Imagining Oneself To Be Another Steven L. Reynolds Noûs, Vol. 23, No. 5. (Dec., 1989), pp. 615-633. Stable URL: http://links.jstor.org/sici?sici=0029-4624%28198912%2923%3A5%3C615%3AIOTBA%3E2.0.CO%3B2-S Noûs is currently published by Blackwell Publishing.
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/journals/black.html. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.
The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academic journals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers, and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community take advantage of advances in technology. For more information regarding JSTOR, please contact
[email protected].
http://www.jstor.org Sun Jul 22 18:41:41 2007
Imagining Oneself To Be Another ARIZONA STATE UNIVERSITY
The imaginative activity naturally described as imagining oneself to be suggests a difficult question. When one engages in this activity, what exactly does one imagine? I shall argue that the difficulties that lie in the way of an answer to this question require us to say that, in spite of appearances, in some cases of such imagining, one is not imagining anything about oneself. Instead one imagines someone else in a first personal way. My question about what one imagines should not be confused with the semantic question about what kinds of things can be values of the variable 'y' in 'x imagines y'. Possible answers to that question include propositions, property objects (Lewis, 1979), states of affairs, events, or perhaps, in some cases, the things named in the report, such as Napoleon. I want to know which propositions, which property-objects, or which states of affairs are imagined in a given episode of transference imagining. Whether the entities imagined are propositions, property objects, or states of affairs does not concern me. The question as to the exact content that is imagined in transference imagining has some broad philosophical interest, for philosophical thought experiments often invoke transference imagination. For example, when discussing personal identity, Sydney Shoemaker asks us to adopt the first person point of view on his brain transplant example. We are asked to imagine ourselves waking up as a person who has Robinson's body and Brown's brain. (Shoemaker 1963) In defending his version of utilitarianism, John Stuart Mill asks us, by implication, to imagine ourselves as satisfied pigs, and to compare that sort of life tr, our present lives. (Mill 1861, 11- 14.) These imaginative exercises are expected to produce
~ 0 6 23 s C
(1989) 615-633 1989 by NoCs Publications
insight into philosophical issues. Yet it is difficult to say just exactly what we are being asked to imagine.
Here is an example of the sort of case I have in mind. Suppose I have been reading a vivid account of Napoleon's conduct during the battle of Waterloo. Putting the book aside for a few minutes, I actively imagine myself being Napoleon in the battle of Waterloo. I imagine that I am mounted on horseback, watching the battlefield. I see my aides come and go bearing messages and I hear the cannons roar in the distance. Typically, while I thus imagine, various mental images occur to me: I "see" the horse on which I am mounted and the armies clashing in the distance, I "hear" the hoofbeats, the shouts and screams, the roar of the guns; I "smell" the smoke of gun powder. Perhaps I also narrate some of the imagined events to myself in words. Active imagining of this sort is plausibly regarded as a kind of story telling. The imaginer tells himself a story using mental images, and perhaps some narration, in the way movie makers use images and sounds to tell stories. In this particular case I tell myself a Napoleonic story using mental images and perhaps some supplementary commentary. When I am asked what I have imagined, it seems that I should respond by saying what happens in this story. An important consequence of this conception of active imagining as a kind of internal storytelling is that, in my reports of what I have imagined, I am not simply reporting the content of a current propositional attitude of mine, as if it were one of my current beliefs or desires. I am instead in the position of an author saying what happens in a story he has written. The author's own account of what happens in the story may be a valuable aid to determining what really happens in the story, particularly if the author has a good memory of his intentions while writing the story. But it may easily be in error, and it is certainly subject to correction. In answer to the question as to what I have imagined I am inclined to say that I have imagined that I am Napoleon. But philosophers have argued that, if this account of what I have imagined is taken literally, that is, as a claim to have imagined that a certain false identity statement is true, then it is mistaken. I haven't really imagined that I am Napoleon in that sense. The following argument for that conclusion has been widely discussed. (Williams 1966, p. 44. Vendler 1976, 1984. Mackie 1980) When I say "I imagine that I am Napoleon" I apparently claim to imagine that a false identity statement is true: "I am Napoleon".
I ~ ~ A C I I V I NONESELF G TO BE ANOTHER
617
It isn't possible for me, this distinct individual person, to be another person, such as Napoleon. For distinct individuals are necessarily distinct, and hence not possibly the same. Furthermore, it is assumed, the impossible cannot be imagined. It follows that I cannot imagine that I am someone else, and hence that I did not imagine that I was Napoleon. Thus, imagination of the sort ostensibly reported is impossible. Therefore the report is not to be taken literally, but is instead to be regarded as a kind of idiom requiring interpretation into plain language. We thus require an account in plain language of what has been imagined, even after we have the imaginer's report. Unfortunately, this argument makes it appear that difficulties arise only if we accept the sssumption that imaginability entails possibility. It seems that we might avoid the alleged difficulties by rejecting this assumption and admitting that we can imagine impossibilities. In answer to the question as to what is imagined in the Napoleon case, we could then accept my apparent claim that I have imagined myself to be identical to Napoleon. No other account of what has been imagined would be needed. However, there is another kind of difficulty, which survives the rejection of the assumption that only the possible can be imagined. This other kind of difficulty was first suggested, so far as I know, by Richard Wollheim. For me to imagine that I am Napoleon is not to imagine that Napoleon is me. (I have changed Wollheim's example.) But identity is a symmetrical relation. Therefore, I do 1984, not imagine that I am identical to N a p ~ l e o n .(Wollheim ~ 72-76.) Let me fill out this argument a little: "I am Napoleon" evidently expresses the same cognitive content as "Napoleon is me." If these sentences express identity statements, then they should differ from one another in cognitive content no more than "Cicero is Tully" differs from "Tully is Cicero". (Purely grammatical changes in pronoun and verb should not affect the cognitive content expressed.) If p expresses the same cognitive content as does p ' , then anyone who imagines p should also imagine p '. Indeed, it seems that anyone who imagines that Cicero is Tully also imagines that Tully is Cicero. Yet, although I imagine that I am Napoleon, it seems false to me, the imaginer, that I imagine that Napoleon is me. Therefore what I imagine is not that I am identical to Napoleon. It is evidently false that when imagining that I am Napoleon I also imagine that Napoleon is me. To speak metaphorically, it gets the direction of fit wrong-it is fitting Napoleon into me, rather than me into Napoleon. I don't imagine that Napoleon is me, for that would, it seems, be imagining that Napoleon is Steven Reynolds. (This is one respect in which imagining that I am Napoleon is not like believing that I am Napoleon. If I believed that I was Napoleon,
I would believe that Napoleon was me, and so I would probably believe, for example, that Napoleon is working on this paper. I certainly wouldn't imagine that Napoleon is working on this paper.) The previous argument held that the claimed content of imagination is not a possible content of imagination, and therefore could not have been the content of my imagining, whatever I say. The Wollheim-inspired argument just given holds instead that the claimed content would not be acceptable to the imaginer himself as an account of what he has imagined, if he were to think that content as expressed in only slightly different language. It thus seems that the content apparently imagined is not the content really imagined. In saying "I imagine that I am Napoleon" I am not reporting having imagined that a false identity statement is true. If that is correct, then what have I imagined? It seems that the answer to this question hinges on the answer to the question as to whom I imagine. When I imagine that I am Napoleon riding a horse, whom do I imagine riding a horse?
