A Theory of Ideology
J.M. Balkin

Yale University Press
New Haven & London

Copyright © 1998 by Yale University. All rights reserved.

Online version

With the permission of Yale University Press, the author has made an online version of this book available under a Creative Commons Noncommercial Sharealike (by nc sa) license, accessible at the author’s website at JackBalkin.Yale.edu



Our discussion so far has been aimed at dissolving the study of ideology into the larger study of cultural understanding. Now we must take the opposite approach: we must break ideology down into smaller and distinctly analyzable parts. The goal is to replace the study of ideology with the study of ideological mechanisms and ideological effects of cultural software, which, taken together, produce what previous theorists have called ideology.

A great failing of traditional Marxist models of ideology has been that they have usually not offered very detailed explanations of how ideological beliefs are formed in individuals and how these beliefs could be produced by psycho­logical and cognitive mechanisms. Instead, traditional models have tended to focus on the distorting content of beliefs and on whether these distortions tend to serve or disserve the interests of various classes.

By contrast, the theory of cultural software focuses on the mechanisms that produce ideological effects and the means by which they spread widely among human minds. This task requires a study both of social psychological mechanisms and of the ways in which people share their understandings of the social world through the spread of language and symbolic forms.

To this end, the next four chapters offer a partial catalogue of the ideolog­ical mechanisms produced by different forms of cultural software. The exam­ples I offer here are not intended to be exhaustive. A comprehensive study of the various devices of human understanding and their possible ideological ef­fects is the work of many lifetimes. Rather, I wish to offer a sample of the wide diversity of tools of human understanding and the many different ways in which these tools can misfire and help produce and sustain injustices.

My goal, however, is not simply to show that different kinds of cultural software can produce ideological effects. It is also to show that these ideological effects are cultural-that they are widely distributed among members of a culture through memetic transmission. When previous theorists have spoken of ideol­ogy, whether in neutral or pejorative terms, they usually have meant shared ways of thinking. For example, a Marxist might hold that many members of the pro­letariat suffer from the same illusions about the inevitability of their lot; a femi­nist might point out that most men tend to view certain occupations as inappropriate for women, and so on. But the idea of “shared” ideology brings us back to the metaphysical puzzles that we encountered in Chapter 1. How can these beliefs be shared without presupposing unworkable theories of causation or implausible supraindividual entities? Because traditional theories of ideology have tended to focus on the distorted content of beliefs, or the interests or func­tions they serve, they have not faced this question squarely.

The theory of cultural software does provide an answer to this question: people experience similar ideological effects because they share similar cultural software and because this cultural software is employed in similar contexts with similar results. If ideological effects are shared, they must be produced by the kinds of memes that can spread widely and reliably through a single population or a group of related populations. This epidemiology produces the effect of shared ideology.

There is no reason to think that memes that produce ideological effects have the same nature or operate in precisely the same way. They need only share an ability to spread widely among human minds. Indeed, the next several chapters will examine very different features of human cognition, including narrative construction, cognitive dissonance reduction, heuristics of decision, metaphor, and metonymy. My examples will be drawn from a wide variety of sources and social scientific models. Often they will have been identified and expounded by theorists with very different theoretical commitments. I shall consider, for example, ideological mechanisms identified by methodological individualists and structuralists, positivists and antipositivists, cognitive theorists and anthropologists, social psychologists and literary critics. It is likely that several of the scholars whose work I draw upon would object strenuously to being discussed together. They would no doubt disagree heatedly about the right way to approach the study of human understanding.

Nevertheless, I must ask the reader to see beyond their various methodo­logical disputes and focus instead on the products of their respective researches. Each of them has, I believe, identified isolated examples of a single, central phenomenon. Each theory, suitably reinterpreted, reveals aspects of human thought with two basic features. First, in each case we have a cognitive mech­anism that is “ambivalent”: it serves human understanding in some cases and contexts, yet hinders or frustrates it in others. Second, each of these cognitive mechanisms is cultural, in the sense that it can be and is spread to many dif­ferent people through communication and social learning. In short, each of these theories reveals a kind of cultural software that, under the right condi­tions, can act like an ideological virus.

This, I believe, is the proper way to approach such diverse theories about human understanding. The advantage of the theory of cultural software is that it allows us to see how very different research projects can be reinterpreted and united under the umbrella of memetic evolution. Appropriately, this approach is itself a form of bricolage, for it cobbles together different ways of under­standing human understanding in the hope of providing a more powerful and unified account. Because human understanding is itself a process of bricolage, we might think of this method as a sort of “metabricolage.”[1]

The next four chapters consider a wide variety of phenomena, each of which operates differently from the others. The present chapter takes up heuristics of decision and strategies of cognitive dissonance reduction. Chapter 9 concerns the cultural software contained in narratives and scripts, Chapter 10 discusses networks of association, and Chapter 11 explores metaphoric and metonymic models. The variety of these examples amply demonstrates that cultural soft­ware does not have to take any particular form to have ideological effects, as long as it can spread widely among a population. Yet as we shall see repeatedly, human language plays an important role. Language is the most effective carrier of memes and is itself one of the most widespread forms of cultural software. Hence it is not surprising that many ideological mechanisms either have their source in features of language or are propagated through language.

