Economics, cognitive science and social cognition

Don Ross

University of Alabama at Birmingham and University of Cape Town

don.ross@uct.ac.za

Abstract

I discuss the role of economics in the study of social cognition. A currently popular view is that microeconomics should collapse into psychology partly because cognitive science has shown that valuation is constitutively social, whereas non-psychological economics insists that it is not. In the paper I resist this view, partly by reference to the relevant history of economic theory, and partly by reference to an alternative model of the way in which that theory complements, without reducing to, psychological accounts of social cognition.

Keywords: economic theory, behavioral economics, sociality, atomism, aggregation

1. Introduction

This essay concerns the role of economics in the interdisciplinary study of social cognition. Increasingly many economists believe that economics has such a role. Most who hold this opinion do so because they think that, to some extent, important parts of microeconomics should collapse into psychology. They think this in part because they are convinced that most human motivation has turned out to be irreducibly social, whereas traditional microeconomics depended for maintenance of its distance from psychology on modeling people as if their social relations were incidental rather than constitutive. Bruni (2005) is a representative instance of the newer view.

The perspective I will defend here agrees that economics can and should contribute to the understanding of social cognition. Economics is an important part of a complementary suite of cognitive and behavioral sciences that accomplish more together than they could do in isolation. However, I do not believe that any part of economics should be collapsed into psychology, and I reject the widespread opinion that economics for much of its history 'went wrong' by ignoring the social dimension of value.

I will aim to do three things in the paper. First, I will describe the origins of the widespread misperception. As will be seen, both the cognitive revolution of the 1960s and the later interanimation of cognitive science and social theory are important parts of this story (see also Angner & Loewenstein forthcoming). Then I will explain why I think the perception is confused. Finally I will indicate why all of this matters: economics can make its distinctive and important contribution to our understanding of social cognition (and social behavior) only if it is recognized to have a different role from that of psychology. Economics is not equivalent to the psychology of valuation, though there is (of course) such a psychology, which is partly social, and economics helpfully informs it. It would reduce confusion, I believe, if much of what is now called 'behavioral economics' were referred to as 'psychology of valuation' instead.

The confusions I aim to dispel are historical in origin. Thus much of the essay will be about what economists call 'history of thought'. Let me therefore note that my motives are not the historian's. That is, I am not concerned per se with the way in which historical thinkers represented their own intentions and views to themselves. I am instead concerned with how we should critically regard previous episodes of reasoning in light of what we think we have learned that participants in these past episodes did not know. To illustrate with a simple example: the question 'Why did Copernicus set the scientific revolution in motion?' is not a question for the intellectual biographer, because Copernicus never imagined he was doing any such thing; but it is a perfectly good question for the historian of science who knows how events turned out and what led to what.

2. The supposed chastening of economics by cognitive science

It is doubtful that many economists were ever unaware, as one might suppose from reading some critics, that social relations are very important to people and to their economic activity. Nevertheless, it is noteworthy that an increasing number of economists have recently been motivated to produce models of socialization and reciprocal interpersonal preference modulation themselves[1], instead of simply acting as consumers of such models produced by their colleagues in psychology, sociology, anthropology, ethology and neuroscience. Most of the economists in question think that this reflects a discovery that in their modeling practice they had been neglecting important considerations. Often this is expressed in tones suggesting that economists should feel chastened by discoveries coming from cognitive science generally and cognitive neuroscience particularly (Camerer, Loewenstein & Prelec 2005).

The discoveries in question can be broadly sorted into four sets: (1) findings that people don't reason about uncertainty in accordance with sound statistical and other inductive principles; (2) findings that people behave inconsistently from one choice problem to another as a result of various kinds of framing influences; (3) findings that people systematically reverse preferences over time because they discount the future hyperbolically instead of exponentially; and (4) findings that people don't act so as to optimize their personal expected utility, but are heavily influenced by their beliefs about the prospective utility of other people, and by relations between other peoples' utility and their own. All of these are taken to threaten the supposed 'dogma' of mainstream (typically called 'neoclassical' or 'Walrasian') economics that people are rational and self-interested. Note that the findings in sets (1) – (3) undermine assumptions about practical (means-ends) rationality, while set (4) undermines assumptions about narrow self-interest. It is thus not obvious that worries about conventional theory motivated by (1) – (3) are directly related to sociality. I will later indicate why I think that they are. That (4) is directly related to sociality will be evident.

Economists, like other scientists busy accumulating new knowledge, often rely for their sense of the history of their discipline on textbooks. Textbook writers tend to like to introduce their subjects by spinning whiggish narratives about their histories, thus recruiting students to purposeful, progressive projects. This understandable aim requires no bad faith, because whiggish histories are not necessarily false. Perfectly reasonable narratives can interpret past cul-de-sacs in theory development as valuable lessons that inspired later progress, and therefore, at the relevant level of abstraction, as not really cul-de-sacs at all. So, let us give two cheers for whiggish history. We should withhold the third cheer only because relentless whiggery makes it difficult to comparatively evaluate the extent to which earlier shifts in doctrine, methods or modeling techniques represented important discontinuities, from which we can inductively learn now. In addition, whiggish attitudes cry out to be taken down a peg for their inevitable simplifications and air of self-congratulation; and this encourages counter-narratives of renunciation and self-chastening.

The view of economics as chastened by cognitive science is in part rebellion against bewitchment by a standard whiggish narrative on postwar economic theory, which celebrates some climactic technical accomplishments that resulted from mathematicization. In 1954 Kenneth Arrow and Gerard Debreu established that, under certain restrictions, perfectly competitive economies have uniquely stable, optimally efficient equilibria. They thereby completed an idea and an argument that had animated Adam Smith, a few predecessors, and most economists after him. Formulating this 'general equilibrium problem' in terms clear enough to permit technical solution required strong general constraints on the nature of the participants in economies. In particular, it was necessary to assume that the participants could rank all possible states of the world with respect to value, and that they never changed their minds about these rankings. Exact specification of these constraints on economic agency also had a long history in economic thought, and Debreu, once again, triumphantly completed this arc of progress (in 1959) by axiomatizing the required formal account of subjective valuation.