An adequate answer to my question should: 1. Permit a reasonable explanation of the ordinary use of the first person pronoun, ' I , , and the name of the other person (e.g. 'Napoleon') in unreflective attempts to say what is imagined in transference imagination. 2. Be compatible with the observation that transference imagining is essentially imagining from inside. 3. Allow that the obvious logical and evidential consequences of what is explicitly imagined are, with some exceptions, also imagined to be true. ("Evidential" consequences are those supported by the available "evidence" together with the appropriate background information.) Let me summarize my reasons for thinking these are reasonable requirements. The first requirement: It is observed that I use 'I' and 'Napoleon' in reporting what I imagine while imagining that I am Napoleon. For example, I might say "I, Napoleon, am witnessing a battle. I see my aides approaching me, and I shift uneasily on my horse, trying to prepare myself for bad news." If I am imagining Napoleon witnessing this battle, then it seems a mistake to use 'I' in this report; if I am imagihing myself, then it seems a mistake to use 'Napoleon'. If I am imagining someone or something else then it seems wrong to use either. Yet it is our practice to thus use such words in reporting what is transference imagined. Furthermore, the use of these words is not only sanctioned by custom, but it seems right. When I am reporting what I im-
IMAGINING ONESELF TO BE ANOTHER
619
agine, my use of 'I' and 'Napoleon' seems to be a principled extension of their correct use in other contexts. The first requirement asks for an explanation as to how these words can be correctly used in reporting what is transference imagined, or, if such use is a departure from the correct usage, why it seems to be an appropriate departure. An account that doesn't permit such explanation is unacceptable. The second requirement is reasonable because transference imagining is essentially imagining from inside, not from outside. The distinction between imagining from inside and imagining from outside is easy to grasp, but difficult to draw in an informative way. Dispositions to re-tell the story in the first person and to react emotionally in certain ways to what is imagined may help to constitute a bit of imagining as imagining from the inside. But I think that the primary difference between imagining from inside and imagining from outside is a difference in what the accompanying mental images represent. Mental images represent or. stand for things in pretty much the way that images in representational painting stand for the things depicted. What a given image stands for will be largely determined by the intentions of the painter, or in this case by the intentions of the imaginer. So two qualitatively indistinguishable images could stand for different things, if the intentions of the imaginer in producing the images were different. (Williams 1973, p. 30) I claim that the difference between imagining from inside and from outside is a difference, not in the qualitative character of the images that are produced, but in what the images are intended to represent. The same qualitative types of images would represent one imagined situation if I were imagining from inside,, and a different imagined situation if I were imagining from outside. For if I said what I imagined by visualizing wounded men when I was imagining from outside, I would say "I imagined that there were wounded men". If I said what I was imagining by visualizing wounded men when I was imagining from inside, I would say "I imagined that I saw wounded men." There need be nothing visualized which corresponds to the second occurrence of 'I' in that last report. The visualization may be qualitatively exactly similar in the two cases. Yet, in imagining from outside, that type of visualization will have the effect that it was imagined that there were wounded men, while in imagining from inside it will have the effect that it was imagined not only that there were wounded men, but also that someone (I) saw them. Genuine transference imagining cannot be imagining from outside. For, in any case of this sort of active imagining, what I im-
agine is determined by my visualizations (in the extended sense of that word) and by my intended background story. Any difference between merely imagining Napoleon, and imagining from outside that I am Napoleon, would have to show up either in the visualizing or in the intended background story. No such difference appears. For there need not be any qualitative differences between visualizing myself from the outside being the real historical Napoleon, just as he was, and just visualizing Napoleon as he might have looked to an observer. If it's the historical Napoleon I imagine in both -ases, then it seems there wouldn't be any substantial difference in the background story either. So imagining from the outside that I am Napoleon would not differ from just imagining Napoleon from the outside. It would not be transference imagining. Thus transference imagining is essentially imagining from inside. An adequate answer to the question about what is imagined in transference imagining must be compatible with this essentially inside character. The third condition holds that the account of what is transference imagined should be compatible with the claim that the obvious logical and evidential consequences of what is explicitly imagined are, with some exceptions, also imagined to be true. As I shall argue, the fictional truth associated with a particular fiction includes all of the obvious logical and evidential consequences of the explicitly asserted or displayed truths of that fiction. Since what is imagined is a species of fictional truth, this principle also holds for what is imagined. How are evidential consequences arrived at in the case of fiction? T o avoid some not very relevant complications, let us just suppose that, with respect to a certain fiction, there is a body of fictional truths that are explicitly asserted or shown in that text, movie, play, etc., and that these truths are not to be controverted by the reader or viewer in working out what really happened in the fiction. These fictional truths function as "evidence". In a novel it may be incontrovertible that the narrator said that p, even if, later on, we shall have to conclude that he was lying when he said it; in a movie it may be incontrovertible that the detective appeared to drop his cigarette case accidentally, even if we must later conclude that it was no accident. The reader or viewer is expected to reason from this "evidence" to conclusiops about what "really happened" in the fiction, just as he reasons from similar evidence in his non-fictional life.(One's reasoning with respect to a given fiction is often also expected to be guided by the known conventions of the fictional genre. But such complications apparently aren't relevant to the case of active imagining.)
