Review of General Psychology

1999, Vol. 3, No. 2, 101-117Copyright 1999 by the Educational Publishing Foundation

1089-2680/99/S3.00

Why Fiction May Be Twice as True as Fact:Fiction as Cognitive and Emotional Simulation

Keith OatleyUniversity of Toronto

Although fiction treats themes of psychological importance, it has been excluded frompsychology because it is seen as involving flawed empirical method. But fiction is notempirical truth. It is simulation that runs on minds of readers just as computersimulations run on computers. In any simulation, coherence truths have priority overcorrespondences. Moreover, in the simulations of fiction, personal truths can beexplored that allow readers to experience emotions—their own emotions—and under-stand aspects of them that are obscure, in relation to contexts in which the emotionsarise.

Fiction means something made, even some-thing made up. As compared with things found,such as the data of science, people aresuspicious of it. This suspicion leaks intocommon usage: Fiction has come to meanfalsehood.

The work from which has sprung muchliterary criticism, and some psychology ofreading and writing, is Aristotle's Poetics (trans.1970). In it poetry, like fiction, meant somethingmade, covering all literary works includingdrama. In what follows, I use the term fictionrather than poetry, but with the same inclusivesense. This article can be thought of as arguingfor a new relevance of the Poetics in modernpsychology.

The Poetics combined literary criticism andpsychology; indeed, it assumed an easy relation-ship between the two.

1 In modern times,

however, psychology has become empiricalscience. As such, it seems to offer no serious rolefor fictional literature. Narrative accounts ofhuman behavior by any single observer, withoutregard to sampling or the subjective bias of any

This article is based on an address I gave as retiringexecutive officer of the International Society for Research onEmotions, at the society's ninth meeting in Toronto, August1996. The writing and research for this article weresupported by the Social Science and Humanities ResearchCouncil of Canada, whom I thank warmly.

Correspondence concerning this article should be ad-dressed to Keith Oatley, Centre for Applied CognitiveScience, Ontario Institute for Studies in Education, Univer-sity of Toronto, 252 Bloor Street West, Toronto, Ontario,Canada M5S 1V6. Electronic mail may be sent tokoatley@oise.utoronto.ca.

individual perspective and without methods thatallow generalizability, fail to meet even minimalstandards for empirical psychology. And if tosuch defects is added the admission that fictionis a difficult-to-disentangle mixture of what hasbeen observed, remembered, and imagined byan author, then the predicament of fiction inpsychology seems hopeless indeed (see Oatley,1992).

Consider, for instance, Carlson and Hatfield's(1992) textbook on emotions, in which excerptsfrom novels are used and in which, every 10pages or so, a literary quotation is offered, set offfrom the text (e.g., "The ancestor of everyaction is a thought, Ralph Waldo Emerson"; p.80). According to the authors, they used artisticand literary "reference points [meaning illustra-tions] . . . to add liveliness and a human aspectto the scientific presentation" (p. x). Carlson andHatfield correctly caught the psychologist'susual sense of the issue, which is that it may beregrettable that scientific psychology is oftendry, and therefore, as a concession to students,some illustrations are allowed. Note, however,that beyond this, no role for fictional literature inpsychology is envisaged.

The assumption that fiction is largely irrel-evant to serious psychology was difficult tocounter until Gerrig's (1998) book. He showedthat fiction is easily accommodated by ordinary

1 To help bring psychology closer to literary criticism, Ipresent much of the argument in this article not in the formof inference from empirical data but as literary quotation,from which I hope certain phenomenological effects may beexperienced.

101

102 OATLEY

cognitive processes such as inferring fromincomplete evidence. His is a subtle argument,one that reaches thought-provoking conclu-sions. He proposed that, for psychologistsinterested in language, fiction is of interestbecause it has the same status that illusions havefor psychologists interested in perception. Be-cause of its clear nonveridicality, fiction canhelp to demonstrate cognitive processes thatunderlie both fictional and nonfictional under-standing. Gerrig concluded, correctly I believe,that both fiction and veridical understanding arebased on schematic construction.2

The primary psychological discovery of theconstructive nature of narrative understandingwas due to Bartlett (1932). He showed that whenpeople read a story, their comprehension andremembering of it are not faithful renderings.They are based on idiosyncratic and societalschemas available to the reader; these schemasassimilate salient details and the emotional toneof a story and can then, if remembering isrequired, generate a construction of it that ismore or less inaccurate. With Bartlett's finding,one can see that it is not just that stories aremade up; in people's understanding of them,they suffer uncontrolled changes and a furtherdrift toward the inaccuracy with which fiction isassociated. In a postmodern context, anotheraspect of this finding is that fictional stories arepolysemous. People necessarily make differentinterpretations of them.

The abilities of cognitive construction ofhuman beings have, of course, led to philosophi-cal suggestions that social psychology mightproperly be based on social construction (e.g.,Gergen, 1994). It is not my purpose to makesuch a case here, but my proposals are easilyassimilated to this movement.

Although Gerrig was persuasive in demon-strating processes such as schematic construc-tion that are common to both fiction andveridical understanding, the tenor of his argu-ment about the truth value of fiction remains oneof suspicion. He and his colleagues (Prentice,Gerrig, & Bailis, 1997) included, in theirintroduction to an experimental article, theobservation that in The Lyre of Orpheus thenovelist Robertson Davies has a charactertalking about fetal alcohol syndrome, but withthe facts wrong. Their experiments were empiri-cal demonstrations that readers encounteringassertions in fiction tend simply to accept them.

They found that, as compared with participantsreading a fictional passage who had firsthandknowledge of its setting, participants who didnot have such firsthand knowledge tended tobelieve information that was false. The problemwith fiction, according to Gerrig—and accord-ing to the conclusions of these experiments—isnot that of explaining what Coleridge called the"willing suspension of disbelief" (Gerrig, 1998,p. 6), which implies that some special effort orsome special psychological process is necessaryto explain how fiction works. The problem isquite the opposite. It is nonfiction, with itsprimary representative science, that requiresspecial effort. Fiction lowers barriers to assimila-tion; all too easily, it can undermine the effortfulprocesses of science. One can imagine thatGerrig would have thought Coleridge morecorrect to have written that science requires thewilling suspension of facile belief.

At the conclusion of his book, Gerrigsuggested that fictions may, nonetheless, beuseful in intermediate stages of thought, becausethey are cancelled when valid results arereached. He speculated that, perhaps, "informa-tion from fictions . . . has provided a positivebalance of utility over the period in which ourmental processes were shaped" (Gerrig, 1998, p.237). So, despite the interest of his argumentthat fiction is supported by the same cognitiveprocesses as ordinary comprehension, in the endGerrig too separated fiction—untrue and poten-tially misleading—from psychology, with itseffortful procedures for reaching valid empiricalconclusions.

In this article, I argue that such separation hasbeen a loss to both fields, and I make analternative proposal. Fiction is dismissed in thiskind of way only if it is understood as defectiveempirical description. I accept Gerrig's conclu-sion that fiction is to be understood by means ofthe ordinary processes of cognition, but Ipropose a conception within which fiction couldhave a wider place in modern psychology.Modern psychology as science has allied itself

2 Gerrig's (1998) proposal that fiction is a kind ofcognitive illusion seems not to be readily compatible withhis idea that fiction might be useful as an intermediate stagein valid thought. I &o not believe, for instance, that anycomparable argument for the utility of illusions has been putforth by perceptual psychologists. Perceptual illusions areregarded as having no intrinsic functions and as by-productsof normal processes.

FICTION AS COGNITIVE AND EMOTIONAL SIMULATION 103

with only one kind of truth: truth as empiricalcorrespondence. This kind of truth is necessarybut not sufficient. If psychology is to be fullypsychology, there must be consideration of twoother kinds of truth as well: truth as coherencewithin complex structures and truth as personalrelevance (see Table 1).

Empirical psychology obeys criteria of thefirst type of truth. Fiction fails this criterion butcan meet the other two. One could say, then, thatfiction can be twice as true as fact.

The proposals that fiction can fulfill thecriteria of truth as coherence (as in simulations)and truth as personal insight both give a centralrole to emotions, so before putting theseproposals forward, let me offer a word about thisrole. Bruner (1986) has argued that narrative isthat mode of thinking in which human agentswith goals conceive plans that meet vicissitudes.He compared it with the paradigmatic mode,which is used to reason about scientific matters.(This is the more effortful mode mentionedearlier, which requires suspension of too-facilebelief and careful use of different kinds ofinference in coordinated sequences [Oatley,1996].)

