imagining and believing

Upload: arkadiusz-baksik

Post on 05-Apr-2018

213 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/31/2019 Imagining and Believing

    1/12

    The Journal of Aesthetics and Art Criticism 62:2 Spring 2004

    SHAUN NICHOLS

    Imagining and Believing: The Promise of a Single Code

    The imagination has always been one of the darkerfaculties of the human mind. Recent work on theimagination in cognitive science scarcely dispels

    all the mysteries of the imagination. But cognitiveapproaches might illuminate one central featureof imaginationthe striking similarities betweenimagining and believing. In this paper I will explorethe idea that the similarities between imaginationand belief are partly explained by the hypothesisthat imagination and belief are in a single code.A good deal of scaffolding is necessary to explainthis idea, but I will argue that the single-codehypothesis provides a surprisingly powerfulaccount of one aspect of the imagination.

    Unraveling the details of the relation betweenimagining and believing has ascended to a prom-inent place on the agenda of researchers workingon theories of imagination, fiction, and belief. Inthis paper, I will not be defending an underdogtheory. Far from itthe single-code theory, inone form or another, is embraced by most con-temporary cognitivist accounts of pretense andimagination.1 However, the explanatory powerof the single-code hypothesis has not, I willmaintain, been adequately recognized. The single-code hypothesis provides a unified explan-

    ation for a wide range of similarities betweenimagination and belief. In the first section of thispaper, I will describe and elaborate the version ofthe single-code theory that Stephen Stich and Ihave developed. Then I will argue that this the-ory offers simple solutions to a diverse set ofphilosophical puzzles about fiction.2

    I. FICTION AND IMAGINATION

    In earlier work, Stephen Stich and I developed acognitive theory of the imagination, based

    largely on empirical findings and empiricalconsiderations about pretend play.3 A briefreview of a portion of our theory of pretend play

    will supply the backdrop. Several features of thetheory can be illustrated with a lovely experi-ment on children by Alan Leslie. In Lesliesexperiment, the child pretends to fill two cupswith tea. The experimenter picks up one of thecups, upends it, and places it next to the othercup. Then the experimenter asks the child to pointat the full cup and at the empty cup. Both cupsare really empty throughout the entire procedure,but two-year-olds reliably indicate that theempty cup is the one that had been turnedupside down and the full cup is the other one.4

    One significant feature of episodes likeLeslies tea party is that children distinguishwhat is pretend from what is real. That is, at nopoint in this experiment do children believe thateither of the cups is full. Their pretense that thecup is full is quarantined from their belief thatthe cup is not full. Another, more interesting,feature of the experiment is that it indicates thata belief and a pretense can have exactly thesame content. When the children are asked topoint to the empty cup and the full cup,

    they maintain that the previously overturnedcup is the empty one. On the most natural inter-pretation of this, the child ispretending that thecup is empty. We adopt the representationalistapproach that is common in this area and saythat such pretending involves a pretense repre-sentation with the content the cup is empty.Although the child is pretending that the cup isempty, she is not blind to the fact that the cup isreally empty throughout; rather, the child alsobelieves that the cup is empty. This suggeststhat the crucial difference between pretenserepresentations and beliefs is not given by the

  • 7/31/2019 Imagining and Believing

    2/12

    130 The Journal of Aesthetics and Art Criticism

    content of the representation. For a pretenserepresentation and a belief can have exactly thesame content. So, pretense representations arequarantined from beliefs, and yet the distinction

    is not driven by differences in content. The naturalcognitivist proposal, then, is that pretense repre-sentations differ from belief representations bytheir function. Just as desires are distinguishedfrom beliefs by their characteristic functionalroles, so too pretenses are distinguished frombeliefs.5 Stich and I exploit the familiar illustra-tive device of using boxes to represent func-tional groupings, and we propose that besides abelief box and a desire box, there is a pretensebox.6 This pretense box, we suggest, is part ofthe basic architecture of the human mind.

    In addition to a pretense box, Stich and Ipropose a mechanism that supplies the pretensebox with representations that initiate or embel-lish an episode of pretense, the Script Elabor-ator. This is required to explain the bizarre andcreative elements that are evident in muchpretend play. However, there are also muchmore staid and predictable elaborations in pre-tend play. This too is well illustrated by Lesliesexperiment. Virtually all of the children in hisexperiment responded the same way when

    asked to point to the empty cup. How arethese orderly patterns to be explained? In every-day life when we acquire new beliefs, we

    routinely draw inferences and update our beliefs.No one knows how this process works, but noone disputes that it does work. There must besome set of mechanisms subserving inference

    and updating, and we can simply use anotherfunctional grouping to collect these mechanismsunder the heading Inference Mechanisms.Now, to explain the orderly responses of thechildren in Leslies experiment, we propose thatthe representations in the pretense box are proc-essed by the same inference mechanisms thatoperate over real beliefs. Of course, to drawthese inferences the child must be able to usereal-world knowledge about the effects of grav-ity and so forth, and so Stich and I also supposethat the inferences the child makes during pre-

    tense can somehow draw on the childs beliefs.7

    Putting this all together generates the functionalarchitecture depicted in Figure 1.

    With the functional architecture in place, wecan finally articulate the core idea that I want toexplore in this paper. Stich and I suggest thatthe representations in the pretense box and therepresentations in the belief box are in thesame code, to adopt a computational locution.Of course, it is far from clear what the code isfor representations in the belief box, so it is not

    possible to be specific about the details or thenature of the putatively shared code.8 But theimportant point for present purposes can be

    FIGURE 1. The pretense box account of the imagination.