1hlAG11\ilNG
ONESELF T O BE A A T 0 T H E R
621
Let F be a fiction. Various things are explicitly "asserted" or shown in F. We will count these as the "evidence" of F. If the counterfactual "If it were the case that . . . (here a statement of all the "evidence" presented in the fiction F) . . . , then it would be the case that q" is true with respect to the relevant world, then q is an evidential consequence of the evidence presented in F. It is controversial in which world that counterfactual conditional must be true. The relevant world is evidently determined by what background information is relevant to understanding the fiction. Some hold that the conditionals must be true, as counterfactuals, with respect to the actual world, while others say that they must be true instead with respect to the world as the author (or his primarily intended audience) believed it to have been. (Lewis 1978) Other views are defensible as well. For my present purposes there is no need to take a side in this controversy. Why should we hold that these obvious consequences are also true in the fiction? Reasoning from the "evidence" is essential to understanding fiction. Consider for example the indirect effects sought by the writer of prose fiction. From a few asserted details, such as that Sherlock Holmes is again using cocaine, Conan Doyle expects his informed audience to conclude something about Holmes which is needed to understand the further development of the story: that Holmes is bored and in need of intellectual stimulation in the shape of an unsolved mystery, however insignificant. The necessity of such reasoning for understanding the story is even more evident in the case of visual media such as plays and movies. Often very few of the important points of the story are asserted in such fictions; we are given images of behaviors and sounds and we have to infer the relevant point: e.g. that the heroine is dissatisfied with her job. A similar sort of reasoning is essential to the kind of pictorial imagining I have been describing. While imagining that I am Napoleon I have an image of an aide approaching me with a panic-stricken face, and, qua audience for this fiction, I conclude that the battle is going badly. (There is an appearance of paradox here. If I am telling myself this story, why should I need to reason my way to conclusions about what is happening in it? The answer of course is that these images frequently occur to me when I do not consciously intend them to occur, and my understanding of the story must then be developed around them. Imagining wouldn't be interesting if there weren't unexpected events in it.) If reasoning is needed for understanding the story, or for collecting what happens in the story, then it follows that the products of correct reasoning from the evidence are true in the fiction (or perhaps the products of whatever reasoning was intended by the author, who must expect his readers to reason
correctly in most respects). In particular, the obvious consequences of what I imagine must also be imagined to be true. Some have claimed that fictional truth is closed under logical implication and the appropriate sort of counterfactual reasoning. (Lewis 1978) But this is too strong. Authors are sometimes inconsistent, and, if they are, then on such a closure principle anything would be true in their stories. Yet if these stories are to be understood at all, it must be the case that not everything counts as true in them.4 In some of the Sherlock Holmes stories Watson is said to have an old wound in his leg, and in others he is said to have an old wound in his shoulder. Yet it does not seem to be fictionally true that he has wounds in his leg and in his shoulder, for the two wounds are never mentioned together. If, as in this instance, the conclusion is clearly contrary to the author's intentions, we refrain from drawing it. So closure fails. Nevertheless, it seems that a weaker rule, admitting as true in the fiction at least the obvious and intended logical and evidential consequences of explicitly asserted fictional truths, is required for a plausible account of the way in which we understand fiction. The account of what is imagined in transference imagining should not have obvious consequences which the imaginer would not accept as part of what he has imagined. If it does, that is at least an indication that the account of what is imagined is incorrect.
Since we have rejected the claim that in transference imagining one imagines oneself to be identical to another, there remain four alternatives as to whom one imagines in transference imagining. One might hold that the imaginer imagines 1) no one, 2) himself, 3) some other person or entity, or 4) the person he imagines himself to be.5 The first alternative, that I imagine no one, seems quite implausible to me, so I won't spend much time on it. When I imagine that I am Napoleon, riding a horse, receiving messages, giving orders, watching the battle, surely I do not imagine all of these actions and events as not being done by or happening to anyone. True, the person to whom these things are imagined to happen has that peculiar offscreen quality that the self assumes when one tries to locate the self. But it is not plausible that in transference imagination we imagine actions without a subject. Even if my uses of 'I' in the reports of what I transference imagine are non-referring, (as some have suggested with respect to all occurrences of 'I7) I still must be imagining someone as the subject of the imagined actions.
