Culture and Cognition Author(S): Paul Dimaggio Source: Annual Review of Sociology, Vol

Home , Cultural psychology, Paul DiMaggio

Culture and Cognition Author(s): Paul DiMaggio Source: Annual Review of Sociology, Vol. 23 (1997), pp. 263-287 Published by: Annual Reviews Stable URL: http://www.jstor.org/stable/2952552 Accessed: 18/10/2010 10:56

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=annrevs.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Annual Reviews is collaborating with JSTOR to digitize, preserve and extend access to Annual Review of Sociology.

http://www.jstor.org Annu.Rev. Sociol. 1997. 23:263-87 Copyright? 1997 byAnnual Reviews Inc. All rightsreserved

CULTURE AND COGNITION

Paul DiMaggio Departmentof Sociology,2-N-2 Green Hall, Princeton University, Princeton, New Jersey08544; e-mail:[email protected]

KEY WORDS: sociologyof culture, social classification,social cognition,schemata

ABSTRACT Recentwork in cognitivepsychology and social cognition bears heavily on con- cernsof sociologistsof culture. Cognitive research confirms views of cultureas fragmented;clarifies the roles of institutions and agency;and illuminates supra- individualaspects of culture.Individuals experience culture as disparatebits of informationand as schematicstructures that organize that information. Culture carriedby institutions,networks, and social movementsdiffuses, activates, and selectsamong available schemata. Implications for the study of identity, collective memory,social classification,and logics of action are developed.

INTRODUCTION The studyof culturein everydaylife remains a virtuosoaffair. Interpretive studiesoffer great insight but fail to buildon one another.Cultural theory has becomehighly sophisticated but not fully operational. These riches ready the fieldfor takeoff, like the study of social stratificationin Sorokin's day (1957 [1927]). But beforethe study of livedculture can becomea cumulativeenter- prise,scholars must clarify the cognitive presuppositions behind their theories ofwhat culture does andwhat people do withit, and the fundamental concepts andunits of analysis (Jepperson & Swidler1994, Wuthnow 1987). Recentwork in cognitive psychology and social cognition provides resources forboth tasks. Afterdescribing recent convergence between cultural sociologyand psychology, this chapter considers lessons of recent work on cognition forpresuppositions about the nature of culture;develops implications of these lessonsfor sociological work on identity,collective memory, social classification,logics of action,and framing;and pointsto keyproblems that remain unsolved.

263 0360-0572/97/0815-0263$08.00 264 DiMAGGIO

Ratherthan offer an exhaustivereview of cognitivesociology per se (see Zerubavel1997) or workin psychologyrelevant to culture(see D'Andrade 1995), I emphasizetensions and affinitiesbetween recent cognitive research andwork in the sociology of culture with the aim of bringing the former into the serviceof the latter. I focus on how people use culture, rather than the production of culture,ideology, or cultureembedded in thephysical environment. The pointis notto psychologizethe study of culture,but to lay a foundationfor a viewof culture as workingthrough the interaction of sharedcognitive structures andsupra-individual cultural phenomena (material culture, media messages, or conversation,for example) that activate those structures to varyingdegrees.

SOCIOLOGY AND PSYCHOLOGY: POINTS OF CONVERGENCE A handfulof sociologistshave appreciatedthe potential of cognitivescience to informsociological work on culture(Carley 1989, Cicourel 1973, Schwartz 1981,White 1992), and some social constructionistshave anticipated impor- tantresults of cognitiveresearch (Berger & Luckman1967, Garfinkel1987 [1967], Zerubavel1991). For themost part, however, sociologists of culture have ignoredrelevant work by cognitivepsychologists, social psychologists, and public-opinionresearchers. This omissionreflects a mismatchbetween themodal intellectual styles of humanistic, interpretively oriented cultural so- ciologistsand experimentally oriented positivistic psychologists, as wellas the disappointinglegacy of Parsons' efforts at disciplinaryfusion, which psychol- ogizedculture, reducing it to sharedvalues, norms, and attitudes. Sociology:More ComplexViews of Culture In recentyears, however, common ground between sociology of cultureand psychologyhas grown. The majordevelopment within sociology has been a shiftto a morecomplex understanding of culture.Thirty years ago, most sociologistsviewed culture as a "seamlessweb" (Swidler1997), unitary and internallycoherent across groups and situations.In effect,culture was por- trayedas a latentvariable influencing incommon such manifestations as media images,responses to attitudequestionnaires, and thevalues embodied in ev- erydaypractices. Individuals were presumed to acquire culture in the course of socializationand, in thepopular oversocialized view (Wrong 1961), to enactit unproblematically.It followed from this perspective that there was littlereason to worryabout constructs used to studyculture, for any kind of "cultural stuff" couldserve as an indicatorof the underlying latent variable. By contrast,recent work depicts culture as fragmentedacross groups and inconsistentacross its manifestations(Martin 1992). The view of cultureas CULTURE AND COGNITION 265 values thatsuffuse other aspects of belief,intention, and collectivelife has succumbedto one of cultureas complexrule-like structures that constitute resourcesthat can be putto strategic use (Bourdieu1990, Sewell 1992,Swidler 1986). This shiftmakes studying culture much more complicated. Once we acknowledgethat culture is inconsistent-thatpeople's norms may deviate from whatthe media represent as normal,or thatour preconscious images and dis- cursiveaccounts of a phenomenonmay differ-it becomes crucial to identify unitsof cultural analysis and to focus attention upon the relations among them. In effect,our measures stop being indicators of a latentvariable (culture), and theirrelationship to culturebecomes analogous to thatof education,income, andplace of residence to social stratification: separate phenomena, analytically relatedto a commontheoretical construct, the relations among them a matterfor empiricalinvestigation (D'Andrade 1995 notes similar trends in anthropology). Similarly,once we acknowledgethat people behave as if theyuse culture strategically,itfollows that the cultures into which people are socialized leave muchopportunity for choice and variation. Thus our attention turns to waysin whichdiffering cultural frames or understandingsmay be situationallycued. Addressingsuch issues requires more elaborate and contestable psychological presuppositionsthan did theculture-as-latent-variable view. Psychology:More ComplexViews of Cognition Suchquestions make it sensible for sociologists of culture to turn to psychology forinsight into the mechanisms through which shared culture enters into cognition.Yet nothing guarantees that psychologists, who have their own research agendas,can helpus. Thirtyyears ago, behaviorismmade psychology essen- tiallyirrelevant to thestudy of culture. Twenty years ago, psychologists casting offthe yoke of behaviorismfocused primarily on theacquisition of skillsand capacitiesof littleinterest to mostsociologists of culture.Even a dozenyears ago, theimplications for cultural sociology of manyof the ideas and research traditionsthat are most useful today were still unclear. Whathas happenedto makepsychology useful to sociologistsof culture? First,psychologists have rejected behaviorism, accepted and demonstrated the existenceof mentalstructures used to perceive,process, and retrieve information,and foundways to makeinferences about such structures. Second, just as sociologicalresearch has demonstratedculture's complexity and fragmentation,psychological research has demonstratedthe complexity of memoryand providedglimpses of thepartitioning of mentalstructures by domain.Third, recentfoci of psychologicalresearch (schemata, categories, mental models, andso on) aremuch richer in culturalcontent than the formal operations or in- tellectualcapacities that once preoccupiedcognitivists and developmentalists 266 DiMAGGIO

(Rogoff& Chavajay1995). Fourth,some psychologists have taken notice of suchsociological topics as cross-culturaldifferences in cognition(Shweder & Bourne1991, Markus & Kitayama1991), elite/popular interaction in cultural change(Moscovici 1984), and "distributedcognition" (i.e. thesocial division of cognitivelabor) (Resnick et al 1991,Salomon 1993). In additionto expanding the grounds of shared interest between the two disciplines,such developments have also softenedtwo important epistemological differences.Whereas most sociologists of culturehave been steadfastlyan- tireductionist,resisting efforts to portray culture as theaggregate of individual subjectivities,psychology has focusedupon the individual. Increasingly, however,as I shallargue, psychological research bolsters and clarifies the view of cultureas supra-individual,and even addresses supra-individual aspects of cog- nitiondirectly [as in workon pluralisticignorance (Miller & Prentice1994)]. Second,some sociologists of culturerejected the subjectivist focus of psychologicalresearch, calling instead for research on externalaspects of culture amenableto direct measurement (Wuthnow 1987). In recentyears, cognitivists havedeveloped ingenious empirical techniques (reviewed in D'Andrade1995) thatpermit strong inferences about mental structures, going far toward closing theobservability gap betweenexternal and subjective aspects of culture. Of course,the fit between the disciplines must not be exaggerated.Most of whatpsychologists do is irrelevantto sociologistsof culture,and much of theculture sociologists' study is supra-individual.Common ground has in- creasedbut will remain limited by the different subject matters of the disciplines (Zerubavel1997), which will remain complements rather than substitutes.