Not all narrative is fiction. Aristotle made thisdistinction: History is about the particular, aboutwhat has happened, whereas poetry (fiction) isabout the universal, about what can happen. Onecould add that empirical psychology, with itsconvention of past-tense descriptions of datathat have been gathered, can be grouped withhistory.

A further distinction of psychological interestis that (as many people have pointed out)whereas nonfiction is primarily informational,fiction is concerned with the emotions. Vicissi-tudes tend to elicit emotions. Thus, one couldadd to Bruner's proposal the following: Fic-

Table 1Types of Truth, Methods to Approach Them,and Criteria for Their Recognition

Type Method Criterion

Correspondence Empirical Valid discovery of facts,reliability, practicalusefulness

Coherence Simulation Production of emergenteffects without adhoc intervention

Personal Reflection Recognition, insight

Author

World

Reader Text

Figure 1. Diagram of the relations among author, world,

reader, and text, after an idea of Abrams (1953, p. 6).

tional narrative is that mode of thought aboutwhat is possible for human beings in whichprotagonists, on meeting vicissitudes, experi-ence emotions.

A typical fictional narrative is based on thefollowing schema (Rumelhart, 1975): agentwith goals and a plan that typically involve otheragents —• vicissitude —• emotion, which mapsonto Aristotle's proposal that a story has abeginning, a middle, and an end. Moreover,because the concern here is with a psychologynot only of writers and texts but of readers too, Ishould note, as Aristotle did, that in fictionemotions tend to be experienced by the reader.

Mimesis: Relation of the Text to the World

My first proposal concerns the relation ofliterary works of art to the world. Abrams (1953)offered a schema of literary criticism during thelast 2,300 years (see Figure 1) that clarifies thisquestion. The relation in question is indicated bythe link between "text" and "world." Thequestion of how writing can relate to the worldhas been a concern of literary critics fromAristotle onward, and the writing of this articleis itself an indication that the problem has notbeen fully solved.

The term used by Aristotle to describe therelation of world to text was mimesis. It was thecentral concept of the Poetics. Over the timeduring which such matters have been writtenabout in English, the term has almost universallybeen taken to mean imitation or representation.Shakespeare's version was that drama holds "as'twere the mirror up to nature" {Hamlet, III, ii,1. 22 [Methuen])hb Philip Sidney (1595/1986),a poet who was Shakespeare's contemporary,mimesis meant "a representing, counterfeiting,or figuring forth" (p. 114), and English transla-

104 OATUEY

tions for the succeeding 400 years havefollowed the same tradition.

The first experiments in photography oc-curred about 200 years ago. There followedother copying technologies, such as film, audiorecording, and video recording, that not onlyproduced images that endure but can imitatemore faithfully than can any human. So,unintentionally, the idea of art as a faithfulrepresentation of nature was finished.

Historically, literary criticism did not remainfixated on this problem. During the Renaissance,for instance, rhetorical issues became moreprominent, as indicated by the link in Figure 1between author and reader, with its question ofhow the author addresses the reader via the text.About 250 years ago, the era of romanticism inliterature began. This involved a further shift ofinterest to the link, in Figure 1, between authorand text, of how the writer produces the work ofliterary art. In the last 30 years or so, althoughgenerally still within the romantic period, thefocus has shifted again, now to the link betweenreader and text, to the questions of readerresponse and how readers interpret a text.

Important in discussions of the romanticapproach to literature was Wordsworth (1802/1984), who explained the relation of writer totext:

Poetry is the spontaneous overflow of powerfulfeelings: it takes its origin from emotion recollected intranquiliry: the emotion is contemplated till by aspecies of reaction the tranquility disappears, and anemotion, kindred to that which was before the subjectof contemplation, is gradually produced, and does itselfactually exist in the mind. (p. 611)

And so, it seemed, with the romantic concep-tion, the problem of art as mimesis might havebeen made to go away. Perhaps the mostarticulate philosopher of this romantic theory ofart was Collingwood (1938). He accepted thatmimesis equaled representation and saw this asthe mark of the technical. He went on to arguethat one needs a representation when one isworking to a plan: representation equals blue-print. In fiction, the idea of writing technically isthat of formula: to achieve a predetermined endof arousing specific emotions in the reader, suchas anxiety in a thriller or tender feelings in aromantic love story. When people see deliberateelicitation of specific emotions in other societ-ies, they call it magic, which they tend to regardas pseudoscience. But this is wrong, according

to Collingwood. Magic is legitimate socialactivity; it can- be seen in the shaping ofpreference by advertisement, in a skilled speakerarousing an audience, and in mass-marketpaperbacks and movies. Such emotional—magical—effects are not pseudoscience butpseudo-art. Art proper is the creative expressionof an emotion in a particular medium such aswords. Because emotions are responses to theproblematic and to the unanticipated vicissi-tudes of life, they basically demand creativeresponses, as Averill and Nunley (1992) havepersuasively argued. By definition, being cre-ative cannot be done to formula in relation toany representation or to determine any specificeffect. It is for this reason that Collingwood,with the typical understanding of mimesis asrepresentation, rejected Aristotle's Poetics astechnical hints for hacks and thereby finessedthe problem of what relation the work of artmight have to the world. With art proper, heargued, there is only the creative expression ofemotions, and there is no issue of representation.

Despite this, the question of the relation ofwords to world did not go away. Also, despitethe limitations of science and the naggings ofpostmodernism, science can bring words andother symbols into correspondence with thingsin the world, but this needs the complex andeffortful procedures discussed earlier.

Let me now, however, offer two accounts ofmimesis that, although related in part to theromantic solution, involve two interconnectedideas that do not deny an important relationbetween world and text but propose a relationother than mere representation. One is in thehumanist tradition, and the other is fromcognitive science. Both show how one can thinkof art as illuminating or clarifying the problem-atics of human action in relation to the emotions.

Fiction as Dream and as Imagination:A Humanist Metaphor

In "The Art of Fiction," Henry James(1884/1951) wrote in favor of representation: Anovel, he wrote, is "a direct impression of life."Robert Louis Stevenson (1884/1992b) made areply, saying it was no such thing. The novel isan abstraction. One cannot do without the ideaof a perfect circle, but it exists only in theabstract realm of geometry. One cannot dowithout the idea of character, but one encounters

FICTION AS COGNITIVE AND EMOTIONAL SIMULATION 105

it in pure form only in fiction. Stevenson(1884/1992b)wenton:

Life is monstrous, infinite, illogical, abrupt andpoignant; a work of art, in comparison, is neat, finite,self-contained, rational, flowing, and emasculate. Lifeimposes by brute energy, like inarticulate thunder; artcatches the ear, among the far louder noises ofexperience, like an air artificially made by a discreetmusician, (p. 182)

In 1888, Stevenson, continuing to ponder theissue, wrote the following:

The past is all of one texture—whether feigned orsuffered—whether acted out in three dimensions oronly witnessed in that small theatre of the brain whichwe keep brightly lighted all night long, after the jets aredown, and darkness and sleep reign undisturbed in theremainder of the body. (1888/1992a, p. 189)

Fascination with dreams was common forwriters of the romantic era. The following isfrom a French contemporary of James andStevenson, Hippolite Taine (1882):

So our ordinary perception is an inward dream whichhappens to correspond to things outside; and, instead ofsaying that a hallucination is a perception that is false,we must say that perception is a hallucination that is ofthe truth, (p. 13, my translation)

(Departing for a moment from the humanistmetaphor, Taine's theory of perception restatesHelmholtz's [1866/1962]. It has been restatedagain in recent connectionist theory that themechanism of visual perception depends on akind of constructive dreaming [Hinton, Dayan,Frey, & Neal, 1995].)

Gardner (1984) has used the same idea: Whata novel does for a reader, he stated, is to provokea certain kind of dream. Think of it like this:Close your eyes and then, instead of yourexperience being guided by signals from thevisual system, think of the novelist whisperingin your ear to offer purely verbal materials tostart up, construct, and sustain the dream. Gerrig(1998), in his book on the psychology ofexperiencing narrative worlds, came to a compa-rable conclusion; he proposed the metaphor of"being transported" (p. 2) by the book toanother time and place.