    Beliefs Desires

    Decision-making

    (practical reasoning)

    system

    Action control

    systems

    BEHAVIOR

    Body monitoring

    system

    Perceptual

    processes

    Pretenses

    Inference

    mechanisms

    Script

    elaborator

  • 7/31/2019 Imagining and Believing

    3/12

    Nichols Imagining and Believing: The Promise of a Single Code 131

    made without giving further detail about whatthe code is. If pretense representations andbeliefs are in the same code, then mechanismsthat take input from the pretense box and from

    the belief box will treat parallel representationsmuch the same way.9 For instance, if a mechan-ism takes pretense representations as input, thesingle-code hypothesis maintains that if thatmechanism is activatedby the occurrent beliefthatp, it will also be activated by the occurrentpretense representation that p. More generally,for any mechanism that takes input from boththe pretense box and the belief box, the pretenserepresentation p will be processed much thesame way as the belief representation p. I willcount any theory that makes this claim as a

    single-code theory.Taking a cognitive approach to imagination,

    it is worth emphasizing, provides the crucialresources for articulating the single-code pro-posal. For the single-code hypothesis is a pro-posal about the relation between certain kindsof representations and certain kinds of cognitivemechanisms. As a result, the hypothesis canonly be framed against a background of cogni-tive architecture. Once one has posited a back-ground of cognitive components, one can then

    explain some processing differences betweenimagination and belief by noting that some ofthese cognitive components take input frombeliefs but not from imagination. The single-code hypothesis maintains that one can alsoexplain many of the similarities between imagin-ation and belief with the proposal that some ofthe cognitive mechanisms do process inputfrom both imagination and belief, and further,that pretense representations are in the samecode as belief representations.

    The core claim of the single-code hypothesisis that a mechanism that takes pretense repre-sentations as input will process those represen-tations much as it would process isomorphicbelief representations. As a result, offlinesimulation theorists are plausibly viewed assingle-code theorists. For they maintain thatseveral mental mechanisms process pretendbeliefs just like real beliefs.10 Offline simula-tion theorists often have additional commit-ments of course. For instance, many prominentversions of offline simulation theory explicitly

    invoke pretend desires in addition to pretendbeliefs, and also maintain that the practical

    reasoning system takes as input pretend beliefsand pretend desires. Those additional stipula-tions are consistent with, but not required by,the single-code hypothesis.11 Although offline

    simulation theory counts as a single-codetheory, the development of the idea in this paperwill lean on the account of pretense that Stichand I developed (the pretense box account)because it will be much easier to investigatethese issues with a detailed account in hand.

    The pretense box account was developed asan account of pretend play, but we also main-tained that the account provided a theory of theimagination.12 If this account of the imaginationis right, it should also accommodate imagina-tive engagement with fiction.

    The extension to fiction is obvious enough.In consuming fiction, one (or ones cognitivemechanisms) constructs a set of representationsin ones pretense box. This construction isguided by a prompt of some sort; the promptmight be a movie, a novel, or a raconteur. Therepresentations being put into the pretense boxare distinguished from beliefs by theirfunctional roles, but they are not systematicallydistinguished from beliefs in terms of content.As with imagination generally, the pretense

    representations generated by fiction are in thesame mental code as beliefs. In consumingfiction, then, the inferential systems process thepretense representations much as the inferentialsystems would process isomorphic beliefs.13

    On the single-code hypothesis, if a mechanismtakes pretense representations as input, thatmechanism will process the pretense represen-tation much the same way it would process anisomorphic belief. This should apply to emo-tional systems as well as inferential systems.According to the single-code hypothesis, then,the emotional systems will respond to pretenserepresentations much as they do to parallelbeliefs.14 That is, if the pretense representationthat p gets processed by an affective mechan-ism, the affective outputs should parallel thoseof the belief that p. This claim gains supportfrom a wide range of empirical findings.

    Imagining that something dangerous ishappening can provoke affective response thatat least closely resembles real fear. In one repre-sentative experiment, subjects were presented

    with an imaginary scenario involving a snakeencounter, and virtually all of the subjects

  • 7/31/2019 Imagining and Believing

    4/12

    132 The Journal of Aesthetics and Art Criticism

    showed physiological signs associated with fear.More interestingly, the responses were strongerin people who were snake phobic than in peoplewho were phobic about public speaking.15 So

    the particular character of the subjects affectiveresponse system seems to be reflected in theiraffective responses to imagination. All of thissuggests that we need to elaborate the functionalarchitecture to capture the fact that affectiveresponse systems can take input from both beliefsand pretense representations, hence Figure 2.16

    It remains an open question on the single-code hypothesis whether all affective systemstake pretense representations as input. But wemight as well begin with the bold suppositionthat all affective and emotional systems that

    take beliefs as input also take pretense represen-tations as input. It might turn out, however, thatcertain affective or emotional systems will takebeliefs as inputs but not pretense representa-tions. As philosophers like to say, that is anempirical matter.17

    II. PUZZLES IN FICTION

    The pretense box account is driven by broadly

    empirical concerns, rather than by any consid-erations about fiction. But the account, if right,should accommodate imaginative engagement

    with fiction. So, how does the account fare withfiction?

    Much work in philosophy is driven byapparent puzzles, and this approach has been

    especially central in the philosophy of fiction.Indeed, Kendall Waltons important account ofhow we appreciate representational art is sup-posed to be measured largely by its success atsolving puzzles surrounding fiction.18 In thissection, I will argue that the cognitive accountof the imagination sketched above, and inparticular, the single-code hypothesis, providessolutions to a wide range of the puzzles sur-rounding fiction.

