; each informs the audience of how items are experienced, whether as immediately presented, e.g., pure demonstratives such as `this' or `there,' or as vicariously confronted through a relation to present experience, as in, `that building [pointing to a photo of the Dome of the Rock] is in Jerusalem.' The temporal demonstratives `now' and `then' not only designate intervals of time, but intervals as related to a present experience, the former to the actual time of that experience, the latter to some preceding interval. Similarly, `this' and `that' represent items as present to consciousness, respectively, in the foreground and background. Each indexical type is associated with a unique manner or form of encounter through which the user is related experientially to a referent.8

What, then, is the form of encounter associated with the production of first-person pronouns? It cannot be what is offered by rule (I), and the irreducibility arguments block the simple rule of reflexivity:

(I') A token of the first-personal indexical `I' is used by the speaker to refer to himself (herself).

Referring to oneself is necessary for a first-person use of `I' but not sufficient. Instead, Castañeda offered,

(I*) A token of the first-personal indexical `I' is used (by the speaker) to refer to oneself qua self.

That is, to refer with a first-person token is to be aware of oneself internally, through an irreducible self mode, and this differs from any variety of external awareness of oneself though non-first-person forms of encounter. While Castañeda spoke of the self-concept as unique, primitive, and inexplicable, chapter 14 below attempts to deepen our understanding of this form of personal encounter.

Still, generic indexical forms of encounter like that specified in (1*) do not perform the identifying tasks associated with the production of indexical tokens. What does? It is not enough to relativize forms to persons or utterances since tokens of the same indexical type belonging to a single utterance can differ in reference, e.g., `This is larger than this' or `You, you, you, and you leave, but you stay here.' Nor is the user's perspective enough, whether we understand this to be the user's spatio-temporal vantage point or a spatio-temporal ordering of contents determined by that vantage point. As long as the temporal component of a perspective has the temporal thickness of an utterance, then tokens of `this' or of `you' in these examples issue from one and the same perspective yet are connected to distinct mechanisms of identification. Perspective alone does not do the job.

II. The Position Within a Perspective. Particularized indexical properties or modes also embody the orientations or positions that referents occupy within a given perspective. In the example above, each of the `you' referents is individuated through a different place in a field of awareness vis-a-vis the speaker's unique vantage point identified with `here' and `now.' The referent itself occupies the position if reference is immediate, whereas a distinct content does when reference is vicarious, e.g., the photo in the Dome of the Rock case. For dynamic thoughts like, This is moving fast, where the referent retains its identity as a this though not its spatio-temporal locale, we can speak of an ordering of positions. For `I' tokens, the position is always that of the perspective's point of origin.

Thus, indexical individuation exploits a unique blend of three factors: (i) a generic form of encounter (an I, a you, a this, etc), (ii) a perspective, and (iii) a position or ordering of positions within that perspective. An item is an indexical referent by having a particularized indexical status, that is, by being encountered through a form of encounter at a position within a perspective.9 Such a finely-grained status is ephemeral, but, as indicated above, a token-solipsism that posits a special individuating mode for each indexical token can be avoided, thereby allowing instances of `I am I' or `this is this' to be trivially true.
 

Some Corollaries of This Analysis

With this account other properties of indexicals are readily derivable. First, indexicals are multiply irreducible. Since non-indexical mechanisms of reference fail to express the essential involvement with immediate experience none can serve as an analysandum of an indexical mode. Nor can the various indexical types be reduced to another, since each differs in associated form of encounter. Further, each individuating mode or property differs from others of the same type because of a difference either in perspective or in position within perspective.

Second, there is an essential subjectivity of individuating indexical modes. One person's `I,' `this,' or `over there' expresses, in part, what is unique to his or her perspective at the time of utterance, making it impossible for another to gain a cognitive fix on the very same item in precisely the same way (see chapter 14). For this reason, intepreting indexicals is not to replicate indexical production. Nor do quasi-indicators reproduce someone's indexical references but, at best, only approximate them.10

Third, because of the essential involvement with experience, indexical status is a relational or an extrinsic property of the referent. Just as something is not a you unless encountered as an addressee, so too, no one is an I apart from internal self-awareness. Of course, a referent of `I' is something intrinsically, namely, a "subject of experience" (see chapter 12), but its status as an I is possessed only qua object of experience, specifically, as the object of an episode of internal self-awareness.
 