Against this background, the basis of the later humbling of economics was the 'insight' that Debrevian economic agents are not plausible models of whole people over the course of their natural lives. The truth of this can be made evident before any considerations having to do with sociality are introduced. On pain of failure of the mathematics, an economic agent cannot change its preference ranking over complete states of the world and yet remain the same agent. Yet no person's identity as an organism, or as a social entity defined partly in terms of obligations to particular institutions and people, is threatened by her discovery that (e.g.) she just doesn't like sweet salad dressing anymore. On the other hand, a person's identity is compromised by a discovery that she no longer likes her spouse, her profession, or her country. Social and behavioral theory is obliged to have something to say about the basis of the difference here, but the theory of the economic agent does not. Thus the theory appears to leave out something important (Davis 2003).

Critics of neoclassical theory often follow Sen (1977) in expressing this complaint as follows. Economic agents' preferences are modeled as logically prior to and independent of social context, whereas this is not true of people. Economics, it is said, is thus atomistic, whereas adequate social theory should not be. In Sen's famous phrase, a person who acted 'rationally' in the sense of the economic agent would be a 'rational fool' because her behavior would be socially dysfunctional and she would be shunned and (unless she were sociopathic) therefore miserable. The point here is not that the economic agent's preferences are restricted to selfish material consumption, a bogus worry with which Sen's complaint is often confused. The utility function of an agent can readily include another agent's welfare or happiness in its scope. Sen's objection is that utility functions are not dynamically mutually sensitive, something which the mathematical point of constructing them indeed forbids. Because social expectations are presumed to be represented as cognitive frames, this links the 'chastening' discovery set (2) above to set (4).

Promoters of chastening read the history of economics as if it missed a boat from the beginning in the way it took up the legacy of Adam Smith (Sen 1999; Sugden 2002, 2005). The tale is told as follows. Impressed by The Wealth of Nations, economists took to showing how atomistic agents could realize gains from cooperative exchange and division of labor in general equilibrium. Ignoring Smith's Theory of Moral Sentiments, they forgot that people's most strongly motivating rewards and punishments are functions of relations between their affective responses and affective responses of others. On this reading, the early builders of neoclassicism (W.S. Jevons and Leon Walras) were motivated to imagine atomistic, self-regarding motivational structures as models of human psychology. Microeconomic theory was then built around these models. At the end of the twentieth century, however, cognitive science showed that the Smith of TMS was right, while Jevons and every pre-chastened microeconomist since, were wrong. Hence we are experiencing a paradigm shift in economics, taking us back to the true Smith.

How relatively sound is this reading? The shift in economic theory around 1870, that gave rise to neoclassicism consisted in abandoning objective concepts of value (based, in different versions, on land or labor or some mix of these and other production factors) in favor of value as subjective preference. Fudging the boundaries around atomistic agents is sometimes seen as partly reversing this shift, since such fudging allows that value is determined socially, and, to that extent, independently of personal preferences. Arguably, however, holding that value is intersubjective is more an amendment to neoclassicism than a return to classical objectivism. In any event, let us agree that the abandonment of objective value was indeed a key discontinuity in economic thought. However, the history just told smooths over a later theoretical discontinuity that was at least as important.

In the late 1930s and 1940s two threads in economic theory that had been developing separately were tied together. One thread was Keynes's focus on aggregate structural features of large economies without regard to the kinds of individual agents or actions that compose them – that is, macroeconomics (though Keynes didn't use the word). The other was the attempt, clearly set in play by second-generation neoclassicists (Pareto and Fisher) near the turn of the century, to squeeze the psychological assumptions about economic agents down to a minimal core – ultimately, to nothing but consistency of preference rankings plus the idea that no agent would be content to consume only one type of good, no matter how cheap it became ('non-monomania'). Note that the second assumption is a substantial psychological hypothesis, and much more plausibly true of human beings than the first. Then, with Paul Samuelson's Foundations of Economic Analysis (1947), the need for even this final plausible human property was eliminated. Samuelson instead showed that certain properties of aggregate demand, which he believed to be generally empirically satisfied, could consistently be interpreted as revealing preference structures that fit the convex indifference curves of standard neoclassical analysis. The individual agent effectively disappears. She could be (and typically was) hypothetically inferred by theorists for the sake of making economics relevant to welfare in the context of a subjectivist philosophy of value. However, her characteristics played no role in the derivation of theory. Instead of incorporating a false psychology of people, economic theory after Samuelson broke its connection with psychology altogether.

The interpretation above is retrospective, and wasn't how Samuelson and his contemporaries understood matters. They were working in the heyday of behaviorism in psychology. They thus took it that ideas about how people internally represented their own preferences – most importantly for previous economists, their supposedly not 'liking' each additional increment to their stock of a good as much as they 'liked' the previous increment – were unscientific claims not just as economics but as psychology. Thus Samuelson and colleagues did not see themselves as abandoning psychology; they instead thought they were conforming to improved and up-to-date psychology. This is an important part of why the narrative that goes with the chastening attitude seems so plausible. One can say: economists meant to remain responsible members of a partnership with psychology, but then they missed the bus at the cognitive revolution in the 1960s.[2] As I have intimated, this has a ready explanation: with Debreu's and Arrow's achievements just in, there was a party in swing, and boarding the Chomsky Express would have meant leaving it just as the action promised to heat up.

Unfortunately, the party fizzled somewhat. That is, the programme of attending to aggregate-level phenomena without regard to their basis in properties of individuals hit snags. There were already intimations of trouble in the Lipsey-Lancaster theory (1956) of the second-best, which shows that one cannot infer from an allocation's getting closer to the perfectly competitive Arrow-Debreu general equilibrium that the allocation in question is necessarily increasing in efficiency.[3] While this undermined the policy relevance of general equilibrium economics, it did not threaten its underlying ontology. A major blow against that was struck by the so-called excess demand literature of the 1970s (the Sonnenschein-Mantel-Debreu theorem: Sonnenschein 1972, 1973; Mantel 1974, 1976; Debreu 1974), which showed that although all general equilibria are efficient, there is no unique one-to-one mapping between general equilibria and vectors of individual demand functions. (Put more directly, for a given set of demand functions there is more than one vector of prices at which all demand is satisfied.) What rightly exercised economists about this result was that, in tandem with Lipsey-Lancaster, it challenged the cogency of attempts by welfare economists to justify policy by reference to merely inferred (as opposed to separately and empirically observed) subjective preferences of consumers. Note that this problem arises whether one assumes an atomistic or an intersubjective theory of the basis of value. Nevertheless, Sonneschein-Mantel-Debreu shook the general postwar confidence that if one attended properly to the aggregate scale then specific properties of individuals could be safely ignored. There is obvious affinity between this development and cognitive science's shifting of the causal levers on individual behavior from exogenous structures in environments, where the behaviorists had located them, back into the heads of agents.

