From: Cervantes: Bulletin of the Cervantes Society of America 4.2 (1984): 109-22.
Copyright © 1984, The Cervantes Society of America

N PART II OF Don Quixote, chapters 22-24, Sancho and the Humanist Cousin lead Don Quixote to the Cave of Montesinos where he falls asleep and dreams of the legendary heroes Montesinos and Durandarte, and of the beautiful Belerma. Certain comments of the narrator, Sancho, and the Cousin, as well as Don Quixote's own remarks, raise one of the most persistent skeptical worries in connection with this adventure: our ability to tell dreams from wakefulness. Don Quixote's dream in the Cave of Montesinos suggests a comparison with the dream argument advanced by Descartes in the Meditations, and in fact Descartes and Cervantes appear to say some very similar things about dreaming. Here, I want to show how they differ. In so doing, I hope to point out some ways in which Cervantes is anti-skeptical: he regards knowledge as possible (there is no room for doubt that Don Quixote was in fact dreaming in the Cave of Montesinos), but not submissible to reason. In this way, the adventure of the Cave of Montesinos helps us see certain flaws in Descartes' arguments. That is important not simply as a potential advance reply to Descartes, but as a critical illumination of the entire project of epistemology modeled in the dream argument of the Meditations.
     In Don Quixote, the narrator makes it perfectly clear that Don Quixote's adventure in the Cave of Montesinos was nothing more

than a dream; this fact is central to any further investigation of skepticism that might be prompted by this episode. He is particularly explicit at the point where Sancho and the Cousin retrieve Don Quixote from the Cave and find him in a deep sleep (“No respondía palabra Don Quijote; y sacándole del todo, vieron que traía cerrados los ojos, con muestra de estar dormido. Tendiéronle en el suelo y desliáronle, y, con todo esto, no despertaba; pero tanto le volvieron y revolvieron, sacudieron y menearon, que al cabo de un buen espacio volvió en sí, desperezándose, bien como si de algún grave y profundo sueño despertara; y mirando a una y otra parte como espantado . . .”).¹ But as Don Quixote begins to recount his experiences in the Cave, he has strong enough doubts to want to reassure himself that what he saw there was real. Like the philosopher fighting with the skeptic in himself, he wants to rid himself of doubts, to verify that he was awake, to be certain that he can distinguish reality from dreams: “‘Despabiléme los ojos, limpiémelos, y vi que no dormía, sino que realmente estaba despierto; con todo esto, me tenté la cabeza y los pechos, por certificarme si era yo mismo el que allí estaba, o alguna fantasma vana y contrahecha’” (II, 23). The assurances he seeks are comparable to those Descartes looks for in the first of the Meditations, when he says “At this moment it does indeed seem to me that it is with eyes awake that I am looking at this paper; that this head which I move is not asleep, that it is deliberately and with set purpose that I extend my hand and perceive it.”²
     Descartes too is convinced that he can tell dream from wakefulness (“what happens in sleep does not appear so clear nor so distinct as does all this,” p. 146, my emphasis). But, as various critics of the dream argument, such as Bernard Williams, have pointed out,³ it is entirely possible that Descartes was mistaken on the very premise of his argument. If Descartes were indeed dreaming, like Don Quixote

     ¹ Part II, Ch. 22. Further references will be incorporated into the text.
     ² The Philosophical Works of Descartes, trans. Elizabeth S. Haldane and C. R. T. Ross, I (1911; rpt. Cambridge: Cambridge University Press, 1975), p. 146. Subsequent references will be incorporated into the text.
     ³ Bernard Williams, “Dreaming,” in his Descartes: The Project of Pure Inquiry (Harmondsworth: Pelican Books, 1978), pp. 309-13. See also George Nakhnikian, “Descartes's Dream Argument,” in Descartes: Critical and Interpretive Essays, ed. Michael Hooker (Baltimore: The Johns Hopkins University Press, 1978), pp. 256-286. C. E. Moore raises the explicit issue of the nature of a proof of wakefulness in his essay “Proof of an External World,” Proceedings of the British Academy, 25 (1939), 273-300.