From Indexical Mechanisms to Indexical Contents

It is tempting to conclude that first-person pronouns have no special ontological significance, that the irreducibility of `I' and other indexicals is rooted in their indispensible psychological roles, not in reference to unique sort of entities.11 But this is not Castañeda's view. While he did not follow Descartes in concluding that the first-person referent is a non-physical substance--preferring to leave this question open--he resisted attempts to confine irreducibility to statements about the psychological roles of referential mechanisms. Let us see why.

The conjunction of the satisfaction principle (above) with Castañeda's understanding of the irreducibility thesis implies that there are first-person properties possessed by the referents of first-person pronouns. I am an I, and so are you and any other internally self-aware being. But while we are the same to this extent, my particular I-ness differs from yours. Moreover, my I-ness now differs from that of last year, yesterday, or an hour ago. Some philosophers balk at this point; indexical mechanisms are just that, mechanisms or modes of apprehension--properties of the thinker not of the contents apprehended--and it is only a "sloppy thinker" who confuses a "privileged perspective" on ourselves with a "privileged picture" of what is seen (Kaplan 1989, 533-534).

Castañeda leaned towards a direct reference account of indexicals in his early writings (chapters 1 and 2) but abandoned it when he realized that by acknowledging indexical elements within attitudinal contents various properties of those contents can be explained. Specifically, first-person elements are integral not only to their motivational force of intentions but also to their inferential powers. Suppose I believe,

(9) I am obliged to award the Medal of Efficiency to Henry at 10 am.

To comply with this self-avowed obligation I must be able to pick out the appropriate items and link them to the appropriate elements in my commitment. How is this achieved? Indexically, of course. I see a small round object on the table and think that This medallion is the Medal of Efficiency; my attention is directed to the candidates seated in a row of chairs and I realize that That man is Henry; I glance at my watch and conclude that Now it is 10 am, the time to act. In each case, I accept observational statements of the form: i is the same as a, where i is an indexical and a is a non-indexical. By their means I infer from (9),

(10) I am obliged, all things considered, to give this medal to that man now.

and from this, I infer the proximal intention,

(11) I shall now give this medal to that man.

My action is explained by my acceptance of (11) and this, in turn, by my acceptance of (10). The inference from (9) to (11) could not be made if the sameness propositions I accept are of the form: a is the same as a. Were (10) the very same proposition as (9) then (9) alone should be sufficient for my inferring (11) and explaining my action. Since it is not sufficient, (10) must differ from (9), and the difference lies with the modes associated with the referring expressions.12

If content is indexical, then the subjective, privileged status of indexical reference is as much a property of what is apprehended as it is of an apprehending. The mere claim that experience is real does not imply this since non-perspectival descriptions of particular experiences are possible. But if there are truths expressible only in indexical language, as the irreducibility arguments imply, then perspectivity and subjectivity demand special ontological attention. Castañeda was not among the reductive naturalists at this point. Instead, he followed in the footsteps of Leibniz and Whitehead by taking perspective as an irreducible aspect of experience.
 

The Metaphysics of First-Person Reference: The Empirical Self

First-person modes of presentation enter into content because they are constitutive of first-person referents, for when we thinkingly refer we immediately grasp referents--individuals, logical subjects--not just Fregean Sinn. However, like Frege, Castañeda did not think that these graspable individuals are the distant, massive, physical objects posited to account for the unity of one's experience: "the total physical object is never before the consciousness of the perceiver, nor before his belief-rehearsing consciousness" (Castañeda 1977, 300; 1989a, 16-17). Instead, each individuating mode of presentation determines a distinct object of referential consciousness, a strict thinking referent. Each time I think about the Washington Monument, for example, I confront what I express with tokens of `the Washington Monument' or `that big obelisk'. I might also believe that there is an external enduring physical structure, but that massive continuant is not what I am directly aware of. At best, it is my doxastic referent, what I believe to be the semantic value of the type `the Washington Monument' in my idiolect. Similarly, while I might believe that `I' is correlated to a person, a disembodied soul or an organic physical system, the enduring me, that continuant is not immediately confronted. What is confronted a more finely-individuated strict referent of a first-person token.