Imagination has been perhaps the favoriteway of understanding poetry in the romanticperiod: being able to glimpse the infinitethrough a strongly imagined nature poem, forinstance, or being able to glimpse the nature ofunconscious forces acting on one in gothicnovels of the same era (Miall, 1997). For

psychologists, the idea of imagination relateseasily to that of constructivism. Because mostpeople have the experience of creating dreamsequences from memories, day residues, preoc-cupations, and more obscure sources, the dreammetaphor, although including the idea of imagi-nation, brings the constructive potential of mindmore strongly into focus.

Human mental life depends strongly onconstructive abilities. What human minds dogenerally is to make models that parallel theworkings of the world (Craik, 1943). Languageis thick with ideas of this function: simile,metaphor, schema, parallel, analogy, theory,hypothesis, explanation, model, and so on. Thedream is just one example of the constructivemachinery running to create worlds in particularways or for particular purposes.

Abrams (1953) has proposed that the writer offiction be thought of primarily as drawingattention to things. The metaphor of the dreamof imagination shows how attention is indeeddirected. A text guides one in dreaming thisdream and not some other dream. Abramssuggested that, for the metaphor of the mirror,one substitute the metaphor of the lamp thatilluminates and clarifies. This lamp lights the"small theatre of the brain."

Fiction as Simulation: A Metaphor FromCognitive Science

One can counterpoise to James's idea of thenovel as "a direct impression of life" anothermetaphor that was not available until recently(Oatley, 1992): A play or novel runs on theminds of the audience or reader as a computersimulation runs on a computer. Just as computersimulation has augmented theories of language,perception, problem solving, and connectionistlearning, so fiction as simulation illuminates theproblem of human action and emotions. Withthe metaphor of mimesis as simulation, howeveranachronistic it may seem, one can rereadPoetics and see that something like this waswhat Aristotle really meant. He stated thatcorrespondences are inessential; great drama isoften in poetic verse, although no one everspoke verse in real life. Aristotle asked what it isthat makes a play moving. It is the plot, which is"the heart and soul, as it were, of tragedy... [it]simulates persons primarily for the sake of theiraction" (my translation, in italics, of mimesis).

106 OATLEY

What tragic drama achieves is a focus onessentials of human action. Nonessentials areexcluded. As Aristotle put it: "A poetic mimesis,then, ought to be . . . unified and complete, andthe component events ought to be so firmlycompacted that if any one of them is shifted toanother place, or removed, the whole isloosened up and dislocated."

Dream and simulation are both metaphors ofthe constructive activities that support fiction.Metaphors draw attention to some facet ofexperience and carry it over to something new orto something difficult to understand. Dreamingis familiar to most people, programming comput-ers only to a few; thus, as a metaphor, simulationworks well only for a minority. It has beenimportant in cognitive psychology, however, soI press it further.

If art were representation, then literary artwould simply figure forth or imitate a world andactions within it. But since the Russian formal-ists, two separable aspects of narrative havebeen recognized, thefabula (events of the storyworld) and the siuzhet (the aesthetic working ofthe plot). Following Brewer and Lichtenstein(1981), I label these aspects the event structureand the discourse structure. Thus, the writermust (a) provide an event structure, whichpsychologists can recognize as the material to beconstructed into a mental model (Johnson-Laird,1983) or situation model that, of course, doeshave aspects of representation (semantics), and(b) guide the reader by means of a discoursestructure that includes speech acts (pragmatics)and cues to the reader as to how this model is tobe constructed and run.

The parallel of narrative with simulation ismade stronger by the fact that two comparablestreams of information are also essential to anycomputer simulation. Consider, for instance,this fragment of a program I wrote, followingSharpies, Hogg, Hutchison, Torrance, and Young(1989) in the language Pop 11, as a prototype forstudents to augment in an artificial intelligencecourse. The program simulates a conversationpartner who can answer questions about the bestroute from anywhere to anywhere else in thedowntown Toronto public transit system. In thisfragment, two variables are declared by thecommand "vars." Then these variables ("trav-time" and "changetime") are given values.Next comes the start of the declaration of aprocedure, "setup ()," built on a database that is

a list of subway and streetcar stations in Toronto,

each joined by -"connects" to indicate which

station directly connects with which other.

vars travtime, changetime;

2 —> travtime; 5 —* changetime;

define setup ();

[[[BloorSubway spadina] connects [BloorSubway st

george]]

[[BloorSubway st george] connects [BloorSubway

bay]]

[[BloorSubway bay] connects [BloorSubway bloor

yonge]]

etcetera.

In this simulation, one stream of informationis a mode that exhibits correspondences with theworld and that could be empirically verified ordisconfirmed. Some of this information can beseen in the database over which the programcomputes within the procedure "setup ()":names of lines (e.g., "BloorSubway") andstations (e.g., "st george") that correspond tonames of lines and stations in the real Toronto.But the other stream of information has itsreality only in the computer world. It activatesprocesses to make the simulation run (e.g.,declaration of variables and definition of aprocedure).

For a writer of fiction, it would be no goodsimply offering something like "event + event +event"; he or she must also offer cues and meansfor running the simulation. For instance, inwriting a suspense story, there are three basicelements: (a) Get the reader attached to a likableprotagonist, (b) create a believable threat to thischaracter, and (c) have the threat originatingfrom some person or agency toward whom thereader will feel antagonistic (Vorderer, 1996). Atthe start, a few pages may cover weeks or yearsof event structure. Once the threat is established,one can devote many pages to each hour ofevent structure. As this suspenseful version ofthe simulation runs, the reader feels dulyanxious and continues to turn the pages until heor she can feel the relief of safety again.

The writer invites the reader to enter,Alicelike, through the looking glass and into theimagined story world. This is like entering aparticular kind of social interaction, as Goffman(1961) has proposed, through a semipermeablemembrane within which is a world with its ownhistory and its own conventions from whichmeanings are constructed.

FICTION AS COGNITIVE AND EMOTIONAL SIMULATION 107

Why Fiction Is Not FaithfulCorrespondence

Equipped with the idea of simulation, let usconsider more closely why imitation is such aninadequate translation of mimesis, why afaithful copy such as an audio recording of aconversation could never be part of literature(well, hardly ever). This section also showswhat is missing from correspondences and whatmust be provided to make them comprehensibleto humans.

The following is a transcript of part of apurposeful conversation. A complex new photo-copier had been delivered to a psychologydepartment, so Charles Button, Stephen Draper,and I took the opportunity to record someconversations. Our only input was to giveanother member of our department (who, in thisdepiction, is given the name of Xavier) the goalof finding someone (Yolande) to show him howto use the copier. Apart from inserting this goalinto his mind, everything else flowed naturally;nothing was contrived.

Xavier: Could you show me how to do the photocopy-ing?

Yolande: Double sided?

Xavier: Eh. Yer. I want to do, to do double sided.Yolande: Uhm, I don't know, some or . . .Xavier: Sorry, what do you do here? [points to buttons]Yolande: This one. [selects button "Duplex 2"] But, eh,some turn the other way round. You must have it.

Writers avoid imitating all aspects of suchconversations: "Eh. Yer . . . uhm," needlessrepetitions ("to do, to do"), and lapses ofgrammar. Note, moreover, that after the firstthree utterances, this real conversation is largelyincomprehensible from a transcript, although itwas not to the participants. Let me write it as itmight occur in a novel. I have preserved asmany words from the transcription as I could,and I have made each paragraph correspond toeach utterance in the transcript.

"Could you show me how to do the photocopying?"asked Xavier.Yolande knew that Xavier must be able to dostraightforward copying; he must want to do somethingmore, perhaps learn advanced features of this irritatingmachine, recently delivered to the Psychology Depart-ment, which she had spent a good deal of time learninghow to operate. Xavier only had one piece of paper:Maybe he wanted her to use it to show how the machinedid different kinds of copying. "You want to do doublesided?" she asked."Yes, I want to do double sided," said Xavier.

"It's more complicated than you might think.""Sorry. What do you do here?" asked Xavier. Hewanted to get started.

"You press this button, but usually you have to thinkabout how many copies you want, and whether youhave got single- or double-sided originals, andsometimes you have to worry about whether the copyon the second side will come out the right way round."

Note how the simulated (novelistic) versiondoes not really imitate. Instead, I showed—asAristotle stated—those essentials of humanaction that do not appear in behavioral copiessuch as tape recordings. These include charac-ters' goals and interpretations. Also, I had tocreate a discourse structure to help the readerconstruct a working event model.