COGNITIVE PRESUPPOSITIONS OF CULTURAL SOCIOLOGY Sociologistswho writeabout the ways that culture enters into everyday life necessarilymake assumptions about cognitive processes. If we assumethat a sharedsymbol evokes a senseof common identity (Warner 1959), that a certain frameprovokes people to thinkabout a social issue in a new way (Gamson 1992),that lessons about the structure of space and timelearned in schoolare generalizedto theworkplace (Willis 1977), or thatsurveys can measureclass consciousness(see Fantasia'scritique 1995), we arethen making powerful cog- nitiveassumptions. Such assumptions,while metatheoretical to sociologists, are keenlyempirical from the standpoint of cognitivepsychology. It is crucial,then, to evaluateour assumptions (or adjudicatedifferences among them) by microtranslatingpresuppositions (Collins 1981) to thecognitive level and assessingtheir consistency with results of empirical research on cognition. CULTURE AND COGNITION 267

Coherencevs Fragmentation Many sociologistshave come to rejectthe latent-variable view of cultureas coherent,integrated, and ambiguous in favorof representations ofculture as a "toolkit"(Swidler 1986) or "repertoire"(Tilly 1992): a collectionof stuffthat is heterogeneousin contentand function. Yet much empirical work on culture stillpresumes that culture is organizedaround national societies or cohesive subnationalgroupings, is highlythematized, and is manifestedin similarways acrossmany domains (Hofstede 1980, Bourdieu 1984). Is culturea latentvariable-a tightnetwork of a fewabstract central themes and theirmore concrete entailments, all instantiatedto variousdegrees in a rangeof symbols,rituals, and practices?If so, thenwe wouldexpect to find thatgroup members share a limitednumber of consistentelements-beliefs, attitudes,typifications, strategies-and that the inclusion of any one elementin thecollective culture implies the exclusion of inconsistent elements. Or is culturea grab-bagof odds and ends: a pasticheof mediatedrepre- sentations,a repertoireof techniques,or a toolkitof strategies?If so, then we mightexpect less clusteringof cultural elements within social groups,less stronglinkages among the elements, and weaker pressures for the exclusion of inconsistentelements. Researchin cognitive psychology strongly supports the toolkit over the latent- variableview and suggeststhat the typical toolkit is verylarge indeed. Partic- ularlyrelevant here is research(summarized by Gilbert1991) on howpeople attributeaccuracy or plausibility to statementsof fact and opinion. Consistent withSwidler's (1986) contentionthat "all peopleknow more culture than they use,"Gilbert reports that "The acceptanceof an idea is a partof the automatic comprehensionof thatidea, and the rejection of the idea occurssubsequent to and moreeffortfully than its acceptance." In otherwords, our heads are full of images,opinions, and information, untagged as to truthvalue, to whichwe are inclinedto attributeaccuracy and plausibility. Researchon memorytells a similarstory, revealing that information (includ- ingfalse information) passes into memory without being "tagged" as to source or credibility,and that active inference is requiredto identifythe source of the informationwhen it is recalled. Such inferencesmay be incorrect,yielding misattributionsof source and credibility (Johnson et al 1991). This workhas severalimportant implications for students of culture.First, itrefutes the notion that people acquire a cultureby imbibing it (and no other) throughsocialization. Instead, it directs the search for sources of stability and consistencyin ourbeliefs and representations, first, to schematicorganization, whichmakes some ideas or imagesmore accessible than others; and, second, to cues embeddedin thephysical and socialenvironment. 268 DiMAGGIO

Second,learning that people retain (and storewith a defaultvalue of "cor- rect")almost every image or idea withwhich they have come intocontact, rendersintelligible otherwise anomalous research findings about inconsistency in expressionsof attitudesacross time, cultural volatility in periodsof rapid change(e.g. thefall of the Soviet system), and the susceptibility of attitudesto framingeffects (Sniderman & Piazza 1993). Third,the research explains the capacity of individuals to participate in multiple culturaltraditions, even when those traditions contain inconsistent elements. Fourth,it establishesthe capacity of people to maintaindistinctive and incon- sistentaction frames, which can be invokedin response to particular contextual cues. Fifth,this work raises the possibility that socialization may be less ex- perientiallybased, and moredependent upon media images and hearsay,than manyof our theories (for example, Bourdieu's habitus [1990] construct) imply. Suchinferences as thesego beyondthe scope of cognitive studies, to be sure, and muchrides on theprecise ways in whichschematic organization imposes orderupon stored knowledge and memory.Nonetheless, recent cognitive re- searchstrongly reinforces the "toolkit" as opposedto the "latent-variable" view of cultureand, at thevery least, places theburden of proofon thosewho de- pictculture as stronglyconstraining behavior or whowould argue that people experienceculture as highlyintegrated, that cultural meanings are strongly the- matized,that culture is binding,and that cultural information acquired through experienceis morepowerful than that acquired through other means.

Institutionand Agency Cognitiveresearch can also enhanceour appreciation of theview that culture bothconstrains and enables (Sewell 1992). Althoughthis position has become virtuallycatechismic among sociologists of culture,we knowlittle about the conditionsunder which one orthe other is thecase. Manysociologists believe, followingGramsci (1990), thatculture, embedded in languageand everyday practices,constrains people's capacityto imaginealternatives to existingar- rangements.At thesame time, we knowthat people act as ifthey use cultural elementsstrategically to pursuevalued ends (Bourdieu1990). Cognitivere- searchcannot answer the essentially sociological question of when culture does each,but it can providedirection to thesearch. Thefinding that culture is storedin memory as an indiscriminatelyassembled and relativelyunorganized collection of odds and endsimposes a farstronger organizingburden on actorsthan did the earlier oversocialized view. The question,then, is howthe actor organizes the information that she orhe possesses. Psychologicalresearch points to twoquite different mechanisms or modesof cognition. CULTURE AND COGNITION 269

AUTOMATIC COGNITION The first,and most important, which I referto as au- tomaticcognition is "implicit,unverbalized, rapid, and automatic" (D'Andrade 1995). This routine,everyday cognition relies heavily and uncriticallyupon culturallyavailable schemata-knowledge structures that represent objects or eventsand providedefault assumptions about their characteristics, relationships,and entailments under conditions of incomplete information. Psychologicalresearch on schematais centralto theinterests of sociologists bothmethodologically (due to advancesin techniquesthat reveal taken-for- grantedassumptions to whichsubjects may not have easy verbalaccess) and substantively,for what it tellsus abouthow culture works. Indeed, for some purposes,it maybe usefulto treatthe schema as a basic unitof analysisfor the studyof culture,and to focuson social patternsof schemaacquisition, diffusion,and modification (Carley 1991 makesa relatedargument). Schemataare both representations ofknowledge and information-processing mechanisms.As representations,they entail images of objects and the relations amongthem. Psychologistsuse theterm broadly [some wouldsuggest too broadly(Fiske & Linville1980)]. It can referto simple,highly abstract concepts[for example, container (D'Andrade 1995)]; to concrete activities (buying chewinggum), or to complexsocial phenomena(group stereotypes or social roles). Eventschemata or scripts(Abelson 1981, Garfinkel 1987) constitute an importantclass of schemata.Special attentionhas also beengiven to self schemata(Milburn 1987, Markus & Kitayama1994, Markus et al 1997),cultur- allyvariable representations ofthe self that provide stability both to individual behavioracross time and to social interactionswithin the group. Schemataare also mechanismsthat simplify cognition. Highly schematic cognitionis therealm of institutionalized culture, of typification, ofthe habitus, of thecognitive shortcuts that promote efficiency at theexpense of synoptic accuracy(Berger & Luckman1967, Bourdieu1990, Kahnemanet al 1982). Muchcognitive research demonstrates that "schematic material dominates other materialin accurate recall, in intruded recall, in recognition confidence, in recall clusteringand in resistanceto disconfirmation....Schemata also facilitate inaccuraterecall when the information is schema consistent" (Fiske & Linville 1980: 545). In schematiccognition we findthe mechanisms by which culture shapesand biases thought.

People are morelikely to perceiveinformation that is germaneto existing schemata VonHippel et al (1993) reportthat experimental subjects are more likelyto perceive correctly terms that are schematically relevant than those that are not. Informationembedded in existingschemata and informationthat is schema-dissonantare both more likely to be noticedthan information orthogo- nal to existingstructures (Schneider 1991). Such laboratoryfindings resonate 270 DiMAGGIO withresults in historicalsociology and cultural studies: for example, the grad- ual andhalting acceptance of information about the New World by early modern mapmakers(Zerubavel 1992); the ways in which archaic physical models con- strainedmedical scientists' interpretation ofnew evidence about syphilis (Fleck 1979); and thepenchant of male biologistsfor seeing dominance hierarchies whenthey watch apes andelephant seals (Haraway1991).