Castañeda analyzed strict referents by means of his theory of guises. Each guise is a concrete entity of the form c{F1,...,Fn}, where {F1,...,Fn}is a set of properties constitutive of, or internal to, it and by which it is differentiated. The operator `c' expresses the transformation of a set of properties into a distinct concrete individual.13 Indexical thinking referents are distinguished by having a particularized indexical property within its core properties. Not being cognitive contents, alleged substrates find no place within thinking semantics, nor are the strict thinking referents massive external objects. All predication of external properties--those not included in the guise core--is analyzed in terms of various sameness relations among guises (Castañeda 1974, 1983c).

The first-person strict referent is an I-guise whose core internal property is a particularized self-property or first-person mode. Every first-person reference posits an I-guise as fully present (chapter 12); there are no past selves or future selves since past and future perspectives cannot be accessed (chapter 9). Nor is there a substrate underlying the history of a self; immediate consciousness is confined to properties and groupings of properties that make up individuals. There is no enduring I, and insofar as we speak of a "self" or "person" we are speaking either of the empirical self--a set of contemporaneous consubstantiated I-guises--or a temporally-ordered series of empirical selves (chapters 9, 11). Of course, there is a sense in which I am the same as the person who wrote a dissertation under Hector-Neri Castañeda's direction over 20 years ago. I wrote that dissertion. Which I? Well this I, the same one who is now writing this introduction. Yet on Castañeda's view, there is no genuine identity here, only a very tight sameness relation--transubstantiation--between present I-guises and previous I-guises that I can now refer to only vicariously.14
 

The Metaphysics of First-Person Reference: The Transcendent Subject

If the I exists only as an object of first-person awareness then a metaphysics of first-person consciousness should begin and end with a theory of the empirical selves, hence, with I-guises. However, Castañeda also accepted the Kantian view that an I think must be capable of accompanying all conscious states. He referred to the "I think" as a transcendental prefix but went beyond Kant in expanding it to include the here and now as well. The `I' of the prefix designates a transcendental I-guise, yet its full significance is not exhausted by a guise-theoretic analysis since it also points--not refers--to a transcendent self or subject existing beyond immediate thinking content (Castañeda 1986a, 106). This subject cannot be an I given his account of indexical status, yet it can be the "same" as various empirical and transcendental I-guises (Castañeda 1986a, 109-110). A similar point can be made for the `here' and `now' of the transcendental prefix.

The presence of `thinks' in the prefix provided Castañeda an opening to a metaphysical examination of the structures underlying self-consciousness. Not everything real can be an I. Since the pointed-to reality--the transcendent self--is something that thinks, then only thinking things are capable of being first-person referents, whatever else they might be. Unlike the ontologically promiscuous `this' and `that,' an `I' is appropriately applied only to subjects of experience, more exactly, to subjects capable of internal reflexive awareness: "the essence and substance of an I is just to conceive itself as subject qua subject" (chapter 9). The interesting metaphysical questions include these: What it is about an entity that makes it capable not only of internal reflexive awareness, but of self-awareness over time? What properties must an experiencing subject have that allows it to be an entity capable of self-consciousness and self-reference?

This last question is of special interest since, on Castañeda's view, the internal self-awareness that underlies first-person reference is not a universal feature of consciousness,15 and he thought that blindsight is a type of visual consciousness without self-consciousness. Accordingly, there must be some account of how internal self-consciousness arises within the broader domain of experiences, including those episodes of self-consciousness that are merely external. Like Kant, Castañeda was very much concerned with the unity of self within a single episode of consciousness composed of several experiences; thus, while the subject "pointed" to by the I of the transcendental prefix is not an internally self-aware being, it does correspond to a maximal unity of the experiences that are "internally unifed into one total experience by virtue of the being co-conscious of a manifold of contents." A detailed examination of this unification is the concern of chapters 12, 14 and 15.

There is no polished finalized theory in these last papers. Instead, we are left with a series of probings into the nature of the consciousness that makes self-reference and first-person thinking possible. On this score, Castañeda followed in the footsteps of Hume, Kant, and Whitehead who also endeavored to give a metaphysics of experiential processes. There are many interesting features of his treatment, and the reader is encouraged to examine them directly. The purpose of this introduction has been to indicate their genesis in Castañeda's novel account of first-person reference.
 


