Cognitive Dimensions of Social Science:
The Way We Think About Politics, Economics, Law, and Society

© Mark Turner 2000.

Draft. Please do not quote without checking the published version

To be published by Oxford University Press, New York

"In extending the idea of 'cognitive studies' beyond psychology, artificial intelligence, information processing and linguistics to the social sciences as a whole, Turner has both deepened that idea and set it free of scientistic rigidities. His discussions of 'conceptual blending,' of 'the descent of meaning,' of analogy and metaphor, and of choice provide a powerful new framework for work in anthropology, literature, and rhetoric on the one hand and politics, law, and economics on the other. 'The second cognitive revolution,' the one concerned with meaning and understanding, seems at last at hand." –Clifford Geertz, Institute for Advanced Study.

"A major frontier of the social sciences is to integrate cognitive science with social science. Mark Turner's pioneering study is an imaginative contribution which will, I believe, force social scientists to turn their attention to this frontier." –Douglass C. North, winner of the 1993 Nobel Memorial Prize in Economic Science.



1. Deep Play

2. Reason

3. Choice

4. Analogy

5. Descent of Meaning

6. Cognitive Social Science







I am grateful to the faculty, members, and visitors of the School of Social Science at the Institute for Advanced Study during 1996-1997, where this book was conceived and formed. I thank Agnes Gund, Trustee of the Institute for Advanced Study and President of the Museum of Modern Art, for her financial support, and the University of Maryland, for assigning me to research leave.

This book is based on theories that are the result of my collaboration with Gilles Fauconnier, who deserves an exceptionally public acknowledgment of his influence.

Of the many scholars who have responded to drafts of parts of this book or who have provided me with superior opportunities to present this work publicly, I especially thank Aaron Belkin, Henry Brady, Per Aage Brandt, Bruce Bueno de Mesquita, David Collier, Finn Collin, Seana Coulson, Gary Cox, Gerald Edelman, James Fearon, Monika Fludernik, Dedre Gentner, Samuel Glucksberg, Joseph Goguen, Adele Goldberg, Joseph Grady, Richard Grush, Paul Hernadi, Douglas Hofstadter, Daniel Kahneman, Kristian Ditlev Jensen, Mark Johnson, George Lakoff, Arthur Lupia, Mat McCubbins, Nili Mandelblit, Harold Morowitz, Chrystopher Nehaniv, Todd Oakley, Andrew Ortony, Samuel Popkin, Alan Richardson, Tim Rohrer, John Robert Ross, Ellen Spolsky, Francis Steen, Eve Sweetser, Sarah Taub, Philip E. Tetlock, Francis-Noël Thomas, and Lisa Zunshine.

Parts of this book draw on earlier talks and articles. Chapter two draws on a talk at the MacArthur conference on counterfactual thought experiments in political science in 1995, on an article in Philip E. Tetlock and Aaron Belkin, editors, Counterfactual Thought Experiments in World Politics (Princeton University Press, 1996), and on parts of an article by Turner and Fauconnier in John-Pierre Koenig, editor, Conceptual Structure, Discourse, and Language II (Stanford: Center for the Study of Language and Information, 1998). Chapter three draws on a talk at the Colloquium on Political Reasoning at the University of California, San Diego, in 1996 and a talk at the annual meeting of the American Political Science Association in 1997. Chapters two and three draw on an article in Arthur Lupia, Mathew D. McCubbins, and Samuel L. Popkin, editors, Elements of Reason: Cognition, Choice, and the Bounds of Rationality (Cambridge University Press, 2000). Chapter four draws on talks I gave at the Institute for Advanced Study in 1996, at the National Election Survey Conference on Cognition, Emotions, and Communication at the University of California, San Diego in 1997, at the annual meeting of the American Political Science Association in 1997, and at the Conference on Computation of Metaphor, Analogy, and Agents at the University of Aizu in Japan in 1998. Chapter five draws on a seminar paper presented at the School of Social Science at the Institute for Advanced Study in 1997, parts of which were published in Mark Turner and Gilles Fauconnier, "A Mechanism of Creativity," Poetics Today 20: 3 (1999), pages 397-418, and also on a talk at a symposium on Body, Mind, and Brain at the University of Oregon Institute for Cognitive and Decision Sciences in 1997.

Excerpts from Clifford Geertz, After the Fact: Two Countries, Four Decades, One Anthropologist (Cambridge: Harvard University Press) are copyright © 1995 by the President and Fellows of Harvard College and are reprinted by permission of the publishers.

Excerpts from Clifford Geertz, "Deep Play: Notes on the Balinese Cockfight" are reprinted by permission of Dædalus, Journal of the American Academy of Arts and Sciences, from the issue entitled, "Myth, Symbol, and Culture," Winter 1972, volume 101, number 1.

Chapter 1: Deep Play

The Institute

Early in April of 1996, my wife and I arrived, curious and invisible, at a research institute we intended, as prospective residents, to study. A small place, about two hundred people, and relatively remote, it was its own world. We were intruders, professional ones, uninvited and unannounced, but also practically unnoticed, since the Institute for Advanced Study's annual purge of most of its population and replenishment with fresh recruits makes it a gathering of interchangeable anonymities. Their status, the only one necessary, is that they are "at the Institute." To all appearances, we were at the Institute, too, where outsiders, to a comfortable degree, become insiders exactly by being there.

Uninvited visitors have no place in this world, so there are few signs to direct them, but the pattern of the Institute buildings is conventional, and the receptionist, conditioned to look right through anything resembling an absent-minded professor, dealt with us as though we were not there. Nobody greeted us, but nobody scowled or said anything unpleasant to us either, and that was fine.

We located immediately the common room, with its wooden racks of newspapers and periodicals, which in other circumstances would have distracted us for hours; the mathematics library, with its high windows, spiritual and restful, where, it turned out, I would pass week after week reading by the natural light; the glass-and-concrete dining hall, where a bust of Einstein impassively oversaw the discreet promotional sale of sweatshirts and T-shirts, each carrying an image of a full-frontal naked Truth heraldically matched by a diaphanously veiled but no less anatomically emphatic full-frontal Beauty; the sloping lawns; the serene, kidney-shaped pond; and the five-hundred-acre wood through which our own two Christopher Robins would later pursue the mallard ducks, the Canada geese, the herd of deer, the legend of the baby black bear, and–the chief attraction, aside from the bow hunters who thinned the herd–the April eruption of frogs, toads, and salamanders.

We found our goal, the Institute preschool, and handed over a check to enroll our three-year-old son in the next term's morning program. My first duty on arriving in September would be to hire someone–she turned out to be a young woman fresh from St. Petersburg, Russia–to come to our residence in the mornings to care for our one-year-old son while my wife, a writer of fiction for young adults, plied her careful art, but my tasks for the day were completed. We drove past the playground, between Einstein Drive on one side and von Neumann Drive on the other, and I nearly ran the car into the curb as we gaped at the apartments. The elegance of the Institute buildings, the pleasure of the woods, and the perfection of the grounds had left us aesthetically unprepared for their full-frontal presentation of ugly. (Actually, I would later come to view them as the appropriate neutral and functional background for work at the Institute, and regret to hear that they were to be gutted.) Before we left that afternoon, it had begun to snow–on us, on the Institute, and on the amphibians.

The School of Social Science in the Institute for Advanced Study had announced its intentions for 1996-1997 in a call for applications:

In 1996-1997 the School will be celebrating its twenty-fifth year. Over these years the School has been associated with the development of "interpretive social science" (the attempt to supplement models of natural science with explanations for social change drawn from humanities disciplines such as history, literature, and philosophy). In an effort both to review our past and anticipate our future, we will be looking for projects that exemplify the best of existing interpretive approaches to the social sciences, or that point the way to new kinds of social scientific interpretation, or that assess the strengths and weaknesses of "interpretive social science." Our interest is in the application of interpretive approaches to substantive issues and examples (with an awareness, of course, of the methods being employed), not in the elaboration of abstract theoretical proposals. We will also welcome proposals that critically examine the history of the social sciences during the past twenty-five years.

My own work consists of trying to make sense of acts of meaning and, especially, of trying to explain the mental abilities possessed by cognitively modern human beings that make those acts of meaning possible. "Modern" in this context means roughly the last fifty thousand years. My method consists of deploying any research instrument that seems promising. My hobby-horse preoccupation is Erving Goffman's "What is going on here?" So I guessed that I was a logical candidate for the School, and it turned out that I was right.

Around me that year were other squirrels working on other nuts: civil society in Argentina, Mexico, Peru, and Cuba from 1780 to 1880; the social dimension of laboratory design; boredom in Germany in the nineteenth century; the writing of ethnography on Central America; reform in Morocco; the history of the history of science in the United States; whistleblowing in organizations; Mayan public intellectuals; theocratic thought in China from 1885 to 1924; the social earmarking of money; a biography of the Tobacco Mosaic Virus; segregation in Yonkers; how immigrants in the banlieue of Paris justify their racism; aesthetic constraints on the development of physics; morality internal to capitalism; and one last project whose bogey was the representation of AIDS but whose principled ambition was to escape being labeled or understood.

A conference on "25 Years of Social Science," to be sponsored by the School and held in the Institute's absolutely gorgeous Wolfensohn Hall, was scheduled for May 1997. The announcement of the conference offered, as its grand finale, a bracing swash of impossibly broad questions about the future of social science, questions which the conference participants–no wonder–found difficult to address, much less to answer.

I have written this book in an attempt to answer those questions: where is social science? where should it go? how should it get there? My answer, in a nutshell, is that social science is headed for an alliance with cognitive science. In the chapters of this book I investigate what "cognitive social science" might look like. Here, in this chapter, I take up questions of interpretive social science. In subsequent chapters, I take up central themes of qualitative social science. Each chapter offers pictures of the kind of research we might expect to see if we supplemented the kind of research done by social scientists with the kind of research done by their cousins in the cognitive sciences. I conclude the book with a view of some prospects for cognitive social science.

My story begins with the Institute for Advanced Study's announcement of its conference on "25 Years of Social Science":

Our invited speakers are men and women who have sustained an interest in the larger society while working successfully in their own disciplines. As former members, they know the School of Social Science well (though at different stages of its history). We have asked them to reflect on their own work–its material conditions, disciplinary approach, intellectual goals–in this doubled context, social and academic. How has their research, their discipline, their world changed in the last 25 years–and what do they see as the critical tasks of the next 25?

The conference should open into a collective intellectual stock-taking, so that we come out of it with a better sense of what the School should be doing right now and in the foreseeable future. Where exactly do we stand, and where do we go from here? What kind of work do we want to sponsor? What kinds of problems should we be addressing, with what kinds of approaches and arguments?

For anyone familiar even in passing with the fabulous, tumultuous history of the School, or developments in contemporary anthropology, or the influence of the "interpretive turn" in the social sciences, or even the Sunday New York Times Magazine, this depopulated prose had to be interpreted as pointing offstage to an overtowering main actor, Clifford Geertz. The Institute for Advanced Study, unlike the Center for Advanced Study in the Behavioral Sciences, which is modeled on the Institute, and unlike the National Humanities Center, which is modeled on both of them, has a permanent faculty. In this way, it resembles the Collège de France, on which it is in fact modeled. Carl Kaysen, director of the Institute, recruited Geertz to its faculty in 1970, with the purpose of creating a new branch of the Institute, a School of Social Science. After an attempt to secure a second appointment died in Dantesque darkling flames, Geertz threatened to resign. The Board of Trustees of the Institute then convened to create the School of Social Science, with Geertz as its lone faculty member.

A quarter of a century later, a few days before the members of the class of 1996-1997 occupied their offices, Clifford Geertz attained the canonical age of three score years and ten. By temperament unlikely to march at all, much less under any banner, Geertz had nonetheless in his solitary eminence cut, sewn, and hoisted an intellectual flag–Interpretatio–found at the front of several academic forces, some of whom were passionately hacking one another to bits.

One might have thought that the 1996-1997 year and its conference would center, implicitly at least, around Geertz himself. To be sure, Albert O. Hirschman, Joan Scott, and Michael Walzer would be equally present as faculty members, equally engaged in interpretive social science. Scott, energetic and solicitous, frank and responsive, would serve as presiding officer and run the May conference; Walzer, polished and thoughtful, would conduct both the large, public, Thursday luncheon seminar and the small biweekly seminar reserved for participants in the School; the astonishingly resilient Hirschman would extend the saga of his prodigious "retirement," and all three would publish constantly. But Geertz was the one who had been with the School throughout its history, the one who had brought the others there and bound them together. For most outsiders, Geertz and the School were synonymous.

Consumers of academic rumor knew how, some years back, the Institute's scientists had scuttled the School's unanimous effort to secure an appointment for Bruno Latour, to assume a professorship in science studies funded in part by the Luce Foundation. Because a school in the Institute with fewer than three active faculty members loses much of its authority over appointments, Geertz could not leave until another appointment was made. After Latour, a second candidate for the professorship in science studies had declined. Throughout my year, external gossip would pass along the hot news of yet a third recommendation, to which the members of the faculty themselves never alluded. Social science was taking stock and so was the School. Geertz, himself relaxed, was at center stage, curtain down, in a tense institutional drama. Readers of Geertz's most recent book, After the Fact, knew that he had a refined edge to his views of both social science and the Institute.

But it was anthropologically impossible, given Geertz and the character of the School he launched, that the School would acknowledge overtly a special role for him, even a temporary one. He was present nearly every day, easily approached, prepared to engage without strain in any intellectual discussion or, if that is how it went, equally without strain in badinage. He was jolly, at times, which amazed me since I have never otherwise known a jolly academic. He scrutinized paper after paper prepared by the members and visitors. He was robust, tireless, sensitive, but the least directorial or presidory of personalities. He never imposed in any way on anyone. During the year, Hirschman, Scott, and Walzer all gave public presentations of their work, but not Geertz.

During the year, the only near-breach in the strong surface fiction that our gathering had nothing special to do with Geertz occurred at the end of April, during my installment of the biweekly small seminar, whose year-long theme was "The Past and Future of Social Science." In advance of the seminar, as was our custom, I distributed some notes I had put together. They were on the subject of evolutionary theory of meaning, and I paired them with some work by Geertz–his 1962 article, "The Growth of Culture and the Evolution of Mind," a review he had written for the current New York Review of Books of a book by Jerome Bruner, and various passages from The Interpretation of Cultures, Local Knowledge, and After the Fact. The theme of my seminar, which preceded by a few days our milestone May conference, was the relation of mind, brain, and meaning in social science, the endurance of that theme over several decades, its association with Geertz, and its importance for the future of social science.

I brought to the seminar, as disciplinary exhibits, Coevolution: Genes, Culture, and Human Diversity, by William H. Durham, professor of anthropology and evolutionary biology at Stanford, and The Adapted Mind: Evolutionary Psychology and the Generation of Culture, edited by two anthropologists, Jerome Barkow and John Tooby, and a psychologist, Leda Cosmides. Durham cites and quotes Geertz approvingly as having provided, in The Interpretation of Cultures, insights that would guide us in combining evolutionary biology with the study of the descent of cultures. But Cosmides and Tooby, in the first chapter of The Adapted Mind, give a fiercely opposed portrayal of Geertz as the "literary" wizard who ruined social science by conjuring up the lethal and false "Standard Social Science Model," or "SSSM" for short. Looking for Geertz in these texts was like looking in a fun-house mirror: Tooby and Cosmides endorse Durham as a rare counter-example to the "SSSM," but Durham says he is following Geertz.