Recently, Scheff (1997) proposed a newmethodology for the social sciences. He arguedthat records such as transcripts of conversationneed to be studied in great detail as biologistsstudy single specimens. The "specimens" areinterpreted under the illumination of theories upto the widest level of generality. The processreminds one of literary criticism, in which a textis given context that enables it to be betterunderstood. In two chapters of his book, Scheffindeed works with literary examples, aShakespeare play and Jane Austen's novels. Heworks in a comparable way with transcripts ofconversations of children and with telegramsbetween national leaders during the preliminar-ies to World War I. The conversations come tobe understood as negotiations in the sustainingof self, with its possibilities of intimacy and thethreat of shame.

To help one construct one's mental models ofthe world under consideration, Scheff s interpre-tations are equivalent to the novelist's guidanceabout how to understand events in the world.They are equivalent, too, to the stream ofinformation in programs that guides a computerin running a simulation. In recounting empiricalstudies in psychology, one follows a similarconvention: Methods and results sections wouldbe difficult for a reader to understand if it werethe first time the genre of the empiricalpsychological study had been encountered, andthey can sometimes be hard to understand evenfor experienced psychologists. They are, there-fore, given the context of introduction anddiscussion sections to suggest interpretiveframeworks.

Thus, not only do many imitative representa-tions require explication, but fiction does not

108 OATLEY

really imitate. It does not even copy conversa-tion, which many assert to be the form of lifemost indigenous to fiction.

As a further example, consider one of theworld's greatest novelists, Tolstoy, and a conver-sation from the novel that many consider hisgreatest, Anna Karenina. The following excerptinvolves the novel's central moments: Anna andher husband are watching a cavalry officers' racewhen a rider falls.

The officer brought the news that the rider was unhurtbut that the horse had broken its back.On hearing this Anna quickly sat down and hid her facebehind her fan. Karenin saw that she was crying, andthat she was unable to keep back either her tears or hersobs that were making her bosom heave. He steppedforward so as to screen her, giving her time to recover."For the third time I offer you my arm," he said.(Tolstoy, 1877/1901, p. 210)

Then, in the carriage on the way home:

"Perhaps I was mistaken," said he. "In that case I begyour pardon.""No, you were not mistaken," she said slowly, lookingdespairingly into his cold face. "You were notmistaken. I was, and can not help being, in despair. Ilisten to you, but I am thinking of him. I love him, I amhis mistress, I cannot endure you; I am afraid of you,and I hate you." (Tolstoy, 1877/1901, p. 212)

What is wrong with this? Empirically, almosteverything. It presupposes an observer eitherwith supernatural qualities or with a taperecorder hidden in the carriage. In any case, thewhole passage is too stagy; people do not talklike this. The whole thing is made up. It isobviously better written than ordinary peoplecould manage, but it has scant relation to realconversation.

If you want to know the correspondence ortruth of what goes on between married people inconflict, you bring a proper sample of marriedcouples into the laboratory, as Gottman andLevenson (1992) have done. You have eachperson complete psychological instruments tomeasure marital satisfaction. You wire them upto polygraphs and videotape them to recordspeech and facial expressions. You ask them totalk with each other, first about a neutral subjectand then about a topic of conflict between them.To obtain records of subjective emotions, youshow each participant the video the next day andhave the participant indicate what he or she wasfeeling at each point.

The point is this: Many, perhaps most,transcriptions would not be intelligible on their

own. There are two routes. Either transcriptionmust be accompanied by interpretative analysis(e.g., of the kind offered by Scheff or Gottmanand Levenson) so that the reader can understandit, or, if utterances are to be understoodsmoothly by the reader at the time of reading,they must be altered (fictionalized). They maythen become parts of a simulation in whichmeanings can become comprehensible. Fictionprovides context to understand the elliptical. Itoffers the context of characters' goals and plans.It gives a sense of how actions lead tovicissitudes. It allows, too, the reader toexperience something of emotions that canarise. All of these elements are omitted from afaithful, empirically unexceptionable copy ofreal life.

In fiction based on simulation it is important,of course, to have some imitative correspon-dence with the real world; some aspects arerepresented in the event model, just as the namesof subway stations were represented in mysimulated conversation partner. Without suchcorrespondence, fiction founders. But piece-by-piece correspondence is in the service ofcoherence within the larger structure and of therelationship with the reader.

Dream and Simulation as AlternativeMetaphors of Construction

Dream and simulation are alternative meta-phors; one rather than the other is likely to bepreferred by each reader. The essence of both isthat mental life is constructive: The machineryof cognitive construction, the reader's share, iswithin every one.

For the literary experience of fiction, theJapanese drama form Bunraku (Inoura &Kawatake, 1981) offers a slightly surprisingglimpse of human constructive abilities. InBunraku, the characters are represented bypuppets that are about two-thirds life size. Eachis activated by three puppeteers, who appear onstage dressed in black cowls. The chief puppe-teer holds the puppet's torso, manipulating itsright arm and expressions of its face. Oneassistant manipulates its left arm and the otherits legs. The woi ds of the play are poetry read bya narrator who is sitting in full sight at the side ofthe stage with an open book. For these dramas,stage settings are colorful and striking, and theplays are accompanied by music.

FICTION AS COGNITIVE AND EMOTIONAL SIMULATION 109

The elements of this drama form, described inthis way—actions represented by artificialpuppets, partly hidden forces that control themdepicted by visible puppeteers, the metricalpoetry read by a narrator, and the music—couldall seem utterly disparate and incomprehensiblystrange formal devices. But the Bunraku formbecame popular during a period of rapidurbanization in which Japan was emerging fromfeudalism and undergoing a humanistic renais-sance comparable to the European Renaissance.The plays include themes of power, revenge,and adultery and were written for this form bygreat poets, including Chikamatsu, "the Japa-nese Shakespeare." The audiences integrate theseemingly disparate elements, constructing andcreating for themselves a seamless, movingexperience.

Westerners think of themselves as preferringrealism in drama. But, lest the idea of integrat-ing the disparate elements of Bunraku seemsstrained, note that film, that seemingly mostrealistic form, is usually made with severalcameras in the same scene. Some cameraszoom, others track smoothly on rails, and yetothers may be located high in the air. Shots areedited to form sequences that no humanobserver could ever experience. The editing isdesigned not to imitate what a real humanobserver might ever see but to give theimpression of being in exactly the right place, atthe right moment, to follow a plot. The stagingand editing are the equivalent of the discoursestructure of a literary text, the housekeeping of aprogram that allows the simulation to run.

Personal Properties of Fiction

The first proposal of this article is that fictionis a kind of simulation that serves as a coherenceform of truth. The second is that fiction can beinvolving; it can serve as a personal truth andgive rise to insight. Although nonfiction such asscience can be interesting, it is not generally partof its function to touch its readers in a personalway.

Larsen and Seilman (1988) argued thatwhereas perception could conceivably be ex-plained in such terms as extracting invariants,the generation of meanings from black andwhite marks on a page can be explained only interms of readers' constructive activities. They

investigated this further by what they called themethod of self-probed retrospection. They hadreaders mark either a fictional or an expositorytext (each of about 3,000 words) when amemory occurred during reading. The readersthen went back to identify and classify thememories. The authors found that twice as manymemories in which the reader was personallyinvolved as an actor (as compared with anobserver of reported events) occurred with thefictional text as with the expository one. Theyargued that this kind of reminding provides thebasis of a personal resonance between themes ofa story and those of the reader's life.

It seems likely that the provenance of fictionis the ordinary conversation (turning things overtogether) in which humans have taken part in allknown societies. Such conversation is mostfrequently about what people have done, whatthey are up to, and what the personal implica-tions of such doings might be. Thus, Dunbar(1993) found that, in conversations in a collegerefectory, approximately 70% of the talk (slightlymore when women were speaking and slightlyless when men were) involved personal relation-ships, personal experiences, and social plans. Itis natural to suppose that fiction derives fromthis kind of (mostly historical) narrative toinclude what one can imagine people doing.Such narratives, of both the conversational andfictional kind, are not just common; they are atthe center of social life.