IMAGINING ONESELF TO BE ANOTHER
623
The second alternative holds that in transference imagining one imagines oneself. When one sees this sort of imagining as a species of fiction this view seems very peculiar. Fictions told in the first person are seldom confused with stories about the author. No one supposes that it is true, according to the Sherlock Holmes stories, that Dr. Watson, the first person narrator of the stories, is Sir Arthur Conan Doyle. But the view that in transference imagining I am imagining myself seems to find more supporters than does the corresponding view about fictions told in the first person. I don't think that it should seem plausible on closer examination. If in transference imagination I imagined myself, what I imagined would frequently be impossible. In imagining that I am Napoleon I may imagine for example that I am the son of Carlos and Letizia Buonaparte. But (it is commonly claimed) I am not possibly the child of those parents. So, if I am imagining myself being the son of Carlos and Letizia Buonaparte, what I imagine is i m p o ~ s i b l e .As ~ we have noted, some fictions do contain impossibilities, and even contradictions. But they are disruptive. They tend to confuse us about what is fictionally true. So other things being equal, we would like to avoid an account of what is transference imagined that says we imagine impossibilities. Another difficulty about the alternative that in this particular kind of transference imagination I imagine myself, is that none of the expected implications about me seem to be fictionally true. Perhaps all I explicitly imagine, all that I provide myself in the way of evidence from which to reconstruct the world of this fiction, is visualizations and narration of the events of the battle. Then it seems that the person referred to by using 'I' in the report of transference imagining ought to be interested in philosophy, at least when not leading the armies of France. For it is true that if I led the armies of France at the battle of Waterloo, gave such and such orders, and so on (all that can be concluded directly from the "evidence" that is presented in the imagining), then when the battle was over, and I had recovered from all of the excitement, I would want to read philosophy. Yet it seems to me that when I imagine being Napoleon I don't imagine that I would want to read philosophy. Perhaps all such implications are barred by my intentions as to how I shall imagine myself. If so, then it seems that I am imagining myself to lack every distinctive property I have, and instead to have every property that Napoleon had. But if I do that, what is left of me in this imagining? How can it be imagining myself? This is not a Fregean claim about the semantics of the first person pronoun. I am quite willing to believe that one does not assert a
contradiction in asserting that F(1) for almost any predicate F. (The predicate 'do not exist' may be an exception, if, as David Kaplan holds, 'I do not exist' is a logical falsehood. (Kaplan 1979)) If Frege was wrong about proper names having senses, then presumably there is a similar lack of contradiction in all assertions of the form F(N) where N is a proper name, such as 'Winston Churchill'. Nevertheless, if someone claims to be imagining Winston Churchill at the Battle of Waterloo, but, as he imagines him, Churchill isn't English, he doesn't smoke cigars, he lacks the bulldog determination, and as to his positive properties, he is French, is called Napoleon, etc. etc., then he is not imagining Churchill. He's just imagining Napoleon. If there aren't at least some fictional implications to the imagined character's being Churchill, which wouldn't equally be fictional implications of his being Napoleon, then he isn't imagined to be Churchill. Likewise, even if I would naturally claim to be imagining myself, surely I'm not really imagining myself if there's nothing about me in what is imagined. The third alternative, which is sometimes motivated as a response to the difficulties about the possibility of what is imagined which were found for the proposal that I am imagining myself, is that I imagine some third thing, neither me nor Napoleon. I will use the term 'non-standard self for whomever or whatever I transference imagine, other than Napoleon and myself, if there is such a thing. I doubt that the view that I imagine a non-standard self will satisfy any of the three requirements for an acceptable answer to my question. There seems to be no reason why I should call this non-standard self 'I' and 'Napoleon', so the view that I imagine a non-standard self will not satisfy the first requirement. Nor will the view that I imagine a non-standard self be compatible with the essentially inside character of transference imagination. Imagining that this non-standard self is playing Napoleon's role at the battle of Waterloo, even in Napoleon's body, need not be imagining that I am Napoleon. I could imagine from the outside that this non-standard self had the relation to Napoleon which it now has to me, thus engaging in transference imagination from the outside, according to this account. But there is only one way of imagining that I am Napoleon, and that is from inside. Nor is it easy to find a candidate for the non-standard self that could satisfy the last requirement, that what is imagined should include all the obvious logical and evidential consequences of what is imagined. What sort of thing could such a non-standard self be?
IA4AGINING ONESELF T O BE ANOTHER
625
As has just been suggested by the special case of myself, any object one imagines will import some of its own character, in the shape of evidential consequences, into what is imagined. Thus my Cartesian Ego, if it exists, has lots of properties compatible with the evidence provided by my imagining that I am Napoleon-an interest in philosophy, for example. Yet I do not imagine that I am interested in philosophy in imagining that I am Napoleon in this battle. There may also be difficulties about the possibility of what is imagined, if my Cartesian Ego has essentially the property of coming into existence at a certain time.
