In 1974, six eminent philosophers gave a series of lectures at Oxford on the general subject of mind and language— that is, on the very subject of analytic philosophy itself. Among these lectures, which were subsequently published as the Wolfson College Lectures, Elizabeth Anscombe's paper on the first person undoubtedly gave rise to the liveliest debate and has continued to be debated ever since.11 G. E. M. Anscombe, “The First Person,” in Mind and Language: Wolfson College Lectures 1974, ed. Samuel Guttenplan (Oxford: Clarendon Press, 1975), pp. 45–65. References to this edition of Anscombe's text will henceforth be given as FP followed by the page number. The text can also be found in G. E. M. Anscombe, The Collected Papers of G. E. M. Anscombe, Volume 2: Metaphysics and the Philosophy of Mind (Oxford: Blackwell, 1981), pp. 21–36. As sometimes happens in analytic philosophy, the debate ended up fixated on a question whose canonical form is this: “Is the word ‘I’ a referring expression?” The import of this question is not self-evident. Why and how does it arise? In truth, it was not Anscombe's analysis as a whole, but only a few points within it, that attracted the attention of philosophers. Reading some critics, one might come away with the idea that Anscombe's essay can be reduced to the following question: “Can the word ‘I’ be interpreted as an indexical expression?” Anscombe is then credited with having imagined a situation in which the speaker would be able to use the pronoun “I” while nevertheless being bereft of anything that could be indicated indexically. If I were in a state of total anaesthesia, I would no longer have the slightest bodily sensation. I would therefore no longer be given to myself as an “empirical self” or “phenomenal body” [corps propre].22 [Translator's note:] The phrase corps propre is a term of art that Merleau-Ponty uses interchangeably with corps phénoménal to refer to the body as both “me” and “my own,” the body that “appears to itself in the course of causing the world to appear” (Pascal Dupond, “Corps” in Le vocabulaire de Merleau-Ponty [Paris: Ellipses, 2001], p. 9) and as distinguished from the corps objectif [objective body], the body as a mere thing. See Maurice Merleau-Ponty, La phénoménologie de la perception (1945; Paris: Gallimard, 1976), Part I. Unfortunately, the existing English translation is insufficiently precise with this term, usually translating it generically as “own body”; see Maurice Merleau-Ponty, The Phenomenology of Perception, trans. Colin Smith (1958; London: Routledge, 2005). Such a predicament would in no way keep me from complaining in the first person about my state or from making my intentions known. The critics then usually object that such a circumstance is exceptional and therefore cannot be of use in normal situations where a speaker lets it be known that he is speaking about himself. In other words, the critics have focused on three pages out of the twenty in the text as originally published. Moreover, nowhere in the text does Anscombe use the phrase “indexical expression.” Of course, Anscombe's reasoning is so dense that it is not always easy to grasp. Even so, it is striking the extent to which the critical readings carried out so far do not even attempt to determine the overarching argument of the text. In fact, Anscombe's essay does not begin with the canonical question about whether the word “I” is a referring expression. It begins with an analytical interpretation of the Cogito, the entirety of which bears on the question of whether the word ego—“I” in English—can serve as a proper name in the logical sense, that is, a term that provides a singular proposition with its subject. The phrase “referring expression” appears only quite late in the text. In a discerning article devoted to Anscombe's book Intention, Richard Moran and Martin Stone wryly draw a distinction between Anscombe the real philosopher who is revealed by reading her book, and a “transformed Anscombe,” an imaginary character more compatible with the dogmatic canons of the dominant philosophy.33 Richard Moran and Martin J. Stone, “Anscombe on Expression of Intention” in New Essays on the Explanation of Action, ed. Constantine Sandis (Basingstoke: Palgrave MacMillan, 2008) and reprinted in Essays on Anscombe's Intention, ed. Anton Ford, Jennifer Hornsby, and Frederick Stoutland (Cambridge, MA: Harvard University Press, 2011), pp. 22–75. I aim to show that the same distinction should be drawn with regard to Anscombe's essay on the first person. There is the Anscombe who said what she actually said and a different Anscombe reflected in the secondary literature. I will begin with a little hermeneutical exercise. Suppose that Anscombe's text had been lost and was only known to us through citations and rebuttals of it. Suppose