People recallschematically embedded information more quickly Most psy- chologicalevidence is basedon laboratoryexperiments, which reveal that sub- jectsremember longer lists of words, or interpret ambiguous stimuli more accurately,and retrieve information about a storythey have heard more effectively ifit is relevantto preexisting mental structures that render the information inter- pretable(Sedikides & Skowronski1991). Butagain, there are intriguing socio- logicalparallels in studiesthat report cross-cultural differences in descriptions of thecontent of thesame novel(Griswold 1987), television program (Liebes & Katz 1990), or movie(Shively 1992) thatreflect collective preoccupations ("chronicallyactivated mental structures" in psychologicalparlance). People recall schematicallyembedded information more accurately When Freemanet al (1987) askedmembers of a facultyworkshop to list the people who hadattended the previous meeting, they found that long-term attenders correctly recalledparticipants who regularlyattended, but forgot the infrequent attenders.Using a verydifferent method (analysis of Watergate transcripts), Neisser (1981) reportedthat Nixon aide JohnDean rememberedschema-consistent eventsmore accurately than events that were schema-inconsistent.

People mayfalsely recall schematically embedded events that did notoccur Freemanet al's (1987) informantsremembered regular attenders as presentat themeeting in questioneven when they hadn't been there. When subjects are toldto code small-groupinteractions and then given questionnaires about char- acteristicsof group members shortly thereafter, the post-hoc evaluations yield muchhigher correlations of schematicallyrelated behaviors (e.g. criticizing orexpressing hostility) than do thereal-time codings (Shweder 1982). Similar confusionof schematicrepresentations for real eventsmay be observedin at leastsome reports of satanic child abuse (Hacking 1995) and in some of former PresidentReagan's speeches. The parallelwith sociological accounts of institutionsis striking.Typifi- cations(mental structures) influence perception, interpretation, planning, and action(Berger & Luckman1967, DiMaggio & Powell1991). Institutionalized structuresand behaviors(i.e. thosethat are both highly schematic and widely shared)are taken for granted, reproduced in everydayaction [Giddens' "structuration"(1984)] andtreated as legitimate(Meyer & Rowan1977). Indeed,an CULTURE AND COGNITION 271 eminentpsychologist (Bruner 1990:58) has writtenexplicitly of the"schema- tizingpower of institutions." Thus the psychology of mental structures provides a microfoundationto the sociology of institutions. Researchon social cognition enhances our understanding ofhow culture con- strainsbut does not support theories that depict culture as overwhelminglycon- straining.Instead, consistent with contemporary sociological theorizing, work in psychologyprovides microfoundational evidence for the efficacy of agency.

DELIBERATIVE COGNITION In contrastto automaticthought, psychologists notea quitedifferent form of cognition,which is "explicit,verbalized, slow, and deliberate"(D'Andrade 1995). Whensufficiently motivated, people can overrideprogrammed modes of thought to thinkcritically and reflexively. Such overridesare necessarilyrare because deliberation is so inefficientin itsrejection of the shortcuts that automatic cognition offers. Consequently, the keyquestion is whypeople are ever deliberative. Psychologists have identified threefacilitating conditions in studiesthat intriguingly parallel work in the sociologyof culture. AttentionPsychological research suggests that people shift into deliberative modesof thought relatively easily when their attention is attractedto a problem. Forexample, experimenters can create false recollections of a videotapeor story amonglaboratory "witnesses" by presenting inaccurate information or asking leadingquestions (Loftus et al 1989). But whenthe task is changedto ask subjectsto thinkcarefully about the source of particularbits of information, theexperimental effect is diminishedor eliminated(Johnson et al 1993). In experimentalstudies of attitude-behaviorconsistency, merely increasing self- awarenessby placing a mirrorin the face of the subject as he orshe completes an attitudequestionnaire significantly increases the attitude-behavior correlation (Abelson 1981:722). Such resultsparallel the insights of studentsof social movements,who have studiedagenda-building and who have also notedthe effectivenessas an organizingdevice of refrainingissues in ways thatcall attentionto problems salient to movement participants (Snow & Benford1992). Motivation People may also shiftfrom automatic to deliberativecognition whenthey are stronglymotivated to do so by dissatisfactionwith the status quo or by the moralsalience of a particularissue. For example,although racistschemata are accessibleto mostwhite Americans, whites can override suchschemata to some extent through awareness and reflexivity (Devine 1989). Marx's theoryof class consciousness-whichcontends that physically prox- imateworkers facing immiseration will overcomefalse beliefs through inter- actionand reflection-isa classicsociological counterpart (and see Bourdieu 1974). 272 DiMAGGIO

Schemafailure Finally,people shift to more deliberative modes of processing whenexisting schemata fail to accountadequately for new stimuli. Research on thepsychology of intergrouprelations suggests that people in taskgroups initiallycode others on the basis of stereotypes but shift to more deliberate evalu- ationswhen faced with very strong inconsistent evidence (Schneider 1991:536, Bergeret al 1980). Moscovici,whose Durkheimiansocial psychologydif- fersin manyrespects from other psychological accounts of mentalstructures (Farr& Moscovici1984, Augoustinos & Innes1990), argues that collectivities confrontedwith disjunctive social change construct new social representations (oftenanchored in analogiesto pre-existingschemata, and oftenconstructed deliberativelyby experts in the social sciences and mass media) in order to inter- pretnew stimuli. Such arguments are paralleled in Garfinkel's (1967) breaching experiments,which forcibly and painfully overrode automatic processing, and in Swidler'scontention that ideologies and other consistent cultural forms are moreinfluential during unsettled times (1986, Jepperson& Swidler1994 on constitutivevs. strategicculture). Psychologistsmay notethat I have paid scantattention to activedebates aboutthe natureof mentalstructures and have drawntoo sharpa contrast betweenautomatic and deliberative processing. Research on culture,however, can alreadybenefit from what research on cognitionhas resolved.The notion of schemais a fairapproximation of phenomena identifiable in fuzzyoutline, if notsharp relief, by experimentalmethods; research on schemataadvances sociologicalunderstandings ofculture, especially institutions; and research on automaticvs deliberativeprocessing may help sociologists determine what to do withthe widely believed but theoretically inert notion that both institution and agencyare centralto sociallife. Cultureas Supra-Individual Itis nonews to sociologists that culture exists, sui generis, at the collective level. (The positiontaken here-that culture is also manifestin people'sheads-is probablymore controversial.) Nonetheless, psychological research can helpus appreciateseveral aspects of culture'ssupra-individual character that sociologistsof culture sometimes neglect.

PLURALISTIC IGNORANCE A livelybranch of social-psychologicalresearch derivesfrom Robert K. Merton'snotion of "pluralistic ignorance" (1957): the idea thatpeople act with reference to sharedrepresentations ofcollective opinion thatare empiricallyinaccurate. Such researchdirects us to distinguish betweentwo sensesin whichculture is supra-individual:as an aggregateof individuals'beliefs or representations, oras sharedrepresentations ofindividu- als' beliefs.Substantial evidence indicates that the latter deviates substantially CULTURE AND COGNITION 273 fromthe former with significant behavioral consequences and that this process representsa basis for the relative autonomy of social norms (Miller & Prentice 1996,Noelle-Neumann 1993).

INTERGROUP CONTRAST AND POLARIZATION The existenceof group-level cultures(shared understanding partly independent of individual beliefs) is also suggestedby thetendency of groupsto adoptpublic positions more extreme thanthe preferences of theirmembers, especially when acting with reference to a contrastinggroup. What is strikingis notpolarization per se, butthe cul- turalavailability of polarized stances (representations ofcollective opinion) on whichmembers of each groupcan converge(Tajfel 1981).

SCHEMATA AS CULTURE Not all schemataare culturalto the same degree. Some schematareflect universal cognitive processes (for example, basic object categorization),whereas others may be quiteidiosyncratic. Many schemata, however,and theschemata of greatestinterest to sociologistsof culture,en- act widelyheld scripts that appear independent of individualexperience. For example,the research, cited above, that found coherence in ratingsof small groupbehavior emerging only after the fact, led theauthor (Shweder 1982) to speculatethat much of what passes as clinicalresearch on personalityis really aboutcultural constructions of personhood(and see Meyer1986).

