3.The cultural and the agentic

Culture is thus an effect as much as a cause…

–Herb Gintis1

The late Pat Bateson used to tell a joke about two philosophers. ‘Thinkers can be divided into two kinds’, says the first philosopher, ‘those who propose dichotomies, and those who reject them’. ‘Nonsense!’, replies the second.

At risk of similar tendentiousness, I contend that there are two major styles of social explanation. They have their origins in lay talk about reasons for actions, and they run through the professional discourses of social researchers. Extremes of both styles fail, and they fail in complementary ways. Our job is to make synthetic theories that capture the valid insights of both styles of thinking whilst also transcending them. I will call the two styles cultural and agentic respectively. I am not sure these are perfect names. There are various synonyms and near-synonyms for each one knocking about. Nonetheless, my chosen names are reasonably memorable and I will stick with them here.

First, let’s point out that human societies have an order, an order that transcends the minute-to-minute decisions, or even the lifetime decisions, of any one individual or interacting dyad. When I wander down the lane to the bakery, I don’t have to devise a strategy for making it clear to the baker that the object of my desire is the leavened wheat product on the shelf. There’s a convention that both of us unthinkingly subscribe to but neither of us invented, of denoting this with the sounds ‘loaf of bread’. I don’t have to offer to write the baker a scientific paper, teach her programming, or tend her garden (about all I could offer in improvised barter) in exchange for her comestibles. We have an institution called money by which exchange of anything for anything is possible and requires no further contestation other than specification of the price. I also understand that I am not to haggle over the price; that she will sell me her bread at the same price regardless of how many loaves she has left; and that it would be completely unacceptable for her to ask different prices of different customers, for example on the basis of the colour of their skins. These last two rules (price unrelated to scarcity, price unrelated to identity of buyer) are particularly interesting. Strategic agents making improvised decisions might well, you’d think, charge more when supply seems short relative to demand, or when selling to people they don’t like, but in the shops on my street, there are social rules that you don’t do that. And these rules have moral force; people treat them as if they were binding and are outraged when they are violated. So there is a lot of social order going on, even in the simple act of procuring my lunch. This order—its structure, diversity, and evolution—is to social theory what fritillaries and swallowtails are to lepidoptery.

You are informally offering what I call a cultural explanation every time someone asks you why we don’t haggle over price in Newcastle, or voluntarily eat horses, or allow polygamy, and you answer: that’s just our culture. That’s what is normative here. For this to constitute anything like an explanation rather than just a restatement of the phenomenon, you must be claiming something along the following lines. The social order itself, or something that encodes it, has a real concrete existence external to individual actions, is causally primal in respect of those actions, and hence explains those actions in a non-trivial sense. Individual actors inherit and reproduce this order, with not much more deliberation and choice than when we inherit and reproduce our DNA. Whereas the inheritance and reproduction of DNA happen by meiosis and mitosis, the inheritance and reproduction of the social order happen by socialization. To the extent to which people seem to be exerting free choice, they are only doing so within the constraints and set of acceptable roles that the social order makes available to them, like alternative expression levels of the same DNA sequence. If we want an explanation for the existence of the social order, we need to move to a different level of analysis, in which the explanatory forces will be something other than individual choices, since individual choices are the consequence, not the cause, of the social order under which they occur. Thus, in cultural thinking as I define it here, the social order, or the cultural rules that encode it, is the upstream source of individual actions.

Different flavours of cultural theory abound in social science. According to certain versions of the ‘cultural evolution’ paradigm, humans have a very general propensity to acquire and internalize whatever is normative in their culture. They automatically adopt the norms of the majority of people they encounter, or of the most locally-prestigious people they encounter. They do so, according to the theory, largely credulously; that is, without regard to how those norms suit their interests. So much so that, according to one hypothesis, whole societies can and do fail through slavish adherence by their members to self-injurious norms, in what has been termed cultural group selection.2 The reason that the societies we observe have fairly sensible norms is not that the people in them exerted good sense, but that all the ones that happened to have bad norms have gone extinct. That’s a pretty strong claim.