Mechanisms Hot and Cold

I begin with heuristics of decision and strategies of cognitive dissonance re­duction. Here I shall build largely upon the work of Jon Elster. Elster, in turn, built on the work of two different theories of social psychology: Leon Festin­ger’s theory of cognitive dissonance, and Daniel Kahneman and Amos Tver­sky’s studies on heuristics and cognitive  biases.[2] My goal in this section is twofold. First, I wish to describe the kinds of ideological mechanisms that Elster identified. Second, I want to show the limitations of his approach, and how it can be reinterpreted more broadly and fruitfully in terms of the theory of cultural software.

Elster’s work is admirable because he is one of the few theorists of ideology working in the Marxist tradition who has attempted to break down ideology into its component parts and ask how beliefs are caused rather than focusing on the interests they serve. He has turned to the theories of heuristics and cognitive dissonance to provide “micro-foundations” for the Marxist theory of ideology and to put it on a firmer scientific footing.[3]Unlike the structuralists I shall discuss in the next chapter, Elster’s approach is motivated by a com­mitment to methodological individualism in the social sciences. Because he hopes to explain as many social phenomena as possible in terms of individuals, their actions, and their beliefs, he tries to explain ideology in terms of individualpsychological mechanisms.[4] His project thus attempts to reduce the Marxist theory of ideology to a theory of individual psychology.

Elster divides ideological mechanisms that distort belief into two groups, which he calls “hot” and “cold.”[5]These correspond roughly to the theories of Festinger on the one hand, and Kahneman and Tversky on the other. “Hot” mechanisms are motivational; they are attempts to reduce cognitive dissonance. ”Cold” mechanisms are cognitive; they involve heuristics or cognitive biases.[6]

Festinger’s theory is only one of several motivational theories that might explain ideological effects. Moreover, the theory of cognitive dissonance has undergone considerable innovation since his original formulation.[7]Cognitive dissonance is now thought to be produced not by the mere fact of conflict or contradiction but by the self’s need to preserve its view of itself.[8] Thus, we might call “hot” mechanisms self-preserving or ego defense mechanisms.

In contrast, Kahneman and Tversky’s approach explains distorted beliefs not by the need for ego defense but through various failures in cognitive pro­cessing. These defects include heuristics or cognitive rules of thumb that op­erate effectively only in certain limited situations but are extended to situations in which they do not apply.[9]These are classic examples of tools of understand­ing stretched beyond their usefulness.

Together these two theories of social psychology offer causal explanations for many of the ideological effects described in Marx’s writings. For example, one can replace the familiar “dominant ideology” thesis by showing that op­pressed groups engage in self-defeating strategies and hold self-defeating be­liefs. These strategies and beliefs result from cognitive and motivational biases of the oppressed group rather than being imposed from above by dominant groups. Nevertheless, these biases and illusions may benefit dominant groups, even if they had no hand in producing them.[10]

Elster’s central example of a motivational bias produced by the need to reduce cognitive dissonance is wishful thinking: people form beliefs because they prefer a world in which the beliefs are true to a world in which the beliefsare false. Elster’s formulation of wishful thinking is inadequate because it relies on Festinger’s original formulation of dissonance theory. Later theorists have pointed out that mere self-contradiction among one’s beliefs does not neces­sarily produce dissonance reduction because people are able to live with all sorts of contradictions. Only contradictions that threaten the self’s view of itself will cause people to reduce dissonance.[11] So people engage in wishful thinking not merely because they prefer a world in which a certain state of affairs is true to one in which it is false. They engage in wishful thinking because accepting that the world is a particular way would significantly threaten their views about themselves. A person might change her beliefs, for example, if holding a certain belief would make her seem less moral, less worthy, less capable, or less in control of her life. People try to preserve belief in states of affairs when they have a personal or existential stake in them.

The need to reduce cognitive dissonance may also cause changes in a sub­ject’s values and preferences. One example of this is the phenomenon of “sour grapes.”[12] People adapt their preferences to value what they believe is poten­tially available to them. Conversely, they tend to undervalue that which they believe to be impossible or unattainable. Another example of dissonance re­duction is the tendency to believe that states of affairs in which we have a stake or which otherwise advantage us are not too immoral or too unjust, or do not show us in a particularly bad light. Lawyers who participate in the adversary system, for example, often have to represent reprehensible clients and argue for positions that they do not believe. They can justify their activities on thegrounds that it is demanded of them by the legal system. But if they believed that the legal system was fundamentally unfair, this justification would be se­riously undercut and their activities would look much more morally problem­atic. Hence they have incentives to believe either that the system as a whole is basically fair and just or that precisely because the system is so unfair and unjust to people like their clients, they are entitled to bend the rules to level the playing field and make it more just.