COHERENT CULTURES AS EXTERNAL TO PERSONS Despite thischapter's focus on subjectiverepresentations of culture, we mustnot forget that relatively coherentcultural forms exist independently ofpersons in thebroader environment.Indeed, one of themore notable characteristics of modernsocieties is theexistence of a culturaldivision of labor in whichintellectual producers in- tentionallycreate and diffuse myths, images, and idea systems(Douglas 1986, Farr& Moscovici1984, Swidler 1997). Otherrelatively coherent representa- tionsexist less formallyas narrativesor storiesrepeatedly invoked in public discourse(Dobbin 1994,White 1992).

AN INITIAL SYNTHESIS Some would arguethat whatever coherence exists flowsfrom such externallyavailable sources, i.e. thatcultural coherence is entirelyexternal to theperson. As we have seen, however,such a position pushesthe healthy shift from the latent-variable to the toolkit one steptoo far. Instead,the research reviewed here suggests that culture works through the in- teractionof three forms. First, we haveinformation, distributed across persons (Carley1991). Such distributionis patterned,but not highly differentiating, due to theindiscriminant manner in whichbits of culture are accumulated and storedin memory(Gilbert 1991). Second,we have mentalstructures, espe- ciallyschematic representations of complexsocial phenomena,which shape 274 DiMAGGIO theway we attendto, interpret, remember, and respond emotionally to thein- formationwe encounterand possess. Such schemataare more clearly socially patternedthan are memory traces. Finally, we haveculture as symbolsystems externalto the person, including the content of talk, elements of the constructed environment,media messages, and meanings embedded in observableactivity patterns. Cultureinheres not in theinformation, nor in theschemata, nor in thesym- bolicuniverse, but in the interactions among them. As we haveseen, schemata structureour use ofinformation. But people acquire many schemata throughout theirlives, and someof theseare inconsistentboth in contentand in implica- tionsfor behavior. How is itthat people invoke one amongthe many schemata availableto themin a givensituation? To simplifygreatly in orderto focusupon the aspect of theprocess most relevantto thesociology of culture, selection is guidedby culturalcues available in theenvironment. Although a fewschemata may be chronicallyavail- able,more often they are primed or activatedby an externalstimulus or frame (Sedikides& Skowronski1991, Barsalou 1992, Gamson 1992:6-8, Schud- son 1989). Framingeffects in social surveys-e.g.the finding that whites are morelikely to accept negative stereotypes of African-Americans ifthe question is precededby a neutralreference to affirmativeaction (Sniderman & Piazza 1993:102-104)-are familiarexamples. But schematacan also be activated throughconversation, media use, or observationof thephysical environment. Understandingthe interaction between two distributions-of the schemata that constitutepeople's culturaltoolkits, and of externalcultural primers that act as framesto evoke(and, in evoking,exerting selection pressures upon) these schemata-is a centralchallenge for sociologists of culture.

APPLICATIONS This sectionreviews work on cognitiveaspects of the sociologyof culture in lightof theperspective developed here. The topicsare identity,collective memory,social classification,logics of action, and framing. Identity Identityhas becomeone ofthe most active research fields in thesociology of culture.It is usefulto distinguish between two quite different kinds of collective identity:the identities of collectives, on theone hand, and collective aspects of theidentities of individuals on theother.

IDENTITIES OF COLLECTIVES At the supra-individuallevel, collective iden- tityis a sharedrepresentation of a collectivity.Research at thislevel portrays collectiveidentities as highlyconstructed (Anderson 1983), throughexplicit CULTUREAND COGNITION 275 messagesand moresubtle elements such as anthemsand flags(Cerulo 1994). Collectiveidentities are chronicallycontested, as groupsvie to producesocial representationscapable of evoking schemata favorable to their ideal or material interests(Moscovici 1984, Zerubavel 1994, Friedland & Hecht1996). Anotherline of research,active in bothpsychology and sociology,views identitiesand selves as collectiverepresentations that vary cross-culturally and historically.Markus et al (1996) reviewresearch on differencesin thecultural constructionofidentity in East Asian and Western societies. Meyer & Jepperson (1996) contendthat the modern self (and its variations in different polities) is a constructedidentity endowed with agency in relationto thecollectivity.

COLLECTIVE ELEMENTS IN INDIVIDUAL IDENTITIES Muchresearch on collec- tiveidentity is actuallyabout the more complex issue of the ways in which social identitiesenter into the constitution of individualselves. Social identitytheory viewsindividual identities as comprisingprioritized identity-sets based on par- ticularisticand role-based group affiliations (Stryker 1986). Self-categorization theoriesalso portraycollective identities as invokedby conditionsthat make particularidentities especially salient (Tajfel & Turner1986). In thisview, individualidentities reflect elaborated group-identity schemata in proportion to strengthand recency of activation.Viewing identities as context-dependentin thisway is consistentwith observations of thevolatility with which identities maygain and lose salienceduring periods of intergroup conflict.

CollectiveMemory Collectivememory is the outcomeof processesaffecting, respectively, the informationto whichindividuals have access,the schemata by whichpeople understandthe past, and theexternal symbols or messagesthat prime these schemata.Like collectiveidentities, research on collectivememory portrays thephenomenon in bothsupra-individual and individual terms. Severalscholars have studied institutional processes that maintain or suppress informationas part of public culture, such as factorsdetermining the reputation and popularityof particular persons or artworks (Fine 1996,Griswold 1986, Lang& Lang 1988). Muchresearch, however, focuses upon the schematic level, studyingstruggles to definethe ways in whichmembers of a societyinterpret widelyshared information about their past, either tracking change in theways in whicha personor publicfigure is understoodover time (Schudson 1992, Schwartz1991) or analyzingconflict over alternative visions of a collective past(Maier 1988,Zerubavel 1994). Littleresearch has focusedon theinteraction between individual and col- lectivememories. An exceptionis thework of Schuman & Scott(1989), who use surveymethods to explorethe possibility that the historical events that 276 DiMAGGIO menand women of different generations remember most vividly structure their understandingofcontemporary social issues. Social Classification The studyof social classification-thesocial construction and use ofcategory schemes-has burgeonedin the last decade. Somework has analyzed processes ofclassification in historical time, describing the emergence of a stronglyclas- sifiedartistic high culture (DiMaggio 1982),or the use ofsocial categorization inthe formation and implementation ofsocial policies (Starr 1992). Ofparticu- larinterest is Mohr's(1994) analysisof "discourse roles," which uses structural equivalenceanalysis to identifythe implicit classification of social problems and clientgroups embedded in self-descriptionsofsocial-service and poverty- relieforganizations in earlytwentieth-century New YorkCity. Otherresearch has focusedupon social differentiation in shorter time spans. Zelizer(1989) describesthe process by whichwomen find ways to differenti- ate evenmoney, the universal medium of exchange,in orderto imbueit with social meaning.Lamont (1992) analyzesthe bases uponwhich men of differ- entregional and nationalorigins make social distinctionsthat reinforce their senseof social honor. Gieryn (1997) describesboundary work within scientific communities,examining how scientists respond when the strong classification science/nonscienceis threatened. Zerubavel,one of fewsociologists to studyclassification from a cognitive perspective,points out that the drive to partitiona continuousworld appears to be a humanuniversal, though the nature of thecategories constructed may varysignificantly among groups (Zerubavel 1991, 1997, Douglas 1966). Rosch (1978), whosework has dominatedpsychological thinking on thetopic, proposes (withmuch experimental support) that cognition is mostefficient when we chunkmany separate features (bits of information)together by thinking witha prototype(complete mental image) of an object.Prototypical constructs emergeat the most efficient level of abstraction: i.e. wherean increasein speci- ficityprovides the greatestmarginal increase in information.Thus we have prototypesfor "chair" but not "furniture" or "divan,"and for"bird" but not for"animal" or "sparrow." Although the level at whichobject prototypes form appearsto be relativelyuniversal, the specific content of a prototypereflects a mixof typicality and availability in a givenlocation (D'Andrade 1995). Rosch appliedher model of prototypes to relativelysimple concepts. Self- categorizationtheory draws on theprototype model (Hogg & McGarty1990), butit remainsto be seen if complexsocial constructsare representedin such unambiguousterms. If so, applicationto roleanalysis may be useful,in light of an intriguingparallel between Rosch's characterization of a prototypeas a CULTURE AND COGNITION 277 coreof essential features and Nadel's (1957) classicdefinition of social role as consistingof a coreof entailments and a penumbraof optional features. Logics ofAction Manyauthors have used theexpression "logics of action"to referto an inter- dependentset of representations or constraints that influence action in a given domain.Sometimes, of course, the term is usedas a synonymfor "ideal type" (Orru1991) or,in rational-actorapproaches, to referto situationalconstraints thatinduce parallel behaviors among players with similar resources given par- ticularrules of the game (Block 1990,Offe 1985). A richer,more cultural, sense of logics has emerged in recent work in political economy,a viewthat embeds them in the interaction between mental structures instantiatedin practicalreason (Bourdieu 1990), on theone hand,and institu- tionalrequirements on theother. Friedland & Alford(1991:248-49) provide themost thorough exposition and definition,describing "institutional logics" as sets "of materialpractices and symbolicconstructions" that constitute an institutionalorder's "organizing principles" and are "available to organizations and individualsto elaborate."According to Friedland& Alford,these logics are"symbolically grounded, organizationally structured, politically defined and technicallyand materially constrained." Similarimagery is apparentin Boltanski& Thevenot'snotion of modes of justification(1991), institutionallylinked discourses embodying specific orientationstoward action and evaluation.Empirical development of similar ideas can be foundin Fligstein's(1990) workon "conceptionsof control"in corporategovernance, and in Stark's(1990) analysisof shop floor politics in a Hungariansocialist factory. Such workrequires a taxonomyof institutions,each of whichentails a dis- tinctivelogic. (For Friedland& Alford,the institutions are capitalism,the state,democracy, family, religion, and science,each of whichhas itsown ax- ial principleand linkedroutines and rituals.)Conflict erupts from the clash of institutionallogics, as whena wifeviews her household labor through a marketplacelogic of explicit exchange, whereas her husband imposes a family logicof selflessservice upon the situation. The notionof logics is immenselyappealing. First, it proposes that external ritualsand stimuliinteract with internal mental structures to generateroutine behavior. Second, it is consistentwith the view thatculture is fragmented amongpotentially inconsistent elements, without surrendering the notion of limitedcoherence, which thematization of clustersof ritualsand schemata aroundinstitutions provides. Third,it providesa vocabularyfor discussing culturalconflict as confrontationbetween inconsistent logics of action. 278 DiMAGGIO