Having rejected the alternative views that I imagine no one, that I imagine myself, and that I imagine a non-standard self, we are left with the view that I imagine Napoleon. There are problems for this view too, but I think they can be solved. There shouldn't be any difficulties about the obvious logical and evidential consequences. If I were imagining Napoleon, then what I imagined when I imagined being Napoleon at the Battle of Waterloo would include the obvious logical and evidential consequences. In principle, what I imagine could even be possible, for the predicates used in an accurate summary of what I imagined would likely be only those that were or might have been true of Napoleon. Of course if I were imagining Napoleon, then there would be no obvious evidential consequences based on my own character. The requirement that the account be compatible with the essentially inside character of transference imagining leads to two difficulties for the view that I am imagining Napoleon. (a) The first objection arises from the qualitative character of these imaginings. While engaging in them, I seem to be involved in what is imagined in the same way that I am involved when I imagine myself doing something. I do not seem to myself in these imaginings to be merely looking on from behind Napoleon's eyes, as it were, but to be thinking, deciding, and feeling, as when I imagine myself doing something. So it seems I am imagining myself engaged in these events, rather than Napoleon. (b) Apparently I cannot imagine someone else from inside, for that would be imagining that I experience his feelings, perceptions, etc. from inside. It isn't possible for me to experience someone else's (e.g. Napoleon's) conscious states from inside, so I cannot imagine that I do. (a) An answer to the first objection is suggested by Bernard Williams's comparison of transference imagining to an actor's playing a part.8 (Williams 1973, pp. 44-45) When Charles Boyer played
Napoleon, events that happened to Napoleon and actions performed by Napoleon according to the story were represented by events that happened to Boyer and by actions performed by Boyer as he was playing the part. Thus Boyer would not have felt himself a mere spectator of the story, but would have felt himself to be active in it, although the story was not about him. Similarly, I imagine in a way that makes it seem to me that I am part of the imagined events. I visualize, and entertain other "images", as I would if imagining myself experiencing, feeling, deciding, and acting. But my visualizations and other images stand for Napoleon's imagined experiences, emotions, decisions, and actions, not mine. Incidentally, Williams's suggested comparison also indicates how I could imagine that, as we would put it, "Boyer is Napoleon". This 'is' of 'Boyer is Napoleon' is not the 'is' of identity, but only indicates that Boyer represents Napoleon in the telling of the story. (As when explaining the plan of battle over breakfast, the general says "This salt shaker is the fourth regiment.") I think that a similar sense of the verb 'to be' is involved in my claim to imagine that I a m Napoleon. It is an assertion that I represent Napoleon in the story. This would account for the asymmetry Wollheim noticed: to imagine that I am Napoleon is not to imagine that Napoleon is me. The representation relation is asymmetrical; I may represent Napoleon although he does not represent me.9 But there is some reason to doubt this interpretation of 'am' in my assertion. Boyer's acting the role of Napoleon entails that certain sorts of events happening to Boyer represent corresponding events as happening to Napoleon. When Boyer raises his arm, Napoleon is represented as raising his arm. But it may be that no rules connect events involving me while I am imagining with events that are imagined to happen to Napoleon. For example, it seems that no rule implies that if I visualize a fish while imagining that I am Napoleon, Napoleon is thereby imagined to have seen a fish. O n the other hand perhaps there are such rules, but we suspend them in the case of an evident mistake, as we do for a stage actor who has obviously missed a line. Another difficulty about imagining actions in transference imagining is suggested by John Perry's work on the relation of first person beliefs and actions. (Perry 1979) Actions are explained by the first person beliefs of the agent. M~ belief that Steven Reynolds is about to be struck by a falling anvil may not lead me to jump out of the way, but my belief that I am about to be struck by a falling anvil will. In transference imagining many of the beliefs I imagine the person to have do motivate imagined actions in just
IMAGINING ONESELF TO BE ANOTHER
627
this way. So the beliefs must be first person beliefs, and hence beliefs about myself. But the comparison with acting reveals that imagined first person beliefs need not be beliefs about the imaginer. In the movie Napoleon's actions will be explained by first person beliefs he is depicted as having, but Boyer wouldn't have been having those beliefs about himself. If he were caught up in his part he might have had feigned first person befiefs which would have helped guide his enactment of Napoleon's actions, but those beliefs would not have been about himself. His feigned belief which would have been expressed "I have lost the battle." would not have been a belief, or even a feigned belief, about himself, Boyer. Likewise, my imagined first person beliefs when I imagine Napoleon from the inside are not beliefs about myself, even though, as with ordinary first person beliefs, they are, as it were, action guiding. (b) It is objected that I cannot imagine Napoleon from inside, for that would be imagining that I am experiencing Napoleon's conscious states, and it is impossible for me to experience Napoleon's conscious states. One might reply,.of course, that the impossible can be imagined, contrary to the assumption made in this objection. But I think it is illuminating to try to find a reply to the objection according to which what is imagined is possible. For in addition to assuming that only the possible can be imagined, the objection makes another interesting mistake. The objection assumes that to imagine a conscious state "from inside" is to imagine oneself having that conscious state.1° But in imagining Napoleon from the inside I do not imagine that I am experiencing Napoleon's conscious states. I merely imagine Napoleon having those states by representing them to myself in a certain way. The temptation to suppose that I imagine that I am having them arises because I represent them to myself in a way that reproduces Napoleon's point of view on them. Another example of reproducing a point of view occurs in the movie "Double Indemnity". As Walter Neff (Fred MacMurray) is rising from behind the seat of the automobile where he has been hiding, the camera moves jerkily up the back of the seat. That sequence represents Neff's visual experience; we are meant to understand that that view of the back of the seat is what Neff experiences as he rises from the floor. It thus reproduces, in a way, Neff's own point of view on the experience it represents. My claim is that in my transference imagining the visualizing and other mental images represent Napoleon's experiences. They reproduce his point of view on those experiences, but it is not thereby imagined that I , the imaginer, am conscious of his experiences,
anymore than that sequence in "Double Indemnity" implies that, according to the movie, the viewer of the movie is aware of Walter Neff's conscious states. It isn't part of what happened in the movie that the audience was aware of Neff's visual experience, for the movie isn't about the audience; rather, Neff's visual experience was represented to the audience in a first personal way, as if it had been their own experience. Similarly I can imagine Napoleon's experiences from inside, without imagining that I am directly aware of his experiences. The last difficulty is how to explain my use of the first person pronoun, '1', in my imagination reports. If I am imagining Napoleon, there is no difficulty explaining why I use 'Napoleon'. But why do I use ' I , , if I am not imagining myself? In brief, the answer to that question is that my visualizations resemble experiences that would justify me in judging (coming to believe) in the first person. But this brief answer requires explanation and defense. It seems obvious (but I have defended it at length elsewhere) that there are some beliefs we come to hold, which we would not have been justified in holding, had we not had the appropriate experiences. (Reynolds 1988) My belief that Sam is wearing a black tie, which I arrive at on having a visual experience as of Sam in a black tie at a party, is an example. It is in virtue of my responding to certain aspects of the qualitative character of that visual experience that I am justified in judging that Sam is wearing a black tie. What aspects? In Sam's case, the qualitative features of the experience that justify me in judging about Sam will be the same qualitative features that would have justified me in attributing to the man I was seeing certain distinctive visible properties-a certain shape and configuration of nose, eyes, and mouth, perhaps. Suppose that I visualize Sam wearing a black tie at a party. That visualization will resemble a visual experience I might have had of Sam wearing a black tie at a party. And indeed, if the visual mental image came unbidden, as images often do, I might have recognized Sam in that visual image in much the same way I would have recognized him had I actually seen him at a party. Of course I don't deny that there will be many ways in which the visual image is unlike the visual experience I might have hadfainter, less detailed, and less stable,,to mention only some of the more obvious consciously available differences. But all I need to claim is that the image visualized resembles the experience I might have had in the relevant ways. It doesn't have to resemble a visual experience as of Sam in every particular.