The next great enthusiasm in microeconomics was game theory (GT), which, during the 1970s and 80s, conquered most of the sub-discipline. GT can be given a behaviorist interpretation, according to which one uses it to guide inferences about what players will behaviorally orient toward through observing which vectors of their actions, in strategic interanimation, are (Nash) equilibria. The power of such inferences is often limited because most games have multiple Nash equilibria. Efforts to derive stronger predictions led a majority of economic game theorists in the 1980s to interpret games as descriptions of players' beliefs instead of their actions. On this interpretation, a solution to a game is one in which all players' conjectures about one another's preferences and (conditional) expectations are mutually consistent. Such solutions are, in general, stronger than Nash equilibria, and hence more restrictive. This approach to GT, which economists call 'the refinement program' because it refines Nash's solution concept, was criticized by a number of game theorists (e.g. Binmore 1990) for muddling the economist's descriptive ambitions with normative / philosophical intuitions about what ideal rationality does and doesn't require. The program also leads demonstrably to logical paradoxes (Bicchieri 1994). Most importantly for the theme of the present essay, 'epistemic' GT involves agents that are computational prodigies, instantly updating all their beliefs, using all valid principles of Bayesian probability, upon receipt of any information.

A theorist who, instead of treating GT as a body of pure mathematics that can be applied in any number of ways as modeling purposes suggest, regards it as a normative theory of the strategic rationality of economic agents, is set up to be chastened. GT as this theorist deploys it is in the business of producing generalizations of the same logical form as generalizations of cognitive science, describing what agents believe, desire, expect and compute. But the capacities it claims for these agents are not plausible capacities of finite human beings who were built by natural selection's incremental tinkering. And, sure enough, when people play the games analyzed by game theorists in laboratories, they do not behave like the agents in the models and they converge on vectors of strategies that are often not Nash equilibria according to the models (Camerer 2003). Thus, it seems, the 'assumptions' of microeconomics need correction by the empirical facts of cognitive science.

The correction in question turns out to be drastic. Real people approximate economic agency behaviorally in that they often accomplish their projects at bearable costs; but they don't exhibit any of the core computational properties of economic agents; 'their' behavioral rationality typically turns out to really be natural selection's rationality, evolution having built rough situational rules of thumb ('heuristics') into people that serve them well as long as their environments are not too strange (Gigerenzer et al 1999). This critique then appears to be reinforced by cognitive neuroscience, which musters evidence for biases, heuristics and framing effects operating directly in the processing systems of the brain (Camerer, Loewenstein & Prelec 2005). Thus, it is concluded,[4] economics collapses not just into abstract computational psychology, but all the way into computational neuroscience. That the word 'collapse' is not too strong is indicated by the sorts of things some new 'neuroeconomists' claim to discover. Recently, a team reported having determined from inspection of dopamine neurons that people do not value rewards by reference to their opportunity cost (Knutson et al 2007); they infer from this that economic theory requires revision. Non-economists may underappreciate the extent to which this must boggle any economist in whom chastening has not gone quite far.

Once economics is taken to collapse into psychology, then discoveries in sets (1) – (4) are naturally interpreted as tearing its standard theory apart. Furthermore, the news seems to have been getting worse since the early days of so-called 'behavioral economics'. Findings in sets (1) – (3) can, at least in principle, be accommodated by constructing new kinds of valuation functions. For example, people can be taken to maximize within frames, even if not across them. Hyperbolic discount curves can be approximated by composing exponential ones of different slopes (Laibson 1997, 1998). However, cognitive science has lately been shaking free of a hyper-rationalistic and atomistic legacy of its own. The past decade has seen enormous upgrading of the significance attached to affect in explaining both mentation and behavior (Damasio 1994; Panksepp 1998). Furthermore, affect itself is increasingly understood as both responding to and conditioning dynamic social interaction, an approach to modeling that seems to be borne out by the discovery of mirror neurons (Frith & Wolpert 2004). As individual people appear less and less to be autonomous bearers and computers of valuations, whose preferences explain their exchanges but are unchanged by them, and come instead to be seen as resembling adaptive nodes in social colonies where valuations continuously modulate one another in interacting cascades,[5] the more hopelessly inaccurate people's identification with economic agents is thought to be.

3. Against reductionism

The high-level historical sketch just given indicates the extent of complexities in the chains of association that have led to self-chastening among some economists. In this short essay on just one aspect of the chain, that bearing on social cognition, I cannot hope to unpack them all. (See Ross forthcoming b for the fuller account.) In one respect, concentrating exclusively on sociality as a supposed refuter of standard economic theory picks on the easiest target, because the view of the theory as committed to psychological atomism is wrong in a more straightforward way than is the view of the theory as committed to humans as computational prodigies. On the other hand, a logical advantage of focusing on sociality is that confusion over its role in economics is basic to the confusions around discovery sets (1) – (3) above, for reasons I will indicate.

I stressed two discontinuities in the history of modern economics, alleging that some of those who urge chastening pay insufficient attention to the second one.[6] Before it, neoclassical economics was indeed atomistic. Furthermore, it was understood as having its foundations in individual psychology. The key property of the neoclassical economic agent, diminishing marginal utility in consumption from a stock of close substitute goods, was thought to be a function of a person's internal 'pleasure registers'. As indicated, however, a tendency toward minimizing psychological commitments became dominant in neoclassicism. In the 1940s, as we saw, there is a complete break with psychology, though this was misunderstood at the time because the kind of psychology from which economics separated was itself in eclipse anyway.