At thesame time,the work remains frankly exploratory and calls attention to gaps in ourcurrent understanding of cultureand cognition,which neither psychologynor sociology can address.These are the topics of the next section.

KEY PROBLEMS IN THE STUDY OF CULTURE AND COGNITION The notionof institutional logics can be reinterpretedas an effortto thematize schemataand link them to social structure.In orderto exploitthe insights this perspectiveoffers, students of cultureneed three things that we nowlack:- an understandingofhow schemata aggregate to more complex cultural structures, or"logics"; an understandingofcultural change, which, in turn, requires a clear understandingofthe way in which actors switch among institutional logics; and a theoryof analogy,which is necessaryif we are to understandprocesses of schematicgeneralization that thematization and switching both require. Modelsof Schematic Aggregation Perhapsthe highest priority for students of culture and cognition is to develop modelsof thematization, by which I meanthe ways in whichdiverse schemata aggregateto more general and sociologically interesting constructs like thought styles,stories, logics, paradigms, and ideologies. There are several candidates forsuch models.

ATOMISTICDECOUPLING Thenull hypothesis is that everyday thought is pop- ulatedby randomly invoked, loosely coupled schemata with little or no higher- levelarchitecture. If so, thematizationis simply imposed post hoc by cultural specialistsor embedded in the environment and in everyday routines. Although thisview is inconsistentwith most work in the sociology of culture, and would seemill-equipped to explain either experimental research on schemata or macro- culturalchange, it cannot now be disconfirmedabsolutely.

NESTED HIERARCHY At the oppositeextreme is the view of cross-cultural psychologiststhat culture comprises a hierarchyof nestedschemata, arrayed fromabstract to concrete,with the latter entailed by theformer. For example, Markus& Kitayama(1994) view a wide rangeof cognitivedifferences betweenJapanese and Americans as flowingfrom fundamental differences in self-schemata.Although they provide compelling evidence of significantin- tergroupdifferences, one neednot assume as muchcoherence as theydo.

DOMAIN-SPECIFICITYThere is considerableevidence that information and schematapertaining to different life domains is storedin distinct areas of memory,with schematic integration occurring within specific domains (Hirschfeld CULTURE AND COGNITION 279

& Gelman1994). In thisview, clusters of schemataare coherentonly within limitedboundaries; taken together, the domains are "morelike thecollected denizensof a tidepool thana singleoctopus" (D'Andrade 1995:249). This viewhas considerableexperimental support, though there is littlecon- sensus as to the size or characterof the domains. It is temptingto equate "domain"with the institutional realms identified by Friedland & Alford(1991) or Boltanski& Thevenot(1990), and to positthat culturally specific "logics of action"are thus embedded in schematicorganization, but there is at present littleif any empirical warrant for doing so.

IDENTITYCENTRALITY Someevidence suggests that affectively hot schemata aremore salient and have more extensive entailments than do emotionallyneu- tralstructures. Work on identity(Wiley & Alexander1987, Hogg & McGarty 1990) suggeststhe possibility that "the self" may be an emotionallysupersat- uratedcluster of schematatending toward consistency and stabilityover time. Schematathat are embedded in the self-schemata, then, are more closely artic- ulatedwith other schemata than those that are not incorporated into the self.

ROLE CENTRALITY By analogy,one can viewroles as situationallyevoked, emotionallyactivated, partial identities that provide integrated chunks of schematicorganization and permit compartmentalization ofdifferent cultural contents. This perspectiveis appealingbecause it identifiesa mechanism (i.e. role activation)connecting schematic triggering to contextualvariation, and because itis consistentwith evidence for domain-specificity ofschematic organization. Moreover,because roles are embeddedin distinctiverole relations, this view pointstoward an integrationof culturaland networkanalysis within a single framework(McCall 1987). Whichof thesemodels of schematicthematization best describes the processes bywhich people integrate schemata is at presentanybody's guess. Sig- nificantmatters-the extent to which ideology enters into conscious experience, thepatterning of culturalstyles or orientations,and thestability of cognition acrosscontext-ride on itsresolution. CulturalChange A secondpriority for sociologists of cultureis to createtheories of cultural changethat integrate ideas from research on culture and cognition with macroso- ciologicalperspectives. At leastfour different change processes are crucialto understand.

THEORYOF ENVIRONMENTALTRIGGERING I have arguedthat culture enters intoeveryday life through the interaction of environmentalcues and mental structures.I have further suggested, by combininglogic-of-action theories in 280 DiMAGGIO sociologyand domain-specificitytheories in psychology,that cultural understandingsmay be fragmentedby domain, so thatwhen persons or groups switch fromone domainto another,their perspectives, attitudes, preferences, and dis- positionsmay change radically. It follows that large-scale cultural changes may be causedby large-scale,more-or-less simultaneous frame switches by many interdependentactors. At themicro level, we need a betterunderstanding of how and whypeople switchamong frames, logics, or domains(White 1995; froma rational choiceperspective, Lindenberg & Frey1993). The paradigmaticwork on this comesfrom language, where research on code-switchinghas documentedthe circumstances(ordinarily changes in context,conversation partner, or topic) thattrigger change in languageor dialect (Gumperz 1982). At themacro level, thechallenge is to createmodels that link environmental change to patternsof switching(White 1995).

THEORYOF SCHEMAACQUISITION, DIFFUSION, AND EXTINCTION Psycholo- gistshave cast substantiallight on theacquisition of schemataby individuals duringdevelopment (Nelson & Gruendel1981, Hirschfeld1994). Sociolo- gistsof cultureshould turn their attention to factorsleading to changein the distributionand level of activationof culturalrepresentations or schemata in thepopulation. Such changemay occur if different cohorts acquire particular schemataat varyingrates; or if changesin thedistribution of environmental cues lead to enhancedactivation or deactivationof particularschemata that havealready been acquired. Diffusionmodels of the sortthat have been used to studythe effectsof mediaexposure on theadoption of new technologies or beliefsmay be useful. Diffusionshould be mosteffective where resonance exists between the new culturalelement and existing schematic organization (Sperber 1985). Work in the historicalsociology of cultureprovides some guidance. Wuthnow's(1989) macro-theoryofideological change, which points to the im- portanceof ecological effects on the life chances of new beliefs, may be usefully transposedto more micro levels. Tilly (1992) hasdeveloped and implemented a valuableapproach to studying change over time in contentious movement reper- toires. Buchmann& Eisner(1996) presentevidence of acceleratingchange in thepublic presentation of selves duringthe secondhalf of thetwentieth century. A particularchallenge is to understandcognitive aspects of major collective eventsin whichlarge numbers of persons rapidly adopt orientations that might haveappeared culturally alien to the majority of them a shorttime before. Some religiousrevivals, the emergence of capitalism after the fall of the Soviet Union, and somespirals of ethnic antagonism are demanding cases ofthis kind. CULTURE AND COGNITION 281