    Before we start, however, a caveat is in order.Each of the puzzles that I describe has attracted

    significant theoretical attention, and for most ofthe puzzles there are a number of competingtheories about how to solve the puzzle.19 For themost part, however, I will not engage alternativeexplanations of the puzzles, which will make thediscussion rather breezy. This is because I amnot primarily trying to score points against othertreatments of the puzzles; rather, the goal is toindicate the explanatory sweep of the single-code hypothesis. The single-code hypothesis isindependently motivated by broadly empirical

    considerations, and I want to show that, across awide range of puzzles, the single-code hypoth-esis provides natural solutions.

    FIGURE 2. The pretense box account with affective systems.

    Beliefs Desires

    Decision-making

    (practical reasoning)

    system

    Action control

    systems

    BEHAVIOR

    Body monitoring

    system

    Perceptual

    processes

    Pretenses

    Inference

    mechanisms

    Script

    elaborator

    Affective

    mechanisms

    Affective

    response

  • 7/31/2019 Imagining and Believing

    5/12

    Nichols Imagining and Believing: The Promise of a Single Code 133

    i. Emotions and Fiction

    I begin with perhaps the most familiar puzzle,our apparent emotional responses to fiction.

    One way to put the puzzle is as an inconsistenttriad:

    I pity Lear.

    To pity something, you have to believe it exists.

    I do not believe Lear exists.20

    An important part of the puzzle here is that weknow that Lear does not exist and yet we havean emotional response that seems to fit theemotional response that we would have if wedid believe that Lear existed. The puzzlingphenomenon here holds across a wide range ofemotions. We respond to fiction with affect thatclosely parallels our affective responses to thereal world. Novels and movies make us feel sad,happy, triumphant, indignant, embarrassed,afraid.

    The single-code hypothesis applies itself inan obvious way to the puzzle outlined, andother philosophers have ably promoted thevirtues of single-code accounts here, so I will bebrief.21 Since the pretense representations gen-

    erated by engaging with the fiction are in thesame code as beliefs, these pretense representa-tions will be processed by affective mechanismsmuch the way isomorphic beliefs would beprocessed. Single-code theories thus promise asolution to at least the psychological puzzle ofhow fiction can cause affective response.Affective systems can receive input from theimagination, and affective systems processinput from the imagination as they would proc-ess isomorphic beliefs. So on the single-codetheory, an affective system can accept as inputthe pretense representation that Lear is watch-ing Cordelia die, and the affective system willgenerate the same affective consequence fromthis input that it would if the input were insteada belief that Lear is watching Cordelia die.Since the pity response would be activated bythe belief, it is activated by the isomorphicpretense representation. The single-codehypothesis also explains why fiction can pro-duce such a vast range of emotional responses.By hypothesis, affective systems that are

    triggered by beliefs can also be triggered by theimagination. Hence, our emotional reactions to

    imagining should be as diverse as our emotionalreactions to believing.

    ii. Emotion and Iteration

    An additional puzzle about emotion arises in thecontext of iterated pretense. Walton describesthe puzzle as follows: Consider a story thatends with the hero waking from a bad dream.The reader, on realizing that it is fictional (inthe story) only that it is fictional (in a dream)that monsters were chasing him, not that theyreally were, heaves a sigh of (fictional)relief.22 One might quibble with this descrip-tion of the puzzle, but the phenomenon is clear

    and interesting. When we read a novel in whichsomething is merely pretended, the fact that it ispretend in the novel mitigates our emotionalreaction. We are relieved when it is onlypretend that the hero is harmedas when welearn that the events were merely fantasies ofthe villain. But this seems oddthe whole thingis fiction.

    The single-code hypothesis promises toexplain this puzzle as well. When we come tobelieve that some danger to another person was

    merely imaginary, we heave a sigh of relief.Our affective response to the danger cues is mit-igated. The single-code hypothesis suggests thatsomething parallel happens when we imaginethat the danger to the fictional hero is onlyimaginary.23 The parallel can be depicted moreconcretely as follows. The belief that Joe isdying creates an emotional response that is thenmitigated by the subsequent belief that it is onlyfictional that Joe is dying; similarly, thepretense representation that Joe is dying createsan emotional response that is mitigated by thesubsequent pretense representation that it isonly fictional that Joe is dying.24

    iii. Fictional Names

    The next puzzle that I want to consider has beenset most fully by Gregory Currie. Most peopledo not believe that fictional characters reallyexist. Yet our intuitions about names offictional characters parallel in a striking way

    our intuitions about names of real individuals.This is particularly evident in our reference

  • 7/31/2019 Imagining and Believing

    6/12

    134 The Journal of Aesthetics and Art Criticism

    intuitions for fictional names. We know thatMacbeth does not refer to a concrete individ-ual, and the play provides the only descriptionof the character. As a result, perhaps the most

    obvious story about the semantics of fictionalnames is a straightforward description theoryaccording to which the meaning of a fictionalname is given by its associated description.However, a description theory of fictionalnames turns out to be quite counterintuitive.Indeed, there are closely parallel antidescriptiv-ist intuitions regarding fictional proper namesand actual proper names. These puzzling paral-lels demand some explanation, and the single-code hypothesis provides one. To begin, I wantto catalogue explicitly some of the parallel

    antidescriptivist intuitions that are generated forfictional names and proper names.

    Uniqueness. We have the intuition that fictionalnames refer to unique individuals, even whenthe description associated with the fictional nameclearly does not pick out a unique individual.Currie concocts the following (very brief) story:

    Jack got up in the morning and ate breakfast.