that we then had to reconstitute her argument from this doxographical data, in the same way as was done with the pre-Socratic philosophers, about whom we know only what other ancient authors said about them. How would we understand her text? It seems to me that we would be dealing with an entirely different text, a text that might be easier to situate on a map of the schools of modern philosophy but that in no way would express Anscombe's position. For most of the critics of Anscombe's essay, three points seem to have been established: (a) Anscombe's thesis in “The First Person” is paradoxical; (b) that thesis is nonetheless not original because in many ways it takes up viewpoints that were already to be found in Hume and Lichtenberg and that were subsequently echoed by Wittgenstein; (c) finally, the thesis can be explained by the will to rule out anything that might result in a dualist conception of human beings (so that, if we could show that dualism could be avoided without falling into paradox, the thesis would be unnecessary). Here then is what a reader of the secondary literature attempting to reconstitute Anscombe's position from that literature might conclude on these three points: Anscombe's essay, it has been claimed, develops a Wittgensteinian argument in support of an unsettling thesis regarding the grammatical first person, that is, the use of the words “I” and “me.”55 The pronoun “I” used by Anscombe usually has to be translated into French as “moi” rather than “je” because it is the same word that she used—where “self” might also have been chosen—to translate Descartes into English. Yet the word “ego” is closer to “moi” than to “je” as Anscombe herself points out when she observes that, in Latin, the form ambulo [I walk] on its own suffices to indicate the first person. If you ask any English speaker what the function of the word “I” is in a sentence like “I am walking,” she will usually respond that the word allows the speaker to indicate that she is speaking about herself. In other words, a first-person utterance (i.e., an egocentric utterance) does indeed bear on someone, it has an object and this object is indicated by the word “I.”66 [Translator's note:] “Object” is here taken in its logical, rather than grammatical, sense: the object of a proposition is what the entire proposition bears upon. By definition, “I am walking” says something about the person speaking, and it is precisely the referential function of the word “I” to create an identification between the object the utterance is about and the person saying it. Anscombe, though, maintains that this is not the case: the word “I” refers to no one and is not used to indicate that the object of one's discourse is also its author, that is, the speaker. If the very definition of “I” includes self-reference, then what Anscombe says here is undeniably paradoxical.77 This assumption is what allows Strawson to write: “That the first personal pronoun ‘I,’ in the mouth or mind, speech or thought of some human being must refer to that very same human being (unless he or she is simply acting as the mouthpiece of some other individual) sounds like the veriest truism” (P. F. Strawson, “Reply to John McDowell” in The Philosophy of P. F. Strawson, ed. Louis Edwin Hahn [Chicago: Open Court, 1998], p. 146). Similarly, Evans writes that Anscombe rejects the axiom according to which “in using ‘I’ a subject is referring to himself” (Gareth Evans, The Varieties of Reference [Oxford: Oxford University Press, 1982], p. 258). The paradox is this: that which by convention is the instrument of self-reference is not in fact referential. Let us call this paradox the “irreferentialist thesis” (on the understanding that it only pertains to first-person pronouns and adjectives): the word “I” does not serve to determine the subject of predication of a first-person assertion; it is not used to make a reference to oneself. As a result, something that should not be a thesis at all, but rather an obvious fact of no particular significance, is transformed into a philosophical position: the thesis, which can henceforth be called “the referentialist thesis,” that endows the words “me” and “I” with a referential function. This paradoxical view derives from Wittgenstein, even though passages in the Blue Book make clear that Wittgenstein himself did not go quite as far as Anscombe in overturning the common view. Wittgenstein did indeed maintain that the use of the first person was referential only in certain cases: those where I use the words “I,” “me,” and “mine” to speak of an object (my body) and to state about that object (i.e., about my own body) something that I have observed. Because the utterance communicates an observation that I have made, it is constructed in the way that assertions about individuals are: it includes