Dissonance reduction seems to explain a number of ideological effects as­sociated with the Marxist theory of ideology. Exploited and oppressed groups, for example, may sometimes believe in the justness, propriety, or adequacy of their fate because this allows them to reduce cognitive dissonance. Such beliefs ”may indeed give short-term gratification, but cannot be said to serve the in­terests of these classes well at all.”[13] More often, oppressed groups may harbor no such illusions: they may bitterly resent the special favors they feel are granted to more privileged classes. But they may still engage in dissonance reduction. For example, they may openly spurn greater income, more privi­leged lifestyles, and the symbols of privilege as morally bankrupt or corrupt. At the same time, they may believe that improving their situation is impossible and that their condition therefore must be accepted. Although these distortions in belief may benefit dominant classes, the dominant classes have not caused them. They are caused instead by the need of oppressed classes to reduce the cognitive dissonance produced by coming to terms with the difficulty of their situation.

Elster also argues that Marx’s account of the ideological character of religious belief-that religion is “the sigh of the oppressed creature”–can be explained as a kind of motivational distortion.[14] Following Feuerbach, Marxargues that oppressed classes project the essence of humanity onto a supernat­ural being; in this way human beings are enslaved to the products of their own imagination. Elster reinterprets this projection as a form of wishful thinking. Human misery causes people to imagine a great and good being who has their interests at heart; this allows them to feel better about their lot because they know that God is watching over them. People create an object corresponding to their wishes and desires; then they understand it as an entity existing external to them so that they can appropriate it through religious devotion or prayer. Again, Elster emphasizes, what is important about this account of religious belief is that it does not explain religious ideology as a function of what serves dominant interests. Rather, it explains religion as the “spontaneous invention of the oppressed, not an ideology imposed by their oppressors.”[15]

Motivational biases also produce ideological effects in the beliefs of domi­nant or ruling classes. As Elster points out, one of Marx’s most important ideas is that “the bearers of a particular class interest tend to represent it as the general interest of society.”[16] Wishful thinking helps explains this phenome­non. The desire to reduce cognitive dissonance causes individuals to believe that what is in their interest is in the interest of society as a whole. It also causes people to have a distorted image of social conditions that support such a view. Thus people can alleviate their sense of guilt or responsibility about poverty by believing that the problems of the poor are exaggerated or that many of their problems are due ultimately to their bad character or immorality.

Finally, wishful thinking can produce strife between classes that might oth­erwise have common concerns. Middle- and working-class people who face economic insecurity caused by economic restructuring may blame the poor and governmental assistance to the poor for their problems. This both alleviates a sense of guilt or obligation toward those even less fortunate and allows middle­- and working-class people to feel morally worthy by comparison.

Cognitive biases explain a different set of ideological effects. For the most part these involve the misplaced use of heuristics; thus we might call them heuristic biases. Like motivational biases, cognitive biases can affect both ourviews about social conditions and our preferences. A cognitive bias that affects our beliefs about facts is the availability heuristic: “the tendency to believe that the world at large is similar to the part of the world one knows.”[17] The avail­ability heuristic assumes that evidence ready to hand is a good source of evi­dence about parts of the social world not directly experienced by or available to us.[18] A cognitive bias that affects our preferences is a shift in the framing of a problem. When a medical procedure or policy program is described in terms of its potential gains, for example, it seems more desirable than when it is described in terms of its potential losses.[19]

Excessive reliance on the availability heuristic is a particularly common cause of ideological effects. Often it leads to faulty generalization, where “the believer generalizes certain features of his local environment, wrongly believingthem to hold in a wider context.”[20] This phenomenon is reminiscent of Mann­heim’s point that members of groups tend to extrapolate their experience to all other situations.[21] A second and related cognitive bias is “conceptual im­perialism,” which occurs when a thinker uses the categories of her own society ”to understand the social structure of other societies or secondary structures within the same society.” For example, a thinker might apply specifically cap­italist categories to understand precapitalist or noncapitalist social structures. Or she might try to apply American constitutional and political structures to solve the political problems of very different societies with very different his­tories. Anachronistic thinking and ethnocentrism are familiar examples of this sort of cognitive bias.[22]

A third form of cognitive bias is the fallacy of composition: “the tendency to believe that causal relations that are valid locally, or ceteris paribus, retain their validity when generalized to a wider context.” A special case of this fallacyinvolves the “natural cognitive tendency to believe that statements which are true from the point of view of anyindividual agent remain true when applied to the totality of all agents.” Because of the fallacy of composition, “there is a natural tendency for the exploited to believe in the inevitability of exploita­tion.”[23] The proletariat commit this fallacy when they assume that because they would be worse off without the particular employer who oppresses them, a society without such employers or employment relations would be even worse.[24] Similarly, even if a working-class woman in a patriarchal society would be worse off if she were not in a traditional marriage relationship, it does not follow that all women would be worse off if the institution of marriage were significantly changed. The fallacy arises from assuming that the conditions of choice for members of the subordinate group would remain unchanged.