I described some disappointments that befell Debrevian triumphalism. What I did not stress is that these disappointments were more strictly theoretical than practical. Just as the de-psychologization of economics began before Samuelson, so did its increasing concentration on policy guidance, which in turn led to steady improvement in techniques for measuring and studying econometric relations among aggregate variables. The Keynesian revolution of the 1930s was an overnight triumph among economists because, in abandoning concern for microfoundations, Keynes was perceived as liberating the profession; thanks in part to his background in philosophy, he could dismiss ontological scruples that his colleagues had regarded more diffidently.[7] In the everyday practice (as opposed to high theory) of economics, there has been no looking back on this liberation. The overwhelming majority of applied economists never estimate an individual utility function after they leave graduate school. They measure elasticity coefficients of aggregate demand functions from changes in prices, interest rates, income distributions, national savings rates, and other index quantities. Most will avow if pressed that all of these things somehow 'boil down to' decisions by 'rational agents' at the micro scale; but by the weight of behavioral evidence this interest is usually perfunctory and the avowel is typically conventional. For example, textbooks in international economics admit that so-called 'community indifference curves' used to represent national welfare cannot be disaggregated into individual indifference curves without destroying the point of using them; most books cheerfully note this and move on without further ado, assuming that the idea of 'national welfare' makes sense in its own right. Becker (1962) shows that the fundamental property of the standard model of the market – downward sloping demand for any good given constant real income – depends on no claim about the rationality of any agent; it depends only on the assumption that households with smaller budgets and therefore smaller opportunity sets consume less. Thus the majority of economics makes no use of the supposed identity of economic agents and biological / psychological people. This fact should be taken at least as seriously as anything said about 'individual consumers' in opening chapters of introductory micro texts.

It is nevertheless true that theory underwent a 'revenge of the micro-scale' in the 1970s and 80s. I have already mentioned what happened in game theory. At the same time, macroeconomic theorists pursued a program of building so-called 'microfoundations' for macroeconomics (as pioneered in, e.g., Lucas 1972). The core idea of the project involves accounting for changes in macroeconomic trends and variables by reference to agents learning to forecast monetary and fiscal policy. Rhetorical defenses of this program have often cited methodological individualism. I call such defenses 'rhetorical' because although they served as attractive political advertisements for the program in the 1970s and1980s (and therefore also attracted completely confused political criticism from non-economists; see Lucas 1978), one never finds a philosophical premise such as ontological individualism invoked to do any logical work in an economic model. Furthermore, the agents who feature in microfoundational models are 'representative' optimizers whose ontological status is indeterminate. In some canonical models whole economies are modeled as though they are single ('infinitely lived') agents whose budget constraints are national GDPs and whose business cycles result from the schedules on which they invest and take profits (Kydland and Prescott 1982; Long and Plosser 1983). The underlying justification for this is the assumption that what are being modeled are markets in which utility functions differ only indexically. If the individualist rhetoric is taken more seriously than is warranted, the result is ironic: the canonical individualist program would depict people as so many ants, differing only in the information that happens to have come their way. But, again, this takes philosophical and political associations seriously in the way the economists themselves don't. It in fact doesn't matter to the formal analyses in the microfoundations tradition what sorts of extra-economic entities the agents map onto; all that matters is that econometric tests, based on measuring aggregate variables, can distinguish between one model and another. From the empirical point of view, the agents are merely sites of consumption; there is no reason at all to assume they're people, rather than firms or countries or pension funds. The microfoundations program was popular with theorists not because of either strong metaphysical convictions of the Thatcherite variety (“There is no such thing as society”), or beliefs about human psychology. It was popular because it promised to bring the axiomatic and mathematical framework of microeconomics to bear on macroeconomic problems, which are otherwise exercises in applied statistics. In practice, it tends to be invoked exactly when it simplifies the art of writing down a model and never otherwise. Commenting specifically on appeal to the good aura of microfoundations among economists as grounds for claiming that they're truly serious about their individualism, Hoover (2001, p. 110) rightly says “[d]espite lip service to [individualism], it is not widely practiced by economists.”

In game theory, the refinement program largely expired by the turn of the century, mainly choking on its own problems rather than being smothered by the activities of economists turning into psychologists. Three main lines of research have taken centre stage among game theorists over the past ten years. One line applies classical game theory to contexts, such as auctions among highly capitalized players bidding for very valuable assets, in which institutional forces incentivize consortia to indeed behave like fully rational economic agents (Klemperer 2004; Milgrom 2004). These consortia are not biological or psychological entities. Of course their representatives are such entities; but they are not imagined as doing their own computations, nor as choosing strategies using native, in-board cognitive resources. (They have external computing equipment.) Second, game theorists have explored investment patterns in distributed markets by modeling them as games involving large numbers of players facing common uncertainty where all know that all know about the extent of uncertainty, and all know what technologies can be used to manage it (e.g., Morris & Shin 2003). Here again is a use of game theory that eschews any appeal to psychological idiosyncrasies: players essentially use their models of the game situation to stabilize their expectations about one another, and they are embedded in institutional settings that are taken to constrain their utility functions, eliminating any special personal properties. Finally, a way of dealing with multiple equilibria that has overtaken appeal to refinements in popularity is use of evolutionary game theory (Weibull 1995; Samuelson 1998; Cressman 2003). This replaces the hyper-sophisticated agents of the refinement program with thoughtless players who simply inherit or copy strategies from others, with the probability of a strategy's getting inherited or copied being correlated with the strategy's success in previous rounds of iterated games. In this approach, strategies themselves, rather than agents, are the players of the games, with agents merely standing in to play their brief turns in a competitive process that continues beyond their individual lifespans. Young (1998) applies such models to what might best be called 'collective cognition'. Evolutionary game theory is well suited to generalizing the dynamics of swarm intelligence implementations of such cognition (for example, simulated ant colonies).

I said earlier that applied economists seldom estimate individual utility functions. Theoretical economists, by contrast, constantly fill papers with them. But I have just argued that the individual utility functions constructed by developers of microfoundations and by game theorists when they're not involved in behavioral economics should not be interpreted as utility functions of individual people. Of course, behavioral economists intend the utility functions they write down to be utility functions of people, But it is precisely my main contention that is unusual use of economics, not the core application, with implications that thus ramify through the whole discipline, it is often taken to be.