in the Cave of Montesinos, when he said “how often it seemed to me that in the night I dreamt that I found myself in this particular place, that I was dressed and seated near the fire, whilst in reality I was lying undressed in bed” (pp. 245-46), then he would be precisely in the position to make false judgments about the difference between these two states. Thus at the very outset, the adventure of the Cave of Montesinos provides one possible critique of Descartes' dream argument; it does not reverse that argument, but nonetheless shows where it may be flawed. From the start of this adventure, Cervantes' reader simply has a more encompassing vantage point on the dream situation than Descartes could possibly have attained; in a certain sense, the reader of the Quixote stands in the position of transcendental knowledge for which Descartes' meditative alter-ego strives.
     But the dream argument in the Meditations is more complex than this, and the perspective on Don Quixote's adventure in the Cave of Montesinos likewise becomes more subtle. In the Sixth Meditation, Descartes rejects his “hyperbolical doubts.” At that point, he finds a new way of telling dreams from reality —not by the criteria of “clear and distinct ideas,” but by recognizing that there is a continuity of experience proper to the wakeful state which is unknown to dreamers: “I ought to set aside all the doubts of these past days as hyperbolical and ridiculous, particularly that very common uncertainty respecting sleep, which I could not distinguish from the waking state; for at present I find a very notable difference between the two, inasmuch as our memory can never connect our dreams with one another, or with the whole course of our lives, as it unites events which happen to us while we are awake” (pp. 198-99). This certainly sounds like a good way to tell whether we are asleep or awake; but again Don Quixote provides the basis for a cogent argument against it. Compare Descartes' observations with Don Quixote's remark about his experiences in the Cave: “‘. . . el tacto, el sentimiento, los discursos concertados que entre mí hacía, me certificaron que yo era allí entonces el que soy aquí ahora’” (II, 23). What Don Quixote says is true, literally true; he is the same person now that he was when he had the dream; but as his very example shows, he may still be mistaken on the essential point: he may be unable to distinguish reality and dream even though he can track his own existence continuously over time. His sensations lead him to a conclusion about personal identity, not about reality; the argument might be used as good evidence for the continuity of the ego through

time, for self-awareness as a function of memory —what Hobbes called “mental discourse”— but not evidence of wakefulness.
     How, then, can we tell reality from dreams? And how can we distinguish them without invalidating the common experience of dreams altogether? The question becomes raised rather pointedly in Don Quixote because Don Quixote insists so strongly that he is certain of what he saw and that what he saw was real: “‘lo que he contado lo vi por mis proprios ojos y lo toqué con mis mismas manos,’” he tells Sancho and the Cousin (II, 23). Sancho is in a fit of laughter about the incident, for he was the one who contrived the adventure to begin with, and Don Quixote has fallen victim to it. But the Humanist Cousin takes a different tack. With a characteristic air of pedantry, he asks Don Quixote how he could possibly have seen and done all that he claims in the Cave of Montesinos in such a short span of time. How could the Knight have survived for three days without food? What demonstrable knowledge does he have about ghosts and underworld spirits to certify that he was really in the Cave of Montesinos? Do these spirits eat and sleep? Is Don Quixote in a position to confirm the whole experience? Even though he is a critic of Don Quixote, pedantic and arrogant, there is a certain similarity between them. Like Don Quixote, the Cousin ignores the distinction between dream and reality that the reader knows instinctively how to make; instead, he wants to take notes on Don Quixote's dream, to evaluate it as if it were fact:

     “Yo, señor Don Quijote de la Mancha, doy por bien empleadísima la jornada que con vuesa merced he hecho, porque en ella he granjeado cuatro cosas. La primera, haber conocido a vuestra merced, que lo tengo a gran felicidad. La segunda, haber sabido lo que se encierra en esta cueva de Montesinos, con las mutaciones de Guadiana y de las lagunas de Ruidera, que me servirán para el Ovidio español que traigo entre manos. La tercera, entender la antigüedad de los naipes, que, por lo menos, ya se usaban en tiempo del emperador Carlo Magno, según puede colegirse de las palabras que vuesa merced dice que dijo Durandarte, cuando al cabo de aquel grande espacio que estuvo hablando con él Montesinos, él despertó diciendo; ‘Paciencia y barajar.’ Y esta razón y modo de hablar no la pudo aprender encantado, sino cuando no lo estaba, en Francia y en tiempo del referido emperador Carlo Magno. Y esta averiguación me viene pintiparada para el otro libro que voy componiendo, que es Suplemento de Virgilio Polidoro, en la invención de las antigüedades; y creo

que en el suyo no se acordó de poner la de los naipes, como la pondré yo ahora, que será de mucha importancia, y más alegando autor tan grave y tan verdadero como es el señor Durandarte. La cuarta es haber sabido con certidumbre el nacimiento del río Guadiana, hasta ahora ignorado de las gentes” (II, 24).