a predicate (what I observed) and a subject (the object on which this observation bears). Wittgenstein gives examples like: “My arm is broken,” and “I have grown six inches.”88 Ludwig Wittgenstein, The Blue and Brown Books: Preliminary Studies for the “Philosophical Investigations,” 2nd edition, ed. Rush Rhees (1958; Malden, MA: Blackwell, 1969), p. 66. There are two ways in which the utterance “my arm is broken” can be false: either the arm I am referring to is mine but not broken or it is broken but not mine. The situation is entirely different in the case of egocentric utterances like “I have toothache” or “I try to lift my arm.”99 Wittgenstein, The Blue and Brown Books, pp. 66–67. I do not have to observe a case in order to make such utterances and I therefore had no need to identify an object as being my own body. Wittgenstein explains that in the latter case, the word “I” is used “as subject,” by which I assume he means that it is used so as to provide the verb with a subject.1010 Wittgenstein, The Blue and Brown Books, p. 66. Shoemaker later derived from these pages his notion of “immunity to error through misidentification”: the use of the word “I” to express a form of consciousness is, as a matter of principle, invulnerable to any possibility of a mistake regarding the person.1111 Sydney Shoemaker, “Self-Reference and Self-Awareness,” The Journal of Philosophy 65:19 (1968), pp. 555–567, reprinted in Self-Knowledge, ed. Quassim Cassam (Oxford: Oxford University Press, 1994), pp. 80–93. A remark: In reality, Wittgenstein did not say that this use of the word enjoys protection against error (as suggested by the word “immunity”). He says that the grammar of “I have toothache” has left no room for the question “Am I really the subject of this toothache rather than my neighbor?” The error is excluded not psychologically but logically, with the result that the “immunity” in question protects against nothing at all. A subject of consciousness cannot mistake herself for another or mistake another for herself. Anscombe is then held to have radicalized Wittgenstein's irreferentialist position in her text so as to maintain that every use of “I” is a use “as subject.” Moreover, this thesis, it has been said, is not new. It is part of a tradition of hostility to the positing of a thinking subject that can be traced back, on the one hand, to Hume, and on the other, to Lichtenberg. As early as his 1968 article, Shoemaker referred to Lichtenberg's view and noted that Wittgenstein expressed his approval of it during his courses in Cambridge.1212 Shoemaker, “Self-Reference and Self-Awareness,” p. 80. In another article, “Introspection and the Self” (1986), Shoemaker writes that Wittgenstein and Anscombe derived from Hume's critique—according to which there is no “introspective awareness of a self or mental subject”—an argument in favor of “the Lichtenbergian view” that “the word ‘I’ does not refer (in Cassam, Self-Knowledge, p. 118).” Our hypothetical reader may well draw from such readings the conclusion that there is something that could be called a Lichtenbergian tradition of thought regarding the ego or self and that Anscombe's text belongs to that tradition. What Lichtenberg, Nietzsche, Wittgenstein, and Anscombe are held to share is the denunciation of the “I” as a linguistic illusion. When we express thoughts using first-person forms, we are tempted to take them to be personal thoughts. By saying “I think,” I speak as though the thoughts that come to me were really my thoughts, as though they could be attributed to a self [un moi], i.e., to a subject thinking within me. Or, in a formulation that is perhaps preferable: to me [à moi] as thinking subject. But where is there mention of a thinking subject or reference to a “self”? Is the presence of “I” in the phrase “I think” justified by the given that we seek to express, namely the phenomenon of thought? No, explain the Lichtenbergian thinkers, it is only grammar that obliges us to say “I think.” The verb “to think” requires a subject—our language insists on it—but that does not mean that the phenomenon of thinking provides us with evidence of a thinking subject. The same is true of “I think” as is true of “it is raining”: just as there is no reason to ask “who” is raining, there is no reason to believe that the word “I” appears in the phrase so as to indicate who is thinking when a thought is formed in my mind. And if there is no self, the thoughts that come and go are in fact impersonal. Since the thought “I think” does not truly provide a subject to which the existing thought can be attributed, one should instead say “it thinks (in me)” in the same way that one says “it is raining,” which would eliminate the illusion that the thought belongs to someone. Here we see the extent to which the philosopher who