I

n this paper, I propose to discuss Descartes' dreaming problem. Descartes supposed that this problem, at least prior to the invocation of quite a bit of philosophical machinery, plunges one into scepticism. The most plausible rebuttal to Descartes' argument has been made by Malcolm;^[1] but the sharp attacks in the last decade on Malcolm's assumptions have led some philosophers to suppose that Descartes' dreaming problem is a cogent support for scepticism.^[2] In this paper, I hope to dispose of the problem without using controversial assumptions of the sort used by Malcolm.

  Like many influential statements of philosophical problems, Descartes' statement of the problem of dreaming is surprisingly brief: it is contained in the fifth paragraph of the first Meditation. Descartes recalls that he is accustomed to "imagine" implausible things in his dreams. "How many times", he asks rhetorically "has it occurred that the quiet of the night made me dream of my usual habits: that I was here clothed in a dressing gown, and sitting by the fire, although I was in fact lying undressed in bed?"^[3] While it seems apparent to Descartes that he is not now merely dreaming that he is writing this fifth paragraph, he should nevertheless recall having been misled, while asleep, by similar illusions".^[4] He concludes that "there are no conclusive indications by which waking life can be distinguished from sleep".^[5] To see why Descartes considered this an astonishingly powerful ground for scepticism, we must be clear about several points in the way he frames the problem. First, it is merely in having the dream, and not in subsequent mental episodes, that one makes the mistake. Secondly, one makes the mistake whenever one dreams that some person—oneself or another—is located in a certain place or doing certain things when, at the time the dreaming occurs, that person is somewhere else or not doing those things. Descartes would have avoided the mistake had he, while being naked and asleep in bed, dreamt that he was naked and asleep in bed. (He implies that one also might avoid the mistake by embracing at all times a thorough-going [page 221:] scepticism towards the senses.) Finally, at the moment one is making the mistake—that is, while one is dreaming—there are no conclusive indications available to tell one whether one is merely dreaming or really seeing and hearing. Thus, supposing that one fails to withhold all perceptual judgements, it is not within anyone's power to avoid making mistakes; if one avoids making a mistake on some occasion (e.g., while awake), it is a matter of luck.

  In criticizing Descartes, both Macdonald and Malcolm have seen the enormity of the alleged problem which Descartes exposes. Macdonald complains that, if it be a mistake, "No precautions can be taken against making it except, perhaps, that of remaining permanently awake".^[6] And Malcolm tries to undercut the crucial premises of the argument by bravely maintaining that one cannot make a mistake while one is sleeping because one cannot have thoughts while one is "sound asleep";^[7] and one cannot make a mistake which one could not possibly discover to be a mistake."^[8] Unfortunately, defending Malcolm's line of attack requires controversial philosophical machinery—some of which resembles now-discredited principles of logical positivism. And while Macdonald's essay is helpful in showing how special and unavoidable mistakes by virtue of dreaming would be, she does not thereby dispose of Descartes' problem. Descartes was evidently intrigued by the very novelty and enormity of the dreaming problem.

  We can dispose of the dreaming problem, however, by following out a suggestion with which Macdonald toyed for a few pages in her essay.^[9] I suggest that we should pursue the analogy between dreams and dramas. It will, of course, not be a complete analogy; nevertheless the partial analogy will be sufficient to show what is terribly mistaken in Descartes' statement of the dreaming problem. The analogy which I propose to explore is that between having a dream and watching a play.

  There is indeed no one kind of dream, just as there is no one kind of play. But a feature belonging to many, and the feature which makes dreams intriguing to philosophers, is their representational nature. Accordingly, I shall throughout this discussion focus on dreams displaying that feature, without however assuming that all dreams do. The personae in such dreams are like the characters in many plays—they represent actual people; and the dream episodes may also represent real events. This similarity is manifested in the similarity between reports of dreams and those of plays: both often resemble reports of actual events.

  But if someone, reporting a dream, says he "dreamt of Nixon and Carter walking arm and arm at Camp David", just what relation has the dream to the actual Nixon? What is the basis of the identification of characters in a [page 222:] dream with actual persons? The identification does not seem to turn on any similarity in physical appearance or build, or in characteristic behaviour or personality. The Nixon in my dream may fail to resemble the actual Nixon in crucial ways (lacks jowls, has an extraverted sense of humour), and may in fact resemble some other person more than he resembles the actual Nixon. Even salient biographical facts may be altered (the Nixon in my dream may have survived impeachment). Nor, evidently, is it any causal connection which determines of whom one is dreaming; one may dream of people and creatures which have never existed and, indeed, of such as were even hitherto unknown to fiction.