     I have quoted the Humanist's reply at length to show how Cervantes parodies his type, but also because the error he makes, so glaring in Don Quixote, stands at the root of the dream argument of Descartes' Meditations. Somehow, we feel that he has misused Don Quixote's dream in assuming that it is factual and true —comparable to the world of reality (of history, of geography, of these realms of fact). He assumes that there is no difference between dream and reality when in fact there is every difference; the two are literally incomparable. If they cannot be compared, it is mistaken to conclude that because dreams are not reality they are false, or to deny having them altogether. This restriction of experience is one of the possible, even natural, results of an overzealous anti-skepticism; but Cervantes is at pains to show his readers that recognizing and accepting our dreams does not limit our capacity for knowledge, that the two are not mutually exclusive, that both are somehow possible.
     The Humanist Cousin assumes that Don Quixote's dreams are both factual and true; in chapter 24, the reader is tempted to judge the adventure in these same ways. The translator of Cide Hamete's Arabic history makes the marginal notation that the episode seemed apocryphal to him because it was utterly lacking in verisimilitude “‘. . . todas las aventuras hasta aquí sucedidas han sido contingibles y verisímiles; pero esta de esta cueva no le hallo entrada alguna para tenerla por verdadera, por ir tan fuera de los términos razonables.’” But at the same time, he notes, “‘pensar yo que Don Quijote mintiese, siendo el más verdadero hidalgo y el más noble caballero de sus tiempos, no es posible; que no dijera él una mentira si le asaetearan.’” So the reader is asked to decide whether what happened in the Cave of Montesinos was true or false, whether Don Quixote was telling the truth or not: “‘. . . si esta aventura parece apócrifa, yo no tengo la culpa; y así, sin afirmarla por falsa o verdadera, la escribo. Tú, letor, pues eres prudente, juzga lo que te pareciere, que yo ni debo ni puedo más.’” What Don Quixote says about the Cave of Montesinos cannot be faulted. He is behaving like a noble knight; he is telling the truth, reporting faithfully what transpired there. (Indeed, what sense could it make to say that he was lying, lying about

his dreams? Only the sense that he might be lying to himself, but then, who could tell? The problem is that the issue really has nothing to do with the truth of Don Quixote's narration, with its reliability.) It is equally tempting, and equally wrong, to judge dreams true or false. To do so is to make the same error as the Humanist Cousin, to fail to see that dreams are resistant to empirical investigation, that they take place outside the region where verification is possible. In this sense, the “conclusions” about the Cave of Montesinos episode are apparent from the start: there is a class of experience which ordinary human beings commonly have, called dreams, and dreams are not illusions and they are not reality either; they cannot be judged true or false. (Cf. J. L. Austin: “Does the dreamer see illusions? Does he have delusions? Neither; dreams are dreams.”⁴)
     I am not claiming that this is really a conclusion to be drawn from the episode of the Cave of Montesinos. How could it be, if there were never any grounds for doubt that Don Quixote was asleep and dreaming to begin with? Like an Austin, Cervantes is not pretending to teach us anything about reality that we did not already know, just to make us realize what, and how much, we know, and to prod us to take stock of our relationship to that knowledge. But to judge the episode true or false, or to take notes on it, as the Humanist Cousin does, is to expect that it is there to demonstrate something we did not already know. That expectation is precisely the one which Descartes brings to bear on his dream argument in the Meditations. When Descartes imagines that he might be confusing sleep and wakefulness, when he considers that he might be in his dressing-gown, seated by the fire and dreaming, he expects to make a discovery: he expects to discover whether he is dreaming or not. As has been said of cases like this, “we know why we are being asked to go over the situation again, in the sense that we know exactly what the problem is, and what a solution to it would be;”⁵ the problem is one of verification, and a solution to it would be some evidence, either for or against it. But the trouble is that Descartes uses the idea of a dream for this purpose; and dreams are of such a nature that they do not admit verification. Hence although the skeptic may worry that he is dreaming, it is unwarranted to take the case of dreams as the basis of