What is the function of such narrative?Dunbar (1996) has argued that conversationabout oneself and others has a function ofsocioemotional bonding. As such, it occupiesthe role of grooming in other primates. Underthis proposal, the sharing of fictional narrative,starting from the stories that parents tellchildren, will be as bonding as the recounting ofremembered events. The private reading ofnovels is less immediately intimate but retainssome of the sense of social connection.

Rime (e.g., 1995) has found that whenever anemotion is remembered a few hours after itoccurred, it is likely to be confided to someoneelse. Confiding such everyday emotions andturning them over with others may be compa-rable to experiencing emotions in fiction in waysthat allow one tci reflect on them from the newperspectives of the stories' circumstances. Bothof these modes, therefore, may help self-understanding.

110 OATLEY

But fiction has yet other functions. Fiction inthe form of myths and cultural themes contrib-utes to the forming of societies and individuals'identities within them. When intense emotionsare elicited by intense adversities, societiestypically offer their members ways of dealingwith these disrupting elements. Most indigenousforms of narrative therapy and healing, forinstance, have cultural functions of reintegratingdisordered individuals and disturbing experi-ences with communal beliefs and practices. Inthe same vein, Spence (1982) has argued that, inpsychoanalysis, analysand and analyst jointlycreate a narrative that replaces the analysand'ssymptomatic account. It is not an empiricallyverifiable history of the analysand's life. It is aform of art, and its therapeutic value depends onthis. One of the most important suggestionsmade in this area is that certain rituals, as well ascertain kinds of drama and other fictional forms,achieve their principal therapeutic value foremotions that have been too overwhelming forpeople to assimilate in ordinary life. Thesenarrative forms prompt individuals to recallsuch devastating emotional circumstances andcome to terms with them (Scheff, 1979; I discussScheff s hypothesis in more detail later).

Emotional Involvement in the Genreof Tragedy

It can be seen from the classical Greektheater, how personal involvement in certainkinds of narratives can help members of anaudience understand the problematic in sociallife and integrate understandings with existingmodels of self and other. By general consent, themost momentous literary discovery of Aeschy-lus and Sophocles in their tragedies was thathuman action, no longer directed by gods, oftenproduced unforeseen results. If there was a timewhen gods directed human action, those gods, ofcourse, had perfect mental models and unlimitedagency. They were, in other words, omniscientand omnipotent, and therefore all turned out asintended. Such a time can be glimpsed inHomer's Iliad (trans. 1987). The psychologicalinsight is that if now humans live in a world inwhich they choose many of their own actions,then their mental models are usually imperfectand invariably incomplete, whereas their agencyis limited by the constraints of embodiment.So—necessarily—although they start with aspi-

rations (goals) and although they contrive plans,people's actions .can have consequences they donot foresee (vicissitudes). As authors of theseactions, people experience their consequencesand must take responsibility for them, sufferingthe changed personal circumstances and theiraccompanying mental states (emotions).

In Aeschylus's (trans. 1966) cycle of plays,the Oresteia, the gods have an intermediateexistence between those powers whose planswere responsible for all human action andsuffering (as in the Iliad) and their seemingabsence from modern life. In the final play of thetrilogy, their roles as aspects of human psychol-ogy become clear. Vengeful anger is personifiedas the "furies" who appear as a chorus that canpursue and drive people in ways that areindividually as well as societally destructive.They pursue Orestes for having killed hismother. But the voice of civic justice ispersonified as Athena. The furies' power is nolonger untrammeled. They must plead their casein a judicial process. Orestes says that hismother, Clytemnestra, slew his father and thatthe outrage of this and the compelling will ofApollo drove him to matricide. In the play,responsibility for resolving this impossible andangry crisis moves beyond family vendetta.Angry vengeance becomes a part (not thewhole) of the resolution and must enter arelation with civic wisdom.

The members of the original audiences forsuch plays sat not in the privacy of cinemadarkness but in full sight of their neighbors,beside them and across the other side of thesemicircular seating area. Such plays were notjust social occasions; they were occasions ofsocial engagement, enactment, and integration.

In Sophocles's King Oedipus (Aristotle'sexample of the perfect tragedy; trans. 1947), theidea of human knowledge being limited isexplored in a yet more personal way. Theimplication is that one must limit too one'sarrogance. Oedipus had come as a prince fromCorinth, to save Thebes by solving the riddle ofthe deadly Sphinx that menaced the city. He wasreceived by the Thebans with joy, became then-king, and married Jocasta, widow of Laius, theformer king. At the start of the play, the city isagain in affliction. Jocasta's brother says it iscaused by the pollution of harboring the killer ofLaius. Oedipus forms the goal of again savingthe city. He will find the murderer, and he starts

FICTION AS COGNITIVE AND EMOTIONAL SIMULATION 111

a plan to do so. The play is full of metaphors ofseeing and blindness, of gods knowing andforetelling the future, of humans rejecting thisknowledge and longing to shape their own fate,thinking they know but being ignorant.

What Oedipus does not know is that he hadbeen adopted by the king of Corinth and that hehimself killed Laius in a skirmish on his journeyto Thebes. As both searcher and suspect, hegradually uncovers the secrets of his own birth,of his having been put out to die, pinioned by hisankles on a hillside, by a shepherd-servant ofLaius, to escape the oracle that Laius's sonwould kill his father and marry his mother. Asthe narrative of his own guilt is pieced together,as finally the shepherd is summoned and tellsthat he was unable to condemn the infantOedipus to death, the full discovery occurs.Oedipus, in frenzy, calls for a sword and goes tofind Jocasta: "Where is that wife, no wife ofmine." As the revelations began to be piecedtogether, when she heard the shepherd was to bequestioned, she recognized the terrible truth,which had been hidden from her too. The chorusrecounts how she fled to her bed chamber,locking the doors, crying to Laius in despair, andhanging herself. Oedipus burst through thedoors.

The King saw too, and with heart-rending groans

Untied the rope, and laid her on the ground.

But worse was yet to see. Her dress was pinned

With golden brooches, which the King snatched out

And thrust, from full arm's length, into his eyes—

Eyes that should see no longer his shame, his guilt.

On rereading this play (for perhaps the fourthor fifth time) for the writing of this article, Ifound myself once again moved, found tears inmy eyes. It is not the psychological descriptionof a process that is moving. It is its content thatgives meaning to the tragic vision of human lifeand leads to the play's famous closing words ofacceptance but with a certain subdued dignity inthe shared human condition.

Then learn that mortal man must always look to his

ending,

And none can be called happy until that day when he

carries

His happiness down to the grave in peace.

The nature of the personal involvement hereand, I would argue, in other literary fiction isemotional. Not only do the characters experi-ence emotions as they meet vicissitudes, but

typically members of an audience, or readers,also experience emotions.

Why Emotions Are Important in Fiction

Emotions are important in fiction becausethey arise at the vicissitudes of life, brought onby actions with unforeseen results. Because ofthe way in which they can monopolize attention,emotions prompt one to concentrate on theseissues (e.g., how this action could have led tothat outcome). Moreover, they can also point togoals and concerns of which one is unaware. Ina study of people who were asked to keep diariesof incidents of emotion, Oatley and Duncan(1992) found that 6% of emotions occurredwithout the participants knowing why, and, indifferent samples, between 5% and 25% ofemotional incidents had aspects participants didnot understand. If one falls in love, why withthat particular person? If one feels grief forsomeone who has died, why does one feel angryas well?

Here is an example: One of Oatley andDuncan's participants, a nurse who worked bymaking home visits, was interviewed afterrecording an incident of disgust in her diary. Shehad felt the overwhelming sense of disgust as apatient cleared his throat. So strong was herdisgust that she had to interrupt what she wasdoing, and her emotion made it difficult for herto perform necessary procedures. Thinkingabout the incident afterward, she found heremotion still sufficiently compelling to make herwonder about whether she should give up herjob. When interviewed, she said that the man'sclearing his throat reminded her of her ex-husband, who used to clear his throat in a similarway, but that she could not understand why thepatient, who was an inoffensive person, madeher feel like this or why the feeling was sostrong. So here are the beginnings of a story of awoman, and her life, and of something to dowith an emotion that had an element of themysterious about it.

Emotions can point to goals. They happenwhen events occur that relate to one's goals, tothose concerns that are important to one. So anemotion not understood means that somethinghas happened wjth an obscure implication.Perhaps one has some goal or concern of whichone is unaware, or perhaps one has an innerconflict of which one recognizes only one side.