puts forward this view takes on the role of a critic of shared illusions. Each of us thinks we are the thinker of our own thoughts, but a deeper analysis shows that this belief must be challenged as a linguistic illusion. If that is Anscombe's reasoning, our reader might well say that there is an obvious weakness that, once pointed out, will reestablish the commonsense view. The weak link in the chain is point (B), which claims to derive from the “immunity from misidentification” enjoyed by the word “I” a proof that the referent of the word “I” is immaterial. There is a more natural explanation, though. If the use of the first person entails such an “immunity from misidentification,” it is not because the subject is infallibly able to identify itself (thanks to its immateriality) but because the linguistic convention governing the use of “I” exempts it from the application of any identity criterion: the word “I” need only emerge from the subject's mouth or pen for him to be established as the subject of predication to which whatever is said in the first person is then linked. In other words, “I” functions as an indexical expression. I do not have to name or describe the place that I am pointing to when I say “put the parcel here” to be able to designate it using the indexical term “here.” I have no need to tell you what moment of the day I call “now” when I tell you that I must leave now. Similarly, I need not identify the speaker about whom I am talking to you in the first person since, by definition, I am that speaker, regardless of the manner—whether correct or erroneous—in which I imagine my own identity. I will call the above presentation of Anscombe's argument the “Lichtenbergian Reading.” It boils down to this claim: the thoughts expressed in egocentric utterances (“I-thoughts,” as Anscombe calls them) are impersonal thoughts. These thoughts are “subjectless,” as she herself puts it (FP 65). Yet I believe that by revisiting each of the three points described above, it can be demonstrated that this reading is in fact untenable. It is true that she does challenge an entire tradition of reflection about the self. Her aim is precisely to reveal what she elsewhere refers to as “the principal root of the philosophical idea of ‘the subject’.”1313 G. E. M. Anscombe, “The Subjectivity of Sensation,” reprinted in Metaphysics and the Philosophy of Mind, p. 55. This does not, however, mean that she upends any commonsense views, for common sense has nothing to say about “referring expressions.” It is said that, by denying that “I” is a referring expression, Anscombe asserted a paradox, by which is meant a view that is contrary to common sense. But in order for there to be a paradox of this kind in denying that the pronoun “I” is a referring expression, common sense would have to have opinions about “referring expressions.” Yet, taken in this sense, the term is a technical one that was introduced only recently for the needs of linguistic philosophy. In its nontechnical use, the term “reference” is more likely to be used for the act of directing someone toward a source of information or to provide confirmation (one commonly asks of a job candidate: “Do you have references?”).1414 Gareth Evans begins his book by drawing up a list of expressions that have “traditionally,” according to him, been classed as referring expressions. But there is no tradition in that area: the term “referring expression” has only been in use since the middle of the 20th century within a philosophical milieu influenced by Strawson. For his part, Edward Harcourt does not hesitate to talk about a “standard” and well-established definition of “referring expression.” See his “The First Person: Problems of Sense and Reference” in Logic, Cause, and Action: Essays in Honour of Elizabeth Anscombe, ed. Roger Teichmann (Cambridge: Cambridge University Press, 2000), p. 26. What would a paradoxical point of view be? It would be to say that we do not know whom a speaker expressing himself in the first person is talking about. In fact, referentialists consider the question from the perspective of the hearer. The common sense that they invoke consists in saying that the word “I” allows the hearer to determine whom the utterance is about. Yet Anscombe never disputed that this is how things are for the hearer. When a speaker states something in the first person, the hearer can always translate it into a third-person statement made by others (other hearers) about that same speaker. Moreover, Anscombe took this possibility into account so as to dismiss it as irrelevant to the problem. To this end, she formulated a semantic rule governing the use of the first person. She even invented a stock character that she sets up as her interlocutor and adversary in this dispute and whose job it is to invoke this rule. She