At that two-hour Wednesday seminar, no one aside from me so much as referred in as little as a pronoun to either Geertz or the large packet of his work before us on the table. (For the sake of ethnographic completeness, I should record that Michael Walzer, as moderator, did once say "Cliff" to acknowledge Geertz's turn to ask a question, and that the faculty of the School was distracted: the following day, the director of the Institute, by declining to carry an ad-hoc committee's positive but troubled recommendation to the Board, would put a stop to the year-long campaign to make the rumored new appointment in science studies. Upon this third failure, the money offered by the Luce Foundation would go somewhere else. The rejected candidate was a Princeton historian of physics, opposed by the representatives of the Institute's Schools of Historical Studies and Natural Sciences. In an unprecedented tactic, the united faculty of the School of Social Science would explain all this and publicly air its collective grievance in the following week's Chronicle of Higher Education.) At my seminar, Geertz also ignored Geertz, but that was expected.

Here was Clifford Geertz, for thirty-five years at the middle of what was now an increasingly pressing debate in social science on the relation of mind, brain, meaning, and culture, an issue which, I argued, should be the focus of the "intellectual stock-taking" to which we were called at the big May conference.

During that seminar, it occurred to me that some of the answers to the questions posed for the conference might be found in what cognitive science and social science might have to say to each other, and that some of what they might have to say to each other might be found in what I have to say to Clifford Geertz, or anyway, in what I have to say about the most famous essay in interpretive social science, "Deep Play," whose author was Clifford Geertz.

Of Meaning and Social Science

In "Deep Play," Clifford Geertz offered a sustained interpretation of a specific sociological entity–the Balinese cockfight. Much more influentially, he laid out–systematically if implicitly–the principles of interpretive social science that have served followers for over twenty-five years.

These principles are now widely institutionalized, and social science as a whole is even more widely institutionalized, not least in having its own School in the Institute for Advanced Study. Social science occupies entire divisions of major universities, a branch of the National Science Foundation, office after office in state and national governments, and what looks like a thousand miles of shelf space in the periodicals section of the library. By contrast, cognitive science is fresh on the scene–the term did not exist until I was a graduate student. Yet mature social science and young cognitive science have begun to flirt, and their intellectual friction is already leading to some fundamental reconsideration of the principles that guide social scientific research. The crux of this reconsideration is "the problem of meaning."

The "problem of meaning" is the riddle of how meaning can come into existence, develop, and descend. What are the basic cognitive operations that human beings use to create new meanings–that is, meanings that do not already exist–and how do those basic cognitive operations work, specifically? Somehow, meanings arise. Somehow, meanings develop. Just as sexual organisms in an environment interact to produce descendents, so meanings in an environment interact to produce new descendent meanings. And somehow, the interplay between existing meanings in contexts creates new, descendent meanings, which may in turn interact with other meanings to create descendents of their own. Meaning descends, and somehow, new meanings are among the descendents.

In conception and nearly in practice, until just lately, biology has chosen to pass over the problem of meaning. Biology's historical achievements are not associated with the problem of meaning–the discovery of the basic neurocognitive mechanisms that human beings use to create new meanings, the specific principles that govern those mechanisms, and how those mechanisms might have evolved during the phylogenetic descent that produced cognitively modern human beings. Biology textbooks do not have chapters with titles like, "How two meanings interact to create new, descendent meaning." Biologists have historically set aside inquiries into the problem of meaning as belonging to an unnamed future branch of research. We do not need answers to the problem of meaning in order to do biology, and biology has plenty of targets at which to aim–viruses, cancer, birth defects, ecology, immune systems–without taking aim at meaning.

Yet meaning is fully biological. It is made by brains, often groups of brains, always in bodies, always in environments. Recently–this is a reversal–neuroscience has begun to take up the study of meaning, and, to an extent, the problem of meaning. During my undergraduate years at Berkeley, it seemed to me that all but a few neurobiologists considered questions of meaning to be in scientific bad taste, premature, given the poverty of our scientific knowledge about the brain. Perhaps they were, and perhaps they still are, but they are becoming professionally acceptable. In the last decade, several adventuresome neurobiologists have begun to work directly on meaning and the brain.

Social science, on the other hand, looks at meaning all the time, but not at the problem of meaning. It offers analyses of meaning as created discursively, or arising between people, or precipitated by interaction between people, or induced according to different "perspectives"–religious, aesthetic, scientific, historical, commonsensical, philosophical, artistic–, or negotiated from self-interest. We have learned from social science how certain meanings are transmitted by culture, or selected for their regency by an upper class that sees them as weapons of class struggle, or constituted at the aggregate level by invisible market summations over individual choices made in that market, or modified by generations caught between norms of their parents and their own insufferable conditions. We might want to praise or disparage any of this work on its own terms, but there is no general principle according to which any of it should be devalued or displaced by cognitive science.

Yet all of these social scientific approaches have, until recently, assumed, taken for granted, left unexplored the neurocognitive level of operation without which these other social operations would be impossible. Social scientists study meanings and their conditions, but with few exceptions not the basic neurocognitive operations that make those meanings possible. Naturally, there have been social scientists, many more of them now than even a few years ago, who have faced the problem of meaning, and we will encounter some of their refreshing work as we go along. For example, the data and phenomena studied since the late 1950s in what has come to be called cognitive anthropology, surveyed in D'Andrade (1995), have led to the emergence of a vital few anthropologists, like Hutchins (1994), who work simultaneously as social scientists and cognitive scientists. But such cognitive social scientists remain a miniscule minority among those who practice political science, economics, sociology, and anthropology.

Just lately, however, both social science and cognitive neuroscience have started to take a new turn: both have rapidly begun to take note of the problem of meaning. In this respect, they are converging and are accordingly fated to combine. That prediction, in a nutshell, is the view of this book and the motivation for each of its chapters: cognitive science and social science are merging, and the future of social science lies in this blend.

Social science for the most part is built on a tradition that takes meanings as achievement to be interpreted rather than as the dynamic results of neurocognitive operations that it is our job to explain. Quite rightly, social scientists usually take it as obvious that people construct meaning–they assign utilities, arrive at conceptions of self, model the behavior of others, categorize and classify, form organizations, set prices, pass laws, create representations, develop rituals, prefer one candidate to another, develop attachments. Also quite rightly, social scientists typically take it as obvious that human beings use mental capacities to do all of this–human beings recognize, see, classify, express, represent, and deploy every other mental capacity conventionally associated with human life. The moment at which social science characteristically unleashes its power comes when there is already some meaning that needs to be interpreted. Social science does not characteristically aspire to explain the neurocognitive mechanisms through which meaning comes to exist.

In this way, the social scientist resembles the textual scholar who takes it as given that a particular text is meaningful and that human capacities were involved in creating its meaning, and who regards it as his task to interpret the text for us. The nature of human neurocognitive capacities, what exactly they are, how their mechanisms work, how it is possible in the first place for meaning to arise and descend–these are not the questions that interpretation addresses. Interpreting a specific meaning–wrestling with it, trying to penetrate and to translate it, looking over the shoulders of those to whom it properly belongs and trying to make sense of it for foreigners–does not require discovering the neurocognitive processes that deliver that specific meaning.

In this respect, social science as a whole is in a position something like biology before the theory of evolution. Biologists, or rather botanists and zoologists, studied flora and fauna in exhaustive detail, in niches, in situ, penetrating the mysteries of their local habitations, measuring them, counting them, tracking cycles, writing all this down in the equivalent of field guides, and developing the ability to predict many natural phenomena, including phenomena of change: if frost falls, the bud is harmed; if the soil is enriched, growth improves, and so on. The world of life forms was a text whose meaning the biologist interpreted. But these interpretations did not explain and were not meant to explain the biological processes according to which these species could exist in the first place, or descend, or develop, or differ. To explain these more basic issues required the theory of evolution, which, once it was available, became an indispensable instrument in the professional study of local, narrowly coordinated, in situ life forms and the niches they inhabit.

Taking human meaning as given and interpreting it, according to one or another social scientific practice, without referring to the neurocognitive level at which these meanings emerge, is like taking the existence of life forms as given and interpreting them without referring to the theory of evolution. Social science does nothing wrong here, since cognitive science has no theory of emergence and descent of meaning that can begin to compare with the theory of evolution of species. We really are in the position of botanists and zoologists before the theory of evolution, and it is indeed something like the theory of evolution that cognitive science is trying, by gists and piths, with setbacks, to discover.

What cognitive science offers social science, at this moment, is an expectation that interpretation of meanings will eventually go hand-in-hand with explanation of neurocognitive processes of meaning–processes that underlie the objects and the acts of interpretation. Cognitive science offers a few initial, provisional proposals for joining with social science, but it is still in its infancy, and if the theory of evolution is our standard for comparison, cognitive science has very far to go.

Once developed, a cognitive theory of meaning will not displace or dismiss social science, any more than the theory of evolution supplanted the local study of zoological phenomena in their full particularity. I announce as a fact that a child can arrive at an astonishingly detailed, organized, exact, and useful understanding of frogs, toads, and salamanders, and of their differences and relationships, without having recourse to the theory of evolution, and it goes without saying that the theory of evolution alone brings essentially none of this understanding with it. The theory of evolution by itself tells us little in detail about its specific products. In each specific case, we must investigate the contingent details of how evolution played out, and those contingent details are complicated. Still, an understanding of amphibians becomes much fuller once the theory of evolution becomes available.

Just so, an explanation of neurocognitive processes of meaning will tell us little by itself about how those processes play out in any actual complicated case of human culture, because actual cases have intricate and unpredictable boundary conditions. It is clear, from the world's cultures and peoples, that there is a great diversity in human acts of meaning, and that, if we want to analyze all this variety in actual performance, we will need at least all of social science.

Basic human mental operations operate over cultural and personal assemblies of knowledge. Some of these assemblies will be widely shared in a culture, and expressions in the culture's language will evoke them. In our culture, for example, such cultural assemblies of knowledge include buying and selling, stopping at a red light, moving into a residence, going to the movies. Cognitive scientists call such assemblies of knowledge "frames." Frames are conventional packets of knowledge that usually include roles (such as buyer, seller, price, location, time, and so on) and various interactions between elements. Frames can be quite abstract (the stronger versus the weaker) or very specific (a team climb of Mt. Everest, a pilgrimage to the Vatican, trying to find a parking place in Manhattan, getting through customs at John F. Kennedy International Airport).

Since basic mental operations operate over cultural frames of knowledge, and those frames can vary dramatically from culture to culture, and purposes and conditions can also vary dramatically, different cultures can and do look strikingly different. Products of cognition vary across cultures even though their members share basic cognitive operations. We need every bit of social science to interpret these quite various products, but we need neuroscience and cognitive science to explain the basic mental operations that produce them, and the interpretations offered by social science should become fuller once neurocognitive theory of meaning is brought in. The rest of this chapter and the rest of this book are my attempt to give snapshots of how social science might look if we tried to do that.


Backstage Cognition and the Balinese Cock Fight


"Deep Play: Notes on the Balinese Cockfight," by Clifford Geertz, has become by an infinite distance the most famous and successful attempt by a modern anthropologist, perhaps by any social scientist, to explicate a sociological entity. Geertz's beautifully written and finely detailed analysis has become the canonical ideograph for the kind of social science whose purpose is "the analysis of the significance of social actions for those who carry them out and of the beliefs and institutions that lend to those actions that significance.". This semiotic and hermeneutic approach sees human behavior, or at least the interesting part of human behavior, as symbolic action. It asks what the import of that action is, what it is that is getting said. "The whole point of a semiotic approach to culture is . . . to aid us in gaining access to the conceptual world in which our subjects live so that we can, in some extended sense of the term, converse with them." This sort of analysis, Geertz explains, is "not an experimental science in search of law but an interpretive one in search of meaning. It is explication I am after, construing social expressions on their surface enigmatical."

This hermeneutical, interpretive impulse results in two features that give interpretive social science its characteristic feel. These features are shared, not surprisingly, with the best historical criticism of texts. The first is particularity–a sustained sensitivity to fine nuance and to local elements that are indispensable to the full meaning of the individual, narrowly situated, contingent sociological entity in situ. In the manner of Erwin Panofsky on Early Netherlandish painting or C. H. Dodd on the Fourth Gospel, Geertz works through every nuance of the cockfight. No detail that is significant for the actors can be insignificant for the social scientist.

The second feature is historical retrospection. This is the feature Geertz emphasizes in calling his book After the Fact (1995), and it is the feature with which he begins that book:

What we can construct, if we keep notes and survive, are hindsight accounts of the connectedness of things that seem to have happened: pieced-together patternings, after the fact.

It is also the feature with which Geertz ends that book:

A sage is squatted before a real elephant that is standing right in front of him. The sage is saying, "This is not an elephant." Only later, as the elephant turns and begins to lumber away, does a doubt begin to arise in the sage's mind about whether there might not be an elephant around after all. Finally, when the elephant has altogether disappeared from view, the sage looks down at the footprints the beast has left behind and declares with certainty, "An elephant was here."

For me at least (and that is the "we" we are talking about here), anthropology, ethnographical anthropology, is like that: trying to reconstruct elusive, rather ethereal, and by now wholly departed elephants from the footprints they have left on my mind. "After the fact," is a double pun, two tropological turns on a literal meaning. On the literal level, it means looking for facts, which I have, of course, "in fact" been doing. On the first turning, it means ex-post interpretation, the main way (perhaps the only way) one can come to terms with the sorts of lived-forward, understood-backward phenomena anthropologists are condemned to deal with. On the second (and even more problematical) turning, it means the post-positivist critique of empirical realism, the move away from simple correspondence theories of truth and knowledge which makes of the very term "fact" a delicate matter. There is not much assurance or sense of closure, not even much of a sense of knowing what it is one precisely is after, in so indefinite a quest, amid such various people, over such a diversity of times. But it is an excellent way, interesting, dismaying, useful, and amusing, to expend a life.

As far as they go, particularity and historical retrospection are unassailable, and they have manifestly led to superb interpretations in the fields of historiography, historical criticism, ethnography, anthropology, and sociology. But they do not go very far at all in some directions, and it is in those other directions that cognitive science has insights to offer.

First, particularity. While it is true that every particular is strictly unique, no particular is isolated. It is instead intelligible only because we bring to bear on it more general operations and knowledge, not to isolate it but to connect it with wide, sometimes very wide, mental arrays. We see an individual cock fight, but to see it, or know that it is a cock fight, or understand anything about it, we must use conventional conceptual frames–that is, conventional schematic packets of shared knowledge–concerning such things as public events, cocks, money, and anxiety, and we must also use very general cognitive operations ranging from vision to categorization. To recognize a particular speckled hen as a speckled hen is not a matter of particularity but rather of connection. To interpret, as an everyday matter, a particular symbolic act as an act, as symbolic, and as having an interpretation is not a matter of particularity but rather of connection.

Second, historical retrospection. The future of human action and meaning is not at all random, and it is therefore misleading to say that we live it forward but understand it only backward. In real senses, we already understand what lies ahead of us and we can make that understanding much more precise and scientific. We understand that what lies ahead of us must start from here and must develop through the neurocognitive processes that human beings have. We know that what can follow from our present point in the historical path depends strongly on the point, on the path, and on human cognitive nature. We want to know not only about particular past events but also about today and tomorrow. What is meaning and how is it constructed and how can meanings that have arisen be further developed? How can meanings interact to give birth to new, descendent meanings? What mental equipment do we have that provides the potential for creating new meaning out of old? The cognitive operations we will use tomorrow and probably twenty thousand years from now are the same as those we used yesterday, a hundred years ago, and probably twenty thousand years ago. We want to know not only the intricacies of the previous products of those cognitive operations, but also what those cognitive operations in fact are and what our prospects are.

Geertz's article, while apparently dedicated to the retrospective interpretation of a highly particular sociological entity–the Balinese cock fight–struggles, just below the surface but not therefore any less powerfully, to discern the basic cognitive operations that make the invention of the cock fight possible. These basic cognitive operations, I will argue, are universal among human beings, fundamental to cognition, and indispensable to reason, inference, and invention. They are also imaginative and creative.

I will take one of these basic cognitive operations–conceptual integration, also known as "blending"–as my exemplary offering from cognitive science to social science. My demonstrations will all for the most part be displays of what happens when something like the cognitive science of conceptual integration is brought to bear on central themes in the social sciences. My hope is that these topical sketches will give representative pictures of what cognitive social science could be.

