The Smith-Nietzsche-Schelling Model of Rationality

Sociologists tend to have a lot of problems with rational choice models of behavior that emphasize the decisions of discrete, rational individuals. In a sense, the discipline itself can be seen as a reaction to liberalism* and its overemphasis of individuals, individuality and choice. The easiest place to see this is in Durkheim or Marx, both of whom explicitly argue that individuals are formed by society and not the other way around. In a way, the whole structure-agency debate (can we call it a debacle at this point?) stems from social theorists trying to grapple with the basic problem of reconciling a liberal belief in the self with a sociological belief in society.

The contemporary debate has moved on from liberalism somewhat, and now focuses more on critiques of economistic models of human nature. Such models are varied (see, for example, Teppo’s four-fold division), and recent advances in behavioral economics have started to take more seriously some of sociology’s concerns – things like limited processing capacity, framing effects, network models of information diffusion and the like (see, for example, March and Simon on bounded rationality, and Kahneman and Tversky on framing effects). But have these improvements to the rational actor model fixed the underlying problems for sociology? Should we give in and become behavioralists?

The old standard model held that individuals had consistent preferences over known sets of outcomes, and perfect information about how their own choices would (probabilistically) lead to different outcomes (perhaps with some strategic interaction thrown in, in a game-theoretic version), and almost always held that people were solely self-interested**. Sociologists had myriad issues with this formulation but most centered on either the exclusion of altruism and more general other-regarding preferences, or the hyper-unrealistic assumptions about the capacity of individuals to know the space of possible outcomes, the probabilities of those outcomes, and to make calculations about what would be best. So, sociology argued that individuals were not solely selfish, and that they were not so informed nor perfectly calculating***.

Economic sociology took on this task as much or more than any other subfield. Granovetter’s foundational article, “Economic Action and Social Structure: The Problem of Embeddedness” critiques the “undersocialized” conception of action in economic theory. Granovetter focused especially on how economic action happens through our immediate social structure – our networks of friends, family, co-workers, and etc. Information flows through these networks, along with resources and opportunities. White (1981) argued that companies in stable markets used available information – not unknowable hypotheticals about preferences and the like – to decide what to produce and how much to charge. And so on. Here, sociology challenged not just the idea of traditional rational action in spheres where it always seemed dicey – like the family – but in its hearth, the economy.

But economics did not stand still from the 1970s to present. While many of the basic models of the field rely on the same sets of assumptions sociologists have criticized for ages, behavioral models are on the rise (as noted above). And many of sociology’s most powerful critiques seem a bit hollow against this set of assumptions – no longer are individuals assumed to be perfect calculators in possession of complete information and immune to framing effects or social influence (though I have still seen relatively little work that uses other-regarding preferences to much effect). Sounds pretty good, right?

I think these improvements are fascinating, and I look forward to seeing how they are incorporated into the basic teachings of the field going forward. But I think they do not address the underlying concern that sociologists have always had: that modern economics really believes in pre-social individuals as the basic units of analysis, and believes those units to be basically stable, coherent and non-problematic. The place where this assumption is easiest to see, and where we have focused our attention the least, is in the assumption of consistent and stable preferences. While there has been much ink spilt over the issues with assuming preferences are consistent in a given moment (that no one prefers A to B, B to C and C to A, basically), I think the larger issue concerns the stability and consistency of preferences across time. And here, I think, there are compelling reasons to think that preferences are not so stable, and perhaps peeking into this instability and its drivers might lead us to a much better model of human action – one that does not require nor assume a stable individual.

The cleanest argument for the lack of consistency in preferences over a short period of time comes from one of my favorite articles, written by a Nobel Prize winning economist no less! Thomas Schelling, a game theorist who helped theorize the cold war, wrote in 1984 a fascinating piece entitled Self-command in practice, in policy, and in a theory of rational choice. Schelling argues that there exists an entire class of behaviors that are common, consequential, and utterly inconsistent – acting on your own future actions. Schelling, perhaps inspired by his own experiences, notes that when someone tries to quit smoking, they often ask their friends to refuse them a cigarette if they should late ask for it. Similarly, someone who is worried about becoming suicidal may ask their friends to check in on them, and to prevent them from doing anything harmful. On the other hand, a terminally-ill patient may ask a friend or doctor to help them end their own life because they lack the will. And so on. These actions on our future actions only make sense in a world where people’s preferences vacillate in relatively short order between diametrically opposed desires – to smoke or not smoke, to live or to die. But these behaviors exist, in small and large form, as anyone who has ever tried to go on a diet or begin an exercise program knows.

So what do we do with that? I wonder if the answer might lie in much older conceptions of human nature. As Albert O. Hirschman argues, in a book I have been citing like crazy lately, in the 17th-18th century, political theorists were concerned not with rationality but with the passions. These passions were varied, and included the lust for power, for recognition, for control, for money, and just plain old lust. Man was an internally divided creature, as the various passions warred against each other to determine our behavior. Political economists hoped that the passion for money – often named “interests” – might restrain the other passions, and thus act as a sort of civilizing force. Thus, they promoted freeing up markets and stimulating commerce – a prince concerned with prosperity and wealth would be less concerned with war and tyranny. Of course, their plans do not seem to have succeeded – the rise of commerce did not mean the end of war or tyranny as the 19th and 20th centuries show – but what I’m interested in here is the basic model of a set of conflicting passions rather than a unified sense of self-interest.

