1 Representing Race: Universal and Comparative Perspectives Both universalist and comparativist accounts have provided fundamental insights into the way racial concepts emerge and are sustained over historical time and cultural space. In this chapter I pursue the ways in which these accounts diverge and converge, particularly with respect to the assumptions they make. On the Notion of Human Kinds Like most things in the world, humans vary one to the other. As with other variation, we take some but not all of this diversity to be meaningful. Relevant to theories of everyday reasoning in both anthropology and psychology is variation diagnostic of the types of humans there are---what might be called human kinds. People everywhere and at all times appear to recognize human-kind variation. The notion of kind is of course a vague one, admitting all sorts of stable differences. There are tall (kinds of) people, grumpy (kinds of) people, and Italian (kinds of) people. It is uncontroversial that all humans recognize human kinds of the tall sort. It is likely, although disputed, that all humans recognize human kinds of the grumpy (or nervous, or generous) sort.1 In this chapter I am concerned with the third sort of human kind: Italian (kinds of) people, or kinds that specify that an individual is funda mentally a certain sort of person. The notion of human kindhood that interests me here is not a construction of modern science or of the centralized state; rather, it is the familiar and eminently commonsense notion of what at times has been called a "primordial group." I mean by this any diffuse group of individuals that a person recognizes as being like him or her in some fundamental and enduring way but whose similarity is not historically traceable to local events (e.g., immediate relations of kinship). Such groups are not organized around specific properties or states of affairs (e.g., a common activity or instrumental function) or shared sentiment (e.g., a common political or moral perspective). Rather, the level at which this cut is made is determined by a commonsense partitive logic or social ontology that picks out the "natural" kinds of people that exist in the world. People of this kind are relevant because they are relevant to what it is like to be you and yours. Members of this kind are not all known to you, nor are they all directly or biographically related---in short, they are neither kinsfolk nor what Geertz (1973), following Schutz, called consocietes. They are people whose identity turns on a higher-than-local level of inclusion. In order to avoid loading this notion unreasonably, for the moment I will call borrow Goldberg's (1993) term ethnorace. How universal is the awareness of ethnoraces?3 How universal are Italian kinds of people? Clearly, if we intend by this the recognition of a human kind based on citizenship in a nation state, the notion is fairly modern. However, if we mean a recognition that people perceive themselves as belonging to "deeply" constituted and primordial populations, then the notion may be less tied to a particular historical period. Members of all human societies appear to be capable of discriminating between the members of their own and other ethnoraces, because all humans apparently display a tendency to favor members of their own ethnorace over others (LeVine and Campbell 1972). More important, people the world over appear to discriminate between members of their own and other ethnoraces in a unique way: they naturalize the difference. "Naturalization" here involves the practice of conceptually identifying social differences with natural ones. Evidence that the social world is naturalized is both varied and widespread. The notion of primordial groups is closely linked to relations of power, and other indigenous names reflect this: many tribal appellations are pejoratives, etymologically derived from a neighboring and dominant population's term for "beast" (Dole 1967). Nor is this sort of naming solely the practice of non-Western and localized populations. Initial contact between Europeans and the newly encountered peoples of Africa and the New World provoked speculation about cultural and racial variation; it also provoked conjecture among Europeans that non-Europeans were sub human---i.e., naturally different from Europeans. Both scientific and popular writers of the sixteenth century debated whether Africans (and the Irish) represented an intermediate population between ape and man (read "real Europeans") in the "great chain of being" (Banton 1987; van den Boogaart 1980; Curtin 1964; Mosse 1978). Finally, there was considerable discussion as to whether the indigenous populations of the New World had redeemable souls and accordingly could be considered fully human (Guillaumin 1980). On its face, this seems compelling evidence of a premodern naturalized notion of human kinds. Yet there is reason to be skeptical. Despite the fact that many of these naming and reasoning practices involve premodern populations (i.e., traditional societies or European populations in premodern periods), it is not clear that the practices themselves are truly premodern. The relatively remote highland Sumatran village in which I did fieldwork has many hallmarks of a traditional society--- kinship-based social organization, swidden agriculture, aboriginal religious system, etc. It also has motorcycles and plastic buckets and is serviced by a modern postal system. Generally the meanings and practices associated with the motorcycles, the plastic buckets, and the modern postal system assume a local and distinct