Conceptual integration–blending–is a basic mental operation. It is at the very center of what it means to have a human mind. It plays a profound role in all areas of thought and action, including deciding, judging, reasoning, and inventing. It is dynamic, supple, and active in the moment of thinking. It yields products that frequently become entrenched in conceptual structure and in grammar. It often performs new work on its previously entrenched products. For the most part, it is a routine, workaday process that escapes detection except on technical analysis. It is not reserved for special purposes, and is not costly. As we will see, some researchers have proposed that the development of this cognitive capacity for conceptual integration was the most important event in human evolution, the evolutionary leap that separated cognitively modern human beings from other species, and in particular from merely anatomically modern human beings. Conceptual integration, in this view, is the basic cognitive operation that makes human culture, science, and art possible, indeed, the one that makes us possible.

The early theoretical work on conceptual integration was done jointly by Gilles Fauconnier and me, and presented in various publications: Fauconnier & Turner (1994, 1996, 1998a, 1998b, and in preparation), Turner and Fauconnier (1995, 1998, and 1999), Fauconnier (1997), and Turner (1996a and 1996b). The model we offered–the "network model of conceptual integration"–has additionally played a role in Collier and Levitsky (1997), Coulson (1995, 1996, and 1997), Grush & Mandelblit (1997), Mandelblit (1996, 1997), Mandelblit and Zachar (1998), Oakley (1995), Ramey (1997), Sun (1994), Veale (1996), Zbikowski (in press), and many others. This work is presented on a website dedicated to conceptual integration: it is available by visiting and following the links to "Mark Turner" and then to "Blending."

A conceptual blend always has at least two conceptual influences, sometimes called its "contributors" or its "contributing spaces," sometimes called its "inputs," sometimes called its "parents," sometimes called its "espaces d'entrée," depending on the culture of the audience. If we say, "This surgeon is a butcher," the influences are the prototypical notions of a surgeon and a butcher. If we say "front-loaded IRA," the influences are the notion of a conventional Individual Retirement Account and the notion of something's being "front-loaded."

Before conceptual integration can proceed, some provisional cross-space mapping must be constructed between the influences: the surgeon corresponds to the butcher; the conventional IRA corresponds to something that is "front-loaded." But the essence of conceptual integration is its creation of a new mental assembly, a blend, that is identical with neither of its influences and not merely a correspondence between them and usually not even an additive combination of some of their features, but is instead a third conceptual space, a child space, a blended space, with new meaning. This new meaning is "emergent" meaning, in the sense that it is not available in either of the influencing spaces but instead emerges in the blended space by means of blending those influencing spaces.

The blend inherits some of its elements and some of its meaning from the influencing spaces, and in this way it is the conceptual descendent of the influencing spaces, just as a child is the biological and cultural descendent of its parents. But like the child, the blend develops its own identity and is not merely a copy of its parents. It has meaning that is its own: "emergent" meaning.

The surgeon who is a butcher is a blended notion–neither a prototypical surgeon nor a prototypical butcher. Incompetence is the central feature of the butcher-surgeon even though incompetence belongs to neither the prototypical surgeon nor the prototypical butcher. The meaning incompetence emerges in the blend: it is not available from either of the influencing spaces, since neither the prototypical surgeon nor the prototypical butcher is at all incompetent. The blend, the "surgeon-butcher," is the descendent of the two influencing spaces surgeon and butcher, but it has its own "emergent" meaning possessed by neither of its influences: incompetence. By now in the history of the language, the word "butcher" can be used conventionally for anyone who does a sloppy job, but a new and unfamiliar blend that has not yet become conventional to us works in just the same way: "This surgeon is a lumberjack" can be interpreted as yielding incompetence for the blended lumberjack-surgeon even though incompetence belongs to neither the prototypical lumberjack nor the prototypical surgeon. "Lumberjack" could become conventional in the way "butcher" has. The surgeon-lumberjack is a blend, and while it is the conceptual descendent of surgeon and lumberjack, it has meaning of its own, new meaning that "emerges" only in the blend: incompetence.

An IRA–Individual Retirement Account–is a financial instrument for investing funds for retirement. It has (or had, until just now) an essential feature: the owner places untaxed income into it, hoping the investment will appreciate, and pays tax on amounts withdrawn later during retirement. It is thus "tax-deferred." "Front-loaded" suggests something whose load is by contrast at the front; metaphorically, with "front" referring to the earlier part of a time line and "load" referring to a burden, "front-loaded" suggests that the burden is suffered at first. So a "front-loaded IRA" is one whose tax burden is paid not at the end but at the beginning: one deposits taxed money into the front-loaded IRA and pays no tax on withdrawals made later in retirement. The emergent meaning here is obvious: one influence, "front-loaded," says nothing about IRAs; the other influence, the conventional IRA, is criterially a back-loaded tax-deferred investment account; but in the new blend, there is a new entity, with a new meaning that emerges only in the blend: an IRA that is not tax-deferred. The central feature of the category IRA–tax-deferral–is replaced in the blend by its opposite: tax up front. This emergent meaning is so profound as to count, in concepts of finance and in tax law, as a permanent revision of the category IRA. The emergence of this meaning has led to a revised set of meanings for IRA in which there are now two subcategories of IRA–"classic" IRAs are the original, back-loaded IRAs, while "Roth" IRAs are front-loaded, and all IRAs are investment accounts into which specifically limited amounts of money can be deposited for retirement.

We usually do not notice the work we are doing during conceptual integration. If we take "red ball" to mean a ball whose surface is permanently red, we must integrate the notion of a ball, which is a physical object with a surface that can be colored, and the notion of the color red, to produce a ball whose surface is red, and this is no trivial feat. But this product of blending is by now entirely conventional. If we take "red light" to mean a traffic sign that requires us to stop, we must integrate not only the notion of light and the notion of the color red, but we must additionally blend in the notion of traffic signs and the management of traffic. Yet in such a case the relevant meaning we need from the notion of traffic signs is usually indicated conventionally, either by the linguistic context ("Stop the car at the next red light") or by the non-linguistic context (the red light is of a very special sort and form and hangs in exactly the conventional spot at a roadway intersection). Because of such conventional indication, we again do not notice that we are doing any mental work to make such a blend. It seems to us as if we are doing nothing, because all the work of blending in these cases happens below the horizon of observation, in robust but unconscious cognition, and only the product of all that work comes into consciousness: we recognize the red light or understand the use of the phrase "red light."

But a phrase like "left-handed pen" will seem mysterious to people who do not already know that it refers to any pen containing special ink that will neither smear nor stain. A "left-handed pen" thus causes no problems for the left-handed writer whose writing hand slides over the ink. When we learn the phrase "left-handed pen," we must integrate our conventional schematic knowledge of a left-hander with our conventional schematic knowledge of using a pen. That is, we must blend two frames. They are the influences that contribute to the blend. But the relevant structure we need to project from those frames to the blend may be hard to locate in memory or find in the context; we may have to ask for an explanation. In any event, as we build this blend, we are likely to notice that we are engaged in a mental act of blending. We have no such awareness when we hear "red light."

The basic cognitive operation of conceptual blending always establishes some set of connecting links, even a very minimal set, between elements in the two influencing spaces. This set of connecting links is called a "counterpart mapping between the influencing spaces." Conceptual blending also always projects structure selectively from the influences to the blended mental space. Through composition, completion, and elaboration, the blend develops structure not provided by the influencing spaces. Blending thus operates according to a set of uniform structural and dynamic principles.

There can be any number of influencing spaces projecting to a blend, and blending can happen repeatedly, so the array of mental spaces involved in any particular conceptual integration network can be quite elaborate. Moreover, blending is a dynamic activity, with spaces and connections formed and reformed as the network is forged. My diagram therefore presents only a poor suggestion of what is going on during conceptual integration:

In "Deep Play," Geertz provides a brilliant demonstration of the intricacy of conceptual blending, even though that is not at all his purpose. His purpose, at least his announced purpose, is instead to interpret, through historical retrospection, a fabulous, complex, particular sociological entity–the Balinese cockfight–and thereby to give us access to its local wonder, to make it intelligible to us who are not Balinese. Put generally, his guiding purpose is historical retrospection of a narrow particularity. What motivates his study is the mystery of the particular sociological entity–the Balinese cockfight. What he offers to explain is the local meaning of that particular sociological entity.

My purpose, motivations, and goals are altogether different, but complementary, and that I take it is the general relationship of cognitive science to social science–cognitive science offers an altogether different but complementary line of analysis. What I mean to explain is how the Balinese cockfight and Geertz's interpretation of it arise from an underlying, general, basic mental operation–conceptual blending. What motivates my study is that basic mental operation. The Balinese cockfight–or rather the distributed conception of it that guides and makes intelligible its many enactments–follows the structural and dynamic properties of blending, and it conforms to the constraints on blending. I take the underlying mental operation rather than the particular sociological product as my subject of analysis.

Like the specific cockfight in April of 1958 during which Geertz and his wife were nearly caught in a police raid, the Balinese cockfight as a sociological entity is highly local, individual, and particular. Viewed as a bundle of past experience, surveyed in retrospect, it is an exotic and distant phenomenon which Geertz, the anthropologist from the field, emerging with notes, finally makes sense of, after, like the elephant, it has disappeared from the scene. Geertz's work, after the fact, directed backward, tells us about something that does not properly belong to us.

But the same Balinese cockfight viewed as I view it is a recognizable product of a mental ability that is permanent, indispensable, and apparently universal to human beings, an ability that runs across all cultures, all histories, all languages, past, passing, and to come, a mental operation that no human being with a standard biological endowment raised in anything like a human environment can fail to develop and deploy widely and powerfully. Conceptual blending is, in this respect, part of human nature–controversial as that expression has become–a mental power characteristic of our species and, I think, the central cognitive characteristic of our species. Conceptual blending is like computation of color constancy in visual cortex (that is, the way we see an apple as red regardless of the fact that the light coming from it varies dramatically from pre-dawn to noon to dusk) or like categorization: just as human beings everywhere develop the intricate mechanisms for computing color constancy, just as they cannot fail to categorize perceptually and conceptually, and just as certain cognitive principles of categorization run across all of the many quite different category structures human beings have in fact constructed, so human beings cannot fail to do conceptual blending, and while the products of conceptual blending are impressively various and intricate, its cognitive principles are uniform.

The Balinese cockfight is local and past, but the mental operation that underlies it is invariant over historical time and is the central engine of human meaning. It is part of us, where the "us" in this case is neither the Balinese of 1958 nor scholars in the year 2001, but all cognitively modern human beings, beginning very far before written history and stretching indefinitely into our phylogenetic future. Blending is basic, not exotic.

The idea of the Balinese cock fight is a conceptual blend. It begins from a conceptual connection between two very different kinds of things: cocks and men. As Geertz explains, "For it is only apparently cocks that are fighting there. Actually, it is men." The outcome of the cockfight cannot alter the status of the men whose cocks are fighting, but the cockfight counts nonetheless as an action of their status rivalry. Geertz makes various attempts to say what kind of mental operation accounts for this conceptual connection–"An image, fiction, a model, a metaphor . . ." The cockfight, he informs us at several points, is at once an act of nature involving cocks and an act of culture involving men. "This crosswise doubleness of an event which, taken as a fact of nature, is rage untrammeled, and taken as a fact of culture, is form perfected, defines the cockfight as a sociological entity."

Geertz's phrase–"crosswise doubleness"–for capturing the essence of the Balinese cockfight is in fact as perfectly descriptive a short name as could be devised for conceptual integration: "doubleness" for the two influencing spaces, "crosswise" for the way in which their independent contributions come together in the blend.

Geertz works up to his magisterial interpretation of the Balinese cockfight blend by degrees, providing first a series of quick treatments of several other cock-based blends in Balinese culture. Here is one: "Even the very island itself is perceived from its shape as a small, proud cock, poised, neck extended, back taut, tail raised, in eternal challenge to large, feckless, shapeless Java." Some principles of conceptual blending can be seen from just this accessory blend, as follows.

The island of Bali, in this blend, is a fighting cock whose adversary is the island of Java. But, as everyone knows, the motivation for this blend comes not from ideas about islands as adversaries, but rather from ideas about the peoples on those islands as adversaries, and about the adversarial opposition of their cultures: Balinese versus Javanese, Balinese culture versus Javanese culture.

The island-cock in the blend is not just an island and not just a cock, or even just an island-cock; it is at once a cock, Bali, the Balinese, and their culture, and this multiple projection to a single element in the blend has many effects, including the feat of turning the Balinese into a single organism, the cock.

The blending of the Balinese with the cock arises, for us, against a background of some pre-existing templates for certain kinds of blending, all of them available as part of the standard mental furniture our culture provides: we can blend a people with an animal species (e.g., the English people with a bulldog); a group of people with a particular animal (the University of Georgia football team is a particular bulldog and is represented as such in cartoons in which the bulldog rips up the players of the opposing team); a national people with a particular person (French-Marianne; Americans-Uncle Sam); and a particular person with an individual animal of another species (my boss is a snake). Canonically for us, when a people is blended with an individual person, the blend has an element that is simultaneously both the "ethnic character" of the people and the individual character of the person (e.g., self-reliant Americans/Uncle Sam). Canonically for us, when a person is blended with an animal, the blend has an element that is simultaneously both the "character" of the person and the "instinctive nature" of the animal ("Bill is a fox"). Canonically for us, when a people is blended with an animal, the blend has an element that is both the "ethnic nature" of the people and the instinctive nature of the animal (indomitable bull dog/English). We also of course routinely blend animals and human beings: talking animals are the mainstay of children's literature.

All of these conventional templates for blending are guided by a much larger template used for blending an entity at one level (such as a person or a people) with an entity at another level (such as an animal or a plant or a physical object), on the constraining principle (first discussed in Lakoff and Turner, 1989) that, other things being equal, it is their "highest" natures that are to be fused. This is the principle that stops us from interpreting "My boss is a spider" to mean that my boss is black, although prototypical spiders are black. It stops us from interpreting "The legal profession is a disease" to mean that we could kill it off through poisons, although perhaps we could. Of course, the constraining principle can be overriden by other cues.

The Bali-as-cock blend has a single element that is a cock with an instinctive nature and a people with an ethnic character. This element also has some of the features of human individuality, since the behavior of the cock is not rigid instinctual behavior but instead something the cock has individually chosen and is to be admired for having chosen. Animals do not choose their instinctive natures.

It is important to see that blending is not just matching of structures that already exist in one influencing space (in this case, the animal cock) and the other influencing space (in this case, Bali and the Balinese). On the contrary, the Bali-Balinese-cock blend has crucial structure that is unavailable from the influencing spaces. In the influencing space with the actual cocks, the actual cocks fight; they engage, and the Balinese disrespect them if they do not engage; what they fight is other cocks. In the blend, the picture is very different: the cock does not fight but is rather frozen in eternal challenge–something entirely unsuiting to an actual cock–and is moreover admired for staying frozen. The object of its challenge is not a cock, is not even a recognizable animal, but is some shapeless and unrecognizable lump twenty or thirty times its size. The influencing space with the cocks cannot supply this structure because an actual cock could not behave this way. Moreover, an actual cock will fight any other cock regardless of its features, and with identical ferocity, but the cock in the blend is dedicated to a single adversary whose suitability depends upon certain features, namely, the fact that the geographical "head" of the island of Bali points at the island of Java and that the Javanese and the Balinese have lived as adversaries. This is a strange blended cock, who isn't interested in fighting either the cock to the east or the islanders to the east.