NOTES

 




1. See chapter 4 below. Castañeda noted that some occurrences of both `he himself' and `I' within oratio obliqua are replaceable, but the replacement sentences must contain other occurrences of those expressions. Quasi-indicators within the scope of a single operator, are ineliminable. He extended the notion of a quasi-indicator to cover expressions that can be used to represent other types of indexical usage, for example, `there' and `then' in `Yesterday, in the trunk, the heiress thought that it was quiet there then' to report a thought that a kidnapped heiress fomulates with `It is quiet here now.'
 

2. To thinkingly refer to something is to "pick it up in thought as a subject of properties and as an object of a propositional attitude" (chapter 14). Castañeda also spoke of singular reference as "reference to individuals insofar as they are thought of as individuals" and as "a fundamental phenomenon lying at the root of every exercise of our thinking powers" (Castañeda 1989a, 3). Yet, reference does not exhaust consciousness. To think referringly to an item is to think it qua logical subject, whereas we think predicatively of a property in attributing it to a logical subject (chapter 8). One can also think of things in a classificatory way, as when I think of blueness in thinking that the blue tie is expensive.
 

3. The linkage of identifying to referring is similar to what Evans called "Russell's Principle" (Evans 1982, 74, 89), namely that in order to think about a particular object one must be able to "discriminate the object from all other things" (p. 89). That Castañeda is committed to what I here call the satisfaction principle is evident in chapter 13 below: ". . . thinking in (perceptual) absence can reach its objects only if it segregates them from other objects by means of a set of uniquely possessed traits." This principle should not be taken as requiring an ability to reidentify the same item in another context; in one setting I might distinguish a square box by means of its color and location, but I may be unable to reidentify that same box if the background is altered.
 

4. Context-sensitivity does not distingish indexical reference since proper names are similarly context-sensitive (Castañeda 1989a, chapter 2). Nor is it enough to say that indexicals express the speaker's point of view. For Castañeda, every thinking reference is from the thinker's "point of view," never from an external or absolute point of view; there is no direct word-to-world connection that does not go through the thinker(s) to whom the words belong. This is the kernel of Kant's Copernican Revolution (see chapter 14 and Castañeda 1981, 285-286).
 

5. This conclusion is explcitly favored by some who offer token-reflexive rules like (I). For instance, Ruth Millikan writes: ". . . a token of "I" does not not tell me who the originator of that token is, that it is, say, Alvin. Rather, if I am to understand a token of "I", I must already know who the speaker is" (Millikan 1993, 270-271).
 

6. This point is overlooked by Carol Rovane who writes that one who uses `I' needs to know that she is the speaker, since one speaks with the intention of getting the interpreter to recognize that its referent is identical to the speaker (Rovane 1987, 161, 166). This view is inimical to Castañeda's insistence that thinking reference is different from communicational reference, and it fails to distinguish between referring with `I' and knowing that one refers with `I'. The latter requires knowing that the referent of `I' is also the speaker, but the former does not.
 

7. This argument is presented in Castañeda 1983b, 323. See also Evans 1985, 320 and Recanati 1993, chapter 4. Castañeda's emphasis upon the creative or executive role of indexical usage (1983b, 1989c, and chapter 13 below) suggests a limitation to the view that our use of language is primarily that of consumers (as advocated in Kaplan 1989, 602-603, and Millikan 1993, 86). To be sure, we are first exposed to language as consumers; the words we consume come with prepackaged meaning and our use of indexicals illustrates this as much as anything else. Yet the speaker produces the tokens, and with indexicals--perhaps more so than with descriptions and names--the speaker invests the tokens with content.
 

8. Indexical forms of encounter are ways that thinkers cognize referents, e.g., as an I, a you, or a now. While distinct from the interpretive forms of being the speaker, being the addressee, and being simultaneous with the time of utterance, they are equally public. Since each corresponds to a relational property of the referent, then we can understand how indexical expressions can be coupled with determiners, e.g., "an I", "a this", "her that," "qua you" and so forth. The common noun usage of indexicals helps explain quasi-indicators: when we say that `Yesterday at noon, Marlo thought that she herself was then happy' the quasi-indicator `she herself' is a proxy for Marlo's `I' because it expresses, for the attributor, `Marlo, thought of by Marlo qua I.' Similarly, `then' expresses `Yesterday noon, thought of by Marlo qua now.'
 