    As Currie notes, we cannot simply gloss this assaying, There is someone called Jack whogot up in the morning and ate breakfast, sincethe story purports to describe the activities of aparticular Jack.25 Curries example here bearsa close analogy to Kripkes Feynman exam-ple against the description theory of propernames.26 Someone might believe that Feynmanwas a physicist and not have any other distin-guishing beliefs about Feynman. Of course, thisdescription, he was a physicist, hardly picksout anyone uniquely. Yet we are still inclined tosay that the person is referring to a particularindividual. Curries story of Jack similarlyseems to pick out a single fictional character,despite the impoverished description.

    Meeting the description is not sufficient. Even ifthere is a unique individual who actually meetsthe description associated with a fictional name,we have the intuition that the fictional namedoes not refer to that individual. Kripke callsattention to this fact when he claims that even if

    there is an actual person who happens to meetthe description associated with Sherlock Holmes,

    the stories are not about this man.27 Althoughit goes unremarked in Naming and Necessity,the Holmes example parallels Kripkes Gdel/Schmidt case.28 Kripke notes that the one thing

    that is commonly associated with Gdel is thathe discovered the incompleteness of arithmetic.However, Kripke maintains, if it turned out thata man named Schmidt actually discovered theincompleteness of arithmetic, we would notassume that Gdel really referred to Schmidt.We have the intuition then, that meeting thedescription associated with Gdel does notsuffice to be the referent ofGdel. Again, thisparallels the intuition that meeting the descriptionassociated with Sherlock Holmes does notsuffice to be the referent ofSherlock Holmes.

    Meeting the description is not necessary. Whenreading the Holmes stories, it seems that wewant to say that Holmes might not have donesome of the things that he did dowe donot regard the description associated withHolmes as necessary. Currie writes: It is truein the story that if Holmes had not done thethings described in the story he would still havebeen Holmes.29 When reading the stories, itseems possible that lots of the descriptions that

    are attached to Holmes are accidental. Forinstance, it seems possible that in the story,Watson would discover that Moriarty haddisguised himself as Holmes in a number of theadventures, while Holmes was tied up inMoriartys den. Kripke makes a similar pointabout proper names: Most of the things com-monly attributed to Aristotle are things thatAristotle might not have done at all. In a situationin which he didnt do them, we would describethat as a situation in which Aristotle didnt dothem.30 So, just as we have the intuition thatthe bearer of a name need not have the proper-ties associated with that name, we also have theintuition when engaged with fiction that thebearer of the fictional name need not have theproperties associated with that (fictional) name.

    The puzzle, then, is that we know that thefictional names do not refer to anything, yet ourintuitions run against a descriptivist treatmentof fictional names. Rather, the antidescriptivistintuitions about proper names find closeparallels in fictional names. Kripkean thought

    experiments seem to work equally well whetherone uses a fictional name or a proper name to

  • 7/31/2019 Imagining and Believing

    7/12

    Nichols Imagining and Believing: The Promise of a Single Code 135

    drive the intuition. The single-code hypothesiscan provide an explanation for the parallelintuitions.

    Our goal here is not, I should stress, to deter-

    mine the right semantics for fictional names.Rather, the goal is only to explain why we havethese parallel intuitions about fictional namesand proper names. To appreciate how thesingle-code account explains the parallel, it ishelpful to consider a related proposal. One ideafor explaining our intuitions about fictionalnames is to maintain that these intuitions derivefrom metalinguistic pretenses; we pretend, forinstance, that Sherlock Holmes refers or thatSherlock Holmes is a proper name.31 On thiskind of proposal, the parallel intuitions follow

    from the fact that we pretend that the fictionalnames have the properties of actual propernames. On the single-code hypothesis, by con-trast, we can explain the intuitions withoutappealing to any explicit pretense that the fic-tional names refer. Since the representationalcode for pretense representations is the same asthat for belief representations, the code forfictional proper names will be the same as thecode for actual proper names. Reference intui-tions about actual proper names are presumably

    generated by the inference mechanisms, and onthe single-code hypothesis, the inference mech-anisms process pretense representations muchas they process belief representations. As aresult, the very same mechanisms that produceintuitions about proper names will also producethe intuitions about fictional names. And, sincethe code is the same, it is no surprise that theintuitions are the same. Our inferences about therelations between Holmes and the associateddescriptions in the pretense box will parallel ourinferences about the relations between Gdeland the associated descriptions in the beliefbox. Again, the single-code hypothesis does nottell us how to characterize the semantics of fic-tional names, but it does provide an explanationfor why we have the puzzling intuitions we do.32

    iv. Tacit Pretense

    The possibility of dispositional imaginingraises yet another puzzle. When reading The

    Great Gatsby, we know that Gatsby could notsurvive decapitation and that he has fewer than

    one billion hairs on his head. But the novelnever says such things explicitly and it isunlikely that we explicitly imagine such things.One reason to think that we have such

    implicit imaginings comes from the fact thatwe can be surprised when apparent implicitimaginings are violated. When reading SirGawain and the Green Knightfor the first time,it is plausible that we implicitly imagine that theGreen Knight could not survive decapitation,for it comes as rather a shock when he picks uphis head after Gawain has chopped it off.