Adam Smith’s own works exhibit some of this flavor as well. I’m right now in the middle of his first book, the much less read Theory of Moral Sentiments. The book argues that men are not selfish, rather they are very much other-regarding. But the details of how we regard each other matter a great deal. Smith argues that because we prefer to sympathize with those are doing well, we end up being ambitious: “It is because mankind are disposed to sympathize more entirely with our joy than with our sorrow, that we make parade of our riches, and conceal our poverty.” (48) It’s a neat formulation, although a bit off-track from my main argument here, which is simply that Smith too concerned himself with the various passions pulling at us – our vices, as well as our sympathies.

Of course, lots of other thinkers have described human beings as internally divided. Freud would be an obvious choice – but one that I know not enough about to want to bring in here, and also one whose system carried with it a lot of seemingly normative baggage about urges and restraints. So, instead I’ll just bring in one other (arguably a precursor to Freud): Nietzsche. Starting with his first work, The Birth of Tragedy, Nietzsche argued that men were constantly at war with themselves, as the Dionysian drive for chaos and the Apollonian drive for order set upon each other (terrible gloss here, apologies out there to any Nietzsche scholars).

So, to get back to the main thread of this post, I think that sociologists concerns with the rational actor model should not simply focus on problems of information or selfishness. Rather, I think we should tackle head on the notion that individuals are somehow obvious, tightly-bounded, coherent actors. One route to this end is to dig deeper rather than abstract up to the macro – that is, by focusing on the inconsistency in human action, and the competing forces that propel us to make seemingly contradictory choices.

We should not confuse the relative predictability of human action (like your tendency to respond to a greeting with a similar greeting) with consistency in preferences, or internal coherency. Rather, to bastardize a phrase from Dahrendorf, perhaps we should view humans as composed of “partly agreed, partly competing, and partly simply different” passions. Which ones win out in a given moment may be a function of the situation – the particularities of how a decision is framed, who is in the room when we are asked a question or told to text 90999 to donate money to the Red Cross. Does our sympathy with an unknown stranger outweigh our desire to hoard and accumulate money for our personal enjoyment? There is no reason to think this question has one answer, because human beings exist across time and in space. At point 1, I may choose not to donate. At point 2, ten minutes later, when prompted again by a different image, a different friend, or simply because I don’t feel quite right, I choose to donate. Is this behavior irrational? Perhaps – if you hold on too tightly to the idea of a unified individual in the classic mode. But why would we do that? We’re sociologists!

* I use “liberal” and “liberalism” here in the old sense – folks like Locke, who argued for individual rights and social contracts and whatnot. Liberalism is not my expertise, so apologies for any gross misinterpretations.

** Of course, what it means to be self-interested is another problem. As John Quiggin has noted, self-interest tends to either be a vacuous notion (since anything I do can be said to be done because it made me feel good, like volunteering or donating to charity) or a strong, but false one (since people clearly do things that are not in their immediate economic interests). Quiggin argues that when confronted with this dilemma, proponents of egoistic rationality tend “to oscillate between the two definitions, in much the manner of the function sin (1/x) as x approaches zero.”

*** The recent emphasis on calculative agencies in the literature on the performativity of economics is an excellent example of sociologists focusing on what needs to happen to permit individuals to calculate the way economists assume they do. We need both theories – like economics itself – and tools – computer, calculators, pen and paper – to do the work that economists often assume happens instantly, costlessly, and without requiring any special equipment or knowledge.



  1. funny, i was just talking about this to my grad students yesterday. i framed the issue using a rubric from viviana zelizer that divides econ soc into:
    * “extensions” of orthodox economics into non-market contexts (i.e., applied micro like Gary Becker or Steve Levitt)
    * “context” to flesh out markets by relaxing but not wholesale rejecting assumptions (e.g., embeddedness)
    * “alternatives” that approach markets from a completely different perspective that may be incompatible with conventional econ (e.g., exchange circuits, identity from networks, etc)

    most of what we think of as econ soc is context or alternative and in practice they bleed into each other. for instance, Joel Podolny is basically using Harrison White’s ideas but with the moderation characteristic of Mark Granovetter. likewise, i see you and Michael Bishop (in his reply post) taking similar ideas but you’re pushing them as “alternative” whereas he is taking them as “context”

  2. limitatiosnandshortcomings

     /  February 3, 2010

    “The easiest place to see this is in Durkheim or Marx, both of whom explicitly argue that individuals are formed by society and not the other way around. In a way, the whole structure-agency debate (can we call it a debacle at this point?) stems from social theorists trying to grapple with the basic problem of reconciling a liberal belief in the self with a sociological belief in society.”

    Perhaps it might be better to rewrite that ‘sociological’ as ‘socialist’? That is Elias’s argument anyway. In any case, Marx is definitely a socialist of some sort and, according to Mauss, Durkheim “sympathized” with socialism. So, its empirically accurate! It also has the benefit, I guess, of making the assertion more symmetrical – rather than placing an ‘ideology’ in opposition to a ‘science.’

  3. Samuel Bowles does a good job dealing with other-regarding preferences. In his chapter on Preferences and Behavior in his “Microeconomics: Behavior, Institutions, and Evolution” he cites Falk and Fischbacher 1998, Fehr and Schmidt 1999, Bolton and Ockenfels 1999, Rabin 1993, Charness and Rabin 1999, and Levine 1998 as having developed utility models “capable of explaining a wide range of experimental behaviors.”

  1. Hirschman on Rational Choice Theory « Permutations