THEORY OF DELIBERATIVE OVERRIDING It is importantto understandnot only how cultureconstrains, but how persons and groupscan transcendthe biasing effectsof cultureon thought.Work on thisproblem by psychologists(noted earlier)must be supplementedby researchon thetypes of social interaction thatlead largenumbers of people to questionand, ultimately, to revisetheir schematicrepresentations ofsocial phenomena. Analogyand Generalization Relatedto thestudy of change, but so importantthat it warrants a section of its own,is theproblem of analogyand generalization.Sociological theories that portraypersons as activelyincorporating culture into cognitive organization invariablyrely on somenotion like the habitus, which Bourdieu (1990) refers to as a "systemof durabletransposable dispositions." The keyquestion for all of thesetheories is: Underwhat conditions are dispositionsor schemata abstractedand transposed from one domainto another? Almostall culturalchange entails the transferof some body of ideas or imagesfrom one contentarea to anotheron thebasis of similarityjudgments. Indeed,any attempt to characterize the culture of a groupor a peoplein abstract terms-i.e.,any analytic effort at thematization-takesfor granted that actors havethe capacity to drawanalogies between classes of objects, actors, events, or actions,and thereby to understandthem in similarways. Thinkof culture as a networkof interrelated schemata, with analogies as the "ties"that create paths along which generalization and innovation occur. How arenew "ties" created? The literatureprovides at leastthree alternatives.

FEATURE CORRESPONDENCE In themost straightforward models, two schemata orrelated structures lend themselves to analogy(and thus to generalization acrossdomains) insofar as theyshare particular features (Lakoff & Johnson 1980) thatcreate a correspondencebetween them. Thus Swinburne'sline, "whenthe houndsof springare on winter'straces," is meaningfulbecause of thecorrespondence between temporal and spatialpursuit and betweenthe destructiveeffects of houndson haresand of springon winter.Two problems withthis view are that the correspondence itself is constructedrather than innate; and thatanalogical power would not seem to varywith the extent of overlap betweentenor and vehicle.

STRUCTURE-MAPPING This view takesas its startingpoint the existenceof someform of content-relateddomain-specificity. Analogies connect not sim- plyschemata but whole domains (Tourganeau & Sternberg1982), deriving their powerfrom the network of entailed comparisons they trigger. The mostpow- erfulanalogies connect domains that are structurally homologous. Put another way,generalizability across domains is a functionnot of theextent to which 282 DiMAGGIO theyshare particular features in common,but of theextent to whichrelations amongfeatures are structurallysimilar (Gentner 1983).

EMOTIONAL RESONANCE Someresearchsuggests that affectively hot schemata aremore likely to be generalizedacross domains than affectively neutral schemata. For example,analogies are likelyto be drawnbetween situations that elicitstrong emotional reactions of a similarkind (Abelson 1981:725).

POLYSEMY AND SEMANTIC CONTAGION A finalpossibility is thatpolysemous expressions-thosewith distinct meanings that resonate with multiple schemata or domains-facilitateanalogical transfer. Bakhtin's work (1986) on textual multivocalityis suggestive in thisregard, as is White's(1992) workon stories and rhetorics.Ross (1992) portraysmeaning as emergingfrom the relations of wordsto one anotherin speech and to activitiesin real time. Because theseconstantly change, meanings are rarely fixed, but instead adapt, diverge, and spreadacross domains through semantic contagion. This perspectiveis particularlyattractive because it acknowledges endemic change in language and othersymbol systems and because it embeds generalization insocial interaction.

SYMBOLS, NETWORKS, AND COGNITION Cognitiveaspects of culture are only one-and notnecessarily the largest-part ofthe sociology of culture's domain. But it is a partthat we cannotavoid if we are interestedin howculture enters into people's lives, for any explanation of culture'simpact on practicerests on assumptionsabout the role of culturein cognition.I haveargued that we arebetter off if we makesuch models explicit thanif we smugglethem in throughthe back door and thatwork in cognitive psychologyand social cognition,although animated by differentquestions, offerstools that we sociologistscan use to pursueour own agendas. Ultimately,the challenge is tointegrate the micro perspectives on culturede- scribedhere with analyses of cultural change in largercollectivities over longer stretchesof time. I haveargued for a perspectivethat privileges schemata and re- latedconstructs as unitsof analysis, and attends to mechanisms by which physical, social,and cultural environments differentially activate these schemata. This argumenthas beggedthe question of which aspects of the environment are mostworthy of study.Without denying the unquestionable importance of researchon howmedia and activity structures interact with subjective cultural representations,I shall conclude by calling brief attention to new research on the relationshipof cognitive and symbolic phenomena to social structures portrayed as social networks. Some researchershave focused on cognitiverepresentations of social structure.[Fiske & Linville(1980) claimthat schema theory is especiallyrelevant CULTURE AND COGNITION 283 to therepresentation of social phenomena;and see Howard(1994).] The idea thatsocial structuresexist simultaneously through mental representations and in concretesocial relationswas centralto Nadel's (1957) role theory.Both theorists(Emirbayer & Goodwin1994, Orr 1995, White 1992) and researchers (Krackhardt1987) areexploring the implications of this view. Networksare crucial environments for the activation of schemata, logics, and frames.In a studyof theParis Commune, Gould (1995) arguesthat political protestnetworks did notcreate new collectiveidentities, but rather activated identitiesthat communards already possessed. Bernstein (1975) demonstrates theimpact of network structures on individuals'tendency to employcognitive abstraction.Erickson (1996), studyingsecurity guards, finds a correlationbe- tweenthe complexity of social networksand thediversity of conversational interests.Vaughan (1986) describeshow people questioningmarriage alter customarypatterns of social relations in orderto create new, independent iden- titiesas prologueto separation.Such studiespoint to a new,more complex understandingof the relationshipbetween culture and social structurebuilt upon carefulintegration of microand macro,and of cognitiveand material, perspectives.

ACKNOWLEDGMENTS Thanksare due to thestudents in myPrinceton graduate seminar on culture and cognitionfor insightful discussions of muchof thework reviewed here; to mypsychology colleagues Marcia Johnson, Dale Miller,and Deborah Pren- tice,who providedvaluable guidance in myefforts to come to speedyterms withculture-relevant literature in cognitiveand social psychology;and to Bob Wuthnow,Dale Miller,Eviatar Zerubavel, Roger Friedland, and John Mohr for opportunitiestopresent these thoughts at meetings and workshops at Princeton, Rutgers,Santa Barbara, and theASA meetings.For valuablereadings of ear- lierdrafts, I am indebtedto RogerFriedland, Michele Lamont, Diane Mackie, CalvinMorrill, Abigail Smith, Ann Swidler, and Eviatar Zerubavel.

Visitthe Annual Reviews home page at http://www.annurev.org.

LiteratureCited

AbelsonRP. 1981. Psychologicalstatus of the Genresand OtherLate Essays,ed. C Emer- scriptconcept. Am. Psychol. 36:715-29 son,M Holquist,pp. 61-102. Austin:Univ. AndersonB. 1983. Imagined Communities: Texas Press TheOrigin and Spreadof Nationalism. Lon- BarsalouLW. 1992.Frames, concepts, and con- don: Verso ceptualfields. See Lehrer& Kittay1992, pp. BakhtinMM. 1986. Speechgenres. In Speech 21-74 284 DiMAGGIO