112 OATLEY

Emotions are at the center of literature becausethey signal situations that are personally impor-tant but that might be either inchoate or justbeyond the edge of easy understanding. Thesimulations that are novels, plays, movies, andso forth can allow people to find out more aboutthe intimate implications of their emotions.They offer a laboratory space that, relative toreal life, is safe and can make the relations ofemotions to goals and action easier to understand.

Fictional Elicitation of Affect and ItsRelation to Insight

There are several ways in which emotionalstates are elicited in fiction (Oatley, 1994). HereI distinguish between the elicitation of generalaffect (discussed in this section) and theelicitation of specific emotions (discussed in thenext). Literary narrative, in contrast to nonliter-ary narrative, has properties that have beendiscussed by literary critics for at least twomillennia. These properties are often referred toin such terms as foregrounding and defamiliar-ization (see, e.g., Miall & Kuiken, 1994).Certain phrases or passages can strike one aneweven when what is referred to is familiar.Literary writers use a range of stylistic devices,from simple alliteration to complex metaphorsand metonymies, that capture one's attentionand prompt one to consider things in new ways.Sikora, Miall, and Kuiken (1998) found thatsuch features were associated with affect inparticipants who read Coleridge's "The Rime ofthe Ancient Mariner" and that they increased thelikelihood of insightful thoughts. Readers be-came involved in what the authors calledenactment. In this mode of personal engage-ment, the boundaries between reader and poetbecame blurred, and the reader reflected onexistential issues. Defamiliarization thus firstprompts dissolution of aspects of a schema.Insightful resolution can occur when the schemareaches a new accommodation.

The question of what prompts insight haslong been associated with the idea that literaturecan be profound. Longinus introduced the ideathat certain expressions or images seem to comefrom some realm beyond the merely human.They can be sublime thoughts on fundamentalissues of existence. Again, as with the lines fromSophocles, I offer a phenomenological demon-stration from Shakespeare. Note that the effect is

not mere surprise, because these are perhapsShakespeare's best-known words. Consider firstan unsublime version that comes from the FirstQuarto of Hamlet, a version that probably waspirated by the actor who played Marcellus inone of Shakespeare's company's productions.The pirate was accurate with his own lines, butelsewhere he ran aground on the shoals of theordinary.

To be or not to be, ay there's the point,

To die, to sleep, is that all? Ay all:

No, to sleep, to dream, ay marry there it goes

For in that dream of death when we awake

And borne before an everlasting judge.

Now compare this with the version from thecollated Second Quarto and First Folio editions,thought to be Shakespeare's own words.

To be or not to be, that is the question:

Whether 'tis nobler in the mind to suffer

The slings and arrows of outrageous fortune,

Or to take arms against a sea of troubles,

And by opposing end them. (Hamlet, III, i, 1.56-60)

A poet who sometimes achieved the sublime,and who also wrote about the issue in his letters,was Keats. The following is from a letter to JohnTaylor, dated February 27,1818:

In Poetry I have a few Axioms, and you will see how far

I am from their Centre. 1st I think Poetry should

surprise by a fine excess and not by Singularity—it

should strike the Reader as a wording of his own

highest thoughts, and appear almost a Remembrance—

2nd Its touches of Beauty should never be half way

ther[e] by making the reader breathless instead of

content: the rise, the progress, the setting of the

imagery should like the Sun come natural too

him—shine over him and set soberly although in

magnificence leaving him in the Luxury of twilight.

(Keats, 1818/1966, p. 46)

Keats's idea of "a fine excess" rather than"singularity" (oddness) seems apt. It is right toofor mimesis as simulation: One simulatesessentials, thus exaggerating them slightly incomparison with inessentials, for the sake ofunderstanding. Just as striking is the idea thatpoetry should be a "wording of [the reader's]own highest thoughts, and appear almost aRemembrance." This was not a new idea incriticism, but Keats phrased it with such tellingsimplicity that some part of oneself leaps tomeet it, with exactly that kind of engagementthat Sikora, Miall, and Kuiken described as themode of enactment, taking on the perspective ofthe poet.

FICTION AS COGNITIVE AND EMOTIONAL SIMULATION 113

Fictional Elicitation of Specific Emotions

There are many methods of demonstratingand investigating the specific emotions thatoccur to readers when reading: emotions ofanger, sadness, joy, and so forth. One way is adevelopment of self-probed retrospection (dis-cussed earlier), as devised by Larsen andSeilman (1988). In our research group, we havereaders mark the text of a short story when amemory occurs, as Larsen and Seilman did. Foreach memory, we have them mark an M. Inaddition, we ask them to mark an E when theyexperience an emotion and, in more recentstudies, also to mark a T when a thought occursthat is not a direct paraphrase of what they haveread. We have now used the method with some400 participants reading short stories. Only theoccasional participant has failed to record bothmemories and emotions in this way.

In one study by Angela Baison and myself(Oatley, 1998), we asked high school students toread a 4,000-word story about adolescentidentity that had either a male protagonist (byCarson McCullers) or a female protagonist (byAlice Munro). The boys experienced a mean of3.9 emotions, and the girls experienced signifi-cantly more, with a mean of 6.7. We agree withLarsen and Seilman that the occurrence ofmemories points to the personal engagement ofthe reader in the story. So, too, does theexperience of emotions. The significantly largernumber of emotions and memories experiencedby the girls is explained, we believe, by the girlsbeing equally able to identify personally withmale or female protagonists (they had about thesame numbers of emotions and memories forboth stories), whereas the boys had feweremotions and memories if they read the storywith the female protagonist than if they read theone with the male protagonist.

Another study showed that emotions canaffect a story's interpretation. Seema Nundy andI (see Oatley, 1996) asked participants to read ashort story called "Sara Cole," by RussellBanks. We found that exactly the same storyengendered different emotions in different par-ticipants, and the different emotions promptedparticipants to different modes of interpretation.Participants who became sad during the first partof the story were significantly more likely toengage in a form of reasoning about the last partthat is known as backward chaining: reasoning

backward to find causes of one's emotionalstate. Those made angry by the story engaged inforward chaining, reasoning forward about whatthe protagonist might do about the emotionalstate.

The finding that short stories generallyproduce emotions, but of different kinds, tendsto confirm the constructivist theory of literaryreading. Emotions that are produced by narra-tive are the participant's own. The finding thatsuch emotions affected reasoning shows one ofthe ways in which qualitatively different interpre-tive readings can be given to the same story (cf.Barthes, 1975).

So general is the finding that emotions can beelicited by narrative that psychologists wishingto induce emotions in the laboratory nowtypically use this method. Usually excerpts fromcommercial movies are shown. For instance, toinduce sadness, an excerpt from Sophie's Choiceis widely used: Sophie is forced to choose whichof her two children to sacrifice to the whim of asadistic Nazi officer. Elicitation of emotions byfilm and by short stories is based on similarprocesses, although film is preferred for reliableand quick induction in laboratory manipula-tions. In contrast to the range of idiosyncraticemotions produced by certain fiction (as in the"Sara Cole" story discussed earlier), somepieces of text or film (such as the excerpt fromSophie's Choice) can reliably give rise to onerather specific type of emotion. This is thetechnical manner of magic, or pseudo-art, asdescribed by Collingwood. Such inductions,then, have an interest for understanding ofemotions such as visual demonstrations inperception (cf. Gerrig's, 1998, argument thatfiction is like perceptual illusion). It is regardedas important among those who study visualperception to have the phenomenological expe-riences of additive color mixing, of the three-dimensional effect of binocular viewing ofstereograms, and so forth. Similarly, I wouldargue, it is important for psychologists whostudy emotions to experience them and theireliciting circumstances both when these condi-tions tend to prompt idiosyncratic emotions andwhen they more narrowly cause specificemotions.

At least three distinct psychological processeshave been identified that lead to the readerexperiencing specific emotions in fiction. Theseare as follows.

114 OATLEY

Identification. The emotions of identifica-tion fall out easily from the theory of fiction assimulation (Oatley, 1992, 1994) discussed ear-lier. The part of the mind on which the reader (oraudience) runs the simulation of a novel, play, orfilm is the planning processor. Ordinarily, it isused in conjunction with one's mental model ofthe world to assemble actions into plans so as toattain goals (e.g., to pick up some milk on theway home from work). By contrast, when one isreading a story, the plot takes over the planningprocessor. One tends to identify with theprotagonist, adopt his or her goals, and take onhis or her plans. Then—straightforwardly, in theterms of this theory—one experiences emotionsas events, and outcomes of actions are evaluatedin relation to the protagonist's goals. But here isthe extraordinary feature: Although the goalsand plans are simulated, the emotions are not.They are the reader's own.