  In this negative way, the representational nature of dreams is similar to that of plays and novels, and different from (say) photographs. A character in a satirical play or novel may bear little similarity to the actual person being represented. In satires it is common for some latent characteristics and tendencies of the actual person being represented to be grossly exaggerated, and for some overt ones to be suppressed. And in a crudely executed satire, there may not be even that correspondence between features of the actual person and features of the character: someone who is being satirized may simply be portrayed as disgusting without the disgusting features being distortions or caricatures of the corresponding features of the actual person. In general, if the author intends this character to represent that person, then this character does represent that person—though we may have (artistic) reservations about how well the satire is carried off.

  We are thus required to qualify the similarity between dreams and plays because of this striking difference between them: plays are intentionally and deliberately contrived, while dreams are not. Hence we cannot appeal—at least not in a straightforward way—to the dreamer's intentions in determining the identity of the characters in his dream. I may enter a small caveat: although dreams are not deliberately fashioned, a person's dreams are frequently autobiographical after the manner of plays and novels—as we all know quite apart from any reliance on Freud; dreams frequently mirror, and so can be used as a basis for making inferences about, the wishes, anxieties, and recent events in the life of the dreamer. Nevertheless, the characters in a person's dream are not intentionally made to represent actual people. But despite this obvious difference the representational nature of dreams is analogous to that of plays in this way: in general, the representations (if any) are established by the avowal of the dreamer in his dream report. In general, the representations which the dreamer recognizes are authoritative.

  Some exceptions to the rule may be noted, so that my argument may not appear to disguise them. First, a person reporting his dream can make a mistake in identifying the characters in his dream; he can make the same sort of mistake which a satirist can make. If, at the time he narrates the dream, he supposes that several definite descriptions refer to the same person [page 223:] when they do not, or supposes that a particular proper name refers to a person characterized in some other way when it does not, then he might wrongly report that he dreamt of the last President who was formerly a Vice-President (when it was the last but one), or wrongly say he dreamt of Prime Minister Baldwin (when it was Balfour). Secondly sometimes a person relates an obviously "autobiographical" dream, and it seems clear to his close friends that a particular character in the dream, given its structure, certainly represents an acquaintance of the dreamer; and yet the person relating the dream sincerely denies that representation. In a parallel way, intelligent and sympathetic readers of a satirical play might feel certain that a particular character represents a particular person, even though the author sincerely denies it. If this sort of example be allowed to pass, then it would belong to the class of self-deceptions, whose members are in some ways like mistakes. It is not certain, however, that this sort of example ought to be allowed to pass. In the analogy between dreams and plays which I am exploring, 'represent' does not mean 'inspired by' or 'modelled on'. Even if a character in a play were inspired by or modelled on J. Edgar Hoover, one could not correctly claim on that account to have seen a play about Hoover. In the same way, suppose that during the Watergate hearings, Nixon dreamed of a character rather like Sam Ervin. And suppose—as it would be natural to suppose—that the activities of the actual Sam Ervin comprised a necessary condition of the dream's occurrence; if Sam Ervin had not been doing what he was doing, Nixon would not have had that dream. Nevertheless, it would not follow that Nixon had dreamt of Sam Ervin. If Nixon sincerely denies that he dreamt of Sam Ervin, while admitting the similarities and the probable cause of the dream, then we should have to say—I think—that he did not dream of Sam Ervin, but only of an Ervin-like person.

  At any rate, except in the case of mistaken identifications which turn on errors regarding the reference of names and definite descriptions, and with the possible exception of self-deception, we do not allow any authority to any identifications of any dream-characters save those made in the dream-report. Perhaps this is a feature which perplexed Malcolm and led him to deny that dream-reports are descriptions of antecedently existing events. Malcolm's mistaken denial may be likened to an attempt to make the reporting of a dream altogether like the writing of a play. It is important to see the limits of the analogy, for if it is pressed beyond, it will explode along with any arguments based upon its over-extension. In general we do not allow third-party identification of a satirist's targets because the representations have been established by the author's intentions. Although there are no parallel intentions of a dreamer which can explain why we should not overrule a dreamer's identifications, nevertheless we do not overrule them. Thus, a person reporting a dream may recognize a character as Nixon even though—as we discover from his description of the character— [page 224:] there are hardly any salient resemblances. And he may recognize the character as Nixon even though he concedes that there is little resemblance between the character and Nixon as he believes him to be.