BergerJ,Rosenholtz SJ, Zelditch M. 1980.Sta- ture:Emerging Theoretical Perspectives, ed. tusorganizing processes. Annu. Rev. Sociol. D Crane,pp. 117-42.Cambridge: Blackwell 6:479-508 DouglasM. 1986.HowInstitutions Think. Syra- BergerPL, LuckmanT. 1967. TheSocial Con- cuse,NY: SyracuseUniv. Press structionof Reality. Garden City, NY: Dou- DouglasM. 1966.Purity and Danger London: bledayAnchor Routledge& KeganPaul BernsteinB. 1975. Social class, languageand EmirbayerM, GoodwinJ. 1994. Network anal- socialization.In Class, Codes and Con- ysis,culture, and the problem of agency. Am. trol: TheoreticalStudies Towards a Sociol- J.Sociol. 99:1411-54 ogy of Language,pp. 170-189. New York: EricksonB. 1996. Culture,class and connec- SchockenBooks. 2nd ed. tions.Am. J. Sociol. 102:217-51 Block F. 1990. Politicalchoice and themulti- FantasiaR. 1995. Class consciousnessin cul- ple 'logics' ofcapital. In Structuresof Capi- ture,action, and social organization.Annu. tal: TheSocial Organizationof the Economy, Rev.Sociol. 21:269-87 ed. S Zukin,P DiMaggio,pp. 293-310. New FarrRM, MoscoviciS. 1984.Social Represen- York:Cambridge Univ. Press tations.Cambridge: Cambridge Univ. Press BoltanskiL, Th6venotL. 1991.De la Justifica- FineGA. 1996.Reputational entrepreneurs and tion: Les Economiesde la grandeurParis: thememory of incompetence:melting sup- Gallimard porters,partisan warriors, and images of BourdieuP. 1974.Avenier de classeet causalit6 PresidentHarding. Am. J. Soc. 101:1159- du probable.Revue FranCaise de sociologie 93 15:3-42 Fiske ST, LinvillePW. 1980. What does the BourdieuP. 1984. Distinction:A Social Cri- schemaconcept buy us? Pers.Soc. Psychol. tique of the Judgementof Taste.Transl. R Bull. 6:543-57 Nice. Cambridge:Harvard Univ. Press FleckL. 1979[1935]. GenesisandDevelopment BourdieuP. 1990 [1980]. Structures,habitus, ofa ScientificFact. Chicago: Univ.Chicago practices.In TheLogic ofPractice, pp. 52- Press 65. Stanford:Stanford Univ. Press FligsteinN. 1990. The Transformationof Cor- BrunerJ. 1990. Actsof Meaning. Cambridge: porate Control.Cambridge, MA: Harvard HarvardUniv. Press Univ.Press BuchmannM, EisenerM. 1996.Images of the FreemanL, RomneyAK, FreemanSC. 1987. self. 1900-1992. Pap. presentedat Ann. Cognitivestructure and informantaccuracy. Meet.Am. Sociol. Assoc.,New York Am.Anthropol. 89:310-25 CarleyKM. 1989.The valueof cognitive foun- FriedlandR, AlfordR. 1991. Bringingsociety dationsfor dynamic social theory.J. Math. backin: symbols,practices, and institutional Sociol. 14:171-208 contradictions.In The New Institutionalism CarleyKM. 1991. A theoryof groupstability. in OrganizationalAnalysis, ed. WW Pow- Am.Sociol. Rev. 56:331-54 ell,P DiMaggio,pp. 223-62. Chicago:Univ. CeruloK. 1995. IdentityDesigns: The Sights ChicagoPress and Soundsof a Nation.New Brunswick, NJ: FriedlandR, HechtR. 1996.To Rule Jerusalem. RutgersUniv. Press New York:Cambridge Univ. Press CicourelAV. 1973. CognitiveSociology. New GamsonWA. 1992.Talking Politics. New York: York:Free Press CambridgeUniv. Press Collins R. 1981. On the microfoundationsof GarfinkelH. 1987 [1967]. Studiesof therou- macrosociology.Am. J. Sociol. 86:984-1014 tinegrounds of everydayactivities. In Stud- D'AndradeR. 1995. TheDevelopment of Cog- ies inEthnomethodology, pp. 35-75. Oxford: nitiveAnthropology. New York: Cambridge Polity Univ.Press Gentner,D. 1983. Structure-mapping:a the- Devine PG. 1989. Stereotypesand prejudice: oreticalframework for analogy. Cogn. Sci. theirautomatic and controlledcomponents. 7:155-70 J.Pers. Soc. Psychol.56:5-18 GiddensA. 1984. The Constitutionof Society: DiMaggioP. 1982.Cultural entrepreneurship in Outlineof a Theoryof Structuration. Berke- nineteenth-centuryBoston. Media Cult.Soc. ley: Univ.Calif. Press 4:33-50, 303-21 GierynTF 1998. CulturalCartography of Sci- DiMaggioP, Powell WW. 1991. Introduction. In ence: Episodesof Boundary Work, Sociolog- TheNew Institutionalismin Organizational icallyRendered. Chicago, IL: Univ.Chicago Analysis,ed. WW Powell,P DiMaggio,pp. Press.In press 1-38. Chicago:Univ. Chicago Press GilbertDT. 1991.How mentalsystems believe. Dobbin F. 1994. Culturalmodels of organiza- Am.Psychol. 46:107-19 tion: the social constructionof rationalor- Gould R. 1995. InsurgentIdentities: Class, ganizingprinciples. In TheSociology of Cul- Commmunityand Protestin Paris from 1848 CULTURE AND COGNITION 285

to the Commune.Chicago: Univ.Chicago Lang GE, Lang K. 1988. Recognitionand Press renown:the survivalof artisticreputation. GramsciA. 1990.Culture and ideological hege- Am.J. Sociol. 94:79-109 mony.In Cultureand Society: Contemporary LehrerA, KittayEF, eds. 1992. Frames,Fields Debates, ed. J Alexander,S Seidman.New and Contrasts: New Essays in Semantic York:Cambridge Univ. Press and Lexical Organization.Hillsdale, NJ: GriswoldW. 1986.Renaissance Revivals: City LawrenceErlbaum Comedyand RevengeTragedy in theLon- Liebes T, Katz E. 1990. The Exportof Mean- don Theatre,1576-1980. Chicago: Univ. ing: Cross-CulturalReadingsofDallas. New ChicagoPress York:Oxford Univ. Press GriswoldW. 1987.The fabrication of meaning: LindenbergS, Frey BS. 1993. Alternatives, literaryinterpretation in the UnitedStates, frames,and relative prices: a broaderview of GreatBritain, and the WestIndies. Am. J. rationalchoice theory. Acta Sociol. 36:191- Sociol. 92:1077-117 205 GumperzJJ. 1982. Discourse Strategies. Cam- LoftusEF, DondersK, HoffmanHG, Schooler bridge:Cambridge Univ. Press JW.1989. Creatingnew memoriesthat are HackingI. 1995. Rewritingthe Soul: Multi- quicklyaccessed and confidently held. Mem- ple Personalityand theSciences of Memory. oryCognition 17:607-16 Princeton,NJ: Princeton Univ. Press MaierCS. 1988. The UnmasterablePast: His- HarawayD. 1991. The past is the contested tory,Holocaust, and GermanNational Iden- zone.In Simians,Cyborgs, and Women:The tity.Cambridge: Harvard Univ. Press Reinventionof Nature, pp. 21-42. NewYork: MarkusHR, KitayamaS. 1991.Culture and the Routledge self:implications for cognition, emotion and HirschfeldLA. 1994.The child's representation motivation.Psychol. Rev. 98:224-53 of humangroups. Psychol. Learning & Mo- MarkusHR. KitayamaS. 1994. The cultural tivation31:133-85 constructionof self and emotion:implica- Hirschfeld,LA, Gelman,SA. 1994. Towarda tionsfor social behavior. In Emotion and Cul- topographyof mind: an introductionto do- ture:Empirical Studies of Mutual Influence, mainspecificity. In Mappingthe Mind, ed. ed. S Kitayama,HR Markus,pp. 89-130. LA Hirschfeld,SA Gelman,pp. 3-35. New Washington,DC: Am. Psychol.Assoc. York:Cambridge Univ. Press MarkusHR, Kitayama,S, Heiman,R. 1996. HofstedeG. 1980.Culture's Consequences: In- Cultureand 'basic' psychologicalprinciples. ternationalDifferences in Work-RelatedVal- In Social Psychology:Handbook of Basic ues. BeverlyHills: Sage Principles,ed. ET. Higgins,AW Kruglanski. Hogg MA, McGarty C. 1990. Self- New York:Guilford categorizationand social identity.In MartinJ. 1992. Culturesin Organizations: Social IdentityTheory: Constructiveand ThreePerspectives. New York:Oxford Univ. CriticalAdvances, ed. D Abrams,MA Hogg, Press pp. 10-27. New York:Springer-Verlag McCall GJ. 1987. The structure,content, and HowardJA. 1994. A socialcognitive conception dynamicsof self: continuities in thestudy of of social structure.Soc. Psychol.Q. 57:210- role-identities.See Yardley& Honess 1987, 27 pp. 133-45 JeppersonR, SwidlerA. 1994. Whatproper- MertonRK. 1957. Social Theoryand Social ties of cultureshould we measure?Poetics Structure.New York:Free Press 22:359-71 MeyerJW. 1986. Myths of socializationand of JohnsonMK, HastroudiS, Lindsay,DS. 1993. personality.InReconstructing Individualism, Sourcemonitoring. Psychol. Bull. 114:3-28 ed. THeller,S Morton,D Wellberry,pp. 212- KahnemannD, SlovicP, Tversky A. 1982.Judg- 25. Stanford,CA: StanfordUniv. Press mentUnder Uncertainty: Heuristics and Bi- MeyerJW, Jepperson RL. 1996. The actorand ases. New York:Cambridge Univ. Press theother: cultural rationalization and the on- Kluegel J,Smith ER. 1986. Beliefsabout In- goingevolution of modern agency. Ms., Stan- equality:Americans Views of WhatIs and fordUniv. Dept. Sociol. WhatOught To Be. New York:DeGruyter MeyerJW, Rowan B. 1977.Institutionalized or- KrackhardtD. 1987. Cognitivesocial struc- ganizations:formal structure as mythand tures.Soc. Networks9:109-34 ceremony.Am. J. Sociol. 83:340-63 LakoffG, Johnson M. 1980.Metaphors We Live MilburnMA. 1987. Ideologicalself-schemata By.Chicago: Univ.Chicago Press and schematicallyinduced attitude consis- LamontM. 1992. Money,Morals, and Man- tency.J. Exp. Social Psychol.23:383-98 ners: The Cultureof the French and Ameri- MillerDT, PrenticeDA. 1994.Collective errors can Upper-MiddleClass. Chicago,IL: Univ. and errorsabout the collective. Person. Soc. ChicagoPress Psychol.Bull. 20:541-50 286 DiMAGGIO