    Although it seems that we implicitly imaginethat Gatsby could not survive decapitation it ishard to see how to account for this. An accountin terms of dispositions seems quite inadequate,

    as Walton notes: A person who has recurrentdreams or daydreams of fame and fortune isdisposed, between the occurrences, to imaginehimself rich and famous. But he need not actu-ally be imagining this nonoccurrently then; hemight even imagine the opposite during theintervals.33 The fact that John is disposed toimagine himself being famous does not meanthat John dispositionally imagines beingfamous.34 Currie goes further and maintains thatall imagining is occurrent:

    With belief and desire we have the distinction

    between dispositional and occurrent states. I may

    believe something I have never thought about and

    which plays no role in explaining my past or current

    behavior. Similarly for desire. In these cases we say

    that the belief or desire is dispositional rather than

    occurrent; the state is possessed in virtue of the fact

    that if a relevant situation were to arise you would

    be likely to behave in a way that would betray the

    belief or the desire. But there is no comparable cate-

    gory of dispositional imagining, and all imagining is

    occurrent. . . . And while I may be disposed to

    imagine certain things, the possession of such a

    disposition does not constitute my dispositionally

    imagining them.35

    Currie proceeds to point out that being disposedto run a mile does not mean you are disposition-ally running a mile.36 Similarly, then, he sug-gests, being disposed to imagine thatp does notmean that you are dispositionally imagining thatp. Yet, again, it seems plausible to say that in

    some sense we implicitly imagine that theGreen Knight could not survive decapitation.

  • 7/31/2019 Imagining and Believing

    8/12

    136 The Journal of Aesthetics and Art Criticism

    This apparent tension about the possibility ofdispositional imagining is resolved on thesingle-code account. The first thing to note is thatCurries complaint about dispositional imagin-

    ing finds a parallel in dispositional believ-ing. Merely having a tendency to believe thatp does not entail that one dispositionallybelieves thatp in the sense we are considering.For instance, if John has a tendency to believethat he is fat, it does not follow that he dispos-itionally believes that he is fat. As a result, theterm dispositional belief is rather misleading.But the underlying idea seems relativelystraightforward. One familiar way to capture thenotion is to claim that dispositional beliefsare, roughly, beliefs that would be produced

    immediately by the inference mechanismsunder the appropriate circumstances. Lycanuses the term tacit belief to capture this sort ofnotion.37 Lycans terminology is useful becauseit allows us to distinguish being disposed tobelieve from tacit belief. Johns disposition tothink that he is fat is not a tacit belief; but hisbelief that Descartes had fewer than a billionhairs is. Similarly, then, being disposed toimagine needs to be distinguished from tacitpretense. If tacit beliefs are simply beliefs that

    would be (under appropriate circumstances)immediately inferred from the current set ofbeliefs, then tacit pretenses are pretense repre-sentations that would be (under appropriatecircumstances) immediately inferred from thecurrent set of pretense representations. Since thepretense representations are in the same code asbeliefs and are processed by the same inferencemechanisms as beliefs, the account of tacitbelief is easily appropriated for an account oftacit pretense.38 Given the other things weimagine when reading The Great Gatsby, wetacitly imagine that Gatsby has fewer than abillion hairs.

    v. Imaginative Resistance

    Richard Moran raises a further, intriguing prob-lem about our emotional and moral responses tofiction, the phenomenon ofimaginative resist-ance. While we seem to be willing to imaginethe most outrageous thingsthat pigs can read,

    that the world is two-dimensional, that Menardtried to rewrite Quixotewe also seem to resist

    imagining certain things. Moran offers a niceexample:

    If the story tells us that Duncan was notin fact mur-

    dered on Macbeths orders, then that is what weaccept and imagine as fictionally true. . . . However,

    suppose the facts of the murder remain as they are in

    fact presented in the play, but it is prescribed in this

    alternate fiction that this was unfortunate only for

    having interfered with Macbeths sleep that night.39

    Although in the former case, we would bewilling to imagine that Macbeth did not orderDuncans murder, in the latter case, Moransuggests, we would not imagine that in thefictional world murdering ones guest is mor-

    ally okay. But, if the story tells us that such amurder is morally acceptable, why do we resistimagining this? Part of the puzzle posed by suchcases, the part that will be our focus, concernsthe psychology of imaginationwhy do weresist imagining that murder is okay?40 Moranindicates that this imaginative resistance isrestricted to moral and emotional features, butI am not so sure. For there are plausibly casesof imaginative resistance in something as amoraland affectless as mathematics. If we are asked to

    imagine that all of mathematics is just as we thinkit is, except that 22, we exhibit imaginativeresistance here as well.41 On the single-codehypothesis, this resistance might be explainedby the fact that the pretense representation that22 would engage our normal inferentialsystems. And just as our inferential systemswould lead us to resist believing that 22, sotoo do they lead us to resist imagining that 22.

    If we suppose that pretense representationsget processed by our moral response mechan-isms, then the single-code hypothesis also pro-vides a natural explanation for moral imaginativeresistance.42 For the moral response mechan-isms will treat pretense representations as theytreat isomorphic belief representations. InMorans case, our moral response mechanismsare engaged by the pretense representationsinvolving the murder of Duncan. Believing thefacts as presented would prompt a reaction ofmoral condemnation. Hence, since pretense rep-resentations are in the same code, imagining thefacts activates a response of moral condemn-

    ation as well.43

    These moral reactions to pre-tense representations are not preempted or

  • 7/31/2019 Imagining and Believing

    9/12

    Nichols Imagining and Believing: The Promise of a Single Code 137

    eradicated if the play tells us that Macbethsactions are morally acceptable. This is just whatthe single-code hypothesis would lead us toexpect. For the same holds of our moral reac-

    tions to what we believe. If we read a newspaperaccount of a murder, our moral disapprobationwill scarcely be eliminated if the journalist saysthat the action was morally acceptable.44