Thus we do not need to turn only to atheoretical, pragmatic, policy-focused economics to find continuation of the tradition that has never looked back on Keynes's rejection of microfoundations or Samuelson's elimination of individual psychological properties from economic modeling and inference. Such economics does not collapse into psychology, social or otherwise: it ignores it (Gul & Pesendorfer forthcoming). This is important because it reminds us that reductionism is not a universal tactic across the sciences. Indeed, many of the same people who advocate increased 'psychological realism' in economics are also fans of applying complex systems theory to social science (e.g. Ormerod 1999; Gintis 2000; Beinhoker 2006). Denial of what philosophers call 'ontological reductionism'[8] – that is, metaphysical atomism – is part of the very point of complex systems theory, with its emphasis on 'emergent' structures, properties and relations which cannot be decomposed into properties and relations of their parts. This new emergentism should, in my view, be approached with caution due to serious problems over stability of state variables across models. However, the simultaneous popularity, often in the same breasts, of extreme anti-reductionism and deflationary versions of neuroeconomics suggests tendencies to favor ideas simply because they rebel against neoclassicism.

In any event, we do not need to appeal to hairy emergentism in order to resist the elimination of the rational economic agent in favor of the social and emotional brain. Behaviorist psychologists have always contrasted 'molar' and 'molecular' scales of description and explanation. Often this has simply been pragmatic, but the philosophers Daniel Dennett (1991, 1994), Andy Clark (1997) and Harold Kincaid (1997) have emphasized its ontologically serious role. On their complementary accounts, the partial autonomy of the molar from the molecular is not mysterious or 'spooky'. It lies in the fact that molar-scale descriptions situate cognitive / behavioral systems in environmental contexts, sorting their dispositions and properties by reference to equivalence classes of problems they face. These equivalence classes can be highly heterogeneous from the molecular point of view while remaining stable objects for scientific generalization due to external pressures that 'capture' molecular processes within distinctive patterns. The logic here is the same as that which explains convergence in evolution by adaptation to niches. At the level of phylogeny, the relevant external pressures are ecological; in the case of people they are mainly social, and frequently institutional. Thus molar-scale and molecular-scale accounts of one and the same system can, up to a point (a point that must be determined empirically in any given instance) vary independently. Ontological reductionism does not describe the typical course of science where biological systems are concerned.

If molar-scale phenomena should not be presumed on philosophical grounds to collapse into molecular-scale ones, then the only allowable motivation for holding that one discipline or sub-discipline should reduce to another is actually observing the redundancy and abandonment, in that particular instance, of molar-scale models and their replacement by molecular-scale ones. I have argued that no such trend is manifest as between economics in general (i.e., outside of the avowed behavioral economics movement itself) and psychology or neuroscience. This does not at all imply that psychology and neuroscience are irrelevant to economics. (I am enthused, not skeptical, about neuroeconomics as modeling of the brain's valuations rather than the molar person's; see Ross forthcoming, Ross et al forthcoming.) Molecular-scale processes typically constrain molar-scale ones without reducing them.

The implication of this form of anti-reductionism is immediate: we can agree that people are not identical to economic agents without this necessarily implying that economic agency as traditionally understood is a useless or confused theoretical construct. 'Necessarily' here needs emphasis. Rejecting an a priori motivation for collapsing economics into psychology does not in itself answer the question implied in the criticism of Debrevian economics based on cognitive and behavioral science. That question is: if economic agents are asocial computational prodigies and people are constitutively social cognitive duffers, then what is the relationship between economic agents and people? To answer that there is no relationship would conjure up a mystery, philosophers' rationalizations notwithstanding.

I will argue in the concluding section of the paper that, far from ignoring the social constitution of persons, attention to this fact about them yields the answer we need to the question just posed.

4. People, agents, socialization and value

The thesis I will summarize here is not new, though in this essay I have led up to it via a new path and set it into a new dialectical context. But the thesis itself has been extensively elaborated in Ross (2005). It synthesizes several ideas.

The first idea is that if people are socially constituted, then they are molar-scale objects[9] by comparison with their brains. This strange statement obviously needs explanation. Human organisms are not born socially constituted. If it is correct to claim that people are constituted socially, this reflects the fact that they are created from human organisms by social development. Of course this process relies on properties of their brains: humans' giant cortex, and dispositions immanent in biases in neural connections and in the architecture of neurotransmitter pathways prepare them, unlike tigers, to be socialized. But the fact that we can distinguish between a very short pre-socialized phase and a socialized phase of a human organism's life supports a distinction between, as it were, the 'raw brain' and the person as a node in a dynamic social network. Raw human brains resemble tiger brains more than they resemble people. That people are socially constituted but their brains are not is the basic reason why behaviorists were right to emphasize the molar / molecular distinction. It doesn't suggest the dualist idea that persons transcend their brains; brains must adapt to socialization during development, and socialization is constrained by what brains can and cannot process.

This is why I think that neuroeconomics is important but will not reduce microeconomics. Neurons indeed compute valuations and are built to maximize utility (Glimcher 2003). But the utility in question is their utility, not the person's socially governed objectives. For example, a group of dopamine neurons maximizes their utility by suppressing competing serotonergic circuits. If they are too successful the result is addiction, which is a disaster for the person and which few people want (Ross et al forthcoming.)

To understand how people are socially created, something must first be said about why such developmental trajectories have been stabilized by selection. Let us distinguish between social animals and herding animals. Whereas the latter – wildebeest, for example, or corals – gain advantage merely by staying close together and coordinating their schedules, the former exploit efficiencies from joint contributions to ranges of projects that individuals can't perform alone, using some degree of specialization, either merely of talent or of dedicated roles. All available evidence suggests that natural selection, given the platforms it has had to work with in terrestrial history, can produce this in two ways: by adapting animals' genetic structures to increase the value of the inclusive coefficient in fitness functions, as in social insects and naked mole rats, or by adapting animals' brains so they develop enough book-keeping capacity to strategically discriminate among conspecifics and can thereby play strategic games involving reciprocal rewards and sanctions. High intelligence (cognitive plasticity) is far from continuously distributed across species, and sociality is far from continuously distributed across clades. It is thus of powerful significance under regression analyses that the entire hyper-intelligent club, which includes apes, elephants, dogs, toothed whales, corvids and parrots along with a few others, is social.

Within this club, humans are ecologically special in navigating an effectively boundless domain of novel collaborative projects. This is made possible by signaling systems – languages – that stabilize ranges of possible signal meanings by digitalizing information. That is, human syntax enables one human to direct another's attention to a specific object of reference even when it is not present to be pointed or gazed at; I can communicatively refer to 'Napoleon' exactly, not just to an indefinite range of things sharing to various degrees Napoleon's analog blend of properties (i.e., 'napoleonishness'). Thus humans can jointly track objects over time and space even when they are not present, and coordinate on future plans involving hypothetical objects picked out by digital contrast with other members of classes into which the grammars of public languages permit them to be sorted (Ross 2007).