IMAGINING ONESELF TO BE ANOTHER
629
If I were visualizing as part of imagining a party, then what I imagined about this party would be, in part, that Sam was present at it. The qualitative character of the visualization, which resembles to an experience I might have had of Sam, is such that I find it appropriate to describe what I imagined using the name 'Sam'. I claim that my use of the first person pronoun 'I' in reporting what I have transference imagined seems appropriate for the same reason. The visualizations (and other mental imagery) resemble experiences, which, if I had had them, would have justified me in making various judgments in the first person. Thus the proprioceptive mental "images" I have when transference imagining Napoleon on his horse resemble experiences which, if I had them, would justify me in judging that I was sitting on a horse. So it seems appropriate to report what I am imagining by saying that I am imagining that I am sitting on a horse. I think this is the correct account of why it seems appropriate to use 'I' in reporting what I transference imagine. But there is an important objection. Consider a parallel case. Suppose I am imagining that Moses is parting the Red Sea. I visualize him as looking a lot like Charlton Heston. But it seems wrong to describe what I imagine by saying "I imagine Charlton Heston parting the Red Sea." If, when I am transference imagining Napoleon, I imagine Napoleon and not myself, shouldn't it also seem wrong to report what I imagine using "I"? In the former case my visualization resembles an experience which would have justified me in judging that Charlton Heston was parting the Red Sea. Nevertheless it seems inappropriate to report what I imagine using the name "Charlton Heston". So even if my visualizations resemble experiences which would justify judgments about myself, shouldn't it still seem wrong to report what I imagine using the first person pronoun? If I visualize Moses with Charlton Heston's features, then using the name "Charlton Heston" to report what I have imagined seems a natural mistake. It isn't an inexplicable mistake, such as I would make if I said that I imagined Peewee Herman parting the Red Sea. Nevertheless it does seem wrong to use the name "Charlton Heston", while, on the other hand, using the first person pronoun continues to seem right even if one is convinced, as I am, that in this sort of imagining one is not imagining oneself. I think there are two reason why this use of 'I' to report what is transference imagined seems right, even though the parallel use of proper names seems wrong.
First, it is our custom to report what we transference imagine using the first person pronoun. Even if, as the comparison to our practice with proper names indicates, that apparently should seem wrong, custom makes errors of expression into acceptable idioms. Perhaps it has become an established convention to report such imagining~in the first person, as if we were imagining ourselves, and that is why it seems right to so report them. The other reason why it seems appropriate to use 'I' derives from the immunity to error through misidentification of experiential first person judgments. (Wittgenstein 1953; Shoemaker 1968) If I judge that Sam is wearing a black tie on the basis of a visual experience, I could be wrong, not because the man I am seeing is not wearing a black tie, but because the man I am seeing is not Sam. In that case my judgment would be in error through misidentification. But when I judge in the usual way that I am in pain, it seems I cannot be in error through misidentification. My experience could not have been an experience of someone else who is in pain, and whom I have mistaken for myself. Many first person experiential judgments are similarly immune to error through misidentification. My judgments that I am seeing a tree, that I seem to be sitting on a horse, that I hear the canons-all of these seem to be immune to this very specific sort of error. Given that one has the appropriate experience, the first person judgment made in response could not be in error through misidentification. My visualizations, however, according to my view of their role in imagination, may represent the experiences of others. So it is wrong (except that conventional usage, allows it) to describe what I have imagined as if it concerned myself, rather than Napoleon. Yet it seems right, and the very possibility that it should be wrong seems a little farfetched, just because these visualizations resemble experiences, which, if I had them, would justify me in judgments about myself which would be immune to error through misidentification. In response to such experiences, I could not be wrong in making a judgment about myself. So my confidence that I could not be misidentifying myself in response to such experiences leads me to feel confident in using the first person pronoun to report what I imagine. It seems that that is a sufficient explanation as to why I use 'I' in reporting what I transference imagine and why I don't feel inclined to stop using it even when I believe that it is not me that I am imagining. In the case of transference imagining that I have been using as an example, the best answer to the question as to whom I im-
IMAGZNZA'G ONESELF TO BE A N 0 THER
63 1
agine is that I imagine Napoleon. The other answers do not work. Transference imagining is not imagining oneself in an impossible situation, nor is it imagining an impossible self in someone else's situation. It's just a first personal way of imagining someone else."