MillerDT, PrenticeDA. 1996.The construction SchwartzB. 1991. Social changeand collec- of social normsand standards.Handbook of tivememory: the democratization ofGeorge Social Psychology,ed. XX, pp.799-829. XX: Washington.Am. Sociol. Rev. 56:221-36 XX SedikidesC, SkowronskiJJ. 1991. The law of MohrJW. 1994. Soldiers,mothers, tramps and cognitivestructure activation. Psychol. In- others:discourse roles in the 1907 Charity quiry2:169-84 Directory.Poetics 22:327-58 SewellWHJr.1992. Atheoryofstructure: dual- MoscoviciS. 1984.The phenomenonof social ity,agency, and transformation. Am. J. Sociol. representations.In Social Representations, 98: 1-29 ed. RM Farr,S Moscovici,pp. 3-69. New ShivelyJ. 1992. "Cowboysand Indians":per- York:Cambridge Univ. Press ceptionsof westernfilms among American Nadel SF. 1957.A Theoryof Social Structure. Indiansand Anglos. Am. Sociol. Rev. 57:725- London:Cohen & West 34 NeisserU. 1981.John Dean's memory:A case ShwederRA.1982. Fact and artifact in trait per- study.Cognition 9:1-22 ception:the systematic distortion hypothesis. NelsonK, GruendelJ. 1981. Generalized event Progressin Exp. Personal.Res. 2:65-100 representations:basic buildingblocks of ShwederRA, BourneEJ. 1991. Does thecon- cognitivedevelopment. In Advancesin De- ceptof theperson vary cross-culturally? In velopmentalPsychology, ed. ME Lamb,AL ThinkingThrough Cultures: Explorations in Brown,pp. 131-58.Hillsdale, NJ: Lawrence CulturalPsychology, ed. R. Shweder,pp. Erlbaum 113-56.Cambridge: Harvard Univ. Press Noelle-NeumannE. 1993 [1980]. TheSpiral of SnidermanPM, Piazza T. 1993. The Scar of Silence: Public Opinion-Our Social Skin. Race. Cambridge,MA: HarvardUniv. Press Chicago: Univ.Chicago Press Snow DA, Benford,RD. 1992. Masterframes OffeC. 1985. DisorganizedCapitalism. Cam- and cycles of protest.In Frontiersin So- bridge,MA: MIT Press cial MovementTheory, ed. AD Morris,CM OrrSW. 1995.Language, values and social net- Mueller,pp. 133-55. New Haven,CT: Yale works:on meaningand the micro-macro link Univ.Press in structuralsociology. Ms., Centerfor the SorokinPA. 1957 [1927]. Social and Cultural Social Sciences,Columbia Univ. Mobility.New York:Free Press OrruM. 1991.The institutionallogic of small- SperberD. 1985. Anthropologyand psychol- firmeconomies in Italyand Taiwan.Stud. ogy:towards an epidemiologyof representa- ComparInt. Dev. 26:3-28 tions.Man 20:73-89 ResnickLB, Levine JM, Teasley SD. 1991. Stark D. 1990. La valeur du travailet sa Perspectiveson SociallyShared Cognition. retributionenHongrie. Actes de la Recherche Washington,DC: Am. Psychol.Assoc. en SciencesSociales 85:3-19 RogoffB, ChavajayP. 1995.What's become of StarrP. 1992. Social categoriesand claimsin researchon thecultural basis of cognitive de- theliberal state. Soc. Res. 59:263-95 velopment?Am. Psychol. 50:859-77 StrykerS. 1987.Identity theory: developments RoschE. 1978. Principlesof categorization.In and extensions.See Yardly& Honess 1987, Cognitionand Categorization,ed. E Rosch, pp. 89-104 B Lloyd,pp. 27-48. Hillsdale,NJ: Erlbaum SwidlerA. 1986. Culturein action: symbols Ross J. 1992. Semanticcontagion. See Lehrer andstrategies. Am. Sociol. Rev. 51:273-86 & Kittay1992, pp. 143-69 SwidlerA. 1997.Talk ofLove: HowAmericans Salomon G, ed. 1993. DistributedCognition: Use TheirCulture. Chicago: Univ.Chicago Psychologicaland EducationalConsidera- Press tions.New York:Cambridge Univ. Press SwidlerA, ArditiJ. 1994. The new sociology SchneiderDJ. 1991. Social cognition.Annu. of knowledge.Annu. Rev. Sociol. 20:305- Rev.Psychol. 42:527-61 29 SchudsonM. 1989. How cultureworks: per- TajfelH. 1981.Human Groups and Social Cat- spectivesfrom media studies on theefficacy egories: Studiesin Social Psychology.New ofsymbols. Theory Soc. 18:153-80 York:Cambridge Univ. Press Schudson M. 1992. Watergatein American TajfelH, TurnerJC. 1986. The social identity Memory:How WeRemember, Forget, and theoryof intergroup behavior. In Psychology Reconstructthe Past. New York:Basic ofIntergroup Relations, ed. S Worchel,WG SchumanH, ScottJ. 1989. Generations and col- Austin,pp. 7-24. Chicago:Nelson-Hall. 2nd lectivememories. Am. Sociol. Rev.54:359- ed. 81 TillyC. 1992.How todetect, describe, and ex- SchwartzB. 1981. VerticalClassification: A plain repertoiresqf contention. Working Pa- Studyin Structuralismand theSociology of per No. 150. Cent.Stud. Soc. Change.New Knowledge.Chicago: Univ.Chicago Press Sch. forSoc. Res. CULTURE AND COGNITION 287

TourangeauR, Sternberg RJ. 1982. Understand- ceptionof man.Am. Sociol. Rev. 26:184- ing and appreciatingmetaphors. Cognition 93 11:203-44 WuthnowR. 1987.Meaning and Moral Order: VaughanD. 1986. Uncoupling:Turning Points Explorationsin CulturalAnalysis. Berkeley: inIntimate Relationships. New York: Oxford Univ.Calif. Press Univ.Press WuthnowR. 1989. Communitiesof Discourse: VonHippel E, JonidesJ, Hilton JL, Narayan S. Ideologyand Social Structurein theRefor- 1993.The inhibitoryeffect of schematic pro- mation,the Enlightenment and European So- cessingon perceptualencoding. J. Personal. cialism.Cambridge: Harvard Univ. Press Soc. Psych.64:921-35 YardleyK, HonessT, eds. 1987.Self and Iden- WarnerWL. 1959. The Livingand theDead: tity:Psychosocial Perspectives. New York: A Studyof theSymbolic Life of Americans. JohnWiley New Haven: Yale Univ.Press ZelizerV. 1989.The social meaningof money: WhiteHC. 1992.Identity and Control:A Struc- 'specialmonies.' Am. J. Sociol. 95:342-77 turalTheory of Social Action.Princeton, NJ: ZerubavelE. 1991. The Fine Line. New York: PrincetonUniv. Press FreePress White HC. 1995. Networkswitchings and ZerubavelE. 1992.Terra Cognita: The Mental Bayesianforks: reconstructing the social and DiscoveryoqfAmerica. New Brunswick,NJ: behavioralsciences. Soc. Res. 62:1-28 RutgersUniv. Press WileyMG, Alexander,CN. 1987. Fromsitu- ZerubavelE. 1997.Social Mindscapes:An In- ated activityto self attribution:the impact vitationto CognitiveSociology. Cambridge: of social structuralschemata. See Yardley& HarvardUniv. Press Honess 1987,pp. 105-17 ZerubavelY. 1994. RecoveredRoots: Collec- WillisP. 1977. Learningto Labor New York: tiveMemory and theMaking of Israeli Na- ColumbiaUniv. Press tional Tradition.Chicago: Univ. Chicago Wrong DH. 1961. The oversocializedcon- Press