Some philosophers have suggested that language plus shared perceptual saliences are sufficient to account for people's ethologically unique capacity to coordinate. This is confused: the range of projects that can be distinguished thanks to recursive grammar makes the human coordination challenge orders of magnitude more complex than that faced by any other species. Game theorists encourage us to underestimate the difficulty of social coordination by solving for equilibria in situations they have already modeled as definite games. They readily forget that their own chief skill is in seeing how to abstract useful strategic models of empirical situations which don't come pre-packaged in terms of utility functions or strategy sets. Real human game players must implicitly construct models of their strategic situations in real time, without benefit of explicit principles, and they must jointly coordinate on these constructions; two interacting people who don't conceptualize their situation in terms of (roughly) the same game should expect not equilibrium but unpredictable chaos. Finally, let us bear in mind that every time a person takes an action she offers a move in a game with everyone whose welfare is potentially influenced by it and who might become aware of it – directly, by observing it or through gossip, or indirectly, by inferring it from outcomes, or second-order, by being influenced by the actions of someone else who is influenced by the original action. The overwhelming majority of human actions are thus simultaneously moves in multiple games with multiple sets of players of multiple n.

This all implies that most human choices of actions, no matter how small in scale, amount to what economists call 'general equilibrium problems'. For example, to determine the best strategic response to my colleague's suggestion that we nominate a third colleague for a certain committee, I should, if I want to implement full rational agency, model the entire strategic history of our species (at least to the point in the future beyond which, due to discounting, I lose interest). This game is self-evidently intractable.

It gets still worse. A person's brain has a trillion neurons and 10¹³ synaptic connections, organized into semi-modular sub-systems that communicate imperfectly with one another, behave semi-autonomously and can no more be micro-managed by a frontal executive system than the President of the United States can plan every postal delivery and sentry assignment. Not only do I not know the exact utility functions and strategy sets of the n other people with whom I'm strategically enmeshed, but I face significant uncertainty in predicting my own utility function and distribution of strategy sets, because much of my behavior is regulated by parts of my brain to which I have no more access than a third-person observer.

People clearly do coordinate, often very smoothly, over substantial stretches of time and place, and across large groups. Even more clearly, they don't do so by solving computationally impossible problems. The model of social coordination as solving a general equilibrium by solving an unbounded-n game must be missing something important. In social embeddedness and language, the very phenomena that lead to the impasse, lie the clues to what this something is. People sensibly insist that others with whom they enter into coordination games narrate comprehensible, publicly manifest stories about themselves and conform their behavior to these stories. Thus they enforce and enable predictability, including self-predictability. They mutually ease the imposed burden of this task by assisting each other as co-authors of narratives, recording expectations, rewarding enrichments of each other's sub-plots, and punishing overly abrupt attempts to revise important character dispositions. Parents initially impose this regime of self-construction on their children, later handing over primary control (often involuntarily) to their offspring's peer groups. Thus people become and remain distinct. The fact that self-creation and self-maintenance are projects requiring effort is what explains prevailing normative individualism, even while ('metaphysical') descriptive individualism is false. Individuals are centrally important to most of us partly because they don't just drop out of the womb.

A crucial enabling aspect of this whole edifice is that humans are biologically adapted to be highly behaviorally sensitive to very cheap rewards (e.g. smiles, laughter, raised thumbs) and punishments (e.g. frowns, eye rolling, refusal of efforts at conversation). Not only are the standard punishments very inexpensive relative to the pain they inflict, but they can be withdrawn leaving almost no damaged infrastructure that then requires a new infusion of capital to put right; a person says “I forgive you” and the other's misery is (typically) instantly relieved. Some leading game theorists make the social coordination problem too hard, thereby motivating extravagently hypothesized genetic adaptations to fix it, by exaggerating the costs of everyday rewards and punishments (Gintis 2006; Seabright 2006). People avoid 'cheap talk' problems, in which their threats and promises would be ignored because it's doubted that these would be followed up if ineffective, by being psychologically adapted to care a great deal about rewards and punishments that cost others almost nothing to dispense (Ross 2006).

The effect of everyday pressures on people to construct and maintain selves is to drastically shrink the ranges of utility functions and strategy sets over which people must coordinate their constructions of games. The structures of these self-narratives then emerge as apparent framing effects and departures from proper Bayesian reasoning when we put people into experimental games and model these games as if the players weren't constrained by their own biographical and autobiographical plots.[10] This is a ubiquitous feature of the experimental literature in behavioral economics. Researchers define their subjects' games as if they were unconstrained by socialization, show that the behavioral outcomes do not match the Nash equilibria of these games, and thereby draw two generic conclusions (as background for various more specific conclusions that constitute the real scientific payoffs). The first sort is unobjectionable: people are constrained by socialization. But that is a truism, and I do not believe that Jevons or Walras or Samuelson or Debreu didn't know it. The second generic conclusion is that therefore standard economic theory is refuted because that theory is necessarily about unsocialized agents. This I reject.

Most economic theory has not been 'about' agents for over half a century, if by 'agents' we mean everyday entities with biologically or psychologically individuated boundaries that are economically rational and consistent. In particular, absolutely nothing in economic theory requires that economic agency be identified with individual people. Economic agency is a theoretical construction. Economists use it to build abstract models of firms, nations, labor unions, consortia in auctions, lineages in evolutionary games and other feedback-sensitive, incentive-driven systems that have no psychological properties at all. The usefulness of the construction is not cast into doubt by behavioral economics or by cognitive science more generally.

It is thus open to us to ask whether economics has any relevance to cognitive science (and hence to cognition understood as social). If the answer were 'no', economists in the spirit of Keynes might shrug this off and leave worries about unification of the sciences to philosophers. But the answer is not, in fact, negative. I just summarized an account of the universal human disposition to construct selves and to enforce such construction in one another. The explanation of this pattern is that it allows people to achieve many of the gains possible for economic agents – gains from trade, from specialization, and from consistent investment over time – despite the fact that their brains are too large and necessarily de-centralized as control structures to pull off economic agency by themselves. Thus economics plays a direct role in explaining the basis, albeit not the computational microdynamics, of social cognition. Furthermore, self-construction is only the first (necessary) aspect of the achievement of large-n coordination. The truly heavy lifting is done by the ultimate self-maintenance engines, institutions.