Elster’s approach to the Marxist theory of ideology is a genuine advance because it tries to provide causal explanations of ideological effects and because it tries to differentiate ideological phenomena according to their disparate causal sources. Instead of attempting to describe some monolithic entity called ideology, Elster’s approach implicitly recognizes that ideological effects result from the confluence of various motivational and cognitive mechanisms.

Nevertheless, Elster’s theory is necessarily limited by two features. The first is his adherence to the basic Marxist problematic of economic class.[25] In ad­dition to the psychological mechanisms just described, Elster also attempts to explain the relationship between ideology and modes of economic production. He views the theory of ideology as essentially concerned with economic class.[26] This is due in part to his ambition, as the title of his book suggests, to “make sense of Marx.” Nevertheless, it is interesting to note that the motivational and cognitive mechanisms that he identifies have no necessary connection to eco­nomic class; they work equally for groups defined in ways other than traditional Marxist criteria. Racial and religious groups, for example, may also engage in strategies of dissonance reduction that cause them to engage in self-justifying views of their present situation; ethnic and other social groups may also mis­takenly assume that the social world is similar to that with which they are most familiar. Thus the irony of Elster’s social psychological approach is that it shows once again how unnecessarily limited is the Marxist approach to ideo­logical explanation.

The second basic limitation on Elster’s analysis is that he pays little atten­tion to culture and language as sources of ideological effects. This is partially due to his reliance on social science models that bracket away questions ofculture and emphasize individual cognitive processes. It may also be due to his general suspicion of supraindividual entities as explanatory factors. Neverthe­less, many ideological effects–including those that Elster is concerned with­–occur through the use of shared linguistic and symbolic meanings and shared forms of cultural understanding. Elster’s theory of ideology is unduly restricted because he cannot easily assimilate the notion of cultural understanding into his model. In fact, as I shall argue, his own psychological model is actually a special case of what it seems to exclude and marginalize. The heuristic and cognitive biases he describes operate through the use of shared forms of cultural understanding.

Cultural Heuristics for Understanding Human Action

To develop this point, I want to turn first to the work of two quite different social theorists, Paul Ricoeur and Clifford Geertz. Ricoeur adopts a pejorative view of ideology, which assumes that ideology is a distortion, while Geertz adopts a neutral view of ideology as a system of social understanding. What unites these thinkers, however, and what differentiates them from the sort of approach we see in Elster, is that both emphasize that social reality is under­stood through (or, in their terms, mediated by) shared cultural symbols. Each argues that people use culture to understand their own actions and the actions of others. And each argues that in order to understand how ideology works one must understand that ideology is inextricably linked to the symbolic char­acter of social understanding.

Ricoeur views ideology as a form of distortion of a very special kind. Ide ology distorts our understanding of social practices (praxis). Ricoeur argues that people need shared cultural meanings about human action before they canunderstand their actions and those of others. Hence people always understand social practices and human action through the use of shared symbols. Indeed, social action itself is always mediated through symbolic understandings. Our understanding of what we are doing–an understanding mediated by shared meanings and symbols–is an important component of the choices we make and the actions we perform. Thus, Ricoeur insists, a system of symbolic un­derstandings of social action, or what Ricoeur calls a “symbolic structure of action,” must be in place before we can even speak of ideological distortion.[27] Shared cultural meanings about human action must already exist before ide­ology can do its work.

Geertz emphasizes that social understanding–and hence ideological un­derstanding–occurs through the use of interlocking and interrelated figures of speech. The symbolic mediation that is necessary to understanding occurs through tropes similar to the classic rhetorical figures of metaphor, metonymy, synecdoche, oxymoron, and personification. As an example, Geertz offers the statement that the Taft-Hartley Act (a piece of labor legislation opposed by organized labor) is a “slave labor act.” A supporter of organized labor who makes this statement is not literally claiming that the Taft-Hartley Act reduces people to a condition of slavery. Rather, she is offering a metaphor.[28]

Metaphor and other rhetorical tropes are tools for understanding social conditions, describing them to others, and persuading people about them. The persuasive power of a rhetorical figure like metaphor is not diminished by its lack of literal correspondence to the social world. To the contrary, its power arises precisely from the fact that it compares things that are admittedly dif­ferent and yet the same in some respect. It begins with an obvious difference and ends with a recognition of similarity. A successful metaphor “transforms a false identification … into an apt analogy.”[29]