The conceit of Bali as a geographical and ethnic cock is a "double-scope" blend. The influencing spaces to a "double-scope" blend have conflicting organizing frames, and the frame developed for the blend draws on both of the organizing frames of the blend. The influencing spaces of the Bali-Balinese-cock blend certainly have strongly conflicting organizing frames– fighting cock versus geographical island. A cock is an animal and an agent, but a geographic island is not. In "double-scope " blends, the organizing frame of the blend borrows heavily from each of the two conflicting frames that organize the two influencing spaces. The more evenly the organizing frame of the blend borrows from the organizing frames of its influences, the more thoroughly "double-scope" it is.

We can see the way in which the Bali blend takes parts of each of the organizing frames of the influencing spaces. From the influencing space with the fighting cock, the blend takes the cock itself, its adversary, the moment of challenge before violence, and the fact that the cock has an instinctual, unchanging nature. However, there is crucial, indispensable organizing structure in that influencing space with the fighting cock that the blend does not take: the blend does not take the fighting cock's challenge as a punctual event that causes an immediate engagement; it does not take the status of its adversary as a cock or even a recognizable animal; it does not take the causal relationship between the extreme fighting spirit of the cock and its prospects of winning; it does not take the possibility that a spirited cock can win, or even that it can inflict serious damage on the adversary; it does not take the certainty that one way or the other, one of the adversaries, maybe both, will be eliminated completely and forever, probably sooner rather than later, or, alternatively, that the confrontation will be broken off and the adversaries will separate so as to have nothing more to do with each other.

Instead, the Bali-as-geographic-and-ethnic-cock blend takes considerable organizing structure from the influencing space with the Balinese people (as viewed by the Balinese people): the Balinese and the Javanese are two entirely different orders of being, and while the Balinese have a distinctive character, the Javanese are so indistinct and ill-defined as hardly to count as a people at all; the lack of spirit and vitality among the Javanese makes them no less inevitably dominant; the conflict between the Javanese and the Balinese is permanent, eternal; Balinese character partly derives from this relationship with the Javanese; and it would be best by far if the defiance deterred engagement, so that no fight ensued.

This blend of Bali-versus-Java also recruits structure from another conventional and abstract template for blending, in which personal character is fused with shapes of objects: we refer to a person's character as "distinct," "angular," "well-rounded," "smooth," "rough at the edges." In this way, character is conventionally understood as having form. In the Bali-versus-Java blend, the formlessness of the island of Java is blended with the character of the Javanese, to produce a character that is "indistinct, shapeless."

There is considerable emergent structure in this blend–that is, structure in the blend that is not available from either of its influencing spaces: in the blend, we have an entirely imaginary kind of cock, who fights on both the physical and spiritual planes, whose highest calling is eternal defiance, who is dedicated to a single adversary, who has human intentional structure and can know that it would lose any physical engagement even as it clearly wins on the spiritual plane.

The use of the island of Bali in the blend illustrates the nature of metonymy in blending. This is a complicated topic, which will take a few steps to introduce. By "metonymy," I mean an organizing relationship between one element in an integrated conceptual assembly and another: "all hands on deck" because hands are metonymically related to the sailors who have them. The hands are the relevant active part of the sailors. Such a part-to-whole relationship is a standard metonymy. Other common metonymic relationships are the relationship between a cause and an effect, between one stage in a transformation and another, and between a location and the institution or activity located there.

Part of the goal of blending is to supply an integrated blended scene that is sufficiently intelligible, compressed, and memorable to be manipulated as a unit. This is a difficult goal to achieve: integrating all the important aspects of the influencing spaces into a unified and compressed blended scene takes imaginative work, not least because important things in the influencing spaces often are only distantly related. One of the great powers of conceptual integration lies in its ability to tighten and enhance those distant relations so they will fit into an intelligible, compressed, useful blended scene.

Consider an example of metonymy in blending to which we will return: in the conception of Death The Grim Reaper, there is a very long causal chain from Death as a general abstract cause applying to all living things, to the individual cause of death (cancer, accident, old age), to the death of the individual, to the corpse, the burial, the decay, the buried skeleton, and the exhumation that produces a visible skeleton. In the blend of Death The Grim Reaper, the skeleton becomes the form of Death. It is part of Death. Thus, in the blend, there is a very tight relation between Death and the skeleton: the skeleton is the overall structuring part of Death. In the blend, there is a tight Death-skeleton metonymy, a form-to-whole metonymy, in place of the very distant and many-step causal-chain metonymy in the influencing space that connects Death to the skeleton. In the blend, the cause (Death) and the effect (the skeleton) are combined, so that the effect (the skeleton) is now the most salient feature of the cause (Death). That is not at all the case in the influencing space. The blend also compresses time: something earlier in time (the advent of Death) and something much later in time (the existence of the skeleton) are collapsed into each other–that is, these two temporal stages now exist simultaneously. The blend also achieves compression of the process of change: the long process of change from dying body to skeleton is now, in the blend, manifest in the cause (Death) itself. This blend of Death The Grim Reaper thus tightens many related metonymies and fits all the elements they involve into a single intentional scene at human scale.

The Balinese cockfight blend uses metonymy compression in the same way. This metonymy compression may be a little harder to see because the compression in this case is maximal, reducing the metonymic relation to zero. That is, two things metonymically related in the influencing space become identical in the blend, as follows. In the influencing space having to do with Bali, there is Bali the island and the Balinese who live on the island of Bali, and there is a metonymic link between them. In this influencing space, Bali and the Balinese are not identical: the Balinese oppose the Javanese, but the island of Bali does not oppose the island of Java. In that influencing space, the Balinese inhabit Bali, and this is a metonymic relation. It is clear in this influencing space that Bali the island is not intentional and cannot attack or oppose.

But in the blend, Bali and the Balinese are fused. They become identical. Thus, in the blend, the opposition of the Balinese to the Javanese is an opposition of Bali the island to Java the island. In the blend, the island and the people are the identical element. And this Bali/Balinese single element is also blended with the cock, making Bali, the Balinese people, and the cock one element. Of course, this does not mean we are confused: when we achieve a blend, we do not (usually) lose the structure of the influencing spaces. In the influencing space with Bali and Java, we maintain the distance and distinction between the geographical islands, their inhabitants, and cocks. But in the blend, we do not. We know what is in the influencing spaces, and we know what is in the blend, and we know the connections between them. Each of these spaces has its uses. One of the uses of the blend is to compress meanings that are diffuse in the influencing spaces.

This process of metonymy compression during blending operates under a constraint: other things being equal, when one element is projected from an influencing space into the blend and a second element in the same influencing space metonymically related to the first is also projected to the blend, it is better to shorten the metonymic distance between them. This shortening produces more compressed blends. In the Bali-as-cock blend, the metonymic distance between Bali and the Balinese is maximally compressed. In Death The Grim Reaper, Death and the skeleton are not compressed all the way to identity, but close: the skeleton becomes part of Death, the salient part.

Compressing a metonymic distance all the way to identity–that is, taking two elements at some metonymic distance from each other in an influencing space and fusing them in the blend–is relatively infrequent in blending since the accidental features that make it suitable or even possible are often lacking. But partial compression of metonymic distance is common in blending. Nearly every political cartoon, for example, involves some compression of metonymic distances to achieve a compressed blend, as when France in the blend is no different from Marianne (a French woman), or the United States is Uncle Sam (an American human being) or an automobile company becomes a particular car, or the "lobbying industry" becomes a single lobbyist and "Congress" becomes a single generic congressman, and so on.

Suppose, to choose one of these examples to dwell on, that one of three competing auto makers is winning in its financial competition with the other two. To present this situation, the cartoonist might show a car race, in which one car is ahead of other two, where each car stands for the company that produces that car, and where each car is in fact the zippiest sports car made by that company. To help readers who don't know much about car models understand the cartoon, the cartoonist might label the cars, in standard cartoon fashion, with the names of the companies. These labels might be applied even if the sporty production model that supplies the image of the winning car is in fact losing money for its company, or even if the winning model car is selling less well than the two models it is beating in the race, because (in this particular cartoon) the cars stand not for the finances of their production models but instead for the finances of the larger corporations that produce those models. This is a severe compression of metonymic distance. In the influencing space with the auto makers, there is a long metonymic distance from the finances of the company, to its operations, to its manufacturing, to its products, to the particular sports car. But in the blend, that metonymic distance is compressed to identity, and the financial aspect of the company is fused with the particular product of the company, indeed, more accurately, with a characteristic performance by that product.

This compression of metonymies is a standard instrument of advertisement. An advertisement for a lemon vodka cocktail using a particular brand of vodka consists of a metal lemon press (with crank and handle) whose top and bottom are simultaneously the top and bottom of the vodka bottle. The vodka and the lemon juice cocktail appear nowhere in the visual representation of the blend. Instead, the bottle, which is metonymically related to the brand of vodka it contains, and the lemon press, which is metonymically related to the lemon juice, are the content of the representation, and their compression into one unit signifies the combination of the vodka and the lemon juice into one cocktail. A different ad, this time for gin and tonic, consists of a bottle cap that is a fusing of one half of the gin bottle cap to one half of the tonic bottle cap.

The compactness of these blends satisfies the "integration" constraint on blending: other things being equal, the blend must constitute a tightly integrated scene that can be manipulated cognitively as a unit. The actual world in which the automakers operate–global finances and marketplaces–does not form a basic, distinct, perceptible, integrated human scene, but blending the auto makers with the frame of competition gives some integration. Much greater integration comes from specifying the competition as racing, in fact automobile racing, and simultaneously exploiting the metonymy between the finances of the auto corporations and the sports cars they produce. This exploitation gives the blend a scene in which financial operation and car racing are fused, and the financial corporations and the individual cars are fused, and the financial corporations are instantly recognizable from the particular labeled cars that represent them. The result in the blend is a compressed, integrated, familiar, distinct human scene. In this case, compressing the metonymy has provided a way to help satisfy the integration constraint, that is, compressing the metonymy helps to make the blend more integrated.

The Bali-as-cock blend takes just this path of compressing a metonymy in an influencing space to achieve a tighter integration in the blend. While blending the Balinese people and the Javanese people with individual adversaries in a combat provides some integration, and yet greater integration comes from specifying the combat as a cock fight, there is additional integration to be had from compressing the metonymic relation between the Balinese, Bali, and the shape of Bali so that, in the blend, Bali, as the location of the Balinese, has the form of the cock. By this conceptual work, the blend can compress the Balinese, Bali, and the cock into identity, giving a compressed, integrated, distinct scene that is culturally appropriate and, if you are Balinese, familiar: the moment of challenge in a cock fight. Again, compressing the relationship between two elements as they are projected to the blend provides a way to achieve greater integration in the blend.

There are other aspects of blending at work in the Bali-as-geographical-and-ethnic-cock blend. Most obviously, blending exploits accidents. It is merely accidental that the side of Bali lying toward Java juts to a point rather than, for example, curving along a concave crescent, and this accident makes it possible for Bali to be viewed as having a head jutting toward Java. Importantly, it is a general principle of blending that the accidental origins of a blend are no argument against its significance or profundity. Indeed, the most profound importance and essence can derive from the sheerest accident. In this way, the phylogeny of conceptual blends resembles the descent of species, in which accidents can set fundamental courses. Social science, not to mention human beings, might not exist had it not been for an accident roughly sixty-five million years ago in which a meteor struck the sea off Yucatan, giving mammals a lucky boost in their competition with reptiles.

The Bali-as-cock blend also shows that blending frequently requires us to reconceive and restructure the influencing spaces. I own a physical map of Bali, which I have presented to many people, including some people who are familiar with farm cocks and even some people who have seen cock fights in the Western Hemisphere. I have asked them what Bali looks like. Nobody has ever answered that it looks like a cock. But if you are disposed to think of Bali and Java as antagonists, and are steeped in cock fighting or otherwise primed to activate a frame for cocks, then, with a little work, the island of Bali is susceptible to being reframed as having the form of a fighting cock. Blending is an active process that can involve extensive work at any point in the conceptual array connected to the blend, including reconfiguration of the influencing spaces, in this case, of the influencing space with the island.

Geertz suggests that there is almost no end of blends in Bali involving cocks. For example, "A pompous man whose behavior presumes above his station is compared to a tailless cock who struts about as though he had a large, spectacular one." Other cock blends have to do with sex and dating, employment, desperation, stinginess, and morality.

The focal blend for Geertz, the one he takes as the subject of his after-the-fact, particular interpretation, the one whose interpretation by Geertz has become the canonical example of how to do interpretive social science, is a blend of fighting cock and social man. It creates the sociological entity that is the Balinese cockfight, which is not a natural event of the animal world at all but instead an intricate, imaginative, and highly sophisticated conceptual blend of cocks and people. Its emergent meaning forms and represents the Balinese.

We can begin to tease a fraction of this emergent meaning into view by looking at an ostensibly obvious, trivial, accessory fact: the fighting cocks wear sharp metal spurs. Cocks in nature have spurs that are natural equipment, but metal spurs exist only in the blend. In the influencing space with Balinese social men, men of prestige have assistants, and the projection of the role "assistant" from the influencing space into the cockfight blend creates a position for a technical helper who cultivates and enhances the natural equipment of the cock in whom the owner has an interest. The metal spurs in the cockfight blend–"razor-sharp, pointed steel swords, four or five inches long"–are the counterpart of the cock's natural spurs. But manufacturing these spurs, gracing them with ritual status, and affixing them to the cocks is the counterpart of performing a service for the owner, a human being. The metal spurs emerge from the blend of something having to do with cocks and something having to do with people. From one influence, the blend takes the fighting cocks; from the other, it takes human social purposive action and interaction. The metal spurs are one result.

The cockfight blend contains a wide variety of additional structure that has no place in a natural cockfight. Some of it is even antithetical to natural fighting. The blend has an audience, a handler for each cock, a context of previous engagements involving these owners and handlers as well as the cocks they have previously brought to the ring, cosmological indications for how and when to fight each kind of cock, a fifty-square-foot ring, a wicker cage under which to gather and infuriate the cocks when they are reluctant, and an umpire steeped in regulations written on palm-leaf manuscripts handed down through generations. The ritual of engagement includes precisely timed rounds and all-important intermissions:

A coconut pierced with a small hole is placed in a pail of water, in which it takes about twenty-one seconds to sink. . . . During these twenty-one seconds, the handlers are not permitted to touch their roosters. . . . Within moments one or the other drives home a solid blow with his spur. The handler whose cock has delivered the blow immediately picks it up so that it will not get a return blow . . . . With the birds again in the hands of their handlers, the coconut is now sunk three times after which the cock which has landed the blow must be set down to show that he is firm, a fact he demonstrates by wandering idly around the ring for a coconut sink. The coconut is then sunk twice more and the fight must recommence. . . . In the climactic battle (if there is one; sometimes the wounded cock simply expires in the handler's hands or immediately as it is placed down again), the cock who landed the first blow usually proceeds to finish off his weakened opponent. But this is far from an inevitable outcome . . .

Not even the motivation for the cockfight comes from the cocks–the Balinese decide that the cocks will fight, when they will fight, and which cocks are appropriate opponents. Bizarrely, but inevitably, the Balinese define "winning" and "losing" in the cockfight in a way that makes no sense for the cocks themselves. In natural cockfights, winning might mean ruling the roost or eating up the food or practically anything that increases fitness, but surely it does not mean falling down in the dust slaughtered alongside your opponent. In the cockfight blend, however, the winner is exactly whichever cock is left standing when the other drops, even if the winner "himself topples over an instant later." The winner wins no less absolutely if he expires immediately. The owner of the "winner" no less certainly takes the carcass of the "losing" cock home in order to eat it.

In the mental idea of the natural cockfight, there are of course two opposed cocks. If cocks did not fight on their own as a natural action, without human intervention, there would be no Balinese cockfight. In the quite different mental idea of Balinese society, there are two owners and the two cocks they own. These two owned cocks are not at all like natural cocks–they have a daily regimen of elaborate pampering that is invented, set, and supervised by human beings. Their breeding is at the pleasure of human beings. Never in their lives are they allowed to fight spontaneously. Nonetheless, different though wild and owned cocks be, they are connected; they are counterparts. In the cockfight blend, these two counterparts are fused: a single cock in the blend is simultaneously the prized and pampered property of a social man and a wild and violent autonomous animal.