'Hoping to avoid confusions suggested by the first person pronoun, some authors prefer "x imagines being y" as the canonical form for attribution of transference imagining. I do not restrict myself to this form of words, however, for I suspect that it is elliptical for "x imagines himself being Napoleon", which includes the pronoun anyway. 2The reflexive pronoun 'oneself' is used here to attribute a state of mind which the person would self-attribute using the first person pronoun '1'. The use of pronouns other than '1', such as "oneself ', "himself ', "herself ', "he" or "she", in the attribution of self-regarding psychological states has been extensively studied by H . N . Castaiieda. (Castaiieda 1966, pp. 130-157, 1967, pp. 85-100) 3Richard Wollheim cites this asymmetry to motivate a view about transference imagination that resembles the one I defend. (M'ollheim 1984, pp. 72-76) A detailed comparison of his view with mine is not possible; Wollheim comments briefly on transference imagination in the middle of a psychoanalytically motivated project, and he does not turn aside from that project to discuss the issues that primarily concern me. But his view of the nature of transference imagination resembles mine, so far as can be judged from his beief outline of it. The main point is that in some cases of "central imagining" certain mental occurrences in the imaginer represent the feelings etc., not of the imaginer, but of the person who is centrally imagined. (Thanks to Gary Kemp for directing my attention to M'ollheim's remarks.) 'Perhaps an inarticulate awareness of this point helps explain why some philosophers have claimed that only what is possible can be imagined. An apparent impossibility in a story seems to demand explanation as a misdescribed possibility. For from an impossibility everything follows, and that would be the death of a story. There are authors who delight in throwing us out of a story by asserting evident impossibilities. Such cases do not conflict with the claim that fictions are typically closed under evident implications, for the device would succeed only if we had a habit of accepting implications as fictional truths. Some have claimed that almost all fictions are impossible. For almost all fictions contain purported proper names that do not have referents (e.g. 'Sherlock Holmes'), and many contain purported natural kind terms which do not pick out a natural kind (e.g. 'unicorn'). These sorts of expressions acquire reference (it is claimed) only by actually being used in (causal) connection with their referents, so if they don't have referents in the actual world, they don't have referents with respect to any other possible world either. This sort of impossibility, if it is such, doesn't seem to be a problem to the reader as other kinds of impossibility would be. One can reason coherently as if these terms had referents. SThe alternative that one imagines both 2) and 4) at once, without imagining oneself to be identical to the person imagined, would face most of the difficulties of each of those alternatives, as well as problems about simultaneous but distinct imaginings that use the same visualizings. 6Zeno Vender's views on transference imagination are impossible to briefly characterize, but since the transcendental self he posits is not, he says, to be regarded as an object, he might answer my question by saying that in transference imagining one imagines no one. (Vendler 1976, p. 118, and 1984, pp. 104 ff.) G.E.M. Anscornbe argues that 'I' is in general not a referring expression. (Anscombe 1975) That might lead us to think that she would hold that transference imagination is not about anyone. But she holds that it is the speaker's properties that make sentences containing 'I' true or false. Thus we can still regard sentences in which 'I' occurs as being about the speaker. As I understand her view, Anscombe holds that 'I' is not a referring expression because a certain sort of knowledge of the referent is necessary for the competent use of a referring expression, and a competent user of 'I' need not have that sort of knowledge
of himself. B u t s u c h a view a b o u t 'I' evidently wouldn't support t h e claim that in transference imagination o n e doesn't i m a g i n e a n y o n e . 'J.L. M a c k i e offers a variant of alternative 2). H e hopes t o avoid difficulties a b o u t t h e a p p a r e n t impossibility of w h a t is i m a g i n e d b y h o l d i n g that, in reports of transference imagining, 'I' is n o t a rigid designator. ( M a c k i e 1980) M y use of ' I ' i n t h e context of a report of transference imagining, instead of referring t o m e with respect t o every possible w o r l d , refers t o whoever in a given world h a s a certain property. I n t h e actual world I a m t h e only o n e w h o h a s this property, b u t it is possible that Napoleon was t h e only possessor of it. If ' I ' is so u s e d , t h e n "I a m t h e son of C a r l o s a n d Letizia Bounaparte" is false, b u t it is n o t impossible. S u c h a s t a t e m e n t m a d e i n t h e context of a transference imagination report is t h u s a b o u t m e , b u t it does n o t express a n impossibility since ' I ' functions i n it a s a non-rigid designator of m e . It is not easy t o guess what t h e relevant identifying property of myself could be, especially if m y i m a g i n i n g reproduces Napoleon's life a s it was. My chief objection to Mackie's proposal however is that if '1', in this context, refers to me nonrigidly, then 'I am the son of Carlos and Letizia Buonaparte' evidently does not capture what is imagined. Let that contingent property of mine by which my use of 'I' picks out its object be expressed by the predicate F. Then for me to say 'I am the son of Carlos and Letizia Buonaparte', using 'I' in this sense, is to say that the F is the son of Carlos and Letizia Buonaparte. But 'I imagine that the F is the son of Carlos and Letizia Buonaparte' evidently need not have the same truth value as 'I imagine that I am the son of Carlos and Letizia Buonaparte'. At least for any candidate for F that I have been able to come up with, (and remember that F must express a property of me which could be a property of Napoleon) it seems that I could imagine that Napoleon is the F without transference imagining. (e.g. "I imagine that the man who lives in apartment 102 is the son of Carlos and Letizia Buonaparte".) 8Williams doesn't develop this comparison as I have, nor does he draw my conclusions from it. correspondence Professor H . N . CastaAeda has suggested, but has not endorsed, a couple of other ways to account for the asymmetry. The order of presentation of the designations in an identity statement in the content phrase of a propositional attitude attribution may indicate that one of the designations is particularly pertinent to the train (or, less one dimensionally, the context) of thought in which the attitude occurs. "Sam thinks that John Kennedy was the 1963 attorney general" suggests a different sort of mistake than does "Sam thinks that the 1963 attorney general was John Kennedy". The former suggests that Sam was primarily thinking about John Kennedy, the latter that he was primarily thinking about the 1963 attorney general (or perhaps attorneys general in general). It is not clear to me that this asymmetry will account for Wollheim's asymmetry however, for that allowed the two cases to differ in truth value, and this asymmetry seems to allow them to differ only in conversational implicatures about the context of thought. Another alternative is that since imagining is not closed under logical implication, the symmetry of the identity relation may be lost in imagination contexts. One might imagine that a = b, without imagining that b = a. But this suggestion evidently does not do justice to the phenomenological facts of imagination reports. As in the case of belief, fictional closure under logical implication presumably fails due to (the author's?) ignorance of implication relations. But I don't reject "I imagine that Napoleon is me" as an account of what I imagine because of a failure to realize that "I am Napoleon" logically implies "Napoleon is me". 'OChristopher Peacocke would probably agree with the claim I am rejecting. In a defense of Berkeley's contention that one cannot (visually) imagine an unperceived tree, Peacocke assumes that to imagine anything visually is to imagine oneself having a certain conscious state, a visual experience. (Peacocke 1985) I think that Berkeley's contention is sufficiently implausible that Peacocke's argument for it from the above mentioned assumption should instead be regarded as a reductio of that assumption. According to Peacocke, the comparison of imagining and theatrical representation on which I rely in the text fails, for although a theatrical performance need not make the audience p a n of the story, in the sense that the story told is about the audience, sensory imagining must always be imagining oneself to be in a certain conscious state. It seems however that his rejection of the theatrical parallel is based on the same assumption that leads to the Berkeleyan conclusion. "I first became interested in the problems of imagining oneself to be another when I read Zeno Vendler's entertaining and influential studies (cited above). I would like to thank the anonymous referees for Noh and Hector-Neri Castaiieda for detailed and helpful comments. Joseph Almog, Keith Donnellan, Michael Griffith, and especially Rogers Albritton gave me' helpful criticism and advice. .They should not be taken to endorse any of the views here expressed. Earlier versions of this paper were presented at UCLA, UC Santa Barbara, and Arizona State University, and the present version benefited greatly from the ensuing discussions.