     ⁴ J. L. Austin, Sense and Sensibilia (New York: Oxford University Press, 1964), p. 27.
     ⁵ Stanley Cavell, The Claim of Reason: Wittgenstein, Skepticism, Morality, and Tragedy (New York: Oxford University Press, 1979), p. 159.

an a priori argument, as a test-case of the possibility of knowledge in general, as is Descartes' intent.
     The objection to the use of the dream as a model for testing the skeptical hypothesis is very different from the objection which says that in case we were dreaming we would be unable to tell; that is a local worry, not a general one, although it does serve as a good rejoinder to the objection that Descartes is not, of course, dreaming as Don Quixote is, just imagining that he might be. The deep difficulty with the dream argument, as I hope to have suggested, has to do with the expectations which the epistemologist brings to it. He expects that an imagined event, a projection, can provide a valid model for assessing the nature of our knowledge of reality, i.e. for making a general statement about what we can and cannot know. The question is not so much whether we are dreaming or not as whether an imagined situation is a good test of the conditions of knowledge; and that, at bottom, is what Don Quixote calls into question. If an imagined situation cannot be verified or disproved, then it cannot give us the evidence we need to confirm or contradict what we know; that is why the supposition of dreaming, as advanced in the Meditations, is invalid as an epistemological argument. It is this insight, which the Cave of Montesinos episode prompts, by which Don Quixote casts doubt over the procedures and expectations of epistemology as such.
     I want to proceed to show Cervantes' criticism of epistemology, which is the form which his anti-skepticism takes, by a look at the defenses and criticisms of the imagination in Don Quixote. Cervantes provides powerful and cogent reasons for relating dreams and imagination in general, showing for instance that treating imagined experience as the Humanist Cousin treats dreams is invalid. But most important, it is the relationship between imagination and dreams which weighs heavily against Descartes' dream argument. With imagination, as with dreams, Cervantes is inclusive rather than exclusive; Don Quixote shows us that knowledge is not limited by these experiences, which lie outside the reach of reason. In so doing he points up some flaws in the effort to place the limitations of certainty on knowledge, as the epistemologist wants to do.

     The terms in which the Humanist Cousin evaluates Don Quixote's dream recall the concerns which the Barber, the Curate, and the Canon of Toledo bring to bear on the romances of chivalry

discussed in Part I of the novel. In chapter 47 of the First Part, for example, the curate objects to the romances because they are “disparados”: they take imaginative liberties and thus appear inane. He thinks for this reason that they fail to serve a worthy purpose, that they fail to edify the reader: “‘son cuentos disparatados, que atienden solamente a deleitar, y no a enseñar: al contrario de lo que hacen las fábulas apólogas, que deleitan y enseñan juntamente.’” The romances of chivalry transgress the time-honored neo-Aristotelian requirements for verisimilitude in literature;⁶ they are long on pleasure, but short on profit: if these books must insist on proffering their lies, then let them do so with at least some semblance of truth. The Curate looks for a literature which would satisfy Horace's demands, one that would please and edify at the same time:

“. . . ¿qué diremos de la facilidad con que una reina o emperatriz heredera se conduce en los brazos de un andante y no conocido caballero? ¿Qué ingenio, si no es del todo bárbaro e inculto, podrá contentarse leyendo que una gran torre llena de caballeros va por la mar adelante, como nave con próspero viento, y hoy anochece en Lombardía, y mañana amanezca en tierras del Preste Juan de las Indias, o en otras que ni las descubrió Tolomeo ni las vio Marco Polo? Y si a esto se me respondiese que los que tales libros componen los escriben como cosas de mentira y que así, no están obligados a mirar en delicadezas ni verdades, responderles hía yo que tanto la mentira es mejor cuando más parece verdadera, y tanto más agrada cuanto tiene más de lo dudoso y posible. Hanse de casar las fábulas mentirosas con el entendimiento de los que las leyeren, escribiéndose de suerte que, facilitando los imposibles, allanando las grandezas, suspendiendo los ánimos, admiren, suspendan, alborocen y entretengan de modo que anden a un mismo paso la admiración y la alegría juntas; y todas estas cosas no podrá hacer el que huyere de la verosimilitud y de la imitación en quien consiste la perfección de lo que se escribe” (I, 47).