flavor, so that the behaviors and beliefs attendant on these things are clearly influenced by indigenous culture. Still, they are decidedly not of traditional culture. For the present argument, the relevant "plastic bucket" may well be the notion of tribe itself, the unit named with this naturalizing vocabulary. Fried (1975) has argued, and many now agree, that the classical notion of a tribe as a politically and culturally bounded unit is modern, and that, hence, many of these pejorative (and "traditional") linguistic practices may be rooted in the political economy of colonialism.4 The concept of race has been incorporated---rather awkwardly--- into this perspective by treating it as one individuating trait or attribute among many. Race is simply one of the salient dimensions along which an individual might form an image of himself or of another individual. As such, the notion is integrated into the cognitive depiction of individuals rather than groups. Group membership is conceived as an aspect of an individual's identity---as a part of the repertoire of charac teristics that make a particular individual distinct---rather than as an aspect of an individual's beliefs about collectivities per se (for exceptions see Brewer 1988 and Hewstone et al. 1991). Even when category-based processes are stressed (see, e.g., Taylor et al. 1978; Fiske and Neu-berg 1990), the emphasis is on information that allows the perceiver to designate an individual's category membership as an attribute of that person (rather than the person becoming a marker for the category). Race becomes a quality of a person, not a means for defining major human groupings and their interrelationships. In view of this tradition, and despite the important role that categories targeting collectivities play in the regulation and organization of social life, it is not surprising that psychology has told us relatively little about how human groups, as opposed to individuals, are mentally represented (Hamilton 1981; Steiner 1974). What we do know raises doubt about conventional wisdom with regard to the relationship between individual and group identity processes. For example, there is reason to believe that self-identity and other- identity are fundamentally distinct cognitive tasks (which is not implausible, since reasoning about the self and about the other is fundamentally distinct, as work on the fundamental attribution error demonstrates).7 Cross (1985, 1991) has convincingly argued that reference group identity and personal identity involve quite separate mechanisms. For example, minority children's preference for representations of majority identity were long interpreted as evidence of black self- hatred. Cross argued that this makes little sense, if for no other reason than that black children do not display evidence of the personal identity pathology (e.g., low self-esteem) that pervasive self-hatred would entail. Children's preferences for representations of majority identity reflect, Cross suggests, issues having to do with reference-group identity, not personal identity. Previous work that conflates these two distinct cognitive spheres misinterprets the domain in which identity conflict and tension occur. Other research suggests that information about groups is actually stored in memory independent of information about individual group members, particularly traits associated with individuals (Wyer and Martin 1986; see also Brewer 1988 and Pryor and Ostom 1981). Milner (1984) suggests that different kinds of learning are involved in acquiring knowledge of individuals and knowledge of groups. My work indicates that people distinguish between social properties (i.e., attributes that are true of an actor) and social kinds (i.e., the sorts of actors there are) (Hirschfeld 1988). In construing racial-group identity principally as an element of an individual's personal identity, standard views of race in social psychology shed little light on why some group identities are more important and bear more causal weight than other identities. Different kinds of individuals are differentially salient, depending both on how salience is construed and on the context in which it is played out. Thus, some human kinds are considerably more important than other kinds, both structurally (in terms of economic, political, and cultural salience) and psychologically (in terms of memorability, attention, and reasoning). Despite the importance in American culture of reasoning about personality traits and/or dispositions, it is widely accepted that, under certain conditions, we are much more likely to remember an individual's racial identity and more likely to use it as the basis for explaining that individual's behavior than his or her personality. In particular, race explains why some people do what they do in ways that personality does not. The reason is that race is conceived as a more natural way to group individuals than personality type is. What conditions produce this effect? The answers are straightforward: race becomes central in contexts in which membership in a group is highlighted. Virtually by definition it is minority race that is highlighted. Race is not an attribute of an individual; it is the attribute of a minority individual. "Majority race," while strictly speaking not an oxymoron, captures a transparent event---something so ordinary as not to be noticed. Minority status, in contrast, is noteworthy. And minority status is defined by the relationship between groups, not individuals. This is not a novel observation---several studies have found that children acquire