To take another example, in cultural theory in the tradition of Michel Foucault, the very ideas that people can conceive of as true, reason about, or discuss, are the products of a symbolic order that pre-exists them, known as an episteme.3 The hegemonic episteme of the age controls what seems right and natural, and limits people’s understanding of the current world or possible alternatives to it. You might think you hold beliefs because they are true, or do things because you want to, but really the episteme has constructed you to think that way. It is insidiously guiding and constraining you at every turn. It is a causal force. And where does the episteme come from? Again, its origins lie at some different level of analysis, upstream from everyday behaviour and apparently voluntary individual choice.

So these are examples of how the cultural corner looks. What about the agentic? Here the leading professional representative is the easiest to characterise: the ‘rational actor’ model of microeconomics and political theory. In agentic theories such as this, the individual is a sovereign decision-maker. She has a set of preferences over different bundles of outcomes. These preferences are causal and primary, though of course they can vary according to the context—effectively, the set of options available—and the state of the chooser. The social order, rather than being causal, is the downstream consequence of the preferences of many agents interacting over time. If I prefer Strauss to Stravinsky, it’s not that the sinister episteme has socialized me with normative conceptions of what music can be. I just like it. If Strauss is popular, it’s because many other agents share my preference. If the divorce rate goes down, it’s not that the social system ‘needs’ more stable families, or subtly coerces people into particular matrimonial roles, or even that a particularly virulent cultural meme has taken hold. It’s simply that under the current set of economic and demographic circumstances, more people are finding that staying married is an attractive strategic option relative to the option of leaving their marriage.

How could an agentic theory deal with the existence of the social order? Social interactions occur, according to agentic thinking, when both parties prefer having them over not having them, and both parties seek the form of interaction that comes highest in their register of preferences. The type specimen is the mutually-beneficial exchange between the buyer and seller of a material good. You do it when it suits you; if it didn’t suit you, it wouldn’t happen. But the rational actor analysis does not need to stop at monetized exchanges between strangers: a scientific collaboration, a romantic relationship, or a commitment to a voluntary organization can all be captured by the same logic. The interactions go on as long as they provide something for which both parties have a preference relative to the available options for the other possible uses of their time and energy.

Taken too literally, rational actor thinking might seem to imply that there are no norms at all, just atomic strategic agents. That is, every time I go into the bakery, the whole business of fixed and universal prices is up for potential renegotiation; or every year in a marriage, each partner has to check their current preference function and negotiate the terms of sexual exclusivity for the next accounting period. But this claim would be unfair on the rational actor model, and since being unfair to the rational actor model is something of a convention in much of social science, it is important to get this right. The rational actor model need have no problem with the idea that there are norms, or even that the norms are in some sense binding. The point is that the norms themselves should be analysed, at a deeper level, as the outcomes of the preferences of interacting individual strategic agents through time. So the social order is the dependent variable, with human preferences as the independent variable, whereas in cultural explanation it was the other way around. We can analyse the norm of fixed and universal prices in the bakery as if it were the product of a voluntary agreement between customers like myself and the baker. In a society where good bread is abundant and affordable, I would prefer not to have to enter a great long discussion about how much money I should hand over every time I want a loaf. The baker for her part values my repeated custom more highly than the opportunity to make an extra pound or two on a one-off occasion when I am desperate, and knows she will gain this by offering me the convenience of fixed prices. So in equilibrium, it’s always £3, and both parties accept the norm.