Sympathy. A second mode of emotionalexperience of narrative has been expounded byTan (1994, 1996). In this mode, the writer offerspatterns of events of the kind that causeemotions. From these, the reader attributesemotions to story characters and experiencessympathetic emotions toward these characters.Tan has argued that film is especially effective ininducing spectator emotions.

Autobiographical memory. A third idea isthat of Scheff (1979). He argued that a primaryfunction of narrative and drama is to allow thereliving of emotional memories from theautobiographical past. In ordinary life, Scheffargued, emotions can sometimes be overwhelm-ing, so their meaning is not recognized. In otherinstances, one distances oneself from emotionsand suppresses them. In either case, the effectsof such experiences can leave one in emotionalarrears, because events have occurred that haveprofound emotional significance but that one hasnot assimilated. Drama, novels, and certainkinds of rituals allow re-evocation of suchemotions, neither underdistanced nor overdis-tanced but at what Scheff called an optimalaesthetic distance at which they can be under-stood and assimilated.

Relation of emotion-producing processes tosimulation. Of these processes, identificationis explained most directly in the simulationtheory of literary experience, but sympathy andthe prompting of memories also derive simplyfrom this theory. Sympathy will occur as a

character's predicament is vividly imagined.Emotional memories are prompted both byemotions elicited by the other two modes and byevents generated in the simulation that depictthemes such as meetings, partings, misunder-standings, and reconciliations with which read-ers might have personal resonance.

Does Fiction Have Properties ThatNonfiction Does Not?

I have argued that fiction has effects that arepersonal and often emotional. There is perhapsno sharp dividing line between fiction andnonfiction. History and biography, for instance,can be profoundly moving, and an implicationof this article is that psychological writing mightalso benefit by moving occasionally beyond thenarrowly informational. Nonetheless, the typicalemotional functions of fiction contrast with thetypical informative effects of nonfiction. Thiscontrast prompts the following question: "Isthere some psychological function that fictionachieves better than nonfiction?" The questionis an old one. One answer is that of PhilipSidney (1595/1986), who wrote that the poet

yieldeth to the powers of the mind an image of thatwhereof the philosopher [one might now say scientist]bestoweth but a wordish description: which dothneither strike, pierce, nor possess the sight of the soulso much as that other [the poet] doth. (p. 119)

One can read experiments, or tables ofquestionnaire responses, on emotions in thepsychological literature, but when one thinks ofexperiencing and understanding an emotion,something like this proposal is needed. Sidneycontinued:

Let but Sophocles bring you Ajax on a stage, killingand whipping sheep and oxen, thinking them the armyof Greeks, with their chieftains Agamemnon andMenelaus, and tell me if you have not a more familiarinsight into anger than finding in the Schoolmen [its]genus and difference. (1595/1986, p. 119)

This kind of statement has generally beenused, as indeed Sidney used it, in adversarialdebates about the status of fiction. In academiclife, the result has been a segregation of thosewho prefer the effects of poetry (and go intodepartments of literature) from those who preferempirical truth (and go into departments of thefaculty of science). Such either-or thinking maybe appropriate in some circumstances, but in

FICTION AS COGNITIVE AND EMOTIONAL SIMULATION 115

psychology it has cut the subject off fromimportant aspects of its subject matter.

One need not choose between what Sidneydescribed as "the sight of the soul" and"wordish description." If we are to be properpsychologists, we need to include with theempirical the Gestalt understandings of simula-tions, and also the personal.

George Eliot, novelist and pioneer socialscientist, put forward one of the most cogentarguments yet made for the relation between theempirical and the personal in psychology:

The greatest benefit we owe to the artist, whetherpainter, poet or novelist, is the extension of oursympathies. Appeals founded on generalizations andstatistics require a sympathy ready-made, a moralsentiment already in activity; but a picture of humanlife such as a great artist can give, surprises even thetrivial and the selfish into that attention to what is apartfrom themselves, which may be called the raw materialof moral sentiment.. . . Art is the nearest thing to life; itis a mode of amplifying experience and extending ourcontact with our fellow-men beyond the bounds of ourpersonal lot. (George Eliot, 1856/1963, p. 270)

A Hypothesis About the Nature of Insight

If, as George Eliot believed, the empirical andthe experiential are complementary, then it is aconcern to psychologists to keep them inregister with each other. As well as beingsimulated constructions of the imagination,novels contain distillations of folk theory.Without continual mapping back and forthbetween theories based on empirical science andtheories based on experientially derived folkunderstanding, these two kinds of theory losetouch with each other. Scientific psychologyloses relevance for the task of explainingourselves to ourselves, and folk psychologybecomes cut off from what is known scientifi-cally. Fiction, with one of its goals being theunderstanding of experience beyond "our per-sonal lot," therefore can provide cultural objectsthat can be shared and discussed amongmembers of a culture. The current renewal ofinterest in reading groups in Europe and NorthAmerica (see, e.g., Long, 1986), as well asdiscussion of fiction on the Internet (Murray,1997), contributes to the development of sharedfolk psychological theory.

There is enough evidence, both from literarycriticism and from empirical studies of litera-ture, to suggest the following hypothesis.Insights of a personal kind when reading fiction

are more likely to occur when the reader ismoved emotionally by what he or she is readingand when the accompanying context helps theunderstanding of the resulting emotions.

Aristotle's statement about katharsis of theemotions of pity and fear in the Poetics was aversion of this hypothesis. The term katharsishas been regarded as obscure and translated as"purification" or "purgation." According toNussbaum (1986), however, there is nothingobscure about it. It has the same root as kathairoand katharos, used commonly by Plato andAristotle, with the meaning of clearing awayobstacles. Katharsis has been translated aspurification, but such a meaning would be thesecondary one of clearing away spiritual impedi-ments. Similarly, it has been given the derivedmedical meaning of purgation, an alimentaryclearing away. It is likely, argued Nussbaum,that Aristotle intended neither of these. Theyboth express a Platonic distrust of the emotionsand imply that the theater might be a place apartfrom ordinary life in which such unfortunateelements could either be transformed intosomething better or be expelled by innercleansing. Nussbaum argued for a straightfor-ward cognitive meaning: clearing away ob-stacles to understanding ("clarification"). Thefollowing is Nussbaum's translation of thesentence in the Poetics that includes the term:"The function of tragedy is to accomplishthrough pity and fear, a clarification (orillumination) concerning experiences of thepitiable and fearful kind" (1986, p. 391).

This is a central hypothesis about thepsychological value of literature. There arealready some empirical suggestions, such as thefinding of Sikora et al. (1998) that literaryexpressions that cause defamiliarization andaffect make insight more likely. As yet, there areno findings that specific emotions, accompaniedby contexts allowing their clarification, havesuch effects. But the idea is susceptible toempirical test.

Specific emotions are primarily evaluationsof events in relation to goals; they are aboutwhat is important to one. If, therefore, emotionsof reading are one's own, not just palereflections of the demotions of fictional charac-ters, insight would be more likely when suchemotional experience is combined with contextsof fictional simulations that allow it to be

116 OATLEY

understood better than is often possible inordinary life.

In simulations that are psychological fiction,the relation of emotion to empirical truth is thatthe story models on which they depend wouldbe well served by being based on empiricalfindings of psychology rather than by being illinformed. At the same time, we should take carethat the empirical part of psychology mightoccasionally move us, rather than being exclu-sively technical.

References

Abrams, M. H. (1953). The mirror and the lamp:

Romantic theory and the critical tradition. Oxford,England: Oxford University Press.

Aeschylus. (1966). The Oresteia (R. Fagles, Trans.).New York: Penguin Books.

Aristotle. (1970). Poetics (G. E. Else, Ed. & Trans.).

Ann Arbor: University of Michigan Press.Averill, J. R., & Nunley, E. P. (1992). Voyages of the

heart: Living an emotionally creative life. NewYork: Free Press.