Most readers of this article will save money for relatively comfortable retirements. You will do this despite the fact that you would, if put in a systematically unfamiliar consumption environment, discount the future hyperbolically and therefore tend to reverse your preferences for prudent investments when temptations to immediate reward presented themselves, then spend still more resources trying to defeat your own myopia. Most of you will avoid this in your actual lives because your behavior is hemmed in and guarded by walls of culturally evolved and collectively designed institutions. If you persistently spend more than your income, this will be reflected in a falling credit rating that will inconvenience you now. Perhaps a recent housing bubble has allowed you to splurge for a few years, but as of this writing market institutions are busy transmitting information about you and hundreds of millions like you that, through still other institutions, will correct your lack of prudence. If you aren't corrected quickly enough, the bank manager who supervises your mortgage may act to speed up receipt of the message. If very many of you are too sluggish responding to the news, the Chairman of the Federal Reserve Bank may reinforce it with an interest rate hike. And so on.

All of these institutions press you to approximate your behavior to that of an economic agent. They can't literally transform you, biological – psychological entity that you are, into such an agent. Even while struggling to save, you may visit a casino. You will buy some items this year that you will disdain and throw away in a year's time merely because your tastes change. But you, together with your fellows in society, have enough in common with economic agents, especially in modern institutional settings, that non-trivial predictions about your individual behavior can be had by modeling you as if, within temporal and institutional constraints, you were such agents. Furthermore, because you live in aggregated markets with dynamics that aren't very sensitive to psychological factors, and because you also play n-person games with other agents who are incentivized to stabilize one another's preference consistency, you can improve your prospects by learning some economic theory and feeding this social knowledge back into your personal planning. Feedback loops of this sort are the very logical essence of social cognition.

Acknowledgement

I thank Erik Angner, Harold Kincaid, and an anonymous referee for the journal for their comments on an earlier draft.

References

Angner, E., & Loewenstein, G. (forthcoming). Behavioral economics. In U. Mäki (Ed.), Handbook of the philosophy of science, volume 13: philosophy of economics. Amsterdam: Elsevier.

Becker, G. (1962). Irrational behavior and economic theory. Journal of Political Economy,

70, 1-13.

Beinhoker, E. (2006). The origin of wealth. Cambridge, MA: Harvard Business School Press.

Bicchieri, C. (1994). Rationality and coordination. Cambridge: Cambridge University Press.

Binmore, K. (1990). Essays on the foundations of game theory. Oxford: Blackwell.

Binmore, K. (2007). Does game theory work? The bargaining challenge. Cambridge, MA: MIT Press.

Bruni, L. (2005). Hic sunt leones: interpersonal relations as unexplored territory in the tradition of economics. In B. Gui and R. Sugden (Eds.), Economics and social interaction (pp. 206-228). Cambridge: Cambridge University Press.

Bruni, L., & Sugden, R. (2007). The road not taken: how psychology was removed from economics and how it might be brought back. The Economic Journal, 117, 146-173.

Camerer, C. (2003). Behavioral game theory. Princeton: Princeton University Press.

Camerer, C., & Loewenstein, G. (2004). Behavioral economics: Past, present and future. In C. Camerer, G. Loewenstein and M. Rabin, eds., Advances in Behavioral Economics, pp. 3-51. Princeton: Princeton University Press.

Camerer, C., Loewenstein, G., & Prelec, D. (2005). Neuroeconomics: how neuroscience can inform economics. Journal of Economic Literature, 43, 9-64.

Clark, A. (1997). Being there. Cambridge, MA: MIT Press.

Cressman, R. (2003). Evolutionary dynamics and extensive form games. Cambridge, MA: MIT Press.

Damasio, A. (1994). Descartes's error. New York: Putnam.

Davis, J. (2003). The theory of the individual in economics. London: Routledge.

Debreu, G. (1959). Theory of value. New York: Wiley.

Debreu, G. (1974). Excess demand functions. Journal of Mathematical Economics, 1, 15-23.

Dennett, D. (1991). Real patterns. Journal of Philosophy, 88, 27-51.

Dennett, D. (1994). Cognitive science as reverse engineering: several meanings of 'top-down' and 'bottom-up'. In D. Prawtiz, B. Skyrms & D. Westerstahl (Eds.), Logic, methodology and philosophy of science (pp. 679-689). Amsterdam: Elsevier.

Durlauf, S., & Young, H.P. (Eds.). (2001). Social dynamics. Cambridge, MA: MIT Press.

Foster, E. & Sonnenschein, H. (1970). Resource allocation and the public sector. Econometrica, 38, 281-297.

Frith, C., & Wolpert, D. (Eds.) (2004). The neuroscience of social interaction. Oxford: Oxford University Press.

Gigerenzer, G., Todd, P., & the ABC Research Group (1999). Simple heuristics that make us smart. Oxford: Oxford University Press.

Gintis, H. (2000). Game theory evolving. Princeton: Princeton University Press.

Gintis, H. (2006). Behavioral ethics meets natural justice. Politics, Philosophy and Economics, 5, 5-32.

Gintis, H., Bowles, S., Boyd, R., & Fehr, E. (2005). Moral sentiments and material interests. Cambridge, MA: MIT Press.

Glimcher, P. (2003). Decisions, uncertainty and the brain. Cambridge, MA: MIT Press.

Gui, B., & Sugden, R. (2005). Economics and social interaction. Cambridge: Cambridge University Press.

Gul, F., & Pesendorfer, W. (forthcoming). The case for mindless economics. In A. Caplin and A. Schotter (Eds.), The handbook of economic methodologies, volume 1. Oxford: Oxford University Press.

Hoover, K. (2001). Causality in Macroeconomics. Cambridge: Cambridge University Press.

Kennedy, J., & Eberhart, R. (2001). Swarm intelligence. San Fransisco: Morgan Kauffman.

Kincaid, H. (1997). Individualism and the unity of science. Lanham, MD: Rowman and Littlefield.

Klemperer, P. (2004). Auctions: theory and practice. Princeton: Princeton University Press.