There’s always a danger that agentic thinking will become Panglossian: whatever social order exists is necessarily for the best, since free agents have brought it about by acting in accordance with their preferences. This is an interesting contrast to cultural thinking, which is very often dystopian and critical of the current order. The impetus behind much cultural theory, for example, is the drive to expose the subtle roots of domination and oppression inherent in the episteme. But we can remain agentic without being Panglossian. The social orders that come down to us, even though they are expressions of past agents’ choices, may not be optimal for the way we live now. People have differed historically in their power and control of resources, and hence social orders have been produced that favour some people at the expense of others. It is then rational to try to change them to make them fairer. Plus, importantly, there are many scenarios where benign equilibria cannot spontaneously be reached without bringing other kinds of institutions into being. For example, in ‘tragedy of the commons’ type situations, individuals acting in an uncoordinated way will end up at an equilibrium that none of them prefers. The solution is to advocate a higher-level mechanism of enforcement, such as a system of fines, laws or customary rules. This higher-level binding social mechanism requires active, deliberate work to bring it about; but it is still in some sense the outcome of the preferences of rational sovereign agents.

§

What are the strengths and weakness of cultural and agentic thinking, respectively? They are the mirror images of one another. The cultural view correctly captures the insight that the social order is cumulative, historical, and has emergent properties; and it does so more naturally than agentic thinking does. The English political system, for example, is of such a Byzantine complexity that no two freely-interacting representative agents could possibly come up with it in a reasonable time. Its peculiar design features—a second legislative chamber that consists partly but not mostly of hereditary aristocrats, the fact that the monarch is the head of state and yet wields no power, and so forth—probably don’t instantiate the preference functions of any of British citizen, living or dead. They represent the current snapshot in a kind of descent with modification process. This process has a historical quasi-life of its own, so much so that the cultural perspective is in some ways right to see the institutional system as the unit of analysis, and individual politicians as partly its current vehicles, rather than the other way around. Many of the properties of the system, although they may have arisen from the voluntary decisions of certain individual actors, were not predictable from those voluntary decisions, and certainly do not represent the outcome of the actors’ intentions. When non-independent, non-omniscient, socialized agents interact through time, this generates emergent properties and historical continuities. And if this applies to the English political system, it applies too to any system of meaning, knowledge, social organization or technology in any human society.

The other great strength of the cultural perspective is that it captures the fact that the social order is, for each of us and to at least some extent, unchosen and empirically real.4 People do often adhere to social norms, even when there is no prospect that a violation will be detected or punished, just because they are the norms. Behaving normatively thus cannot be completely reduced to some immediate strategic or prudential calculation, such as the desire to avoid getting a bad reputation, although such concerns do of course exist. And the available ways of talking about a topic do undoubtedly influence the possible actions we entertain: this is true in science as much as in political life. These ways of talking are, in a broad sense, inherited from our culture, and often go unquestioned much of the time.

But the weakness of the cultural way of thinking is in its inability satisfactorily to account for social change; and hence, in the end, to explain which properties of the social order endure. People follow the norms they are socialized into, except when they don’t. Sometimes they say: I don’t want to do that any more! It’s lame! It’s not right! I protest! It may be what my parents did, what most people do, what the prestigious people do, but I don’t feel it suits my interests and I will abandon it/modify it/flout it. So in an important sense, people are faithful replicators of norms only when those norms suit their current perceived interests and opportunities to a reasonable extent; otherwise, they try to change them in decidedly non-random ways. If you don’t allow for this in your social theory, allow for the order-transforming, purposive exercise of human agency, you really have no useful account of how societies develop, or how we end up with the historical conflicts and compromises that we do.

Another interesting difference between cultural and agentic thinking concerns the extent to which social groups are conceived of as heterogeneous. Cultural thinking leads us to think of each society as having a perfectly shared norm structure, ethos, or episteme.5 This might well be different from that of a different society or historical period, but, within the society or historical period, everyone subscribes to it; within-group homogeneity and between-group variation. Hence the anthropologist’s fantasy that you can talk about ‘the culture’ or ‘the norms’ of the Fuegians or the Russians as more than a statistical summary of a distribution. But this really is a myth: careful study actually reveals that there is, in many domains, much more variation in values and behaviour within cultural groups than there is between them.6

For agentic thinkers, by contrast, the central and most important truth about human societies is that they are made up of diverse individuals with different ideas, preferences and interests. The social order is partial and contested; to the extent that it exists, it is the uneasy truce in innumerable arguments and compromises between people with different wants and values. It is always, therefore, unstable and provisional, imperfectly agreed, and will be obeyed unevenly. At every moment it will be challenged, disputed, violated, transformed and renegotiated, usually in small ways and occasionally in larger ones. It is this ceaseless seething of variation and challenge that gives the system its dynamism, but also which makes it alight over historical time on consensual, or at least hard-to-overturn, solutions to the problems of communal living.