9. See chapter 13 which speaks of perspectival properties, and Castañeda 1990d, 303 where a contrast is drawn between determinable and determinate indexical properties. The situation is more complex for the modes associated with indexical descriptions like `the man in the corner' which include an additional sortal factor. Also, for vicarious reference it is not the position of the referent that enters the mode but that of the immediate content--the photograph in the example of the Dome of the Rock--to which the referent bears a special relation.
 

10. This is a matter on which Castañeda's views changed. In the 1980s he gave up his earlier view that the quasi-indexical that-clause in an attribution like (5) expresses the very same proposition expressed by the attributee's (1) (see Castañeda 1983a, 307). Lacking the essential perspectivity and subjectivity of indexicals, `he himself' does not express the full content of `I.' This consequence came to Castañeda "embarrassingly slowly" as he once put it, but it is a natural consequence of the fine-grained distinctions made in response to the puzzles surrounding thinking reference. In Kapitan 1998 I argue that quasi-indicators are abstractive devices, best construed as variables bound by implicit quantifiers. Although quasi-indicators do not capture the exact content of what the attributee thinks indexically, they do tell us what is being referred to (qua anaphors) and how the reference is being made.
 

11. See, for example, Lycan 1996, which urges that the failures of substitutivity for `I' can be "rendered unsurprising and ontologically harmless" (p. 56). Lycan argues that concepts can have the same extension but differ functionally by playing different roles in the guidance of behavior given various stimuli. Also, concepts can have similar functional profiles but differ in extension. Now a self concept or mental I is one which functions as an I, has its owner as the sole member of its extension, and no one else could use a computationally parallel concept of their own to designate the original owner. But this distinctness of the self-concept "reveals no ontological specialness of the self" (p. 58), and there is no warrant for positing "intrinsically subjective or perspectival facts" that serve as the objects of self-regarding attitudes (p. 68).
 

12. See Castañeda 1975a, chapters 6 and 10, and 1989c. Castañeda's account differs from direct reference views of indexicals that regard only the referents of indexicals as significant for semantic evaluations, not their mediating senses, concepts or modes (see Kaplan 1989). To be sure, these accounts do not share Castañeda's concern with immediate thinking reference, and in their view of indexical referents as enduring external objects it is understandable that they take indexicality as negligible in semantic evaluations.
 

13. See Rosenberg 1986a, 157-161, and Kapitan 1990b, 478-479 for more extended discussions of the concretizing operator.
 

14. See Castañeda 1974 (chapter 13 below) on transubstantiation. Despite the appeal to guises at this juncture, it is important to understand that Castañeda's account of first-person reference is compatible with a variety of ontological schemes, and does not stand or fall with the viability of Guise Theory. There are other accounts of objects qua modes that make fine-grained distinctions among thinking referents, for example, Whitehead's, to whose metaphysics of experience Castañeda's theory is similar. Both thinkers took episodes of experience as fundamental units and explained larger spatio-temporal structures in terms of relationships among experiences, relations that involve not only an integration of experiences but an amalgamation of their contents (see Whitehead 1929, Part III). Also, both found experience to be saturated with indexicality; in Whiteheadian terms, this is the vector character of experience whereby there is a transmission of content from a particular point to a percipient unification.
 

15. From 1970 on (see chapter 6) Castañeda argued for the possibility of an Externus consciousness, one without self-consciousness. In chapter 8 below he rejects the "Self-Ascription" views of Chisholm 1981 and Lewis 1979. Chisholm has since acknowledged the force of some of Castañeda's criticisms (Hahn 1997, 531). Even if one agrees that not all consciousness involves first-person awareness it remains open whether a consciousness of perspective--the perspective that happens to be one's own--accompanies every state of perceptual thinking. I have argued for this claim in discussing the views of Castañeda, Manfred Frank, and others in Kapitan 1998.