Knutson, B., Rick, S., Wimmer, G.E., Prelec, D., & Loewenstein, G. (2007). Neural predictors of purchases. Neuron, 53, 147-156.

Kydland, F., & Prescott, E. (1982). Time to build and aggregate fluctuations. Econometrica, 50, 345-370.

Laibson, D. (1997). Golden eggs and hyperbolic discounting. Quarterly Journal of Economics, 112, 443-477.

Laibson, D. (1998). Life-cycle consumption and hyperbolic discount functions. European Economic Review, 42, 861-871.

Lipsey, R., & Lancaster, K. (1956). The general theory of second best. Review of Economic Studies, 24, 11-32.

Long, J., & Plosser, C. (1983). Real business cycles. Journal of Political Economy, 91, 39-69.

Lucas, R. (1972). Expectations and the neutrality of money. Journal of Economic Theory 4: 103-124.

Lucas, R. (1978). Unemployment policy. American Economic Review 68: 353-357.

Mantel, R. (1974). On the characterization of aggregate excess demand. Journal of Economic Theory, 7, 348-353.

Mantel, R. (1976). Homothetic preferences and community excess demand functions. Journal of Economic Theory, 12, 197-201.

Milgrom, P. (2004). Putting auction theory to work. Cambridge: Cambridge University Press.

Morris, S., & Shin, H. S. (2003). Global games: theory and applications. In

M. Dewatripont, L. Hansen & S. Turnovsky (Eds.), Advances in economics and econometrics, the Eighth World Congress (pp. 56-114). Cambridge: Cambridge University Press.

Ormerod, P. (1999). Butterfly economics. Bungay, Suffolk: Faber and Faber.

Panksepp, J. (1998). Affective neuroscience. Oxford: Oxford University Press.

Ross, D. (2005). Economic theory and cognitive science, volume one: microexplanation. Cambridge, MA: MIT Press.

Ross, D. (2006). Evolutionary game theory and the normative theory of institutional design: Binmore and behavioral economics. Politics, Philosophy and Economics, 5, 51-79.

Ross, D. (2007). H. sapiens as ecologically special: what does language contribute? Language Sciences, forthcoming.

Ross, D. (forthcoming a). Integrating the dynamics of multi-scale economic agency. In H. Kincaid and D. Ross (Eds.), The Oxford handbook of philosophy of economic science. Oxford: Oxford University Press.

Ross, D. (forthcoming b). The economic agent: not human, but important. In U. Maki, ed., Handbook of the Philosophy of Science, Volume ?: Economics. Dordrecht: Kluwer.

Ross, D., Sharp, C., Vuchinich, R., & Spurrett, D. (forthcoming). Midbrain mutiny: the picoeconomics and neuroeconomics of disordered gambling. Cambridge, MA: MIT Press.

Samuelson, L. (1998). Evolutionary games and equilibrium selection. Cambridge, MA: MIT Press.

Samuelson, P. (1947). Foundations of economic analysis. Cambridge, MA: Harvard University Press.

Seabright, P. (2006). The evolution of fairness norms: an essay on Ken Binmore's Natural Justice. Politics, Philosophy and Economics, 5, 33-50.

Sen, A. (1977). Rational fools. Philosophy and Public Affairs, 6, 317-344.

Sen, A. (1999). Development as freedom. New York: Random House.

Sonnenschein, H. (1972). Market excess demand functions. Econometrica, 40, 549-563.

Sonnenschein, H. (1973). Do Walras identity and continuity characterize the class of excess demand functions? Journal of Economic Theory, 6, 345-354.

Sugden, R. (2002). Beyond sympathy and empathy: Adam Smith's concept of fellow-feeling. Economics and Philosophy, 18, 63-87.

Sugden, R. (2005). Fellow-feeling. In B. Gui & R. Sugden (Eds.), Economics and social interaction (pp. 52-75). Cambridge: Cambridge University Press.

Weibull, J. (1995). Evolutionary game theory. Cambridge, MA: MIT Press.

Young, H.P. (1998). Individual strategy and social structure. Princeton: Princeton University Press.

Zak, P. (forthcoming). Review of Don Ross, Economic Theory and Cognitive Science: Microexplanation. Journal of Economic Methodology.

---

[1] For examples from an exponentially growing literature see the following collections: Durlauf & Young (2001), Gui & Sugden (2005), Gintis et al (2005).

[2] Bruni & Sugden (2007) offer an alternative narrative, according to which they did mean to break with psychology and therefore weren't interested in the bus.

[3] Though see Foster & Sonnenschein (1970).

[4] It might reasonably be asked here: “Concluded by who?” I think that Camerer, Loewenstein & Prelec (2005), all highly influential figures in economics, are forced to this conclusion by their general perspective, whether they would acknowledge it or not. Camerer & Loewenstein (2004) say explicitly that conventional economics still has an important role to play in behavioral science, but by this they seem only to mean that economics' analytical and mathematical techniques will go on being useful, while its accounts of economic phenomena are all open to revision based on psychology. In any case, my claim is that anyone who thinks that economics is fundamentally about the decisions of individual people – a view clearly held by many – are implicitly driven by the empirical findings of behavioral economics to the belief that economics collapses into psychology. I think this applies, for example, to Angner & Loewenstein (forthcoming). If in addition someone believes that psychology will reduce to computational neuroscience, then they must believe that economics reduces to neuroscience. See Zak (forthcoming) for an instance of forthright assertion.

[5] Such cascades are simulated in so-called 'swarm intelligence' models; see Kennedy & Eberhart (2001).

[6] This doesn't apply to Bruni & Sugden (2007).

[7] Keynes is sometimes cited (e.g., by Angner & Loewenstein forthcoming) as a precursor to psychologistic economics because he attributed business cycles to contagious emotions. However, this suggestion plays no direct role in his theory, which requires only that high-unemployment states be disequilibria. As later economists made much of, it is important that his theory assumes incomplete expectations on the part of consumers, producers and investors. But this was more of an oversight than an insight.

[8] This locution is required to distinguish between reducing composite objects into parts, and reducing so-called 'high-level' theories to less abstract theories ('intertheoretical reduction'). Philosophers of science have generally been more interested in the latter than the former.

[9] In fact, people are better conceived as processes than as objects.

[10] It's possible to induce people to escape from these constraints, in which case they tend to act much more like economic agents; but this requires deliberate effort in experimental design. See Binmore (2007).