knowledge about minority races before knowledge of majority races (Radke et al. 1949; Proshansky 1966). Because most theorizing in social psychology is about commonsense attributions within groups, not across them, psychologists tend not to grasp what this evidence implies: race cannot be about individuals; it has to be about groups. The pervasive tendency to opt for personality-trait and/or disposition explanations in this context (at least among members of Western secular cultures) is well documented. But this sort of mentalistic explanation tends to overemphasize trait-based and disposition- based explanations of behavior and to obscure the societal and corporate aspects of identity that anthropological accounts stress. When used to characterize race, this theory (really an ethnotheory) about the masterful, bounded, and individuating self paradoxically becomes a theory (or ethnotheory) of society that ignores and effaces the corporate and collective nature of social thought, particularly the role played by social-category identity. Psychology and the Reality of Racial Categories To summarize: According to a range of psychological theories, racial thinking is a by-product of the human propensity to construct categories in the face of any identifiable difference (Fiske and Taylor 1991; Hamilton and Trolier 1986; Ashmore and Del Boca 1981; Allport 1954). Under one interpretation---and, as was noted above, this is a dominant interpretation in the field---there is nothing unique to racial differences themselves (in comparison to, say, gender, disability, or other physical- ized dimensions). Strictly speaking, however, the perspective adopted by virtually all psychologists entails that racial differences are distinctive in at least one respect: they are real. That is, with strikingly few exceptions (Betancourt and Lopez 1993; Jones 1991; Zuckerman 1990), psychologists are inclined to construe race not only as concrete but as a real part of the natural environment. According to the bottom-up model that animates social-category theorizing, a conceptual representation of difference is consistent and coherent to the extent that the difference encountered is consistent and coherent. Most psychologists, certainly, treat racial categories as both consistent and coherent. Actually, the evidence that most psychologists accept the realist interpretation is more persuasive, albeit indirect, than I suggest. Consider how putative racial differences in mental ability have boon treated. The fact that differences in performance are regularly reported and interpreted in terms of racial-group affiliation implies a commitment to the realist interpretation. As intuitive as such comparisons often seem, recall that if racial groupings are arbitrary and contrived it is bizarre to expect racial comparisons in performance to yield interesting results. Clearly those who propose that intellectual potential is biologically linked to race---i.e., who believe that race somehow causes differences in intellectual potential---necessarily have adopted a realist view. What is more striking is that the bulk of the work challenging the claim that race biologically regulates mental ability also proceeds from the same realist assumption. Critics of the racial- intelligence claim have approached the problem largely by seeking alternative explanations for differences in test performance, arguing that previous researchers have taken too little account of bias in testing and inequities in educational opportunity (see, e.g., Klineberg 1975). In many respects, this is a curious strategy. A more direct attack would involve questioning whether race could regulate the biological distribution of psychological traits like intelligence (leaving aside whether intelligence is a sensible concept in itself). Obviously, if race lacks biological coherence it could hardly control the inheritance of any property not specifically tied to the category's definition.8 As straightforward as this seems, the rejection of the racialist claim in psychology has never included a sustained challenge to the notion of racial classification itself (Guillaumin 1980). That this challenge did not emerge earlier in the twentieth century, when the debate initially surfaced, is perhaps not surprising. Remarkably, however, the lack of any questioning of the realist claim is no less evident today than it was five decades ago. For instance, in the flurry of attention surrounding the publication of Herrnstein and Murray's (1994) rehearsal of the old canard about race and intelligence, virtually no psychologist has publicly suggested that their claims about a racial-ized or biologically cognitive underclass are implausible in the extreme. The one collaborative and putatively "mainstream" appraisal of current wisdom in psychometrics about race and intelligence, signed by 52 leading and not-so-leading scholars, never doubts that racial comparisons are a sensible way to explore biological differences (Arvey et al. 1994).9 On the Historical Specificity of Race In view of the foregoing, it is hardly astounding that nonpsychologists have generally found psychological accounts of race lacking. Consider the following comments, one by an anthropologist and two by historians.10 [Race cannot be explained] by reference to psychological processes that we speculate may be taking place within individuals. (Smedley 1993, p. 24) Subliminal and deeply rooted psychological factors were undoubtedly present [in early forms of racial thinking in North America], but they can hardly explain the extent to which racial feeling and ideology