Moreover, the ability of a metaphor to ring true or false depends upon an existing set of cultural associations in which it can be located. Preexisting cul­tural software determines how arguments are to be framed, how comparisons can be made, and how rhetoric can be wielded. It provides the framework in which the apt description and the inappropriate comparison can occur. Thus, for Geertz, ideology is a cultural system of interrelated associations, symbols, and figures. In other words, ideological effects depend on an individual’s par­ticipation in a system of cultural meanings and associations. Hence, Geertz argues, “the sociology of knowledge ought to be called the sociology of mean­ing, for what is socially determined is not the nature of conception but the nature of the vehicles of conception.”[30]

These analyses suggest why Elster’s bracketing of cultural and linguistic sources of ideological effects is too limited. Indeed, what is ironic about Elster’s oversight is that he uses metaphors to convey the difference between his two varieties of causal explanation. “Hot” connotes effects produced by emotion or affect, while “cold” implies effects produced absent emotion. Hence we speak of “cool reason” as opposed to the “heat of passion.”[31] Elster’s own use of these terms is the most telling demonstration that ideological effects can occuras much through metaphor, figure, symbol, and rhetoric–which involve shared categories of meaning–as through individual preservation of the self system or individual cognitive biases or heuristics. These mechanisms cannot easily be fit into his system.

The limitations of Elster’s analysis become most apparent in his treatment of tradition and the practice of borrowing traditional symbols in political dis­course, a practice that produces “the apparently conservative character of many revolutions.” In his discussion of Marx’s essay The Eighteenth Brumaire of Louis Bonaparte,Elster describes borrowing from tradition as a source of ideological distortion. He attempts to fit it into his model as an example of conceptual imperialism. He argues that it is produced by misuse of the availability heuris­tic. Nevertheless, Elster admits that this causal explanation of ideological dis­tortion seems different from the others he has advanced because “it offers an explanation of men’s conception of the future in terms of the historical tradi­tion, not their present position.” The problem in these cases is that people ”have to do with the conceptual luggage they carry with them, even at the very moment they grope around for a way to jettison it.”[32]

The ideological distortions created by the use of traditional symbols involve more than idiosyncratic failures of cognitive processing. Elster is describing tools of understanding that involve or employ shared cultural meanings. Thus these tools of understanding are more than individual heuristics–they are shared orcultural heuristics. As heuristics, they cannot be wholly detrimental. As noted previously, a heuristic and a bias are simply two sides of the same coin; what is a cognitive heuristic or aid to understanding in one context or situation can also be a cognitive bias or distortion in another. This is simply another version of the argument about conceptual bricolage that underlies the ambivalent conception.

Moreover, as cultural heuristics, these tools are partly constitutive of indi­viduals. To say that people are situated in culture is also to say that cultural tools are situated in them. As we saw in Chapter 1, this puts the concept of ”tradition” in a very different light. Tradition is not simply something we live within; it is something that lives within us.

Being part of a cultural tradition is a condition of historical existence. To exist as a historical being is to have a set of tools available to hand that are the legacy of the past. Existence in human history (as opposed to the natural existence of a mountain or a glacier) is existence in culture. It means that one is composed of cultural heuristics shared by others who are similarly constituted by them.

In the Eighteenth Brumaire, Marx takes a largely negative view of the cultural tools that constitute tradition. Near the beginning of the essay he offers his famous statement that “the tradition of all the dead generations weighs like a nightmare on the brain of the living.” Marx thought that “precisely in such epochs of revolutionary crises [people] anxiously conjure up the spirits of the past to their service and borrow from them names, battle slogans and costumes in order to present the new scene of world history in this time-honoured dis­guise and this borrowed language.”[33] Although in this particular essay Marx is critical of the use of the past, and of the mawkish and opportunistic use of political props from the past, there is a larger sense in which (as Marx himself emphasized) the phenomenon of borrowing from the past is unavoidable and inevitable. Indeed, political grandstanding–like a French senator dressing in a toga to call up the idea of the Roman Republic, or a modern American poli­tician identifying himself with the Founding Fathers–works only because there is a shared stock of cultural symbols that facilitates a shared political life.

This suggests an important connection between cultural heuristics and the public rhetoric that we use to persuade and influence others. As Aristotle noted long ago, the successful rhetorician builds upon what the rhetorician and the audience have in common.[34] And what the two have in common are shared cultural meanings and symbols. Thus rhetoric is also a sort of bricolage, and the skillful orator is a sort of bricoleur: out of the old and familiar, she con­structs the new and persuasive. That is why metaphor and figural language ingeneral are effective–because they relate the new to the old, the strange to the familiar, what we come to know to what we already grasp. Indeed, even Marx uses metaphor to make his points: He speaks of the past as weighing ”like a nightmare on the brain of the living”–a mixed metaphor to be sure, but a metaphor nevertheless.