This blending of natural and owned cock is merely a background achievement that invites the essential blending in the Balinese cockfight, a blending in which the owner is projected into his cock. The cocks in the Balinese cockfight are "surrogates for their owners' personalities, animal mirrors of psychic form." They include elements of their owners. Cocks "are symbolic expressions or magnifications of their owner's self, the narcissistic male ego writ out in Aesopian terms." Most important, the cock in the blend carries the owner's social status, making the cockfight blend, in a phrase Geertz borrows from Erving Goffman, "a status bloodbath."

This does not seem so remarkable until one recognizes that, in Balinese society, status is strictly inherited and cannot be changed, certainly not by a cockfight. The cockfight, says Geertz, makes nothing happen. "No one's status really changes. You cannot ascend the status ladder by winning cockfights; you cannot, as an individual, really ascend it at all. Nor can you descend it that way." But in the blend, you can. You can rise or fall, defeat or be defeated. In Balinese society, open altercation is impermissible, and public display of social rivalry is so thoroughly masked as to be treated as if it never occurs. "The Balinese are shy to the point of obsessiveness of open conflict. Oblique, cautious, subdued, controlled, masters of indirection and dissimulation–what they call alus, 'polished,' 'smooth'–they rarely face what they can turn away from, rarely resist what they can evade."

But when men are cocks, which is to say, in the blend, they can attack each other furiously, and status can be gained or lost. "[H]ere they portray themselves as wild and murderous, with manic explosions of instinctual cruelty." Outside the blend, in the human social world and in the domain of natural cockfights, a man's status in society is entirely different from the status of a natural, autonomous cock in a fight. But inside the blend, the status of a man in Balinese society is blended with the performance of his cock, and this has psychosocial consequences: when his cock is victorious, his prestige is affirmed by the harmony between his status pride outside the blend and the status of his cock in the blend. Inversely, if his cock loses, his prestige is insulted by the discord between his status pride outside the blend and the status of his cock in the blend. The events in the cockfight blend thus have influence on the world of Balinese society. I will say that this is a "backward" influence, because usually the influence is from the influencing spaces to the blend, but in this case, meaning that develops in the blend has an influence on one of the original influencing spaces.

Crucially, inferences in the blend do not project back to Balinese society identically or even simply: the life or death of the cock in the blend does not entail the life or death of the social man outside the blend; the cock's gain or loss of status in the blend does not entail the gain or loss of social status for the social man; the fact that the cock cannot heal from death does not mean that the social man cannot recuperate from insult; the inability of the dead cock to get up ever again does not entail an inability of the social man to perform his roles. Yet inferences in the blend do project back powerfully, if temporarily, to the psychological uplifting or abasement of the social man. The Balinese compare heaven to the mood of a man whose cock has just won, and hell to the mood of a man whose cock has just lost. "A man who has lost an important fight is sometimes driven to wreck his family shrines and curse his gods, an act of metaphysical (and social) suicide."

The social man is of course a member of the social groups to which he belongs. By virtue of their metonymic link to him, those groups also project into the cock, so that the

cockfight is–or more exactly, deliberately is made to be–a simulation of the social matrix, the involved system of cross-cutting, overlapping, highly corporate groups–villages, kingroups, irrigation societies, temple congregations, "castes"–in which its devotees live. And as prestige, the necessity to affirm it, defend it, celebrate it, justify it, and just plain bask in it (but not, given the strongly ascriptive character of Balinese stratification, to seek it), is perhaps the central driving force in the society, so also–ambulant penises, blood sacrifices, and monetary exchanges aside–is it of the cockfight.

The blend of social man and cock may look on first blush unusual, but it is, in the principles by which it arises, utterly familiar. The everyday mind has an effortless power to blend self and other. The man who says to a woman who earlier in her life declined to become pregnant, "If I were you, I would have done it," presents a mental blend that has impressive emergent structure: in the blend, which has the appropriate past time, there is a special blended person who has the capacities and situation and appearance and social identity of the woman but the judgment of the man, and this blended person becomes pregnant. The pregnancy is impossible for the two people involved in the influencing spaces: the man cannot do it; the woman did not do it. But the resolute counterfactuality of the blend does not at all suggest that it is an irrelevant fantasy. On the contrary, it is meant to illuminate the real world, the nature of the real man, the real woman, and real human life.

Such a blend creates a mental simulation, often extremely useful as a guide for potential enactments, judgments, and insights. Part of our ability to blend self and other is probably evolutionary: it is adaptive, in observing a predator or prey, to be able to make a blend that includes some of our reasoning and some of its situation and instincts. We can run a simulation to see what it might do. Certainly this is exceptionally useful in dealing with members of our own species, and in fact we blend ourselves and other people as a matter of routine interaction, a baseline capacity, part of what it means for a human being to be social.

Some of these blends of self and other involve what literary critics and psychologists call "identification." When we daydream that we are someone else, enacting what they have enacted, or what they might enact, and experiencing the emotional effects, we are creating an identification blend, a fictional simulation. Outside the blend, self and other remain perfectly distinct, unmistakably separate, not to mention unidentical, but in the blend, they join as a third person who is not merely a patchwork of some of their parts. The blend does not supplant or erase the spaces that influence it but rather exists with them inside a conceptual integration network. Like the influencing spaces from which it derives, the blend has its uses.

Blending offers the indispensable conceptual opportunity to disintegrate an integrated unit by projecting aspects of it to separate elements in the blend. This fission may seem schizophrenic, but it is only what we already do when we lament that our head is saying one thing and our heart another–in the blend, conflicted reason and emotion are separate people engaged in dispute. We can have three people in this blend if our judgment is made into a third person advised by reason and emotion: "my head is telling me one thing and my heart is telling me another." Like Faustus between the good and evil angels, our judgment receives conflicting lectures from the opposed advisors.

The Balinese social man, in exactly this way, is disintegrated in the blend. The status part of him, his psychological security, and his social "face" project from the influencing space with Balinese social man to the cock in the blend. His internal aggressiveness and his impulse to explicit social affront–completely inhibited in his life outside the blend–not only project into the cock but also leave their inhibitions behind as they go. The owner's virility and the virile activity of his penis also project into his cock. But other parts of him project to the owner in the blend, for the Balinese cockfight is a matter of human agency as much as animal action. The owner in the blend, if not the handler and spur expert himself, at least hires those people, discusses strategy with them, agrees to the terms of the fight, and speaks with the umpire. The owner's body and his speech, his power to own, and his normal social interaction all project to the owner in the blend. All the intentional actions that are permitted to him in society project to the owner in the blend; indeed, the only ones that project to the cock are those–some of them, anyway–forbidden to him in society. Last but not least, the owner's money in the influencing space projects to the owner in the blend: he places the (sometimes enormous) central bet. The owner outside the blend undergoes a disintegration as he is projected into the blend, since parts of him can now be found in two different places: the cock is more than a cock because it carries projections from the owner and is liberated in the blend to perform actions whose counterparts the man cannot perform outside the blend.

The Balinese cockfight blend has as its central purpose to comment on one of the influencing spaces–the men, not the cocks. It wants to say something about the space with the Balinese social men. It may be that the cockfight began originally as an entertainment, devoid of the blending that deliberately makes it into a "status bloodbath," but the Balinese cockfight of Geertz's interpretation exists to enact social aggression in a way that is impossible in Balinese society but vivid and pure in the blend, and to project consequences–slighting, affronting, insulting–back to the actual world of Balinese society, where those consequences are deeply felt even though the actions that create them are disallowed.

I will say that such a blend has as its purpose, or at least one of its important purposes, to "spotlight" or "solve for" or "say something about" one or both of its influencing spaces. As a consequence of this purpose, the blend must match the relevant influencing space in appropriate ways, for otherwise inferences that develop in the blend cannot be matched to counterpart influences for that influencing space. Meaning that emerges in such a "spotlight" blend is meant to influence the influencing space, but the blend must "match" the influencing space in the relevant ways for this to happen. This "matching" has to do with relations between elements in a space. In general, other things being equal, when the purpose of a blend is to spotlight an influencing space, it is best if the relations between elements in the blend match the relations between their counterparts in the influencing space from which they were projected. Relations between elements in a space are referred to as "topology," and this constraint on building a blend is accordingly called the "Topology constraint." The Topology constraint is particularly weighty when the purpose of the blend is to "solve for" or "spotlight" or "say something about" an influencing spaces, because it is difficult to use the blend to say something about an influencing space if the relations in the influencing spaces are garbled as they are projected to the blend.

To be sure, blends can have many other purposes–among them humor, entertainment, heightened memorability, event integration–and for many of these purposes, backward projection to the influencing spaces is less important. For example, personal computers now standardly use the "desktop" interface, analyzed in Fauconnier and Turner (1998a), which is a blend of operations involving real desktops (opening and closing folders, putting files in folders, and so on), manipulation (moving a pointer around by hand, pressing buttons to lock and unlock), selecting from lists (pull-down menus), interpersonal commands (which the computer "executes"), and alphanumeric coding (programming languages). The purpose of the desktop blend, with its many influencing spaces, is to organize action–that is, the actual use of the computer–into a single integration; the purpose of the desktop blend is not to say something about desktops, manipulations, menus, commands, or programming. The desktop blend is not constructed in order to solve for, spotlight, or say something about its influences. Our integrated action in the desktop blend is meant to be guided by our intuitions projected from the influencing spaces, and that can happen only if the intuitions can flow into the blend, and that can happen only if the blend matches the influencing spaces in ways needed for the intuitions to survive and work appropriately in the blend.

Yet the desktop blend does not have as its purpose to project inferences and consequences back to the influencing spaces, so it need not match the influencing spaces on that account. For example, it is not important that the blend puts the trashcan on the desktop instead of under it, or that its trashcan never fills up, because we do not think the blend is trying to tell us something about how to deal with real trashcans. The purpose of the desktop blend is not to say something about actual desktops and trashcans, or menus, or manipulations, or commands, or coding languages.

But for the Balinese cockfight, the case is entirely different: consequences from the blend must project back to the actual social world. Accordingly, the blend must match the social world in certain ways. In cases of such backward projection, it is usually the structure of the blend that is manipulated to create the appropriate topological match between blend and influencing space; it is somewhat rarer that the influencing space is significantly altered to make the match.

To retreat and take up these complexities a little more slowly, let us consider a case in which we are forbidden to revise any of the established structure of one of the influencing spaces. This is the case in reductio ad absurdum arguments in mathematics (or logic). In such an argument, we prove that some assertion must be false by showing that a contradiction follows if we assume that it is true. These arguments proceed by blending. One influencing space has the established structure of the mathematical system with which we are working. The other has the assertion we think must be false. We keep the mathematical system, including its deduction procedures, intact, but blend some of it with the new and suspect assertion. We then run the blend according to the deduction procedures of the mathematical system until we prove something that conflicts with something else in the system.

Because the blend has been made to match the mathematical system perfectly except for the assumed hypothesis, we view the emergent contradiction in the blend as proof that the assumed hypothesis is false for the original mathematical system. (This assumption proceeds, of course, from the deeper assumption that the mathematical system itself is not internally contradicted.) Crucially, in setting up this reductio ad absurdum argument, the influencing mathematical system is not at all adapted to match the blend; rather, the blend is adapted, constructed, engineered so as to make an all-but-perfect match with the influencing mathematical system. At the outset of running the blend, the one imperfection in this match is exactly the assumed hypothesis.

Similarly, when a blend is constructed to solve a problem or riddle that exists in the contributing spaces, preserving relevant matches between the blend and the contributing spaces is indispensable. Consider the riddle of the Buddhist monk, analyzed in Fauconnier and Turner (1998a): "A monk rises at dawn and begins to walk up a mountain path, which he reaches at sunset. He sits and meditates through the night, rises the next morning, and walks down the path, reaching the bottom at sunset. Prove that there is a place on the path that the monk inhabits at the same hour of the day on the two consecutive days." One interesting way to solve this riddle is to treat the ascent and descent as two influencing spaces and then to blend them, essentially by superimposition. Suppose you have a film of each monk on each day, beginning at dawn and ending at sunset, and you run the two films on two projectors simultaneously so the images superimpose on the same screen, making the two paths perfectly coincident. In the superimposed images, which combine the ascent and the descent, we will see at dawn two monks (that is, two instances of the same monk), one of them at the top of the path and one of them at the bottom of the path, who begin to traverse the path in opposite directions, each completing the journey at sunset. If one of them starts at the bottom at dawn and the other starts at the top at dawn, and by sunset they have switched places by walking along the same path, then they must have crossed paths at some point. The monk must have met himself someplace, and that is exactly the place we seek.

There is an emergent event in the blend: the two monks meet. In the space with the ascent there is no meeting, and in the space with the descent there is no meeting. This emergent, inferential event of meeting in the blend projects back to give correlative inferences for the influencing spaces: now we know that in each influencing space, there exists some location on the path that the monk occupies at the same hour of the day in the two spaces, the ascent and the descent. The blend is able to deliver this solution only because, as the blend runs, the clocks in the three spaces match each other exactly, and, for each of the monks in the blend, his location and clock-time match exactly the location and clock-time of his counterpart in the relevant influencing space. In sum, the location and the clock-time in the influencing space are projected identically into the blend: the location and the clock-time for a monk in an influencing space match exactly the location and the clock-time for his counterpart monk in the blend. If the blend were not engineered to preserve these exact relations from the influencing spaces, the inference of the encounter in the blend would not have clear counterpart inferences for the influencing spaces, and we would therefore not be able to solve the puzzle posed for those influencing spaces. In this case, where the blend is obviously built to "say something about the influencing spaces," matching the relations between influencing spaces and blend is achieved by engineering the blend in the right ways. This is the typical strategy.

Certainly the cockfight blend follows this typical strategy: the blend is manipulated to achieve relation matches between the blend and the contributing space of Balinese man. To begin with, Balinese society is finely governed by traditional conventions, and so the cockfight blend is engineered to be governed in this way, even though such traditions would be meaningless, unintelligible, and alien in a natural cockfight. Traditional governance is manufactured inside the blend for the purpose of increasing the match between the blend and Balinese society.

But what I want to focus on now is the conceptual engineering in the opposite direction, that is, changing the influencing space so it matches the blend better. This, as I say, is not the usual strategy for satisfying the Topology constraint, that is, "Other things being equal, if two elements from an influencing space are projected to a blend, it is better if their relationship in the blend aligns with their relationship in the influencing space."

Remarkably, the influencing space with the Balinese men is itself changed to achieve a better match between the two influencing spaces and a better match between the influencing space with the men and the blend. It is changed in two ways. The first way arises as an attempt to fix a profound mismatch between the Balinese social matrix and natural cockfights. The Balinese social matrix is manifold, multivariate, and nonlinear. A single man is at once a member of several interacting social groups, some of them partially allied, some of them overlapping, some of them opposed, thus giving any particular social action significance simultaneously in many of these groups, and making social life far more complicated than a sum of discrete and punctual face-to-face encounters. Cocks, on the other hand, engage in just these punctual face-to-face encounters. To make the match between men and cocks better, the influencing space with the men is engineered so as to select two men, two owners, and to make them stand out against the backdrop of the Balinese social matrix, to profile them against their full social context. For a moment, their one-on-oneness is cast into relief in the influencing space. This is a perfectly acceptable casting of particular social relations in Balinese society, and this casting of the influencing space makes it easier to match that influencing space with the other influencing space and with the blend. The inaccuracy of this profiling as a picture of the Balinese social matrix is softened, remedied by having each man backed up, as it were, by his social groups when he is involved in a cockfight. It is also remedied by some direct engineering of the blend: Balinese cockfights are scheduled one after another, so that the suite of cockfights gives us not a single sharp profiling of one-on-one opposition but instead many different oppositions, and this multiplicity of oppositions gives us a much fuller set of snapshots of the social matrix, from which we can deduce a larger and more nuanced system of relationships. For example, two men who take opposing sides in one Balinese cockfight might take the same side in a subsequent cockfight. The picture we get of allegiance in one cockfight is thus clarified, nuanced, and specified by situating it in the context of several other cockfights and the various allegiances they present.