ZMAGZNIrVG ONESELF T O BE ANOTHER
Anscombe, G.E.M. "The First Person" in Samuel Guttenplan (ed.) Mind and Language: Wolfson College 1975 Lectures 1974. Oxford, 1975. Reprinted in Metaphysics and the Philosophy o f M i n d : Collected Philosophical Papers of G.E.M. Anscombe. Minneapolis: University of Minnesota Press. (1981). 21-36. Castaiieda, H.N. "He*: O n the Logic of Self-Consciousness" Ratio 8 (1966) 130-157. 1966 1967 "Indicators and Quasi-Indicators, "American Philosophical Quarteriy 4 (April 1967): 85-100. Goodman, Nelson Languages ofArt: an approach to a theory of symbols. Indianapolis: Bobbs-Med, 1968 1968 Kaplan, David "On the Logic of Demonstratives," The Journal of Philosophical Logic 8 (February 1979 1979): 81-98. Lewis, David
1978 "Truth in Fiction" American Philosophical Quarterly 15 (1978): 37-46.
1979 "Attitudes De Dicto and De Se" The Philosophical Review 88 (1979): 513-43.
Mackie, J . L . "The Transcendental 'I' " in Philosophical Subjects: Essays Presented to P.F. Strawson, 1980 ed. Zak Van Straaton. Oxford: Clarendon Press, 1980. 48-61. Mill, John Stuart
Utilitarianism, Indianapolis: Bobbs-Merrill (1957, originally published 1861).
1957 Peacocke, Christopher 1985 "Imagination, Experience, and Possibility: a Berkeleian View Defended" in John Foster and Howard Robinson eds. Essays on Berkeley Oxford: Clarendon Press, 1985. Perry, John
1979 "The Problem of the Essential Indexical" N o i s 13 (1979): 3-21.
Reynolds, Steven
Experience, Justification, and First Person Judgment. Dissertation, UCLA 1988.
1988 Shoemaker, Sydney 1963 Self-Knowledge and Self-Identity Ithaca: Cornell University Press, (1963). pp. 23-24. 1968 "Self-reference and self-awareness" The Journal of Philosophy 65, 19 (October 1968) 555-567, reprinted in Identi&, Cause, and Mind: Philosophical Essays. Cambridge: Cambridge Univesity Press, (1984). p. 12) Vendler, Zeno 1976 "A Note to the Paralogisms" in Contemporary Aspects of Philosophy, ed. Gilbert Ryle. London: Oriel Press, (1976). 111-121 The Matter of Minds. Oxford: Clarendon Press, 1984. 1984 Williams, Bernard 1966 "Imagination and the Self' British Academy Annual Philosophical Lecture, reprinted in Problems of the Self Cambridge: Cambridge University Press, 1973. 26-45. Wittgenstein, Ludwig 1953 Philosophical Investigations, Transl. G.E.M. Anscombe. New York: Macmillan Publishing Co., 1953. Section 404, p. 122eff. Wollheim, Richard
1984 The Thread of Life. Cambridge Mass, Harvard University Press. 1984. 72-76.
http://www.jstor.org
LINKED CITATIONS - Page 1 of 1 -
You have printed the following article: Imagining Oneself To Be Another Steven L. Reynolds Noûs, Vol. 23, No. 5. (Dec., 1989), pp. 615-633. Stable URL: http://links.jstor.org/sici?sici=0029-4624%28198912%2923%3A5%3C615%3AIOTBA%3E2.0.CO%3B2-S
This article references the following linked citations. If you are trying to access articles from an off-campus location, you may be required to first logon via your library web site to access JSTOR. Please visit your library's website or contact a librarian to learn about options for remote access to JSTOR.
References Attitudes De Dicto and De Se David Lewis The Philosophical Review, Vol. 88, No. 4. (Oct., 1979), pp. 513-543. Stable URL: http://links.jstor.org/sici?sici=0031-8108%28197910%2988%3A4%3C513%3AADDADS%3E2.0.CO%3B2-O
The Problem of the Essential Indexical John Perry Noûs, Vol. 13, No. 1. (Mar., 1979), pp. 3-21. Stable URL: http://links.jstor.org/sici?sici=0029-4624%28197903%2913%3A1%3C3%3ATPOTEI%3E2.0.CO%3B2-F
Self-Reference and Self-Awareness Sydney S. Shoemaker The Journal of Philosophy, Vol. 65, No. 19, Sixty-Fifth Annual Meeting of the American Philosophical Association Eastern Division. (Oct. 3, 1968), pp. 555-567. Stable URL: http://links.jstor.org/sici?sici=0022-362X%2819681003%2965%3A19%3C555%3ASAS%3E2.0.CO%3B2-1