Our search for ideological mechanisms thus takes us far beyond the limited confines of Elster’s model, important as that model remains. Indeed, our anal­ysis places the heuristics and biases that Elster identifies in a new light. If human action and social practices are always understood symbolically, then Kahneman and Tversky’s heuristics and cognitive biases are already embedded in and make use of a set of shared meanings about human action. These heu­ristics and biases already possess a cultural component; they are already a kind of cultural heuristic.

Elster tried to view the cultural heuristics of traditional thought as a special albeit exceptional case of his psychological model of ideology. In fact, it is quite the other way around. We should rather try to think of social psychological mechanisms, cognitive heuristics, and biases as special cases of a larger category of cultural software that includes many different types of shared cultural mean­ings and symbols.

Consider, for example, the mechanism of dissonance reduction. The liter­ature on cognitive dissonance has gradually come to recognize that what lies at the root of dissonance reduction is the preservation of the “self system.” Mere contradiction or conflict does not lead to dissonance reduction unless the self’s view of itself is threatened. But what is the source of the self’s view of itself, and what is the source of what the self regards as a bearable or unbearable conflict? In large part the source of both must come from culture and cultural norms internalized by the individual. Put another way, the preservation of the self system involves the self’s looking at itself through the eyes of what it imagines others in its culture would think about it. This process is in manyways reminiscent of Mead’s idea of a “generalized other” that shapes individual behavior and conscience.[35] Thus, although the basic mechanism of dissonance reduction is individual, the content and context of what drives this mechanism is cultural and social.

Moreover, we can describe the mechanism of dissonance reduction in ex­plicitly memetic terms. Each individual mind is a kind of ecology, more hos­pitable to some memes than to others. Beliefs that do not fit well into theexisting ecology of the mind are more likely to be altered, rejected, suppressed, or forgotten. Although the ecology affects the kinds of memes that will survive within it, the memes it absorbs also can affect the ecology itself. Thus new experiences sometimes alter existing beliefs, and new beliefs sometimes are altered to conform with beliefs already held. Strategies of dissonance reduction adjust beliefs and attitudes so that they can survive together in the existing ecology of the individual mind. People whose ecologies are similar-because of their common interests, their common situations, and the commonality of their previously existing beliefs-will provide similar ecologies for new memes. Thus they will tend to engage in similar strategies of dissonance reduction.

We can also interpret “cold” mechanisms consistently with the theory of cultural software. In their research on heuristics and cognitive biases, Kahne­man and Tversky do not discuss whether these heuristics or cognitive biasesare “hardwired” or whether they are culturally generated. Nevertheless, there are good reasons to think that at least some of them are a kind of cultural software transmitted from person to person. First, individuals can learn to avoid these cognitive biases when they are pointed out. Second, only a certain per­centage of individuals fall prey to these errors and biases in psychological ex­periments. This suggests that these heuristics are learned, adapted, and adopted in new situations and, conversely, that people also can learn when these heu­ristics are badly adapted to solving particular types of problems.

Of course, neither of these facts conclusively proves that all of these heu­ristics and biases are produced by transmissible cultural software. To the con­trary, some heuristics and biases may produce ideological effects because of the absence of cultural software. Some meme complexes may act like cultural ”patches” that allow people to work around the deficiencies of their hardwired heuristics. In that case the reason why only some people fall prey to errors is that they have not yet assimilated the necessary patches through social learning. So we must assume that the group of cognitive heuristics and biases compre­hends some combination of the cultural and innate. Nevertheless, even so-­called hardwired heuristics and biases depend on shared cultural meanings and concepts of human action to do their work. Transmissible cultural software still may be a necessary condition for most ideological effects to occur, even if it is not a sufficient condition.

Nothing in what I have said suggests an abandonment of Elster’s basic thesis: a theory of ideology must attempt to offer causal explanations of ideo­logical effects. My point is that causal explanations cannot bracket away the realm of culture. As Jerome Bruner puts it, “In the end, even the strongest causal explanations of the human condition cannot make plausible sense with­out being interpreted in light of the symbolic world that constitutes human culture.”[36]

The many ideological effects produced by language and culture fall into the category of cognitive or “cold” mechanisms. They are ideological effects produced by the mediation of social understanding through language, meta­phor, narrative, and other symbolic forms. Indeed, they constitute a much larger category of effects than the examples that Elster offers as paradigmatic of “cold” or cognitive mechanisms. Like other cold mechanisms, the tools of linguistic or cultural understanding operate as a kind of heuristic that can pro­duce ideological effects in particular situations. Moreover, it is not easy to separate out their beneficial from their harmful uses. The possibility of ideo­logical effects is built into the very concepts and structure of symbolic under­standing and discourse within a culture. Yet once again, while studying these tools for the ideological effects that they produce, we must not forget the extent to which such tools are empowering or enabling. Psychological and culturalheuristics are just that-heuristics that under some conditions perform well enough as rough guides to reasoning but that are misleading in other contexts. The study of ideology, then, might be summarized as the study of “when good heuristics go bad.”