Culturally-oriented writers do sometimes talk about within-group heterogeneity in values, and of individuals challenging the social order. For example, people may contest or resist the dominant episteme rather than just accepting it, according to their own context and perspective. This immediately raises the question: what are these people standing on whilst they try to rip up the floor? Where is the vantage point outside of an episteme from which you can critique it and formulate a different one more suited to your interests? What resistance to dominant ideologies tells us is that it can’t be socialization all the way down; the cultural explanation, to be complete, requires that there is an agentic bedrock people can stand on to sometimes contest the episteme. Malleable as people may be to the norms and discourses of their time, there are some properties of natural agency—some no-doubt fuzzy set of recurring intuitions, motivations, judgements, and interests—that stand beyond the local social order, and which people can use to resist, criticise or reinforce that order. Maybe these properties of natural agency are only intermittently visible, as it were distant mountains often lost in the epistemic fog, but they must be there somewhere. Their roots, it is reasonable to suggest, lie in the legacy of the deep natural history shared by all living humans.7

§

As things stand, the social sciences are balkanized between agentic parts and cultural parts. Economics is largely agentic in its style of reasoning, and much of political science follows suit. My own disciplinary watering hole, behavioural ecology, also favours the agentic.8 By contrast, parts of Sociology and Anthropology are more cultural. Clearly this split is a problem. Some kind of agreed division of labour persisted for a few decades. For example, for large-scale societies, the discipline of Economics dealt with monetised, rational, instrumentally-motivated exchanges of goods and services between strangers, using agentic thinking, whilst Sociology dealt with non-monetised, normatively-governed features of life such as family formation, households, and voluntary associations, using, with exceptions, mainly cultural thinking. But this division of labour does not really work. For one thing, the rational-actor reasoning of Economics can be readily applied within households and to non-monetised interactions; for another thing, it is clear that morality and normative concerns play a big part even in monetised transactions amongst strangers.9 When you are in a restaurant in a foreign city, do you pay a tip? If you are anything like me, your first question will be: what do people round here generally do? I want to do that.

Cultural and agentic thinking both capture something real, and satisfactory social theories need to encompass both. It’s hard to find clear roadmaps for how to do this: usually authors start off in the camp belonging to one perspective, and at most hint at the other. The most useful source I have encountered is Herb Gintis’ Individuality and Entanglement.10 Gintis’ book is a serious call to arms to reintegrate Economics with Sociology and Anthropology, and hence bridge the agentic and the cultural. Central to Gintis’ analysis is the rational actor model. At the primary level, humans are agentic decision makers. Typical economist, you say, this is just the same old individualistic, agentic stuff. How do you explain morality? How do you explain why people follow norms when no-one is watching? Hang on—you are in danger of making a common conflation of rational actor with self-regarding actor. The micro-economists’ rational actor model only says that people can be modelled as having preferences, and that when they do things, they follow these preferences in a consistent way. For example, if they prefer A over B and B over C, then they will choose A over C. The model nowhere says that the preferences people hold have to be selfish ones. The preferences could be aesthetically, morally or socially motivated. They could be altruistic or self-harming. The rational actor model only makes a claim about how preferences will be translated into decisions. Thus, it’s a fallacy whenever you read (and you often do read): ‘The facts that many people give blood, or donate money to strangers, violate the predictions of the rational actor model’. Not necessarily. The donors could be acting perfectly rationally, just as the rational actor model requires, if their preferences include improving the welfare of others. What generosity violates is the hypothesis that people’s preferences are exclusively self-regarding. But you should have noticed that anyway.