The second manipulation of the influencing space to improve the match with the blend is the essential one, the one Geertz analyzes thoroughly and brilliantly–namely, the betting. Through wagering, which brings real consequences to the real social men, the psychological reality in the influencing space with the social men is altered so that it matches the blend better. This introduction of wagering is prompted by the central problem that the cockfight, being just a cockfight, does not have direct natural significance for the state of mind and prestige of the two owners and their groups. The cockfight has, after all, no actual effect on status. To be sure, ritual representation can be powerful, but its power in this case comes partly from strong engineering that makes the cockfight heavily consequential in social reality for the principals, since they bet (and must bet) heavily on their own cocks, and it is real money–not just funny "blend" money–that they really lose or really win through this betting.

Geertz explains the gradient between "shallow" cockfights, in which the principals have relatively low social status, bet little money, and may be most interested in the money gambling, and "deep" cockfights, the ones that are the prototypes of the game, that everyone takes seriously, and that Geertz is interested to interpret. In "deep" cockfights, the principals have high social status, bet enormous sums, and are engaged mostly in status gambling. "Deep play" refers to these deep cockfights; it is a term Geertz borrows from Jeremy Bentham, who uses it, Geertz explains, to mean

play in which the stakes are so high that it is, from his [Bentham's] utilitarian standpoint, irrational for men to engage in it at all. If a man whose fortune is a thousand pounds (or ringgits) wages five hundred of it on an even bet, the marginal utility of the pound he stands to win is clearly less than the marginal disutility of the one he stands to lose. In genuine deep play, this is the case for both parties. They are both in over their heads.

Winning or losing a relatively large amount of money is a real social event. In Balinese society, wealth correlates with status, and a change of wealth powerfully represents a change in status without actually entailing it. Since, in the Balinese cockfight, the "house" does not bet, and excises only a ten percent vigorish to cover the cost of the event, and since winners can never get out of the game for long, since cockfighting is, for people of high status, an obligatory part of cultural life, this dramatic, sudden change of wealth is, for all but irrational bettors, only temporary. The invention of the betting is a manipulation of the social world to improve the match between the cockfight blend and the influencing space with the Balinese social men: the cockfight is no longer just a cockfight, not even just a cockfight with strong ritual connotations, but instead the cause of impressive and visible, thrilling and disturbing, changes in wealth for the Balinese social men.

Geertz, in so many words, provides a superb analysis of the way in which the betting serves to make a better match between the relations in the blend and the relations in the influencing space that has Balinese society:

But for the Balinese, although naturally they do not formulate it in so many words, the explanation lies in the fact that in such play, money is less a measure of utility, had or expected, than it is a symbol of moral import, perceived or imposed. (433). . . .

It is because money does, in this hardly unmaterialistic society, matter and matter very much that the more of it one risks, the more of a lot of other things, such as one's pride, one's poise, one's dispassion, one's masculinity, one also risks, again only momentarily but again very publicly as well. In deep cockfights an owner and his collaborators, and, as we shall see, to a lesser but still quite real extent also their backers on the outside, put their money where their status is.

It is in large part because the marginal disutility of loss is so great at the higher levels of betting that to engage in such betting is to lay one's public self, allusively and metaphorically, through the medium of one's cock, on the line. . . .

What makes Balinese cockfighting deep is thus not money in itself, but what, the more of it that is involved the more so, money causes to happen: the migration of the Balinese status hierarchy into the body of the cockfight.

Much of the power of Geertz's analysis comes from his elaborate "theory which sees cockfight wagering as the link connecting the fight to the wider world of Balinese culture," or, as I would put it, his explanation of the way in which the betting system changes Balinese social reality (one influencing space) so as to make a better match between relations in that influencing space and relations in the cockfight blend.

I have mentioned only the rudiments of Geertz's analysis. I have also skipped entirely his discussion of the principles according to which the center bet is made up of funds pooled by the owner and his social associates, and skipped his thorough analysis of the extraordinarily complicated system of side betting, which is not, in its appearance or financial structure, anything like the central betting, and which allows those not participating in the central bet to make public announcements–in the form of shouted bids–not of their evaluations of the cocks but rather of their allegiances to those engaged in the central betting. Geertz explains the complicated interaction between these two formally asymmetric kinds of betting as aspects of a larger non-linear cultural system of signification.

Betters in the cockfight blend are guided by this strong match between the social matrix and the cockfight blend. Placing any bet prefers one cock over another, and, via the match between the blend and the Balinese social matrix, it prefers one man over another. This presents an impossible situation to a bettor whose loyalties are evenly divided between the two men. A participant caught in this bind simply ejects himself from the blend: "where a man is caught between two more or less equally balanced loyalties, he tends to wander off for a cup of coffee or something to avoid having to bet . . ." Because there is a strong match between betting on a cock and preferring its owner, a change in social allegiance can be signaled by placing a certain bet: "There is a special word for betting against the grain . . . The institutionalized hostility relation, puik, is often formally initiated (though its causes always lie elsewhere) by such a 'pardon me' bet in a deep fight, putting the symbolic fat in the fire." Conversely, the cessation of hostility is signaled when one man bets on his previous enemy's bird.

But using betting to make a better match between Balinese society and the blend–so that which cock wins and which loses has immediate and large financial consequences for the actual Balinese men–creates a dicey problem. Social man must avoid open altercation, even confrontation. To be sure, the Balinese cockfight very nicely solves the problem by projecting the social aggression into the blend, where it is the cocks, not the men, who do the work. But this partitioning, while highly satisfying, cannot be entirely successful because the betting–the central mechanism for making events in the blend consequential for prestige and status in the society–actually has another consequence: it creates confrontational roles for the actual men inside the blend. Inside the blend, the men must confront, in some fashion, if they are to make a match, place bets, or collect winnings, and these are the very men who are socially constrained to avoid open, face-to-face confrontation.

Therefore, the blend is engineered as much as possible to deal with the sticky but unavoidable problem of confrontational roles for men who are supposed to avoid open confrontation. For example, while in many cultures it is common in betting for a neutral party to hold the stakes to make sure that the bets are actually paid, in the case of the Balinese cockfight there is an additional motivation for the umpire's holding the stakes in the central bet and awarding them to the winner: to make it possible for the two opponents to be completely disengaged at the moment of payment. This disengagement avoids, among other things, "the intense embarrassment both winner and loser would feel if the latter had to pay off personally following his defeat." The procedure for making the match and placing the central bet is designed to export as far as possible all the confrontation to the cocks:

After a fight has ended and the emotional debris is cleaned away–the bets have been paid, the curses cursed, the carcasses possessed–seven, eight, perhaps even a dozen men slip negligently into the ring with a cock and seek to find there a logical opponent for it. This process, which rarely takes less than ten minutes, and often a good deal longer, is conducted in a very subdued, oblique, even dissembling manner. Those not immediately involved give it at best but disguised, sidelong attention; those who, embarrassedly, are, attempt to pretend somehow that the whole thing is not really happening.

A match made, the other hopefuls retire with the same deliberate indifference . . . .

The center bet . . . is thus the most direct and open expression of social opposition, which is one of the reasons why both it and matchmaking are surrounded by such an air of unease, furtiveness, embarrassment, and so on.

Even side bets are paid hurriedly, furtively, with embarrassment.

The umpire is an index of this structural problem, which arises because men socially constrained to avoid confrontation are inserted into a blend where the central act is pure, violent, punctual slaughter, and where they bet in a pattern that cannot be leeched of all its confrontation. Uninhibited, overt, bloody confrontation can be left to the cocks, but the cocks cannot set rules or enforce them, keep time or hand over money, nor, for example, can two dead cocks decide which of them has won. The role of umpire is created in the blend to handle all of these potential sources of open conflict and thus to remove them as points of tension between two men who would, to decide them, have to confront face-to-face. The umpire's role is created in the blend by projecting to it all of the strong constraints against altercation that prevail in society. It is not just that the umpire's official authority is absolute; no one even protests or reacts to it:

I have never seen an umpire's judgment questioned on any subject, even by the more despondent losers, nor have I ever heard, even in private, a charge of unfairness directed against one, or, for that matter, complaints about umpires in general. . . . In the dozens of cockfights I saw in Bali, I never once saw an altercation about rules. Indeed, I never saw an open altercation, other than those between cocks, at all.

At the same time, and in the same way, the umpire indexes another fascinating aspect of the Balinese cockfight blend, namely, that the governing structure of the Balinese cockfight blend has been negotiated by many social actors over generations. This elaborate negotiation by many social agents is unusual, or at least, it marks the Balinese cockfight as a certain kind of blend, because other kinds of blends are not governed in this way. A daydream or thought experiment blend, for example, of course arises according to structural and dynamic principles of blending, and is governed by the set of competing guiding principles, but the mind doing the blending is free to try various projections without consulting the entire tribe about their acceptability. In the Balinese cockfight blend, however, there is a lot at stake, there will be a winner and a loser, so naturally, everything from making the match to engaging the cocks is governed by traditional negotiated rules. The carrier of this tradition of distributed negotiation, and the enforcer of its rules, is the umpire. His role requires exceptional authority of every kind, and predictably is created by blending together as many pre-existing authorities as the Balinese can find: "Likened to a judge, a king, a priest, and a policeman, he is all of these, and under his assured direction the animal passion of the fight proceeds within the civic certainty of the law." The holder of the role must be suitable for this authority, and so "Only exceptionally well trusted, solid, and given the complexity of the code, knowledgeable citizens perform this job, and in fact men will bring their cocks only to fights presided over by such men."

Again, not to belabor the obvious point, the Balinese social matrix has no role for "umpire," and certainly the domain of natural cockfights has no role for "umpire." The role of umpire, launched by the blending of "judge, king, priest, and policeman," is a new pivotal and indispensable role which arises in the cockfight blend as emergent conceptual structure. The meaning of the space with the Balinese men and the meaning of the space with the autonomous, natural cocks descend into the blend and create a new descendent, the Balinese cockfight blend, which has new, emergent meaning, including the new role for an umpire.

Having It Both Ways

The Balinese cockfight blend has peaceful social men and violent animal cocks blended into a contradictory package that may seem at first impossible or at least anthropologically exotic but that, in its principles, is normal and universal for human beings. When we daydream about being another person, or imagine ourselves better or worse, or wonder whether we would be happy doing this or successful doing that, or speak from our experience to advise a child, we are running a mental simulation in which, inside the blend, there is an element that is both us and not us. When we think about how things might be if only a particular event had not occurred, or had occurred, we are running a mental simulation in which, inside the blend, we have our life but do not have our life. In the blend, we have it both ways. If someone says, "Why don't you have a fax machine at home? I do. You could be reading my paper now," we are prompted to construct a "crosswise doubleness"–"doubleness" for the two influencing spaces, "crosswise" for the intersection of some of their meaning in the blend. We are asked to construct a conceptual integration network, a having-it-both-ways no different, in its mental operation, from the Balinese cockfight.

Of course, there are always astounding examples of any mental operation, blending included, and having it both ways can produce some hilarious scenarios. At a cocktail party following a presentation I gave on blending, a professor revealed that he had once, as a graduate student, returned to his apartment building so drunk that he could not get his key into the door, and despaired, but in a flash realized that he could solve the problem by pushing the buzzer for his apartment, thereby alerting himself that he needed to be buzzed in–an admirable blending of two scenes (himself at the door and himself inside his apartment) with a beautiful logic that indeed solves the problem because it replaces a difficult, precise action (insert key) with two easy, gross actions (push big buttons), yet which happens to fail in practice because its particular impossibility stops it from applying in a useful way to reality. You can't be both downstairs at the door to the building and inside your apartment. Blending in a goofy case like this does not differ as an operation from blending in "Paul believes he'll get his daughter admitted to Princeton because he thinks Mary is the dean of admissions," which prompts for a blend of some of what we believe to be true about Paul, his daughter, Mary, and admission to Princeton, but with emergent structure, true only in the blend, that Mary is dean of admissions and Paul's daughter is admitted. The falsity, perhaps even impossibility (Mary could be dead, for example), of this blend is now not funny, not even a hilarious mistake on Paul's part, but instead an explanation of his dispositions and actions contingent upon his beliefs and desires. "I wish I'd had your house; I'd still be living here," said a speaker to a listener as they viewed the remains of the speaker's home, which had been burned to the ground, both knowing that the listener owns an adobe home four hundred miles away, virtually fireproof but completely impossible for the speaker's lot.

All of these examples and, in fact, much of what we think, say, and do in an ordinary day require intricate, orderly, and impressive conceptual blending. Mundane blends are often as intricate in their conception as angels, marionettes, and the Balinese cockfight.

The Balinese cockfight has it both ways (slaughter and peace, cock and man) in the blend, but it also, like all conceptual integration networks, has it both ways in another sense: it allows us to think both inside and outside the blend. The blend does not eliminate the influencing spaces. On the contrary, the blend exists inside a conceptual integration network of different and interacting mental spaces, all of them with their uses. We can work inside the blend, or outside the blend, or in both simultaneously and interactingly. This having-it-both-ways by being both inside and outside the blend is recognized by the Balinese and analyzed by Geertz:

Fighting cocks, almost every Balinese I have ever discussed the subject with has said, is like playing with fire only not getting burned. You activate village and kingroup rivalries and hostilities, but in "play" form, coming dangerously and entrancingly close to the expression of open and direct interpersonal and intergroup aggression (something which, again, almost never happens in the normal course of ordinary life), but not quite, because, after all, it is "only a cockfight."

Operating at the same time in different mental spaces, inside the blend and outside the blend, is not irrational, not even unusual. The person cursing the spare tire, or the telephone, for refusing to comply with her wishes is not deluded. Although she finds it useful to have emotions and actions that make sense inside the blend, in which the phone or tire is intentional, she will, if asked whether she believes the object is literally refusing to comply, find it hard to take the question seriously. Working inside the blended space does not preclude working with the entire integration network, including of course the space in which it is "only" a tire or a telephone. We create mental blends to see whether we want to make them real, or to create emotional states, or to draw inferences that impinge upon reality, or to solve problems, or to achieve a compressed version of more diffuse knowledge, or to supply a global insight into diffuse knowledge, or to create new meaning, or to help us reason to choices, or for other purposes, and in doing so we often work inevitably, simultaneously, having it both ways, with a blend and an influencing space that are incompatible or even, sometimes, centrally opposed.

Blends let you do what you cannot do, be what you cannot be, not always so you can escape your situation, but instead, often, so you can learn about, make decisions about, and develop consequences for your situation, especially your mental and social reality, through events in a blend that, sometimes, for one reason or another, cannot or will not in fact be real. The Balinese, for example, when they are working outside the blend, enforce the sharpest possible distinction between man and animal: "The Balinese revulsion against any behavior regarded as animal-like can hardly be overstressed," writes Geertz. "Babies are not allowed to crawl for that reason." Even eating is regarded as disgusting, "to be conducted hurriedly and privately, because of its association with animality." And yet, in the blend, men can be animals:

In identifying with his cock, the Balinese man is identifying not just with his ideal self, or even his penis, but also, and at the same time, with what he most fears, hates, and ambivalence being what it is, is fascinated by–"The Powers of Darkness." . . .

In the cockfight, man and beast, good and evil, ego and id, the creative power of aroused masculinity and the destructive power of loosened animality fuse in a bloody drama of hatred, cruelty, violence, and death. It is little wonder that when, as is the invariable rule, the owner of the winning cock takes the carcass of the loser–often torn limb from limb by its enraged owner–home to eat, he does so with a mixture of social embarrassment, moral satisfaction, aesthetic disgust, and cannibal joy.