To study the many features of human cognition and cultural understanding that can produce ideological effects, we have to cast a wide net. We have to bring together many different fields of study and many diverse types of theories about cultural understanding and the social construction of thought. This should hardly be surprising, for the theory of ideology is a necessarily and fundamentally interdisciplinary endeavor. In addition to the Marxist tradition, the sociology of knowledge, and the various theories of social psychology just discussed, we might also include structuralism and semiotics, Pierre Bourdieu’s theory of habitus, Hans-Georg Gadamer’s theory of tradition and horizon fu­sion, Wittgenstein’s theory of language games, Foucault’s theory of discourses, and the classical theory of rhetoric, as developed by Aristotle and many others.

Nevertheless, this synthetic project of metabricolage is complicated by two factors. First, the motivations behind these various theories are often consid­erably different. Some of these theories (like Wittgenstein’s or Gadamer’s) are philosophical accounts of cultural understanding. They attempt to evaporate certain confusions concerning how understanding works. Nevertheless, they do not offer anything like a causal or evolutionary explanation of ideological ef­fects.

Second, many of these theories (for example, Wittgenstein’s and Foucault’s) are either unconcerned with or actively hostile to offering accounts of the internal processes of the human mind. Instead, they view features external to the mind–like social behavior or symbolic forms–as the object of their study. This, in my view, is the most serious failing of the theory of discourse that has come to replace the theory of ideology. When discourse is viewed as the struc­ture or content of messages or practices, it casts the study of cultural under­standing out of the mind and into the world of behaviors, writings, and articulated symbols. Yet one must do more than identify particular discourses and their structures and effects. It is also necessary to ask how these discourses could be produced by individual human minds, and how what produces them could, in turn, be produced and reproduced in many other individual human minds.

There is a great irony here. The interpretive turn in the human sciences understood itself in part as a rejection of behaviorism. It emphasized the cul­tural features of human action and the importance of culture and symbols in structuring human behavior. Nevertheless, it is quite possible to re-create a sort of behaviorism within an interpretivist approach if we focus only on be­haviors and symbolic forms that are external to the mind, or if we treat the mind as a black box that simply produces and is affected by these symbols and behaviors. The interpretive turn, which we find in thinkers from Geertz to Foucault, has emphasized the role of symbolic forms and culturally meaningful behavior. Yet if these forms are symbolic, they must be symbolic to someone who is able to process and use symbols. If behavior is culturally meaningful, it must be culturally meaningful to a particular person who has some mechanism for making and understanding meaning. To understand the phenomenon ofideology, then, we must marry two separate movements of the second half of the twentieth century. The first is the interpretive turn, which emphasized the importance in human life of culture and the symbolic. The second is the cog­nitive revolution, which emphasized the internal processes of human under­standing. Each of these movements offers something that the other has either downplayed or disregarded. The theory of ideology–which is only a subset of a general theory of cultural understanding–must make use of each of these approaches and bring them together.

[1] I am indebted to Bruce Ackerman for the insight as well as the term.

[2] Jon Elster, Making Sense of Marx (Cambridge: Cambridge University Press, 1985), 466; Leon Festinger, ATheory of Cognitive Dissonance (Stanford: Stanford Uni­versity Press, 1957); Daniel Kahneman, Paul Slovic, and Amos Tversky, eds., Judgment Under Uncertainty: Heuristics and Biases (Cambridge: Cambridge University Press, 1982).

[3] Elster, Making Sense of Marx, 460-61; Jon Elster, Sour Grapes: Studies in the Subversion of Rationality(Cambridge: Cambridge University Press, 1983), 142.

[4] Elster, Sour Grapes, 142.

[5] Ibid., 141. Elster traces this distinction back to R. P. Abelson, “Computer Sim­ulation of Hot Cognition,” in S. Tomkins and S. Messick, eds., Computer Simulation of Personality (New York: Wiley, 1963), 277-98.

[6] Elster, Making Sense of Marx, 466-67; Elster, Sour Grapes, 141.

[7] See, e.g., Anthony Greenwald and David L. Ronis, “Twenty Years of Cognitive Dissonance: Case Study of the Evolution of a Theory,” Psychological Review 85 (1978): 53-57.

[8] See, e.g., Elliot Aronson, “The Theory of Cognitive Dissonance: A Current Perspective,” Advances in Experimental Social Psychology 4 (1969): 1-34, at 16-17; J. Rich­ard Eiser, Social Psychology: Attitudes, Cognition, and Social Behavior (Cambridge: Cam­bridge University Press, 1986), 93.