    III. CONCLUSION

    I am sure the basic explanatory strategyafforded by the single-code hypothesis is bynow painfully evident, so I do not propose tobelabor the matter any longer. But with a little

    work, I expect that the single-code account canalso explain several other puzzles about fiction,including puzzles about the attribution of truthin fiction, the assertion of fictional claims, andwhy fiction seems to admit of degrees oftruth.45 So its potential explanatory sweep isconsiderable. The single-code hypothesis seeksto explain the similarities between believingand imagining by proposing a strict isomorph-ism between belief representations and pre-tense representations. But obviously there are

    salient and profound discontinuities betweenimagining and believing, and the single-codetheorist must somehow accommodate those dis-continuities.46 That project cannot be pursuedhere, but I hope to have shown that the single-code hypothesis does have considerable pro-mise. Indeed, the single-code hypothesis seemsto provide us with a strikingly powerful accountof one corner of the imagination.47

    SHAUN NICHOLS

    Department of PhilosophyCollege of Charleston

    Charleston, SC 29412

    INTERNET: [email protected]

    1. See, for example, Gregory Currie, Imagination andSimulation: Aesthetics Meets Cognitive Science, inMentalSimulation: Evaluations and Applications, ed. MartinDavies and Tony Stone (Blackwell, 1995), pp. 151169;Gregory Currie, The Paradox of Caring, in Emotion andthe Arts, ed. Mette Hjort and Sue Laver (Oxford University

    Press, 1997), pp. 6377; Robert Gordon and John BarkerAutism and the Theory of Mind Debate, in Philoso-

    phical Psychopathology: A Book of Readings, ed. GeorgeGraham and G. Lynn Stephens (MIT Press, 1994), pp. 163181; Paul Harris, The Work of the Imagination (Blackwell,2000); Alan Leslie, Pretense and Representation: The Ori-gins ofTheory of Mind,Psychological Review 94 (1987):

    412426; Adam Morton, Frames of Mind(Oxford UniversityPress, 1980); Shaun Nichols and Stephen Stich, A Cogni-tive Theory of Pretense,Cognition 74 (2000): 115147.

    2. The single-code hypothesis also has serious problemssince there are clear cases in which we respond differentlyto imagining and believing. In a companion piece to thisarticle, I set out some of the asymmetries that are particu-larly vexing for the single-code hypothesis, and I attempt toprovide a way of accommodating the cases without relent-ing on the claim that imagining and believing are in thesame code (Nichols, Just the Imagination [forthcoming]).

    3. Nichols and Stich, A Cognitive Theory of Pretense.See also Shaun Nichols and Stephen Stich, Mindreading(Oxford University Press, 2003).

    4. Alan Leslie, Pretending and Believing: Issues in theTheory of ToMM,Cognition 50 (1994): 211238.5. See also Currie, Imagination and Simulation.6. In A Cognitive Theory of Pretense, Stich and I use

    the term Possible World Box to pick out the pretense box.But the notion of Possible World that we invoke differsfrom the notion familiar to philosophers. Since this paper isaimed primarily at philosophers, I have adopted a term withless philosophical baggage.

    7. One possibility is that a large class of the childs beliefsthat are consistent with the pretense scenario are simplyadded to the set of pretense representations. See Nichols andStich, A Cognitive Theory of Pretense, pp. 124125.

    8. Indeed, as Barbara Von Eckardt has reminded me, it is

    possible that there are actually multiple kinds of codes inthe belief box. That is, some beliefs might be stored in onecode, while other beliefs are stored in a different code.Obviously, this possibility complicates the picture that thereis a single code for pretense and belief. But the importantpoint is simply that each pretense representation is in thesame code as some (possible) parallel belief representation.This qualification does not affect the main points of thispaper, so for expository ease, I will continue to cast the pro-posal as the single-code hypothesis.

    9. This does not, of course, mean that the mentalprocessing of pretense representations will be exactly paral-lel to the processing of isomorphic belief representations.After all, the pretense representations do not feed into all the

    same mechanisms as beliefs. Another obvious difference isthat when the inference mechanisms (for instance) receiveinput from the pretense box, the outputs are typically sentback to the pretense box rather than to the belief box. Thesequalifications do not, I think, compromise the single-codehypothesis, insofar as the qualifications seem independentlyplausible and well motivated.

    10. Currie, Imagination and Simulation; Robert Gordon,Folk Psychology as Simulation, Mind & Language 1(1986): 158170; Alvin Goldman, Interpretation Psycholo-gized,Mind & Language 4 (1989): 161185; Paul Harris,From Simulation to Folk Psychology: The Case for Devel-opment,Mind & Language 7 (1992): 120144.

    11. Representative defenses of pretend desires can be

    found in Gordon, Folk Psychology as Simulation, andGregory Currie and Ian Ravenscroft, Recreative Minds:

  • 7/31/2019 Imagining and Believing

    10/12

    138 The Journal of Aesthetics and Art Criticism

    Imagination in Philosophy and Psychology (Oxford Univer-sity Press, 2002). Skepticism about pretend desires isdefended in Nichols and Stich, A Cognitive Theory of Pre-tense, and Nichols, Review of G. Currie and I. Raven-scrofts Recreative Minds: Imagination in Philosophy and

    Psychology,Mind(forthcoming).12. Pretend play typically involves overt play behaviorthat raises additional complications (Nichols and Stich, ACognitive Theory of Pretense, section 3.3). But since thefocus here will be on imagination without overt behavior,we need not dwell on these complications.