Representations and Instances


The cognitive features of the Balinese cockfight blend I have run through so far–influencing spaces, counterpart mappings, partial projection to the blend, emergent structure in the blend, possibilities for backward projection so that inferences and emotions developed in the blend can influence the original influencing spaces, and competing guiding principles like topology, integration, and metonymy compression–apply to all instances of conceptual integration, but the Balinese cockfight blend has an arresting exceptional feature that, however common, is not at all required for blending: it has an external, material representation or anchor.

My using the word "external" with the connotation "exceptional" conjures up for me, unbidden, a vivid blend in which Clifford Geertz, reading my prose, shakes his head in slow dismay. Geertz has spent a lifetime arguing that meaning is public. He has attacked cognitivist, behaviorist, and idealist "fallacies" for their misconceptions of culture. He has argued that "thinking as an overt, public act, involving the purposeful manipulation of objective materials, is probably fundamental to human beings; and thinking as a covert, private act, and without recourse to such materials, a derived, though not unuseful, capability."

Modern cognitive science, far from recycling the simplistic views Geertz has attacked, instead provides the best argument for the complex interactions upon which he has insisted. Human capacities result from interacting suites of genetic mechanisms, cellular mechanisms, developmental programs, physical environments, and social, cultural, and intentional environments. Cognitive science has generally recognized and demonstrated that only through empirical investigation can we come to know which suites and interactions actually make human beings what they are. The role of bodily, interpersonal, and public action in the development of minds that are able to engage in covert thinking is a common issue in cognitive science, from developmental psychology to cognitive linguistics. Cognitive scientists who study "distributed cognition" have shown the ways in which successful cognition often requires many functionally interacting agents and instruments, no one of whom conducts the thinking entirely or even mostly, as is the case when an aircraft carrier is successfully and coherently navigated by rotating teams of navigators, some of whom interact with other people and instruments on the ship. Edwin Hutchins' "How a Cockpit Remembers Its Speeds" is a fine, short study in this tradition. Cognitive science routinely explores the ways in which bodily states influence thinking, environments influence thinking, and human beings arrange their environments to serve, extend, and alter their thinking, or, metaphorically, how they rely on their environments to do some of their thinking for them.

"Public" and "distributed" are compatible with "mental" and "neurobiological." The natural conjunction of these terms, which have often been used as if they were antithetical, is part of what cognitive science has to say to social science. Meaning, attributed by intentional beings, indeed, by the brains of those intentional beings (in bodies, in environments, in cultures), can be crucially public, distributed, and indispensably dependent upon objects and situations, but it is still the people, not the tools or the spaces, who find things meaningful.

In the case of the Balinese cockfight, one can have the mental blend without having ever seen or engaged in the overt, public acts of its representation. Indeed, anyone reading this chapter has by now developed a mental blend of men and cocks, but Geertz is probably the only reader among them who has ever had the chance to develop some of that blend by learning it from its public representation in the fifty-square-foot ring in a Balinese village. And most of the very many mental blends a human being constructs in a day have no external representation at all. For example, I can have my blended Geertz–which, interestingly, is not my image of how Geertz responds to my assertions but instead my image of how other people's mistaken conception of Geertz responds–without anyone's being aware of the blend except me and without there being any perceptible mark of it and without there being any perceptible text of any sort that serves to prompt others to construct it and without my developing it through public, overt action.

But, by contrast, the Balinese cockfight blend does have a representation, or rather, particular instances of it have representations, with particular people and cocks in its various roles, and particular events that happen in and around the fifty-square-foot ring, events understood by everybody as a representation of the particular mental blend of these owners and these cocks.

It is characteristic of Geertz that he would select for analysis a blend with an external representation that is maximally public and maximally populated by an anthropological group, and that moreover, in its serial instances, involves the entire community. It is also characteristic of Geertz's preferences that the representation would, on its surface, look exotic to his scholarly audience, and that he would investigate it as a matter of historical retrospection, emphasizing its particularity. Interpretive social science does its work on this plane of historical retrospection and particularity, of overt public acts, of cultural meanings on their face inscrutable. But Geertz's analysis is compatible with working on a different scientific plane, one in which it is natural to see the Balinese cockfight as a product of a universal, common, routine mental operation that only exceptionally pivots on external representation, public performance, or group involvement. Blending these two planes gives us a fuller story of human meaning.

Geertz emphasizes that some of the purposes of the cockfight representation are shared in general by representational art, from Macbeth to David Copperfield. If your purpose is to prompt someone to construct a mental blend, then there has to be a prompt, perhaps a representation, and the prompting must be effective. What is most remarkable in the case of the Balinese cockfight, something that separates it from Macbeth and David Copperfield, is that it makes no sense to ask what "the inventor" of the Balinese cockfight hoped to prompt the audience to construct (in the way we can intelligibly ask, using the traditional principles of historical criticism, what the author of Macbeth might reasonably have hoped the mental reception of its performance would be). There was no inventor or author of the Balinese cockfight representation in the canonical sense. It seems instead to have developed by gradual accretion and refinement into a representation that prompts for a cultural conceptual integration network and that educates the community in its intricacy, without having anything in its history that resembles an intentional author. If that is the case, then the selection of the Balinese cockfight is in another way appropriate to Geertz's tastes: it shows not only maximal distributed cognition in the moment of the cockfight but also maximal distributed invention over its history.

Even more, it is a case where aspects of the representation (cocks fighting) existed before they were a "representation"–that is, before they had been given a constructed significance inside the conceptual integration network of the Balinese cockfight. The notion of the Balinese cockfight could have come up quite incrementally–all that is needed to start it is some minimal conceptual conjunction of social man and cock. After that point, any knowledge or fact having to do with cocks and their fights becomes a potential carrier of meaning in the blend. Whatever belongs to the representation tends to acquire conceptual significance: we are disposed to take elements of representations as prompting us to construct some counterpart structure for the conceptual blend, and to take structure in the blend as connected to the influencing spaces. This makes the Balinese cockfight a particularly fine example of the way in which the public actions of an artform can be prior to their significance, prior even to their status as artforms. The Balinese cockfight, historically and developmentally, begins where Geertz prefers, in the public and distributed arena.

Much of the ingenuity and appeal of the cockfight arises from an unusual feature: the representation deploys realities to which its influencing spaces refer. Specifically, one of the influencing mental spaces has cocks (as mental elements, of course) and refers to real cocks; the other influencing space has social men (again, as mental elements, of course) and refers to real social men. But the representation has real cocks and real social men.

To see how strange that is, consider a cartoon representation of a politician as a spider. The representation–the physical cartoon itself, printed on the page–does not deploy a real flesh-and-blood politician-spider (an element to which the blend refers), nor does it deploy a real flesh-and-blood politician or a real flesh-and-blood spider (elements to which the influencing spaces refer). Representations typically do not deploy real elements to which the influencing spaces refer. Consider, for example, the central panel of Rogier van der Weyden's Altarpiece of the Seven Sacraments, which naturally shows the Eucharist as located inside the church, but through a double representation: the first, minor representation of the Eucharist in the painting is the mass that is being celebrated inside the church, at the altar, in the background of the painting, by the priest who is raising a consecrated host; the other, theological representation of the Eucharist in the picture is Christ crucified, attended by the Virgin, Saint John, and the Holy Women, who are reacting to crucifixion as an immediate event, and who are located inside the church, in the foreground of the painting. We are to understand that the consecrated host is a blend of the object of transubstantiation and the body of Christ: as wafer, it has a human manufacture, and, as body of Christ, it has a divine origin. But Rogier van der Weyden's double representation does not deploy the real elements to which the blend refers or to which the influencing spaces refer–simply, the physical painting (paint, frame, et cetera) does not use wafer, Eucharist, something transubstantiated, something edible, the mouth of a communicant, something of divine origin, and so on. (It does use something of human manufacture, but not the human manufacture referred to by the influencing space with the wafer.)

Similarly, if we survey high canon representations of the Annunciation, we "see" the Virgin holding, anachronistically, a lectionary, opened to the narrative of the Annunciation. We have no trouble interpreting this representation as evoking a blend of a young girl and the Mother of God. The Virgin's bedroom may additionally have features of a church–the lectionary stand and veil that are part of the furniture of an altar, trinitarian tracery windows in Broederlam's version, a full Gothic church interior as in one of Jan van Eyck's versions. Annunciations may have a representation of God in the upper left, although we do not interpret this to mean that God was just up and to the left of the bedroom. They may also have lines interpreted as the "breath" of God, even though breath is invisible and the substance issuing from God's mouth is not only breath but also spirit and the creation of life. This inspiration may carry a dove that is simultaneously the Holy Spirit, or perhaps, as in the Mérode Altarpiece, a homunculus already tolerating his own miniature cross. There are lilies, a fountain (or kettle or pitcher and bowl), the snuffed candle, and, always, scenery familiar to the artist. The representation evokes a blend of girl with Mother of God (itself a blend, of course), bedroom with church, breath with life, and so on, but the physical painting (paint, frame, et cetera) that is the representation does not use a real girl, her real bedroom, a real church, God, or the Mother of God, nor does it use real breath, real lilies, a real dove, and so on. No doubt, the representation of the Annunciation represents elements to which the blend and the contributing spaces refer, but it does not actually use those real elements.

The Balinese cockfight blend, by contrast, has a representation that does use elements to which the influencing spaces refer. It uses real cocks. It has particular real owners placing a particular real central bet in a particular real village on a particular real day, and particular real cocks who really fight, with a particular real outcome. What all this reality represents is the particular mental blend of these particular owners and these particular cocks. That particular mental blend is of course an instance of the general Balinese cockfight blend, which has roles for owner-cocks and outcomes but no values in those roles. The general Balinese cockfight blend is the one we have been discussing all along, since we have made no mention of particular owners, particular cocks, or particular outcomes. But an enacted Balinese cockfight is always a representation of a specific instance of that mental blend, and–the surprising part–it always has real cocks that really fight. It is as if Rogier's representation of the Eucharist used a real crucifixion. It is as if the Mérode representation of the Annunciation somehow used a real dove, or, odder still, used real elements from Mary's real bedroom and a real church. The representation of a particular Balinese cockfight blend is of course quite real–real things really happen in the real fifty-square-foot ring. That reality always includes real elements–cocks–that are referred to by one of the influencing spaces.

This reality–real particular cocks really fighting–has a dynamic structure all its own. Cocks are live tactical agents. Deploying them in the representation has the amazing consequence that the central event in the representation is not scripted; instead, it is handed over largely to the unpredictable actions of real cocks. The general Balinese cockfight blend comes with structure, as Geertz puts it, "joining pride to selfhood, selfhood to cocks, and cocks to destruction," and this structure is entirely inherited by any particular instance of the Balinese cockfight blend, but in the particular instance, what gets destroyed and by whom is largely a matter for the real birds to settle. It is essentially the real cocks in the representation that are in charge of how the representation plays out, and how it plays out has strong influence on the structure evoked for the conceptual blend. In turn, the structure it evokes for the conceptual blend has effect for the actual pride of the actual men in the influencing space. Through this chain, the unpredictable dynamism of the representation has a profound effect on the influencing space that has the particular social men (as mental elements, of course), and, indeed, on the real psychology of the real men to whom that space refers. This makes the Balinese cockfight appropriate in yet another way for Geertz's tastes, since it portrays meaning, even the most vital understandings of self and other and society, as induced through public action whose root causality–which cock's spur bolts into which cock's body–is beyond covert thought, and certainly beyond individually-controlled covert thought. Individual inner meaning having to do with selfhood is, in this case, brought on by the actions of other agents in public.

Actually, blends that are set up like this, to be driven by their representations, are not all that rare. A particular Tarot reading of a deck of cards, for example, evokes the blending of elements of a life story with elements of a deck of cards. The representation, that is, the real turning over of the particular cards in a particular sequence, includes something referred to by one of the influencing spaces, namely the real deck of cards and its order. How that representation plays out is a matter largely of the accidental facts of that particular deck of cards–some cards will come up, in some order. These accidents in the representation evoke specifications of the blend, which in turn can induce revision of the influencing space with the person whose life is being read. Astrology is another case in which fate, the universe, chance, and systems beyond human control are allowed to have their say about human reality, through blending, and in which elements referred to by an influencing space–that is, the heavenly bodies and their dynamics–are actually deployed in the representation.

Leaving aside the possibility that various match sports are already blends of combat and cooperation, enemies and friends, and looking only at those cases where the matched teams are additionally national–America's Cup sailing, World Cup soccer–we again have a representation of the blend (in this case, a blend of national status and sports-team status) that includes real agents whose actions drive the specifics of the representation, thereby evoking structure in the blend, thereby affecting the influencing spaces. I was once surprised to find myself a little unsettled when it became evident that "we"–Americans in this case–were going to lose the America's Cup. I had thought I was interested exclusively in the sailing, and until that moment had been certain I wanted the Kiwis to win, so the next Cup could be held in the Hauraki Gulf in Auckland. I have been in European countries where the entire population has seemed to go into a psychological subduction near mourning because a few random guys–"their" soccer team–lost to a few other random guys "representing" a neighboring country. All of this makes more believable historical accounts of, as Geertz might put it, "the thrill of risk, the despair of loss, the pleasure of triumph" attendant upon a medieval joust. Battle by champion, if it ever in fact happened, would be an even more dramatic case in which the blend of political competition with man-to-man combat had a representation that used elements referred to by an influencing space, and whose dynamics actually altered the reality referred to by the political influencing space.

Geertz especially emphasizes the way in which a representation can give a public revelation of meanings otherwise veiled. In the case of the Balinese cockfight blend, the representation publicizes a set of social attitudes and dispositions that are otherwise strongly masked and never directly enacted: "The slaughter in the cock ring is not a depiction of how things literally are among men, but, what is almost worse, of how, from a particular angle, they imaginatively are." Public revelation is a medium of education:

Attending cockfights and participating in them is, for the Balinese, a kind of sentimental education. What he learns there is what his culture's ethos and his private sensibility (or, anyway, certain aspects of them) look like when spelled out externally in a collective text; that the two are near enough alike to be articulated in the symbolics of a single such text; and–the disquieting part–that the text in which this revelation is accomplished consists of a chicken hacking another mindlessly to bits.

Although some blends can be realized ("If he goes to the plastic surgeon in Rio, you won't recognize him"), the Balinese cockfight blend cannot be fully realized, not least because a single organism cannot be both man and cock, but also because the Balinese do not want their social ethos and private sensibilities enacted as confrontation. Yet its representation can make some of the elements of the blend real and can also make some symbols connected to other elements of the blend real, and this is important, for powerful as mental simulation is, it does not have the same force as perception and action. The reality of the representation gives a taste of an imaginative domain that the Balinese do not want realized. Having a bit of something, especially if we have it regularly, can be enough to let us see that (and why) we want only that much.


The Human Mind


In analyzing the Balinese cockfight as a cultural text, Geertz has equally exposed it as a product of a basic mental operation. That basic mental operation is universal to all cognitively modern human beings, everywhere, in all cultures, past, passing, and to come.

This claim, unbearably broad no doubt for the typical interpretive social scientist, understates what Fauconnier and I have argued since 1995, that conceptual integration, especially conceptual integration of the double-scope sort evident in the Balinese cockfight, is the mental capacity that makes human beings human, the one that separates them, and phylogenetically did separate them, from other species and from earlier anatomically modern human beings. Some species appear to be perfectly capable of perceptual and conceptual categorization, short- and long-term and episodic memory, social and natural intelligence, framing of even novel situations according to existing frames, and the adjustment, tuning, and refinement of frames. What they do not appear to be able to do with facility is double-scope blending. Fauconnier and I have proposed, for example, that double-scope blending was indispensable for the development of language or any systematic and flexible symbolism.