calls this character “the logician.” Here is the rule: “If X makes assertions with ‘I’ as the subject, then those assertions will be true if and only if the predicates used thus assertively are true of X” (FP 55). The same rule is evoked later in the text: “If X asserts something with ‘I’ as subject, his assertion will be true if and only if what he asserts is true of X” (FP 60). According to Anscombe, this rule merely reminds us—the hearers—of the connection that we make between an assertion in the first person and one in the third person. It is by following the rule that we hearers move from the first sort of expression to the second. It does not follow from this that speaker X has made a reference to himself. And, indeed, the speaker did not mention or name X. The name “X” appears nowhere in his sentence. The hearer therefore must have observed that the first-person utterance came out of the mouth of individual X and this allows him to attribute the fact of its assertion to X rather than to Y or to Z. It is therefore the hearer who identifies the subject of predication, while the speaker has only provided an indication allowing that identification to occur. Anscombe's final remark about first-person expressions in Latin is along the same lines: when it is identified by the inflection of the verb, as is the case with ambulo [I walk], it is the “-o” ending that is the mark of the first person and it is more natural to see the inflexion as a signal provided to the interlocutor than as information about someone's identity. Anscombe herself alludes to this in a passage where she states in a positive mode what the function of “I” is: a signaling rather than a referential function (FP 56). Does Anscombe have an impersonalist conception of thought? That is what we are led to believe by the reading that evokes the tradition leading from Hume to her by way of Wittgenstein. Yet from the very beginning of her essay, the entire argument is organized around a remark on the distinction to be drawn between “Descartes thinks” and “I think” (when spoken or written by Descartes). Anscombe emphasizes that the argument of the Cogito—in Augustine as in Descartes—is irreducibly in the first person and cannot be transposed into the third. Finally, the entire second part of the text, which develops the consequences of the “general thesis” (FP 60), bears on what she calls “I-thoughts” or thoughts in the first person. These thoughts cannot be wholly reconstituted in the third person. The problem is not that they are “subjective” in the sense of being incommunicable (or communicable only by an analogy between what happens to you when you think of yourself and what happens to me when I think of myself). For they are both intelligible and communicable, but the means for transmitting them is either to report them in indirect style or to cite whoever produced them. This is why Anscombe feels it important to reflect on reported speech as a technique of attribution, and, in particular, on the difference between reflexive pronouns in direct discourse, as in “Peter locked himself in his room” and reflexive pronouns in indirect discourse, as in “Peter recognized himself as the author of his predicament.” In the latter case, moving to the third person does not eliminate the egocentric character of the thought since “himself” is not anaphoric (i.e., does not refer to the subject of the sentence, “Peter”) but corresponds to the “I” of an assertion that would say: “I am myself the author of my predicament.” Thus, Anscombe's position in her article is not that our first-person thoughts are in reality impersonal even though we express them as though they were personal. The fact that they are “subjectless” does not mean that they are not reflexive personal thoughts. For one thing, the I-thoughts that Anscombe is discussing are personal thoughts, in the sense that they express the consciousness a person has about herself. Anscombe provides examples of these: “I have from time to time such thoughts as ‘I am sitting’, ‘I am writing’, ‘I am going to stay still’, ‘I twitched’” (FP 61). For another, such thoughts are reflexive because the only way for us to attribute them to someone is to have the person speak. We cannot report them using the third person, that is, by taking the pronouns “he,” “she,” “himself,” “herself,” in their anaphoric function. If we were to attempt it, we would have to attribute to Descartes, when he is thinking “I am not Descartes,” the thought that Descartes is not Descartes; or to Anscombe saying “I am Anscombe,” the thought that Anscombe is Anscombe (FP 53 and 64). The only way to recreate the content of such declarations is to cite the speakers directly or to report their words in indirect style; the reflexive pronouns “herself” and “himself” would in that case have the value