have been developing and changing, subject to situational variations in intensity and character. (Fredrickson 1988, p. 205) Race is neither the reflex of primordial attitudes nor a tragically recurring central theme . . . [but is] the ideological medium through which Americans confronted questions of sovereignty and power. (Fields 1982, p. 168) Over the past three decades, a broad consensus has emerged (out side of psychology) that race is a historically and culturally specific notion, embedded in a constellation of economic, political, and cultural discourses and uniquely linked to specific relations of power and authority. Further, race is widely believed to be a modern concept, traceable to the Enlightenment impulse to classify the natural world and naturalize the human one. This does not mean that historically race was an abruptly new concept. Racial thinking has antecedents in earlier ideological forms. It is even acknowledged that underlying varied forms of racial thinking may be a stable core concept, an enabling frame of reference (Goldberg 1993). What is widely rejected (outside of psychology) is the idea that this underlying frame of reference is interestingly mediated by mental processes. The argument runs like this: (i) Race is about relations of power. (ii) Power is about aggregate structural (e.g., political, economic, or cultural) relations, not mental ones. (iii) Structural relations are mental only incidentally, in that domination and subordination as lived practices minimally have to be both represented and representable (typically in some hegemonic guise). (iv) Hence, a society's race concept is shaped by actually existing race relations, not vice versa. Although there is agreement about these four points, it is important not to overstate the unanimity among comparativists. Scholars have overwhelmingly converged on a metastrategy for studying race, namely that it should be grounded in particular historical and cultural contexts. There is considerably less accord about which cultural and historical contexts played the most important role in shaping the modern concept. Consider the following range of influential opinions: * Gossett (1963, p. 3), citing ancient Biblical, South Asian, Chinese, and Egyptian descriptions of "the tendency to seize upon physical differences as the badge of innate mental and tem peramental differences," argues that the race concept can be traced to widely distributed premodern xenophobic beliefs. * Banton (1978, 1987), among others, grounds the concept in the sixteenth century, in originary systems of natural classification (Ray, Linneaeus, Buffon, et al.) and in French historical accounts that interpreted intra-European historical rivalries (e.g., the Franks vs. the Gauls or the Teutons vs. the Latins) in racial terms.11 * Anderson (1983) derives race from eighteenth-century European class relations, specifically the need of aristocracies to legitimate their rule in virtue of an inborn superiority. * Jordan (1968) and van den Boogaart (1980) claim that racial consciousness is a discovery of the British Age of Exploration. The notion of race, on this view, developed out of sixteenth- and seventeenth-century overseas encounters between Northern Europeans and peoples of African descent as "one of the fairest- skinned nations suddenly came face to face with one of the dark est peoples on earth" (Jordan 1968, p. 6). * Takaki (1992) contends that a nonracial image of (educable and redeemable) savagery became racialized (i.e., invested with the notion of an innate, unchanging, and physicalized savage character) in the convergence of competition between settlers and Indians over land and the religious demonization of the indigenous population by colonists in the seventeenth-century Massachusetts Bay Colony. * Morgan (1975, p. 328 ) traces the notion to colonial Virginia and the separation of "dangerous free whites from dangerous slave blacks by a screen of racial contempt." * Harris (1964) and Fields (1990) argue that race consciousness is a New World, eighteenth-century invention that arose from the need to justify and explain slavery. * Guillaumin (1980) and Smedley (1993) see nice as a later- emerging concept, directly bound up with the social elaboration of nineteenth-century science (particularly biology). Why is there such controversy about chronology and site in the advent of racial consciousness? In significant measure, the diversity of opinion can be traced to three problems. All three have to do with the way the race concept is embedded in systems of cultural and political belief and with the complexity of the concept itself. First, as I have already noted, race was not a de novo discovery but a subtle modulation of preexisting beliefs. Second, race is often identified with the use to which it is put, and in view of the range of plausible functions that race and race-like cognitions have served it is unsurprising that scholars have had difficulty linking the notion to a single type of instrumentality (Hall 1980). Third, most traditional approaches have equated the race concept (or race theory) with racial thinking (or racial consciousness). To the contrary, I will argue that the race concept under-determines racial thinking, and that racial thinking can be explained only by reference to the instantiation of the race concept in a particular cultural environment. Racial Thinking and Racial Theories Although racial thinking clearly figures in the interpretation, the representation, and the explanation of social conflict, it may not always be useful to infuse the race concept with