Barthes, R. (1975). S/Z (R. Miller, Trans.). London:

Cape.Bartlett, F. C. (1932). Remembering. Cambridge,

England: Cambridge University Press.Brewer, W. F., & Lichtenstein, E. H. (1981). Event

schemas, story schemas and story grammars. In J.Long & A. Baddeley (Eds.), Attention and

performance, 9 (pp. 363-379). Hillsdale, NJ:Erlbaum.

Bruner, J. (1986). Actual minds, possible worlds.

Cambridge, MA: Harvard University Press.Carlson, J. G., & Hatfield, E. (1992). Psychology of

emotion. Fort Worth, TX: Harcourt BraceJovanovich.

Collingwood, R. G. (1938). The principles of art.

Oxford, England: Oxford University Press.Craik, K. (1943). The nature of explanation. Cam-

bridge, England: Cambridge University Press.Dunbar, R. I. M. (1993). Coevolution of neocortical

size, group size, and language in humans. Behav-ioral and Brain Sciences, 16, 681—735.

Dunbar, R. I. M. (1996). Grooming, gossip and the

evolution of language. London: Faber & Faber.Gardner, J. (1984). The art of fiction. New York:

Knopf.Gergen, K. J. (1994). Realities and relationships:

Soundings in social construction. Cambridge, MA:Harvard University Press.

Gerrig, R. J. (1998). Experiencing narrative worlds:

On the psychological activities of reading. NewHaven, CT: Yale University Press.

Goffman, E. (1961). Encounters: Two studies in the

sociology of interaction. Indianapolis, IN: Bobbs-Merrill.

Gottman, J. M , & Levenson, R. W. (1992). Marital

processes predictive of later dissolution: Behavior,physiology and health. Journal of Personality and

Social Psychology, 63, 221-233.Helmholtz, H. (1962). Treatise on physiological

optics, Vol. 3 (J. P. C. Southall, Ed.). New York:Dover. (Original work published 1866)

Hinton, G. E., Dayan, P., Frey, B. J., & Neal, R. M.

(1995). The wake-sleep algorithm for unsupervisedneural networks. Science, 268, 1158-1161.

Homer. (1987). The Iliad (M. Hammond, Ed. and

Trans.). Harmondsworth, Middlesex, England:Penguin Books.

Inoura, Y., & Kawatake, T. (1981). Traditional

theater of Japan. New York: Weatherall.James, H. (1951). The art of fiction. In M. D. Zabel

(Ed.), The portable Henry James (pp. 391-418).New York: Viking. (Original work published 1884)

Johnson-Laird, P. N. (1983). Mental models: Towards

a cognitive science of language, inference, and

consciousness. Cambridge, England: CambridgeUniversity Press.

Keats, J. (1966). Letter to John Taylor. In R. Gittings

(Ed.), Selected poems and letters of Keats (p. 46).London: Heineman. (Original work published

1818)Larsen, S. E, & Seilman, U. (1988). Personal

meanings while reading literature. Text, 8, 411—

429.Long, E. (1986). Women, reading, and cultural

authority: Some implications of the audienceperspective in cultural studies. American Quar-terly, 38, 591-612.

Longinus. (1965). On the sublime. In T. S. Dorsch

(Ed.), Aristotle, Horace, Longinus: Classical liter-ary criticism (pp. 99-158). Harmondsworth, Mid-dlesex, England: Penguin Books.

Miall, D. S. (1997). Gothic fiction. In D. Wu (Ed.),

The Blackwell companion to romanticism (pp.345-354). Oxford, England: Blackwell.

Miall, D. S., & Kuiken, D. (1994). Foregrounding,

defamiliarization, and affect: Response to literarystories. Poetics, 22, 389^07.

Murray, J. H. (1997). Hamlet on the holodeck: The

future of narrative in cyberspace. New York: FreePress.

Nussbaum, M. C. (1986). The fragility of goodness:

Luck and ethics in Greek tragedy and philosophy.

Cambridge, England: Cambridge University Press.Oatley, K. (1992). Best laid schemes: The psychology

of emotions. New York: Cambridge UniversityPress.

Oatley, K. (1994JL A taxonomy of the emotions ofliterary response and a theory of identification infictional narrative. Poetics, 23, 53-74.

Oatley, K. (1996). Inference and emotions in

FICTION AS COGNITIVE AND EMOTIONAL SIMULATION 117

narrative and science. In D. R. Olson & N.Torrance (Eds.), Modes of thought (pp. 123-140).New York: Cambridge University Press.

Oatley, K. (1998). Meetings of minds. In S. Janssen &N. van Dijk (Eds.), The empirical study ofliterature and the media: Current approaches andperspectives (pp. 58-72). Rotterdam: Barjesteh vanWaalwijk van Doom.

Oatley, K., & Duncan, E. (1992). Incidents ofemotion in daily life. In K. T. Strongman (Ed.),International review of studies on emotion (pp.250-293). Chichester, England: Wiley.

Pinney, T. (1963). Essays of George Eliot. New York:Columbia University Press. (Original work pub-lished 1856)

Prentice, D. A., Gerrig, R. J., & Bailis, D. S. (1997).What readers bring to the processing of fictionaltexts. Psychonomic Bulletin and Review, 4, 416-420.^

Rime, B. (1995). The social sharing of emotion as asource for the social knowledge of emotion. In J.Russell, J.-M. Fernandez-Dols, A. S. R. Manstead,& J. Wellenkamp (Eds.), Everyday conceptions ofemotions: An introduction to the psychology,anthropology, and linguistics of emotion (NATOASI Series D 81, pp. 475^89). Dordrecht, theNetherlands: Kluwer.

Rumelhart, D. E. (1975). Notes on a schema forstories. In D. Bobrow & A. Collins (Eds.),Representation and understanding: Studies incognitive science (pp. 211-236). New York:Academic Press.

Scheff, T. J. (1979). Catharsis in healing, ritual, anddrama. Berkeley: University of California Press.

Scheff, T. J. (1997). Emotions, the social bond, andhuman reality: Part/whole analysis. New York:Cambridge University Press.

Sharpies, M, Hogg, D., Hutchison, C , Torrance, S.,& Young, D. (1989). Computers and thought: Apractical introduction to artificial intelligence.Cambridge, MA: MIT Press.

Sidney, P. (1986). An apology for poetry. In C. Kaplan(Ed.), Criticism: The major statements (2nd ed.,pp. 108-147). New York: St Martin's Press.(Original work published 1595)

Sikora, S., Miall, D. S., & Kuiken, D. (1998, August).Enactment versus interpretation: A phenomenologi-cal study of readers'responses to Coleridge's "TheRime of the Ancient Mariner." Paper presented atthe Sixth Biennial Conference of the InternationalSociety for the Empirical Study of Literature,Utrecht, the Netherlands.

Sophocles. (1947). King Oedipus. In E. F. Watling(Ed.), The Theban plays (pp. 23-68). Harmonds-worth, Middlesex, England: Penguin Books.

Spence, D. P. (1982). Narrative truth and historicaltruth: Meaning and interpretation in psychoanaly-sis. New York: Norton.

Stevenson, R. L. (1992a). A chapter on dreams. In C.Harman (Ed.), R. L. Stevenson: Essays and poems(pp. 189-199). London: Dent Everyman's Library.(Original work published 1888)

Stevenson, R. L. (1992b). A humble remonstrance. InC. Harman (Ed.), R. L. Stevenson: Essays andpoems (pp. 179-188). London: Dent Everyman'sLibrary. (Original work published 1884)

Taine, H. (1882). De I'intelligence (Vol. 2). Paris:Hachette.

Tan, E. (1994). Film-induced affect as a witnessemotion. Poetics, 23, 7-32.

Tan, E. (1996). Emotion and the structure of film:Film as an emotion machine. Mahwah, NJ:Erlbaum.

Tolstoy, L. (1901). Anna Karenina (C. Garnett,Trans.). London: Heinemann. (Original workpublished 1877)

Vorderer, P. (1996). Towards a psychological theoryof suspense. In P. Vorderer, H. J. Wulff, & M.Friedrichsen (Eds.), Suspense: Conceptualizations,theoretical analyses, and empirical explorations(pp. 233-254). Mahwah, NJ: Erlbaum.

Wordsworth, W. (1984). Preface to Lyrical ballads of1802. In S. Gill (Ed.), William Wordsworth (pp.595-615). Oxford, England: Oxford UniversityPress. (Original work published 1802)

; Received May 4, 1998Revision received November 2, 1998

Accepted November 5, 1998