of the first person, meaning that they cannot be replaced by nouns or demonstrative expressions. The thesis Anscombe defends in this text is thus not that first-person thoughts, being subjectless, are therefore impersonal (i.e., are not anyone's thoughts). Instead, the thesis is this: what makes it that my thoughts are mine is not that there is the right connection between them and a subject identified as being the self [le moi] that thinks them; rather, it is that there is the right connection between them and me (me, the human individual, which is something entirely different from an ego or self [un moi] that is supposedly mine and that is meant to be found within the content of my experience). In other words, the entire purport of this text is to give a measure of the gap that separates the ordinary use of first-person forms—those that allow me to speak about me [moi], this individual that I can also name for you (save in cases of amnesia or madness)—from the use instituted by philosophers, one that seems to authorize references to the ego (ego ille), the self, the moi, and so forth. Lichtenberg, in his famous aphorism, states that it would be more accurate to say “it thinks” than “I think.” Anscombe says nothing like this. At no point does she suggest that the ordinary use of first-person forms is a source of mystification. The distinction she draws is not between “I think” and “it thinks,” but between “I think” and “Descartes thinks.” The myth of the self arises not out of ordinary usage but from philosophical speculation that, while claiming to interpret this usage, in fact misapprehends it. We become conscious of certain representations that are not dependent upon us; others, at least we believe, are dependent upon us; where is the boundary? We know only the existence of our sensations, representations, and thoughts. It thinks, we should say, just as one says [in German], it lightnings [es blitzt]. To say cogito is already too much if we translate it as I think. To assume the I, to postulate it, is a practical necessity.1515 Georg Christoph Lichtenberg, Philosophical Writings, ed. and trans. Steven Tester (Albany: SUNY Press, 2012), p. 152. The reasoning that is often imputed to her begins by conceding to the referentialist that the word “I” appears to be referential. It then further concedes to Cartesian philosophers that, if the word refers, it must refer to a Cartesian ego. It is therefore for want of an ego to name or designate in one way or another that the word fails to refer. Why is “I” unable to have a referential function? One can only respond that what prevents the word “I” from being referential is that there is no self, or ego, or moi. The word “I” is then an empty name, the name of a fictive entity. It is as if one were saying: if the name “Santa Claus” were to name something, it would have to designate a bearded character, living in the North Pole, and so forth. Yet this character is fictional. Therefore the name is empty. However, Anscombe claims that the word “I” is not a logical name. She does not claim that it would be a logical name if one could only believe in the existence of a Cartesian ego. It is not because we cannot believe in the Cartesian ego that we must stop interpreting the word “I” as a referring expression (in the way—or so the critics believe—common sense does). The reason is the contrary: because we cannot say how “I” could perform a referential function—whether to the body or to a Cartesian ego—we must abandon Cartesian dualism of the self and the body and, especially, the contemporary dualism that arises out of Locke between the person and the human individual. Suppose now that we have just come into possession of Anscombe's original text and are able to get acquainted with what it actually says for the first time. We note that the article begins with reflections on the argument of the Cogito. It is good practice to assume that such an opening is not a mere appetizer but sets down the object of the entire discussion, which is to provide a philosophical explanation of the first person (FP 64), in other words, of self-consciousness. Anscombe presents three versions of the argument known as the Cogito. This argument, in her understanding, is in fact in two parts, beginning with the establishment of a first truth that the sceptic cannot challenge and then presenting a corollary argument in favor of a dualism of the soul and the body. She reminds us that, in Augustine, the argument is expressed as follows: si fallor, sum [If I am mistaken, I exist]. From this truth, Augustine concludes that the soul is immaterial, having observed that the mind knows itself directly. In Descartes, the argument is also twofold since it must establish the existence of the thinker and, from there, the real distinction between the soul and the