  A dream is like a play in a further but connected respect: it has its own internal structure. Dreams often have plots—the reports of such dreams relate stories. Even when one is dreaming of actual people, the characters may behave in ways the actual people have not, and would not, and even could not. Perhaps in one's dream people fly; perhaps they know of the existence of noteworthy events without having had to read dispatches or to go to the scene to observe them. Perhaps they do things which not only would be miraculous in real life, but which are considered miraculous by the characters in the dream. And, contrary to what Chappell says,^[10] the characters in a dream can do what in actuality is impossible—as can characters in a play or a novel.

  While a dream or a play frequently does presuppose a background which mirrors things taken for granted in the actual world, still, what happens in the actual world is of limited relevance in determining what has happened in a dream or in a play or novel. In most dreams and plays, the laws of gravity and familiar causal connections hold: characters speak in recognizable languages; characters have the sorts of desires and aversions, capacities and limitations, which actual persons have. If a Martian could make no sense out of ordinary human affairs, he would probably find texts of plays and reports of dreams equally unintelligible. The narrations of dreams as well as those of plays frequently resemble the narrations of episodes of real life. But despite that, a dream is self-contained in the way that a play or novel is. Suppose that, in my dream, Nixon seems to be presented to the Prince of Wales; after Nixon makes a little speech expressing his pleasure, everyone in the retinue begins to laugh while the presumed heir-apparent strips a flesh-like mask from her face, revealing Margaret Thatcher. I could report my dream in this way: Nixon (at first) mistakenly supposed he was being presented to the prince, but as he subsequently discovered, he was in fact being presented to the masked Margaret Thatcher. In the dream, Nixon discovered his mistake in a way parallel to the way in which an actual person might discover a similar mistake: the dream presupposed a familiar backdrop. Yet there is no way to "get behind" the story or plot revealed in a dream or play. The question, "But was it really Margaret Thatcher who masqueraded as the prince?" has to be answered, "yes". The fact that the actual Margaret Thatcher would never indulge in such shenanigans is quite beside the point of what did happen in the dream. The report of the dream, like the text of a play, authoritatively answers the question.

  Further, a character in a dream, or one in a play, might succeed even in squaring the circle. Since, of course, one cannot intelligibly imagine a [page 225:] circle's being squared, one's dream is not likely to focus on the details of how the feat was accomplished. All that is needed is for various pieces of the story to fit together in the way they might in actual life: in one's dream, someone might run through certain arcane rituals, and then know^[11] that he had squared the circle; the expected accompaniments to such an earth-shaking feat might then occur—excited newspaper reporters, astonished mathematicians, huge public notoriety, and so forth. We may advert to a brief episode of Monty Python's Flying Circus in which the characters had discovered a joke so enormously funny that anyone hearing it would literally laugh himself to death. The television audience was (cleverly) never allowed to hear the entire joke; rather they saw the results of various characters' reading or hearing the joke. Similar devices are used by authors such as John Barth; the point behind them is that while we readers cannot imagine the unimaginable, we can imagine how people would react when they discovered that what everyone had thought to be unimaginable turned out to be imaginable—and even executed.

  Moreover, even the details of how the circle was squared might be dreamt. Though they will not in actuality constitute a recipe for squaring the circle yet within the dream they may be a complete recipe for squaring the circle. For to dream of someone's squaring the circle is to dream of something which, in the dream, is acknowledged by all to have been the squaring of the circle.^[12]

  Finally, neither a dream nor a play needs to be internally coherent. A play or a novel might have unresolvable inconsistencies (which might or might not make it poorly or sloppily written—it might be done intentionally to some purpose, in which case a critic would have to assess whether the pursuit of that purpose fits with or detracts from the structure of the play). And a dream, too, may have unresolvable inconsistencies: perhaps in the dream Nixon entered the room but yet, at the next moment, though he had not left or changed, Carter was then in his place. Just as in a play or a novel, there is no way to "get behind" such inconsistencies in a dream, no way to discover what happened: that is simply the way the story was dreamt or written.