In 1994, Fauconnier and I published a technical report on blending, followed by some articles, and two years later, we each published a book–mine was The Literary Mind: The Origins of Thought and Language; his was Mappings in Thought and Language–presenting the elements of the theory. At nearly the same time, other scholars, working independently of us, began to advance similar claims emphasizing the singular importance of these integrations. In 1996, Steven Mithen, a cognitive archeologist, published The Prehistory of the Mind: The Cognitive Origins of Art, Religion, and Science, in which he argues that our species did not come into its own, did not invent what we think of as culture, until our ancestors developed a sophisticated ability to blend together structure from different conceptual domains. He calls this ability "cognitive fluidity" rather than "blending," and the flavor of his work is quite different from ours in its focus on archeology and the stages of hominid evolution, but it still offers an unmistakable and arresting overlap with our claims that double-scope blending was, as Mithen puts it, "the big bang" of human evolution. Even his diagrams and his examples are congenial with ours: he discusses people-animals in totemism, animal-people in anthropomorphism, animals with human social behavior in jokes ("A kangaroo walked into a bar and asked for a scotch and soda . . ."), people as objects to be manipulated, and dining as a blend of eating and social communication.

Also in 1996, Terrence Deacon, a comparative neuroscientist and evolutionary anthropologist, published The Symbolic Species: The co-evolution of language and the brain. Without explicitly addressing conceptual blending, Deacon argues that the essential leap for human culture and the human brain was the development of an ability to forge relational networks of symbols that hold together other relational networks, in a cascade, thus making it possible for human beings to draw together elements from many different domains. This evolutionary leap was hard. The incremental transition to nascent symbolic culture was made possible by the invention of ritual, which, in Deacon's view, is a pedagogical device for supporting the difficult learning of cross-domain associations and suppressing the disposition to attend exclusively to immediate tokens. The cognitive power of ritual lies in its public repetition–over and over again in the same scene, immediate and memorable–of prompts to forge and hold a specific set of cross-domain conceptual blends. In fact, his analysis of ritual is almost exactly parallel to the one Geertz gave in "Religion as a Cultural System":

In a ritual, the world as lived and the world as imagined, fused under the agency of a single set of symbolic forms, turn out to be the same world, producing [a] transformation in one's sense of reality . . .

Mithen and Deacon disagree on so many points that they might be surprised to see themselves shaking hands here in my text, but it is intriguing that such different thinkers, coming from such different fields, have proposed that the defining story for our species–culturally, intellectually, and neurobiologically–is the story of how we came to develop the ability to forge conceptual integration networks.

This story is neither triumphal nor joyful. Double-scope blending carries grave pain, not for genes, but for the emotional human minds routinely obliterated when human bodies die. A human mind lives in a dynamically shifting weave of many blends and through them constitutes its existence and imposes meaning, not always pleasant, on its life. A child who died horribly a decade ago is still with us, never leaves, is always there to cast his shadow on the day, even though our days have changed radically over the decade. In the blend, we can imagine him living and appropriately aged, and we do. We can cringe or smile at what we imagine to be our dead grandmother's reactions to our daughter's decisions, although in life our grandmother never met our daughter. We often take our cues for action, feeling, or belief from these blends. We assemble blended futures and choose between them, or blended counterfactual presents and grieve at their counterfactuality.

"Who has twisted us like this?" asks Rilke. "Wer hat uns also umgedreht?"

. . . the shrewd animals
notice that we're not very much at home
in the world we've expounded.
und die findigen Tiere merken es schon,
daß wir nich sehr verläßlich zu Haus sind
in der gedeuteten Welt.

No person, thing, idiosyncratic culture, or local event has twisted us like this, but rather our common phylogenetic development for a mental capacity that brings unprecedented power but no guarantee of pleasure–blending.

Blending is the deepest play of all. What I mean by "deep play" is the mental operation that makes us distinctively human. What Geertz means by "deep play" is a particular historical, and fairly bizarre-looking, sociological entity, optional, punctual, and relatively infrequent, somewhere on an island in Indonesia. But Geertz's Balinese deep play is a product of cognitive deep play, the mental operation of blending, which is neither particular nor past, neither punctual nor infrequent, neither bizarre nor optional, but constant, the working web of the human mind, the phenomenon that distinguishes the human mind from minds in other species. It is special to no culture and no epoch. It is universal, as old as cognitively modern human beings, and must continue to be characteristic of human beings for as long as the species exists. While its operation can be altered by conditions–during dreams and meditations for example, blends are decoupled from bodily action, and especially during dreams their governing principles seem to be relaxed–yet blending operates by the same structural and dynamic principles in art, science, religion, and everyday thought, consciously, unconsciously, during sleep, in covert contemplation and in public action.

To make such a claim is to take an intellectual flight away from Geertz's analysis of the Balinese cockfight. Yet in his own analysis Geertz does claim that many of the representational functions of the Balinese cockfight are shared by representation across cultures, and it may seem to some who have studied Geertz as if my claims are already implicit in his work. It sometimes seems that way to me. If Fauconnier and I had not spent four years hatching the theory of conceptual integration and making our way through technical analyses of its operation and principles before I encountered Geertz's article, it seems to me that his article should have provided me immediately and easily with many of the discoveries Fauconnier and I had managed to uncover only through great mental labor, often after many stages of reflection and revision. It is curious that in 1994, twenty-two years after the publication of "Deep Play: Notes on the Balinese Cockfight," when Fauconnier and I first published our work on conceptual integration, there was no prior theory of conceptual integration, not even a recognition of the mental phenomenon, for us to use as a springboard. Hadn't Geertz already laid it out in 1972?

But then, I feel the same way when I read The Runaway Bunny, published in 1942, one of the two most popular and successful picturebooks for two-year-olds. In The Runaway Bunny, a bunny talks with his mother (already a blend, if one of the most routine). He says that he is going to run away, and the mother says she is going to come after him. He says, "If you run after me, I will become a fish in a trout stream and I will swim away from you." His mother responds, "If you become a fish in a trout stream, I will become a fisherman and I will fish for you." Some of the blends are highly intricate, and depend on previous blends. An especially complex blend occurs when the little bunny says, "I will become a little boy and run into a house," and his mother says, "If you become a little boy and run into a house, I will become your mother." Two-year-olds have not the slightest difficulty putting together the blends and drawing the appropriate inferences. If a two-year-old who knows that fishermen use hooks and bait to fool fish, to snag them, to hurt them, to haul them in, and to eat them is looking at the illustration of the mother-bunny-fisherman fishing for the baby-bunny-fish with a carrot-hook on the end of the line, and you begin to ask questions, the dialogue goes like this: "What is this?" "A carrot." "What is it for?" "To catch the baby bunny." "What will the baby bunny do?" "Bite the carrot." "Will he swim away down the river." "No. He bites the carrot." "What is the mommy bunny doing?" "Fishing for the baby bunny." "What is she?" "She's a fisherman." "Does the baby bunny know his mommy is fishing for him." "No. He wants the carrot." "Can the baby bunny swim?" "Yes. He's a fishie." "Does he have a fishie tail." "No. He's a bunny." "Will the carrot hurt the baby bunny?" "No! The mommy doesn't hurt the bunny!" "What will happen when the baby bunny bites the carrot?" "The mommy bunny will pull him in and hug him and kiss him." "Will he smell like a fish?" "No! He's a baby bunny!" The two-year-olds effortlessly conceive one amazing blend after another and project the relevant inferences back to the influencing space of mothers and toddlers, of which of course they and their own mothers are an instance. Perhaps this is why the book speaks profoundly and memorably to two-year-olds, and why many thousands of copies of this book are sold every year. An analysis of The Runaway Bunny reveals the operation of conceptual blending in much the same way an analysis of the Balinese cockfight blend does.

Unlike The Runaway Bunny, however, Geertz's article has explicit commentary in nearly every paragraph, about "crossed conceptual wires," "joining w to x, x to y, and y to z," and the disarrangement of "semantic contexts in such a way that properties conventionally ascribed to certain things are unconventionally ascribed to others, which are then seen actually to possess them." This is the Geertz who has argued in support of the hypothesis of "the psychic unity of mankind"–that is, the proposition "that there are no essential differences in the fundamental nature of thought processes" (a noun phrase that presupposes that there is such a fundamental nature of thought processes) "among the various living races of man", and who, in what may be the most famous and influential essay of the second half of the twentieth century, "Thick Description," writes:

What, in a place like Morocco, most prevents those of us who grew up winking other winks or attending other sheep from grasping what people are up to is not ignorance as to how cognition works (though, especially as, one assumes, it works the same among them as it does among us, it would greatly help to have less of that too) . . .

Not, I grant, a ringing endorsement. Geertz has said of the Balinese that they "never do anything in a simple way that they can contrive to do in a complicated one," and that preference for complication graces Geertz's sentence. He does not say that knowledge about how cognition works would "greatly help," but rather that less ignorance would greatly help. His endorsement of cognitive science, or at least what I take to be cognitive science, is buried in a parenthesis, and I had to look hard through Geertz's works to find even a parenthesis this clear. Most damagingly, his backhanded endorsement is set in the context of telling us that what we most of all need is not an understanding of universal human cognition.

Nonetheless, and I hold onto this, it does say clearly, and Geertz maintains this clarity in some other places, that cognition–by which I take it he means basic mental operations–works the same in us and in Moroccans, and, by implication, in all human beings.

How then do we account for the fact that interpretive social science and cognitive science have not yet enthusiastically joined forces to study human meaning? The impediments between them seem to me to be not matters of scientific principle but instead matters of aesthetics and ritual, vocabulary and emphasis. The cognitive scientist does not emphasize retrospective interpretation of historical particulars, after the fact, for their own sakes, yet the interpretive social scientist is strongly disposed by temperament to do exactly that: he "strains to read" cultural texts "over the shoulders of those to whom they properly belong." The cognitive scientist tries to explain the mental operations that underlie those texts and those readings, but the accomplishments of cognitive science remain unfulfilling to the interpretive social scientist because they are not interpretations of specific subjects of the sort the interpretive social scientist is trained to give and wants to give.

To be specific, the interpretive social scientist may remark that the theory of conceptual integration, in itself, does not even begin to constitute an interpretation of the Balinese cockfight. That is quite true, in exactly the same way, and for exactly the same reason, that the theory of evolution does not, in itself, even begin to constitute an explanation of the wild cherry tree outside my window, or even distinguish it from the dogwood, the cedar, the pine, the yew, the Japanese maple, the oak, and the locust, or for that matter, distinguish the trees from the Salvadoreans, the African Americans, or the Armenians I can see in the three houses beyond the trees. To explain species requires retrospective investigation into the actual historical paths of evolution–over niches, under accidents, through divisions now irreversibly entrenched–the many paths where reproduction developed its bag of tricks, and the human mind formed its dispositions.

But the theory of evolution is indeed part of the explanation of each of these things; it helps explain the wild cherry tree in explaining the dynamic biological principles according to which it came to exist. The theory of evolution allows us to connect this particular tree to another of its same kind, and one kind of tree to another kind, and trees to other species, not for purposes of simplifying but because, as a matter of fact, in their process and in their descent, they are connected to one another.

In the identical way, the theory of conceptual integration is part of the explanation of the Balinese cockfight. The theory of blending attempts to explain the nature of the basic cognitive operation by which the Balinese cockfight blend arose and descended. The theory of blending additionally lets us connect the Balinese cockfight to a vast set of other products in other cultures, and thus–perhaps countervailing the preferences of interpretive social scientists–to show the ways in which the Balinese cockfight is not at all exotic but instead shares a category with various other conceptual meanings, some of which belong properly to us.

Consider, for example, the current American infatuation with playing the stock market for the ostensible purpose of "wealth accumulation." The mood of the country has changed in the last thirty years, and many people who in the nineteen-sixties and nineteen-seventies either despised capitalism outright or threw a social cordon around it as vulgar, tedious, and above all to be prevented from sullying their personal identity, now invest themselves proudly and publicly in following the daily ups and downs of their equity portfolios. Sometimes they do this hourly. They digest quick news reports on financial matters, follow the ticker, keep an eye on foreign markets tied to U.S. markets. Once in a while, they even make a trade. Their roller-coaster emotional reactions in a single day, and their reactions to the end-of-the-day closing stock prices, are frequently entirely irrational by their own standards. These investors are not day traders; they do not strap on the seat belt and forbid themselves lunch to stay on top of puts and calls; they are not engaged in market arbitrage. Instead, they have bought securities they mean to hold for a long time, and they know that the prices of these securities must wiggle around, insignificantly, inside trading ranges; they know that they haven't made or lost any money until they actually execute a trade; and they know that the closing price has no special significance as long as nothing interesting happens overnight. Yet slight and insignificant changes in prices that are quite unlikely to signify gain or loss in the long term can make an investor feel, at the end of the day, subdued and anxious, or, in the other direction, elated, eager to dine out and brag. If some of these investors are attaching pride to selfhood, selfhood to stock picks, and stock picks to market prices, then we have a blend that not only in its operations but also in some of its content is similar to the Balinese cockfight blend. As The Wall Street Journal put it on March 30, 1999:

With its unpredictable movements, second-to-second pulse and, lately, air of invincibility, the stock market has become a living entity for many of the 80 million or so Americans who own equities. It's there ticking away, at the breakfast table, the gym, the office. Sweeping indicators like market shifts and hot stocks somehow connect to the intensely personal–the retirement account, the college fund–as calculations of net worth blend into notions of self-worth. (Suskind, 1999).

So, yes, I claim that there are basic mental operations that unite the yuppie investor in 2001 with the Balinese cockfight participant in 1958, and that those basic mental operations have intricate systematic and dynamic principles. But, no, I do not claim that investing and cockfighting reduce to blending or to each other. That would be absurd. The theory of blending is by no means a reductive formula professing to explain the products of blending, any more than the theory of evolution is a reductive formula professing to explain frogs, toads, and salamanders.

"Man," writes Geertz, in his most famous phrase, "is an animal suspended in webs of significance he has spun." Yes, and blending is his main way of spinning them. Blending is to human beings what web-spinning is to spiders. To analyze individual webs, each tailored to its local situation, is of course indispensable and illuminating, but particular, retrospective, local analysis of individual webs should be combined with the study of what makes web-spinning possible, a theory of the nature of web-spinning. Interpretation shows us the web; cognitive study connects past webs to each other and to future webs in its attempt to explain the underlying capacities that make all of those webs possible. It is only a natural widening or narrowing of our focus, not a toggle between opposites, when we move back and forth between the study of the Balinese and the study of human beings. With enough of this back-and-forth, the two varieties of research may grow together.

"I take culture," Geertz continues, "to be those webs, and the analysis of it to be therefore not an experimental science in search of law but an interpretive one in search of meaning." Here, at last, Geertz and I part ways, or apparently so, depending on how heavily he means to emphasize "search for law," since cognitive science does not search for law either, in the senses that physics and chemistry do. Cognitive science is at once an interpretive science in search of meaning and an experimental science weighing linguistic, behavioral, genetic, sensory, and neurobiological data, making hypotheses, building models, offering explanations, sometimes offering even predictions or tactics for intervention. Interpretive science and experimental science are, for Geertz, a dichotomy. For me, they are compatible and mated parents only occasionally at one another's throats, with a baby called "cognitive social science."

At the conference at the Institute for Advanced Study on 25 Years of Social Science, in May 1997, during the intellectual stock-taking about the future of social science, I offered the view that the future of social science lies in the blend of cognitive science and social science. Where exactly do we stand? Between two different research agendas, our focus divided between them. Where do we go from here? Toward bringing them together as sharing the identical object of study. What kind of work do we want to sponsor? The integrative kind, finding ways in which cognitive science and social science supplement each other. What kinds of problems should we be addressing, with what kinds of approaches and arguments? The integrative problems, with whatever approaches and arguments from cognitive science and social science look as if they might be useful.

What should the School of Social Science in the Institute for Advanced Study do? That seems clear. It should find a way to support this integration. I even bet–here comes the blend–that the Luce Foundation would put up some of the money.