     In the Poetics (1640 a), Aristotle said that probable impossibilities were better than improbable possibilities. Generally, Cervantes takes care to keep the fiction within these limitations. He will for example explain the extraordinary or marvelous by reference to dreams (as in the case of Don Quixote in the Cave of Montesinos, or the Ensign's

     ⁶ For a discussion of Cervantes and neo-Aristotelian literary theory, see Alban K. Forcione, Cervantes, Aristotle, and the “Persiles” (Princeton: Princeton University Press, 1970), and E. C. Riley, Cervantes's Theory of the Novel (Oxford: The Clarendon Press, 1962).

talking dogs of the Novelas ejemplares), by avowals of sheer madness (as in the case of the man of glass, the “Licenciado Vidriera”), or by reference to spurious magic or cases of pretend (as in the episode of Maese Pedro's puppet show, the enchanted head, or Sancho's flight on Clavileño, in Don Quixote). But the bizarre occurrences of the romances of chivalry cannot be explained in these ways. Still, as Don Quixote shows, Cervantes is convinced that imaginative literature could be edifying, indeed, that it should be so, and that the exercise of the imagination is a necessary part of human nature and a valuable contribution to human knowledge. To deny the imagination, or to limit it in order to circumscribe human experience within limits set by the “possible” or the “probable,” is to restrict an innate human capacity in unwarranted ways. What Cervantes seeks is a defense of the imagination that would not sacrifice the distinction between imagination and reality.
     It is not that Don Quixote's critics are deeply unsympathetic to him or to the romances of chivalry. These examples of the work of the imagination are sources of aesthetic pleasure; that much cannot be denied: the town Barber admits to having read some romances of chivalry, and the Canon of Toledo says that he tried to write one himself (I, 45). The difficulty comes when, in an effort at arriving at empirically verifiable or predictable knowledge, we try to apportion imaginative experience between “true” and “false,” “reality” and “illusion.” These categories, and their codification in the literary precepts adapted from the Poetics by Renaissance theorists, are simply too confining to admit the large class of human experience which we call imaginative and which we customarily tell in the fictional modes. In the Quixote, a narrowness in literary judgment is indicative of a narrowness in appraising human experience, and it is this, over and above the judgments of these critics, which was Cervantes' deep concern; his interest in literary theory is part of a greater concern to understand the scope of valid ways of knowing human experience.
     Don Quixote's reply to his critics is important because it shows how the imagination can be given free reign without running the risks of skepticism, without losing confidence in our ability to distinguish imagination from reality; this is done by characterizing the imaginative experience in such a way that it is free from the rational objections that might be brought against it, free from the epistemological means by which we customarily indict it. On the heels of the Curate's objections to the books of chivalry, Cervantes