the uses to which it is put. For one thing, the concept does not always change with perturbations in the uses to which it is put. If conditions of political, economic, and cultural existence fundamentally shape racial ideology, why does racial thinking often remain unchanged when there are shifts in the underlying material basis and cultural context of the theory? Indeed, there is ample evidence for a lack of contingency between systems of racial thought and conditions of political and economic existence. Fields (1982, p. 153) provides a prime example. She argues that racial ideology is rooted in the historically specific conditions of slavery in the British North American colonies. Yet, contrary to what one would expect, she finds that racial ideology changed little after slavery was abolished. "There would be no great problem," she writes, "if, when things changed, the vocabulary died away as well. But far the more common situation in the history of ideologies is that instead of dying, the same vocabulary attaches itself, unnoticed, to new things. It is not that ideas have a life of their own, but rather that they have a boundless facility for usurping the lives of men and women." Where does this "boundless facility" come from? Fields, following in a long tradition of interpretive scholars, is sure that it is not the result of psychological process acting on history "from outside" (ibid., p. 149). By now it should be evident that it requires considerable effort not to conceive of this boundless facility in psychological terms. Everyday thinking about race engages a complex of elements: a theory of intrinsic difference, a system for classifying such differences, and the instrumental purposes to which this knowledge is put. Relative to other systems of racial thinking---and relative to historical changes in any given system---each system of racial classification focuses on a particular set of characteristics, picks out a peculiar set of individuals, and fulfills a specific set of goals. There is no a priori reason to imagine that these throe elements are contingent. The idea of race---the notion that there are intrinsic differences between groups of humans---may well be independent of any specific system of racial thinking (Appiah 1990a; Goldberg 1993). Different scholars, adopting differing perspectives, have empha sized contradictory aspects of racial thinking, sometimes laying stress on the ways in which it is concrete, referential, and bounded (and thus capable of regulating a broad range of political, economic, and other power relations) and at other times drawing attention to the ways in which it is "fluid" (Goldberg 1993), "promiscuous" (Fields 1982), "protean" (Staler 1992), and "uncertain" (Omi and Winant 1994). That race is at once concrete and immutable yet fluid and changing is paradoxi cal only if racial thinking and the race concept are taken to be the same thing. I suggest that they are not. The race concept is a property of mentation. Racial thinking, in contrast, is an aspect of cultural formations, derived in part from aggregate relations of power. Peoples the world over hold a very similar race concept that includes beliefs about pri mordial identities. This concept is embedded in a commonsense theory that holds that humans can be partitioned into enduring types on the basis of highly correlated clusters of naturally grounded properties. This theory, in turn, figures in widely varying systems of racial referencing that pick out, name, and coordinate relations among these supposedly distinct human types.12 The idea that racial discourse can be characterized in terms of theory is of course not new; a number of scholars have proposed it. Typically, however, little effort has been made to distinguish between the theoretical basis for racial thinking and racial thinking itself. Indeed, the two are usually taken to be one and the same thing. For example, in Banton's (1987) influential book Racial Theories the theories of the title are part of formal public discourse not individual mental life. These explicit and systematic theories are often construed as embedded within scientific theories of biology. Banton's analytic strategy involves describing and examining the implications of changes in racial theories over time and space. This means analyzing patterns of social relations as a way of explaining the racial theory. Omi and Winant (1994, p. 11) make much the same argument: "Racial theory is shaped by actually existing race relations in any given historical period. Within any given historical period, a particular racial theory is dominant--- despite often high levels of contestation. The dominant racial theory provides society with 'common sense' about race, and with categories for the identification of individuals and groups in racial terms." To the contrary, I will suggest that there is a set of abstract principles that constrain the construction of particular theories of race (i.e., the conceptual basis for common sense) as distinct from the systems of racial referencing themselves (i.e., the systems of racial partitioning that have achieved cultural currency in a particular time and place). Further, I suggest that there is no reason to believe that the abstract principles derive from the particular theories. Indeed, it is plausible that the for mer not only constrain the latter but also constrain the uses to which the latter are put. I will try to show in the next chapter that compelling evidence for this position comes, ironically perhaps, from the findings meant to challenge such universalist speculation.