    13. There is one significant elaboration of the pretensebox theory concerning the storage of pretense representa-tions. Most people have a vast knowledge of particularstories. It is plausible that the pretense box contains not onlycurrently developing pretense episodes, but also stores pastpretenses. That is, like beliefs, pretense representations canplausibly be stored. For instance, most of us have clusters ofHamlet-representations from our viewings ofHamlet. Of

    course, we do not have these stored as beliefs about aformer Prince of Denmark. Rather, the representations thatwere generated by past viewings ofHamletare likely storedin the pretense box. But clearly some of these stored repre-sentations in the pretense box will contradict others. Forexample, in Steve Ericksons Tours of the Black Clock,Hitler lives well past 1945 and grows to be an old man; butin Don DeLillos White Noise, Hitler dies in 1945. As aresult, the clusters of stored representations must be keptdistinct to avoid inferential catastrophe. Stored pretense rep-resentations might be distinguished by indexing, and, with anod to David Lewis, we can maintain that stored pretenserepresentations can by indexed by worlds. So, for instance,there might be a pretense representation specifying that in

    the Tours of the Black Clockworld, Hitler lives to be an oldman. There is another pretense representation that says thatHitler did not live to be an old man, but that representationis indexed to the White Noise world.

    14. A number of writers have emphasized this point. See,for example, Paul Harris, From Simulation to Folk Psych-ology: The Case for Development,Mind & Language 7(1992): 120144. Alvin Goldman, In Defense of the Simu-lation Theory,Mind and Language 7, (1992): 104119;Harris, The Work of the Imagination; Nichols and Stich,

    Mindreading.15. For representative experiments on imagination and fear,

    see Scott Vrana, Bruce Cuthburt, and Peter Lang, Process-ing Fearful and Neutral Sentences: Memory and Heart Rate

    Change,Cognition and Emotion 3 (1989): 179195; PeterLang, Daniel Levin, Gregory Miller, and Michael Kozak,Fear Behavior, Fear Imagery, and the Psychophysiology ofEmotion: The Problem of Affective Response Integration,

    Journal of Abnormal Psychology 92 (1983): 276306. Forreviews of the literature, see Peter Lang, Cognition andEmotion: Concept and Action, inEmotions, Cognition and

    Behavior, ed. Carroll Izard, Jerome Kagan, and RobertZajoncs (Cambridge University Press, 1984), pp. 192226;Harris, The Work of the Imagination.

    16. Of course this figure, like the previous figure, leavesout many important functional relations. For instance, thebody monitoring system likely plays an important role inemotion generation. The figure is intended to emphasize

    functional relations that will be particularly important inthis paper.

    17. Another open question concerns the possible role ofdesires in the emotion-processing system. The central scien-tific tradition on emotions, the affect program tradition,does not typically appeal to desires in the explanation ofhow one goes from a situation to an emotion (see, for exam-

    ple, Ron Mallon and Stephen Stich, The Odd Couple: TheCompatibility of Social Construction and EvolutionaryPsychology, Philosophy of Science 67 [2000]: 133154).Rather, the standard explanation of how affect is generatedappeals to appraisal processes that seem to take informationprimarily from the subjects beliefs about the situation. So,for instance, it is appraisal of the details of the situation thatproduces the affect. Of course, it is possible that desiresreally play a crucial role here as well, but that the role ofdesires has not been adequately articulated in contemporaryaffect theory. If it turns out that desires are important, thatwill raise additional issues, of course. But it need not pose aproblem for the single-code hypothesis.

    18. Kendall Walton, Mimesis as Make-Believe: On the

    Foundations of the Representational Arts (Harvard Univer-sity Press, 1990), p. 214.19. The literature is vast, but for some of the important

    discussions, see Nol Carroll, Beyond Aesthetics: Philo-sophical Essays (Cambridge University Press, 2001);Gregory Currie, The Nature of Fiction (Cambridge UniversityPress, 1990); Gregory Currie,Image and Mind(CambridgeUniversity Press, 1995); Susan Feagin, Reading with Feel-ing: The Aesthetics of Appreciation (Cornell UniversityPress, 1996); Tamar Gendler, The Puzzle of ImaginativeResistance,Journal of Philosophy 97 (2000): 5581; Hjortand Laver,Emotion and the Arts; Peter Lamarque and SteinOlsen, Truth, Fiction, and Literature (Oxford UniversityPress, 1994); Jerrold Levinson, The Pleasures of Aesthetics

    (Cornell University Press, 1996); David Lewis, Truth inFiction, in Philosophical Papers, vol. 1, ed. David Lewis(New York: Oxford University Press, 1983); RichardMoran, The Expression of Feeling in the Imagination,Philosophical Review 103 (1994): 75106; Walton,Mimesisas Make-Believe; Kendall Walton, Spelunking, Simulationand Slime: On Being Moved by Fiction, in Emotion andthe Arts.

    20. This formulation of the puzzle follows Currie, TheParadox of Fiction, p. 65.

    21. In The Paradox of Fiction, Currie offers an offlinesimulation explanation of the puzzle of fiction and emo-tions, and since I take offline simulation theory to be a kindof single-code theory, Curries treatment counts as a single-

    code account. Aaron Meskin and Jonathan Weinbergprovide an alternative to Curries simulation approach to theproblem that is very much in the spirit of the accountpresented here. See Aaron Meskin and Jonathan Weinberg,Emotions, Fiction, and Cognitive Architecture,The Brit-ish Journal of Aesthetics 1 (2003): 1834.

    22. Walton,Mimesis as Make-Believe, p. 284.23. It is not entirely clear what the mechanisms are sub-

    serving this effect in the case of belief. But the single-codetheorist need only suggest that the mechanisms that producethis effect for belief are also implicated by pretense repre-sentations.

    24. For further discussion of iteration puzzles, see ShaunNichols, Imagination and the Puzzles of Iteration,Ana-

    lysis 63 (2003): 182187.25. Currie, The Nature of Fiction, p. 151.