offers the example of Don Quixote's narration of the fantasy of the Knight of the Lake. Don Quixote asks his audience to consider “un gran lago de pez hirviendo a borbollones, que andan nadando y cruzando por él muchas serpientes, culebras y lagartos, y otros muchos géneros de animales feroces y espantables” (I, 50), then that the Knight of the Lake is challenged to demonstrate his valor by plunging into a pool of bubbling tar. As he does, he miraculously finds himself amidst flowering meadows comparable to the Elysian fields. He sees a castle with walls of gold and diamond turrets, is welcomed by a score of beautiful ladies, and is entertained by a celebration worthy of a god. Under any neo-Aristotelian system of evaluation, this narration would have to be ruled out, eliminated as invalid, not because it did not occur and was false or falsified, but because it could not occur; it is neither probable nor possible, but a gross impossibility.
     Cervantes is especially careful about the mode of narration that Don Quixote uses in this passage; the mode is that of a consideration or supposition (“‘. . . dígame, ¿hay mayor contento que ver, como si dijésemos, aquí ahora se muestra delante de nosotros, un gran lago de pez hirviendo a borbollones;’” i.e. suppose that, right here and now there were a lake of bubbling pitch before us . . .). His critics nonetheless insist on interpreting situations like this as if they were predictions, not suppositions; they insist on judging them as “possible,” “impossible,” “probable,” or “improbable.” They miss the point that a supposition simply asks us to consider certain circumstances, to project a situation, not to match it to reality.
     The distinction between suppositions and predictions is crucial to the encounter between Don Quixote and his critics, and it is basic to the flaws in Descartes' dream argument, so I want to make it rigorously clear.⁷ If an imagined situation is simply considered as a supposition, there is no evidence which could serve to prove or disprove it; this is internal to the concepts of “supposition” and “evidence.” No effort at counting reeds, for example, no matter how exact, could ever be evidential proof of the following case: Consider that if I have a pile of three reeds and bring to it three more, then there will be six. In the same way, nothing, no eventuality, could possibly disprove the proposition that this case is intended to state; it would be a misunderstanding, or a weak joke (as happens so often in

     ⁷ For the distinction between “suppositions” and “predictions” I rely on Stanley Cavell. The Claim of Reason, pp. 145-59.

Don Quixote) to break in with the observation that one of the reeds might break and then there would be seven, or that the piles might blow away and then there would be none. The mistake, or the joke, comes from reading a supposition as if it were a prediction. Don Quixote's critics do exactly this when they expect that there might be some evidence to prove or disprove the imagined situations of the romances of chivalry, and thus allow them to be judged as more or less “possible” or “probable”; when adding reeds, as in the example, it makes no sense to say that there will “probably be a pile of six” or that it is “improbable” that the piles will blow away and there be none. A supposition, the kind of situation Don Quixote imagines, could not be ruled out by any eventuality; any objection we could make to it, even on the basis of some eventuality which might contradict it, would only serve to change the supposition, alter the hypothesis, deflect the narration from supposition to prediction. (Reflecting on the nature of supposition, Stanley Cavell recalls a gag which shows this deflection. The kind of world it suggests is very much the world of Don Quixote: “A soldier being instructed in guard duty is asked: ‘Suppose that while you're on guard duty in the middle of a desert you see a battleship approaching your post. What would you do?’ The soldier replies: ‘I'd take my torpedo and sink it.’ The instructor is, we are to imagine, perplexed: ‘Where would you get the torpedo?’ And he is answered: The same place you got the battleship,’” The Claim of Reason, p. 151). As a fictional, imagined event in the mode of a supposition, Don Quixote's narration of the Knight of the Lake is impossible to exclude as false; being a mere supposition, there is no possible evidence which could count against it, but it is also impossible to verify as true. Don Quixote's exemplary reply to his critics shows that the imagined or supposed situation is not set up in defiance of knowledge, or of the science of knowledge which they profess, simply that there are claims to which the concerns of epistemology are irrelevant. Here, the problem is not that Don Quixote confuses reality and imagination but that there is an asymmetry between him and his critics, a point of incongruity between his world and theirs; they stand, as it were, on different sides of the joke. Where he has the prodigious ability to project situations, and to project himself into situations, where his ability to flesh out imagined circumstances is so great that he seems mad, his critics lack this ability. They suffer a failure of the imagination. Whereas Don Quixote's difficulty is only in knowing how and when to stop imagining, theirs is knowing how and when to start.