  Thus, although a dream is not intentionally or deliberately designed, as is a play, nevertheless the analogy between them is strong. And this analogy will illuminate the central confusion in Descartes' dreaming problem if we note a further, crucial point: in most dreams, there is a character who represents the dreamer. When one narrates a dream by saying "I dreamt that I saw Nixon tricked", the two occurrences of the pronoun have different referents: the first refers to the dreamer, while the second refers to the [page 226:] protagonist in the dream representing the dreamer. This systematic ambiguity ought not to seem outrageous, for it is of a piece with the parallel systematic ambiguity of the proper names and definite descriptions used to narrate the dream. When I report what I did or saw or heard or thought in my dream, I am talking about a character in my dream. This use of the first person singular in the narration of the dream is no more unusual or paradoxical than the use of proper names of actual persons. The relation between the character and the actual person is one of representation: in his dream, a character representing the dreamer may differ as completely from him, and from his conception of himself, as the other characters may differ from the actual persons they represent. When Walter Mitty dreams, the Walter Mitty in the dream may be athletic, courageous, bold, cheeky, unused to being pushed around, devastatingly handsome, and so on; he may do and think things which the actual Walter Mitty would never do or could never do.

  When, in my dream, "I saw a tortoise shot with an arrow, and knew at once that it was dead", or when "believing it was dead but not being sure, I investigated and confirmed the casualty", it is the character representing me who is said to have, seen, known, believed, and investigated. To understand reports of dreams, the distinction between the characters in the dream and the actual persons who may be represented by those character must scrupulously be respected. And that requires distinguishing between the dreamer and the protagonist who represents him. Yet this simple and obvious distinction undermines the apparently enormous problem of scepticism which Descartes raised. For it is by illicitly comparing the facts of the dreamer with the beliefs of the character representing him that Descartes is led to his conclusion that mistakes in dreaming are ubiquitous.

  Suppose I watch a performance of Jumpers; I subsequently relate that in the play, Moore shot the tortoise with an arrow. And suppose I am challenged in this vein; "But you must be mistaken since, in his lifetime, Moore never shot a tortoise; moreover, you couldn't have seen such a thing since Moore was dead and buried by the time the performance occurred." The objection would be patently confused: it is, of course, the character who I am claiming shot the tortoise; facts about the actual Moore are irrelevant to whether I saw the things in the play which I claimed to have seen. Now suppose that Descartes, while he is naked in bed, dreams that he is awake, fully clothed, and sitting by the fire. Is dreamer Descartes mistaken? In what belief might he be mistaken? The character Descartes presumably believes that he is fully clothed, etc.; in believing this, he is correct. But whether character Descartes is correct or mistaken in his beliefs does not affect the credibility of dreamer Descartes.

  What beliefs may we attribute to Descartes the dreamer? Malcolm would evidently argue, "None. Descartes the dreamer gives no signs that he has any beliefs—he is inert. And even if the dreamer sweats and cries out in his sleep, that does not show that the dreamer believes things to be [page 227:] threatening him, the dreamer". In the same way, if one, watches an effective horror film, one may sweat and fidget while the protagonist is being chased by ghouls. But that would not show that the film-goer believes himself to be in danger. To find a play or film realistic and exciting is not necessarily to be so moronic as to forget that one is watching a play or a film. And to "identify with" the protagonist is not to believe, even for a while, that one is the protagonist or that the ghouls threatening the protagonist are loose in the theatre.^[13]

  Still, Malcolm goes too far in denying dogmatically that the dreamer has any beliefs. It is true, of course, that the beliefs of the protagonist may not be attributed to the dreamer whom he represents. Yet whenever a theatre-goer can subsequently relate the plot of a play he has seen, some beliefs are attributed to him as of the time of the performance; for example, while watching the performance, he presumably believed that the character, Moore, shot the tortoise, or that the character was being pursued by ghouls. Thus the theatre-goer has the sorts of beliefs necessary for our explaining his subsequent recall of the story. And so, by analogy, we may allow the possibility that a person—while he is dreaming—has similar beliefs, the sorts of belief which would account for his subsequent ability to recall the story of the dream. Yet even if these sorts of belief be attributed to Descartes' dreamer, sceptical conclusions remain groundless. One is not automatically making a mistake simply in "following" the story of a dream. Thus Descartes is mistaken in concluding that dreaming inevitably or typically leads to error.