Gintis suggests that we have three broad kinds of preferences, self-regarding (when we are hungry, we want food); other-regarding (when we see someone in distress, we want to help); and normative (we want to follow the rules, because they are the rules). But how do we integrate these very different preferences, when so often it seems that they might point us in different directions? There is no fundamental problem here. After all, we have no difficulty with the idea that we might integrate conflicting self-regarding preferences: we simply weigh them against one another, trading off if need be. I have to decide how much of my money to save, and how much to spend on running shoes. I might well be influenced in my trade-off by the prevailing rate of interest on savings, or the prospect of a looming expense next year, or a sale at the running shop. The same logic of weighing up and trading off applies when integrating self-regarding and non-self-regarding preferences. I want to help others, but the amount I invest in doing so will reasonably depend on the effectiveness of help I can deliver to them, the other calls on my time, and any rules I would have to break to do so. On my bicycle I like to stop at red lights, even when no-one is around—I think it’s a good norm—but I might sacrifice obeying it if I was in a hurry, or in order to come to someone’s rescue. And if I thought it was a bad norm, I would have a lower threshold for breaking it. So, in short, all of our varied motivations, self-regarding, other-regarding and normative, can simply be thought of as producing preferences of a unitary kind, preferences that get weighed against one another and traded off in moment-to-moment decision making, according to the context.

Because we have normative preferences—desires to do the appropriate thing, not for instrumental reasons, but for its own sake—a social and moral order can emerge, be largely obeyed and be somewhat stable, even when there are moment-to-moment self-regarding incentives for violating it at every turn. The order becomes internal and psychologically real. This sounds like a good thing, but is not always so. If the local norm seems to be that other people behave selfishly, then selfishness can spread rapidly and become a locally stable anti-social order.11

The existence of normative preferences allows Gintis to capture what is good about cultural thinking. However, our normative adherence is not slavish; because we also have self-regarding and other-regarding motivations, if the norms get too costly to our self-interest or the welfare of others, we may prefer to violate or abandon them. This captures what is good about agentic thinking. Thus, norms both structure our behaviour in the short term, and yet are structured by our behaviour in the long term, in the sense that bad norms can be changed.12 Norms are empirically fixed, but transcendentally negotiable. And there is an emphasis I would like to add: when people find norms too costly, often they do not just individually abandon them. Instead or in addition, they talk to others about changing them. They produce, in the public sphere, reasons and arguments for why the social order should be different: political actions. Sometimes they even manage to persuade one another, and social change ensues. This aspect of social life, the use of reasoning and conversation as a means to change the rules of the games we live by, all the while accepting that there must be rules, is neglected in many contemporary treatments of culture, which focus instead on automatic, unreasoned socialization. But it is the possibility of reasoning and conversation that gives me, as a scholar and a member of civil society, most hope for our common future.13

Gintis’ analysis has the capacity to generate everything from extreme cultural thinking—if you assume that people’s intrinsic preference for upholding norms is very strong relative to their other preferences—through to business-as-usual agentic analysis, if you assume that normative concerns have a trivially low weight compared to other concerns. The strength of normative motivation is thus like a slider with which you can fade from the very agentic to the very cultural and back again. Gintis does not solve where the slider is actually set for the typical human. Indeed, different passages of his book have quite different emphases in this regard. But surely, if we can take this as a framework, we can go beyond simply calling one or the other style of explanation wrong, and instead design a unified empirical programme that tries to find out where the slider is set, and what affects this setting.