     The imagination shows up in two different ways in the Cartesian Meditations. These correspond to its use in suppositions and predictions. At certain points in the Meditations, Descartes says that the imagination has no bearing on rational certainty and that imagined things and situations are not themselves to be considered true or false: “whether I imagine a goat or a chimera, it is not less true that I imagine one than the other” (p. 159). This is the fictional imagination, like the “suppositions” or “considerations” of Don Quixote's narration of the adventure of the Knight of the Lake, or his account of what he saw in the Cave of Montesinos. Such imaginings are neither true nor false; they are unverifiable, and should have no bearing on rational certainty; they should be immune from it. Thus Descartes can say with pride that “the knowledge of my existence taken in its precise significance does not depend on those which [sic] I can feign in imagination” (p. 152).
     But the imagination appears in another guise in the Meditations, this one connected with the dream argument. Here the similarities between Descartes and Don Quixote should allow us to see how the dream argument is most deeply limited. When Descartes imagines that he might be mistaken about his existence, or that he has confused sleep and wakefulness (“Now let us assume that we are asleep and that all these particulars, e.g. that we open our eyes, shake our head, extend our hands, and so on, are but false delusions; and let us reflect that neither our hands nor our whole body are such as they appear to us to be,” p. 146), or when he supposes that he is a man with no body, or asks us to review with him the fact that he is there, seated by the fire, in his dressing gown, he is using the imagination as a mode of supposition, but he expects to be able to evaluate it as a prediction; that is why he asks us to review these situations with him, to go over them to see that they have not been mistakenly described or that some pertinent aspect of them has not been missed. The reason for conjuring them up in the first place is that if he is to arrive at unimpeachable knowledge, if he is to be certain of what he knows, then he must submit these suppositions to verification. Although they are only suppositions, Descartes treats them as if some eventuality, even the remotest eventuality, could prove or disprove them. It is the skeptic's business to deal in such eventualities, always to hold out for the remotest chance that he may wake up from the dream or somehow find that he is sleeping.

     In the dream argument of the Meditations, Descartes does as Don Quixote when he rubs his eyes and checks himself in the Cave of Montesinos (“‘Despabiléme los ojos, limpiémelos, y vi que no dormía, sino que realmente estaba despierto’” . . . , II, 23). Not only do we know that Don Quixote is utterly mistaken, we know that his dream is as unverifiable from “inside” as from “outside.” Dreams are like suppositions, essentially unverifiable. (In this sense the episode of the Cave of Montesinos is very much indeed like the narrated fantasy of the Knight of the Lake, and the criticisms with which they meet and to which they respond suggest the affinities between them.) And if dreams are unverifiable like suppositions, then what good is it to es imagine a dream, as Descartes does, and expect that some eventuality might confirm or disprove it? Any kind of evidence which might in serve the purpose, indeed, any evidence which might seem to bear on it at all, would be inapplicable. To expect that a dream, or an imagined or simulated dream, could be confirmed or denied by some eventuality is to misread fictions, like the neo-Aristotelian critics in Don Quixote, or like the soldier who didn't catch on (or rather, who caught on too much, too literally) to his instructor's example. Don Quixote of course does this all the time: he reads about knights-errant and strikes out to fight evildoers and giants. Through the comic incongruities which result, Cervantes provides his readers with enough evidence to spot the mistakes.
     The dream argument, and the use of the imagination which supports it as illuminated by these episodes of Don Quixote, are the bases for an insight into the limitations of epistemology as such, not simply to the flaws in one argument. Thus it would not, for me, do any good to object that Cervantes in Don Quixote seems to refute an argument which he could not have known. I see his engagement as with problems of skepticism and epistemology, and more specifically with the use of fiction as a mode of knowledge of the world. His response to skepticism and to its complement, epistemology, is to reject epistemology while remaining anti-skeptical; but this is only another way of saying that his purpose is to affirm the role of fiction in our relationship to the world (which, it might further be said, is an affirmation of the role of fiction in the task of philosophy). Cervantes shows that we relate to the world, including the “world” of our own experiences, in ways other than what the epistemologist calls “knowledge,” and that all we know of the world cannot be characterized in

terms of certainty.⁸ Cervantes' will to include the imagination and dreams within the range of valid human experience —within what we call the “world” in the broad sense— free of the caveats of reason, points this up. Don Quixote shows that in yielding to the temptation of certainty, epistemology is led to expect more of the world than it can possibly provide; Cervantes' rejection of epistemology, by contrast, provides a basis on which a discovery of the world, as such, may begin.

     ⁸ For a further discussion of this and related questions in relation to Cervantes, see my “Cervantes and Skepticism: The Vanishing of the Body,” in Essays on Hispanic Literature in Honor of Edmund L. King, ed. Sylvia Molloy and Luis Fernández-Cifuentes (London: Tamesis, 1983).