  But we may push the question beyond Descartes' example. In the analogue to dreaming, in watching a play, a theatre-goer is liable to an interesting sort of mistake. He may, e.g., falsely believe that the butler did it (whereas the chambermaid turns out to have been the culprit). A theatre-goer is not necessarily passive—he can formulate hypotheses about the plot and about the likely outcome of a character's actions; and when he believes such hypotheses, he is liable to mistake. Of course, being a theatre-goer does not logically require one to be active in this way, it is not one's being a theatre-goer which by itself makes one liable to error, but one's being all active theatre-goer.

  If one can be an active theatre-goer, then might a dreamer, while he is dreaming, formulate hypotheses about what is going to transpire in his dream? I see no way to rule out the possibility. But if someone wishes to claim that this possibility is ever actualized, he will have to distinguish it from the character's wondering what will happen to him (N.B. not "the character's wondering what he will dream next"); for only the dreamer's hypothesizing will expose the dreamer to error. Perhaps something like this possible case of an active dreamer does occur on occasion: people sometimes [page 228:] report thinking "this is only a dream and it will end soon" as part of their reports of some dreams. At any rate, if one ever does follow one's dreams in the manner of an active theatre-goer, one presumably then has the further capacity to withhold judgement or to hedge the hypothesis with "possibly"—just as a theatre-goer can. So if this analogy with the active theatre-goer can be applied, then just as the active theatre-goer has the power to avoid error, so does the "active" dreamer. Thus even in this case a dreamer would not be condemned irredeemably to error.

  What made Descartes' problem so intriguing was his claim that error is unavoidable just in case there is any difference, at the time the dream is occurring, between the character in the dream and the actual person represented by that character. For Descartes wrongly supposed that for a person to dream that he is fully clothed is for him to believe that, at the time the dream is occurring, he—the dreamer—is fully clothed. Yet there is no good reason to attribute such a belief to the dreamer. His capacity to recount the dream later does not imply that the dreamer had such a belief; and neither does his use of the first person singular in narrating the dream.

  Finally, there is one last sort of error which we should mention. Sometimes when a man has seen a particularly striking episode in a play or has read a particularly vivid passage from a novel, he subsequently mistakenly believes that he (once) actually saw such a thing happen, rather than that he saw such a thing portrayed or read an account of it. And a person who has had a dream is liable to the same kind of mistake. After he awakes—perhaps long after—he may recollect some episode and mistakenly suppose that he is remembering something which he once saw happen; or he may actually be recalling something which he once saw, but mistakenly suppose that it is merely a recollection of a dream. Such mistakes are not peculiarly mistakes of dreaming: they are mistakes of memory. And there is nothing devastating about them; since they are made while the person is awake, it is within his power to avoid making them.

  Lawrence University

Footnotes:

¹ "Dreaming and 'Scepticism", Philosophical Review, 64 (1956), reprinted with minor changes in C. Dunlop (ed.), Philosophical Essays on Dreaming (Ithaca, 1977), pp. 103- 26; Dreaming (London, 1959).

² E. M. Curley, "Dreaming and Conceptual Revision", Australasian Journal of Philosophy, 53 (1975), reprinted with changes in Dunlop, op. cit.

³ R. Descartes, Meditations on First Philosophy, tr. L.J. LaFleur (Indianapolis, 1951), p. 18. (I have omitted the translator's marks which show the slight variation between the Latin and French texts.)

⁴ Op. Cit., p. 19.

⁵ Ibid.

⁶ M. Macdonald, "Sleeping and Waking", Mind 72 (1953), reprinted in Dunlop, op. cit., p. 77.

⁷ "Dreaming and Scepticism", in Dunlop, op. cit., pp. 108 and 121- 2.

⁸ 0p. Cit., p. 115.

⁹ "Sleeping and Waking", in Dunlop, op. cit., pp. 75- 9.

¹⁰ V. C. Chappell, "The Concept of Dreaming", Philosophical Quarterly, 13 (1963), reprinted in Dunlop, op. cit., p. 303.

¹¹ Compare a Dostoyevsky novel in which a sensitive character can know the character and intentions of another simply by gazing into his eyes. Does the hero thereby know the other's intentions and character? Certainly; the author tells us that he does.

¹² Thus in the dispute between Malcolm and Chappell, Malcolm is correct. See Chappell, "The Concept of Dreaming", in Dunlop, op. cit., pp. 303-8. Cf. also H. G. Blocker, "The Truth about Fictional Entities", Philosophical Quarterly, 24 (1974).

¹³ See Samuel Johnson's discussions of the unities of time and place in the preface to his edition of The Plays of William Shakespeare.