§

I worry that we are too cultural when reasoning about people who are very different from us, and too agentic when reasoning about ourselves and our friends. Let’s say anthropologist A claims, in her academic work, that the behaviour of the tribal people she studies is well explained by their propensity to internalize norms through socialization, leading to slavish within-group homogeneity in values and beliefs. Presumably, she doesn’t believe these same principles account for her own behaviour in adopting this view. When asked, she doesn’t answer that the reason she holds this view is because she was socialized to do so, and she slavishly accepts whatever she is taught. She says she adopted her view because she thought about it deeply and thinks it is a good theory, possibly in spite of the view being non-normative in the field. Her daily professional life shows theorist A that her fellow anthropologists don’t easily accept the norms of the disciplinary community either: they take great pleasure in ceaselessly and idiosyncratically quibbling with her, for substantive as well as strategic reasons. As a consequence, far from there being cultural homogeneity in her discipline, there are as many shades of opinion as there are scholars, if not more. So the cultural group that theorist A actually has most experience of—her academic community—doesn’t seem to fit the very theory she espouses. She and her fellow academics live in an agentic world, where different individuals have different values; beliefs require substantive justification; everyone is intensely sceptical about the claims of others; and the result is a myriad of clashing and shifting opinions within the same cultural group. Her study subjects, by contrast, seem to live in a cultural world where everyone credulously and automatically accepts what they are taught, and there is stable within-group consensus. So either: academics in Western societies are profoundly different from other humans; theorist A is right about her study subjects, but deluded about herself and her fellow academics; or theorist A is right about herself, but deluded about her study subjects.

It’s easy to see where this double-think comes from. When we don’t know much about a category of people, that category looks homogenous, and we represent it cognitively with a few sweeping and static generalizations. That’s all our direct experience allows us to do. So that’s how, if we are not very careful, we end up thinking about people in faraway places or distant times: more culturally. But where we have more direct personal experience, we build a richer representation, with more room for individual heterogeneity, the diversity of motivations, the within-group conflicts, and the ever-shifting dynamics. So that’s how we think about ourselves and the other academics within our own discipline: more agentically. Thus, a bullshit test for social-theory frameworks that I rather like is to ask: do I find that framework rich enough to account for the social lives of the people I actually know? If the answer is no, then I see no reason it should be rich enough for anyone else either.

A version of double-think that I encounter in my own work concerns the behaviour of poor people, and why it differs from the behaviour of rich people. Poor people within developed countries are relatively likely to do various things that harm their health, such as smoking and overeating. A frequent mode of explanation for this in the literature is to say that, because their adverse lives grind them down, poor people are in less of a position to exert agency or free choice than rich people are, so they end up manipulated by the tobacco or sugar industries.14 I find this argument uncomfortable. I accept of course that poor people have fewer options than the rich (there are fewer ways to furnish 3000 calories per day on a small budget than on a large one), and may also experience different benefits from health behaviour (no point in avoiding smoking if you are likely to be killed in an industrial accident by age 50 anyway). I also accept that people can be manipulated by commercial interests. But the argument that rich people are somehow more agentic than poor people, rather than equally agentic but with a different set of constraints and incentives, disturbs me. In effect, it seems to be saying, the rich may be fully human rational actors responsible for their decisions, but the poor are just credulous patients doing what they are told; a lower stage on some kind of scala rationis humanae. They need to be helped, as children must be, up to the point where they will be capable of choosing for themselves, but they are not yet there. The fact that these ideas come from scholars who are sympathetic to poor people, critical of capitalism, and progressive in intent, does not for me completely mitigate the discomfort of the double standard. I could live with the idea that we are all the passive victims of commercial interests; or the idea that we all choose our health behaviours, according to the constraints and incentives of our circumstances. But I can’t get entirely comfortable with the idea that there are different modes of explanation for rich and poor.15

What I take away from this is the following principle: we should have a strong prior that all humans are just as cultural and just as agentic as each other. I don’t mind where you place your thinking on the cultural-to-agentic continuum—though I am more disposed to the agentic than the cultural, I can see the merits of several different positions. Wherever you place it, though, I think it should start at the same point for everyone, regardless of: the colour of their skin; their level of education; whether they are a hunter-gatherer, a subsistence farmer, an unemployed welder from Glasgow, a university professor, an antique Roman, or a Dane. Either we are all agents, it seems to me, or none of us is.

---