  • 7/31/2019 Imagining and Believing

    11/12

    Nichols Imagining and Believing: The Promise of a Single Code 139

    26. Saul Kripke, Naming and Necessity (Harvard Uni-versity Press, 1980), p. 81.

    27. Kripke, Naming and Necessity, p. 157. Kripkespoint here influences David Lewiss account of truth infictionsee Lewis, Truth in Fiction, pp. 265, 276.

    28. Kripke,Naming and Necessity, pp. 8384.29. Currie, The Nature of Fiction, p. 165.30. Kripke,Naming and Necessity, p. 61.31. These sorts of views are suggested in Walton,Mime-

    sis as Make-Believe, and in unpublished work by Kripke.32. Whether this explanation of the intuitions impacts

    the semantic project is a separate question that I will notconsider here. For one account of the semantics of fictionalnames that dovetails with the account of the imaginationpresented here, see Stacie Friend, The Great BeetleDebate: A Study in Imagining with Names (forthcoming).

    33. Walton,Mimesis as Make-Believe, p. 17.34. Walton does maintain that there are nonoccurrent

    imaginings, and merely raises the above as a problem for a

    simple dispositional analysis of nonoccurrent imaginings(Walton,Mimesis as Make-Believe, pp. 1618).35. Currie, Imagination and Simulation, p. 156.36. Ibid.37. William Lycan, Judgement and Justification (Cam-

    bridge University Press, 1988). See also Jerry Fodor, Psy-chosemantics (MIT Press, 1987), p. 22.

    38. Since my focus is not tacit belief, I do not really wantto commit to one particular account of tacit belief. Rather,the suggestion is that whateverthe right story is about tacitbelief, there should be an analogous story about tacit pre-tense.

    39. Moran, The Expression of Feeling in the Imagin-ation, p. 95.

    40. But this is not the only puzzle raised over suchcases. Indeed, Moran couches the problem largely in termsof what we accept as true in the fiction, and Walton hasargued that the issues concerning truth in fiction need to bedistinguished from the question about the psychology ofthe imagination. Waltons own discussion of these mattersfocuses primarily on our attitudes concerning what is truein the fiction (Morals in Fiction and Fictional Morality,

    Aristotelian Society Supplementary Volume 68 [1994]:2750. See also Walton, Mimesis as Make-Believe,pp. 154155).

    41. Compare Edward Craig, The Problem of NecessaryTruth, in Meaning, Reference, and Necessity, ed. SimonBlackburn (Cambridge University Press, 1975), pp. 131.

    42. We must also suppose that our moral judgmentmechanisms differ from our etiquette judgment mechan-isms, for we typically exhibit no resistance to imaginingalien etiquette (compare Hume, On the Standard ofTaste). But it is independently plausible that the mechan-isms differ, and one explanation for the difference appealsto the role of emotional responses in moral judgment (for

    example, Shaun Nichols, Sentimental Rules [Oxford Uni-versity Press, 2004]).

    43. However, the phenomena surrounding imaginativeresistance are a complex lot. For instance, in some cases, wefind imaginative compliance. For example, we seem to be

    able to imagine things that are immoral in cases like thefollowing: If it were okay for Macbeth to order Duncansmurder, would Lady Macbeths machinations also beokay? These kinds of cases might count as a prima faciedifficulty for the single-code hypothesis (see Nichols, Justthe Imagination).

    44. In her recent discussion of this problem, TamarGendler tells a rather different story. She maintains that weexhibit imaginative resistance because we do not want toimagine morally objectionable perspectives. She writes,the primary source of imaginative resistance is not ourinability to imagine morally deviant situations, but ourunwillingness to do so (Gendler, The Puzzle of ImaginativeResistance, p. 56). Gendler provides compelling examples

    in which we do indeed resist imagining certain thingsbecause we do not want to (see, for example, pp. 7374).But the single-code hypothesis suggests that there is also amore basic architectural explanation for much imaginativeresistance. The inferential and emotional devices that leadus to resist certain beliefs are also triggered by pretense rep-resentations. (Derek Matravers offers an account of imagin-ative resistance that resembles the single-code explanationin some ways [see Derek Matravers, Fictional Assent andthe (So-Called) Puzzle of Imaginative Resistance, in

    Imagination, Philosophy, and the Arts, ed. Matthew Kieranand Dominic Lopes (Routledge, 2003), pp. 91106]. How-ever, to explain the phenomena, Matravers invokes a fic-tional narrator and maintains that resistance occurs because

    we judge that the fictional narrator lacks authority in the tar-get domain. The single-code account need not advert tothese controversial resources to explain imaginative resist-ance.)

    45. For presentations of these puzzles, see Currie, TheNature of Fiction, pp. 53ff., Walton, Mimesis as Make-Believe, pp. 207208, and Currie, The Nature of Fiction, p. 90.

    46. For a broad taxonomy of symmetries and asym-metries in imagination and belief, see Tamar Gendler,Quarantining and Contagion, Mirroring and Disparity: Onthe Relation between Pretense and Belief, inImagination,Philosophy, and the Arts, ed. Matthew Kieran and DominicLopes (Routledge, 2003).

    47. This paper was presented at the 2002 NEH Institute

    on Art, Mind, and Cognitive Science. I would like to thankthe audience for useful comments. I am also grateful toGreg Currie, Stacie Friend, Tamar Gendler, James Harold,Aaron Meskin, Adam Morton, Jenefer Robinson, BarbaraVon Eckardt, Ken Walton, Jonathan Weinberg, an anony-mous referee, and especially Steve Stich for discussion andcomments on earlier drafts of this paper.

  • 7/31/2019 Imagining and Believing

    12/12