[9] Daniel Kahneman and Amos Tversky, “Judgment Under Uncertainty: Heuristics and Biases,” rpt. inJudgment Under Uncertainty, 1-20, at 1; R. Nisbett and L. Ross, Human Inference: Strategies and Shortcomings of Social Judgment (Englewood Cliffs, NJ.: Prentice-Hall, 1980), 6-7.

[10] Elster, Sour Grapes, 164-65; Elster, Making Sense of Marx, 505.

[11] Claude M. Steele and Thomas J. Liu, “Dissonance Processes as Self Affirma­tion,” Journal of Personality and Social Psychology 45 (1983): 5-19; Ruth Thibodeau and Elliot Aronson, “Taking a Closer Look: Reasserting the Role of the Self-Concept in Dissonance Theory,” Personality and Social Psychology Bulletin 18 (1992): 591-601; Green­wald and Ronis, “Twenty Years of Cognitive Dissonance,” 55; Aronson, “The Theory of Cognitive Dissonance,” 27.

[12] Elster, Sour Grapes, 148.

[13] Ibid., 156.

[14] Elster, Making Sense of Marx, 510; the quotation is from Karl Marx, “Contri­bution to the Critique of Hegel’s Philosophy of Right: Introduction,” in The Marx­Engels Reader, Robert C. Tucker, ed. (New York: Norton, 1972), 11-23, at 12.

[15] Elster, Making Sense of Marx, 482.

[16] Ibid.

[17] Ibid., 466.

[18] On sample bias and availability heuristics, see Nisbett and Ross, Human Inference, 77-89; Kahneman, Slovic, and Tversky, judgment Under Uncertainty, 163-208.

[19] See Amos Tversky and Daniel Kahneman, “The Framing of Decisions and the Rationality of Choice,” Science211 (1981): 543-58; Elster, Making Sense of Marx, 466.

[20] Elster, Sour Grapes, 144.

[21] Karl Mannheim, Ideology and Utopia: An Introduction to the Sociology of Knowledge (New York: Harcourt Brace Jovanovitch, 1936), 118-46.

[22] Elster, Making Sense of Marx, 490.

[23] Ibid., 487. Nevertheless, Elster also notes that “the exploiting classes can be victims of similar illusions. Cognitively based ideologies do not always operate to the benefit of the ruling classes.”

[24] Ibid., 488. Elster draws here on Paul Veyne’s work. The basic argument is that ”since I would be worse off without a master, it follows on this logic that a society without masters would be intolerable, for who would then provide employment and protection?”

[25] See ibid., 322.

[26] See, e.g., ibid., 464-65, 468-72. Although his discussion focuses almost exclu­sively on effects of class interests and class positions, it is interesting to note that his actual definition of ideology does not specifically refer to economic class.

[27] Paul Ricoeur, Lectures on Ideology and Utopia, George H. Taylor, ed. (New York: Columbia University Press, 1986), 8-10, 156-58. Thus, according to Ricoeur, ideology distorts “praxis as something symbolically mediated” (157). This argument provides yet another reason to abandon the familiar base-superstructure model of ideology inherited from Marxism. What that metaphor places in the so-called superstructure (culture) is actually basic to human existence and meaningful human action. Moreover, the super­structure does not exist purely for the purpose of distortion; it is not exhausted by its distorting effects. For example, Ricoeur argues that capitalist understandings of wage labor involve a distortion of praxis because the juridical concept of contract is applied to a situation of domination. But this does not mean that the idea of a contract is merely a fantasy or wholly an element of distortion. Rather, this tool of understanding is more than its distorting effects; it has independent uses, functions, and consequences. Nev­ertheless, it has been applied to a situation to which it is not fully appropriate; hence it gives a social situation an air of legitimacy that it does not deserve (155-56).

[28] Clifford Geertz, “Ideology as a Cultural System,” in The Interpretation of Cultures (New York: Basic, 1973), 209-13. In Chapter 11 we will consider metaphor once again through the work of George Lakoff and Mark Johnson, who argue that these rhetorical devices arise through a process of evolutionary development originating in the move­ments of the human body.

[29] Ibid., 211.

[30] Ibid., 212.

[31] See George Lakoff and Mark Turner, More Than Cool Reason: A Field Guide to Poetic Metaphor (Chicago: University of Chicago Press, 1989).

[32] Elster, Making Sense of Marx, 492-93. See Karl Marx, The Eighteenth Brumaire of Louis Bonaparte, in The Marx-Engels Reader, 436-525.

[33] Marx, The Eighteenth Brumaire, 437.

[34] Aristotle, On Rhetoric: A Theory of Civic Discourse, George A. Kennedy, trans. (Oxford: Oxford University Press, 1991).

[35] George Herbert Mead, Mind, Self, and Society, Charles W. Morris, ed. (Chicago: University of Chicago Press, 1962), 154-56, 178-226.

[36] Jerome A. Bruner, Acts of Meaning (Cambridge: Harvard University Press, 1990), 138.