This post offers a little piece of intellectual autobiography that I hope will place some other posts in a clearer light — especially the posts related to the later Wittgenstein, and the posts on economics. For me, it’s a chance to sort out some of my current thinking by considering what preceded it.
There was a time, shortly after my first exposure to the history of economic ideas, following years of being focused almost exclusively on the humanities, when I thought that what the scholarly world really needed was a kind of new grand unified theory of the social sciences. All I look for from a social science — from any science — is an increase in the power to predict and control nature in ways that serve our purposes, whatever they are. The intellectual run-up to the global financial crisis seemed to show that orthodox economics, as practiced by the world’s leading economists, was failing by this standard.
And economics appeared to be at the vanguard of the social sciences. If economics was driven by “physics envy” — the scientistic desire to emulate the mathematico-deductive rigor of theoretical physics — then other social sciences, such as political science, seemed to be afflicted with “economics envy.” But the global financial crisis called into question whether the emperor was wearing any clothes. Under such circumstances, it seemed to me, wasn’t it worth questioning the reigning assumptions? Might it not be time for some revolutionary science?
Once I began reading about the history of economic ideas, along with critiques of contemporary economic thought, my enthusiasm for this idea grew. To begin with, the secondary literature on economic thought is full of persuasive critiques of the intellectual underpinnings of a great deal of contemporary academic economics, especially the kind practiced in “freshwater” economics departments and by business school professors teaching finance. The more one reads about rational choice theory and the assumptions of quasi-omniscient, hyper-mathematical rationality that dominate so much of mainstream academic economics, the more the field seems ripe for a paradigm shift based on a skeptical rethinking of the basic phenomena under investigation.
In fact, it occurred to me that the predictive successes of modern economics, such as they are, might be largely attributable to the fact that when one is investigating human behavior related to money and closely related subjects — the core focus of economics as a subject matter — the single most important factor in human behavior is calculated self-interest, or, as economists sometimes call it, “rationality.” When making money, people will generally try to make as much as they can with as little effort as possible; when spending money, people will generally try to spend as little as they can for the greatest possible return; and so on. If you’re trying to predict money-related human behavior using as simple a model as possible, a model based on the assumption that individual actors are more or less rational agents (in the economic sense of rationality) is probably your best bet.
But even if you achieve good predictive results with this model in the context of money-related activity, this success obviously does not imply that rationality will always be the most useful model for predicting human behavior, especially in contexts less directly related to money, or where we have good reason to believe that non-pecuniary concerns may trump pecuniary ones.
For example, when we try to imagine what contemporary American political life would look like if all the political actors behaved purely based on calculated self-interest — without gaming the results ahead of time by redefining “self-interest” to include all sorts of ad hoc preferences and motivations that we would not ordinarily view as “self-interested” — the thought experiment leads to absurd results. Do we live in a world with no voters, where politicians run for office without any ideological commitments, tribal affiliations and moral commitments play no role, and officials attract the public’s support by offering generous populist benefits, such as lavish infrastructure and a guaranteed minimum income, with no concern for the deficit? Not at all. Many of the central features of our political life are phenomena that one would not expect to see if the relevant actors were behaving purely as rational actors — unless, again, the idea of rationality is transformed beyond recognition or usefulness.
So, when one discovers that the rational choice methodologies of economics have expanded, perhaps based partly on economics’ scientistic allure, to other domains in the social sciences, the case for a new grand unified theory of the social sciences seems even stronger. If the use of rational choice theory in economics invites skeptical questioning, the use of rational choice theory in, for example, political science — in so-called “public choice theory” — can sometimes seem not only absurd but useless. What unexpected predictive successes can public choice theory claim, against the countless instances where its models would lead us astray? The same could be asked of many rational-choice-based forays into sociology, such as the study of family life.
Certainly, focusing on calculated self-interest may help to dispel comforting illusions about human behavior — for example, if anyone thinks that crime results from some kind of mental pathology, it could certainly be useful to show the contexts in which rationality helps explain crime. But how many comforting illusions are there left to dispel today? Hasn’t the Machiavellian assumption of cold, calculating rationality as the driving force in all human behavior become our own dominant illusion — comforting us not by flattering our moral characters, but by flattering our cold-eyed realism, our courageous perceptiveness and freedom from childish illusions — even where an equally tractable alternative model might yield superior predictions?
With these thoughts in mind, I asked myself: why doesn’t someone develop a better alternative to rational choice theory that can displace its imperialistic role within the social sciences? Why, for example, doesn’t someone follow the lead of Thucydides, who recognized the great importance of self-interest to human behavior, but saw self-interest as one only one of human beings’ three central motivations — the other two being fear and honor?
(Oddly enough, there is apparently a long tradition of analyzing Thucydides’ Melian Dialogue through the lens of rational choice theory and game theory, rather than drawing on Thucydides’ writing as a resource for the critique of rational choice theory. I’ve always thought this is the kind of thing that makes a piece of writing a classic: it provides materials for varied and possibly even contradictory readings.)
During my initial exposure to the history of economic thought, I was also surprised to discover just how young the field of economics is — and how recently the pillars of contemporary orthodoxy were established. Given the dominance of sophisticated mathematical modeling in academic economics today, it’s remarkable for a newcomer to learn that this dominance only arose in the post-war era, in the wake of Paul Samuelson’s Foundations of Economic Analysis. Even the intersecting supply and demand curves that every student learns in Econ 101 are only from the later nineteenth century (although see this paper on predecessors of the “Marshallian” cross).
Another example: someone unfamiliar with the history of economic thought might assume that Adam Smith remains so prominent because of the valuable contributions he made to the field — the division of labor, the invisible hand — when in fact, in many ways he helped constitute the field — and with a work published in 1776. Of course, the mercantilists and physiocrats addressed economic issues before Smith, and government ministers have always made economic decisions based on implicit or explicit assumptions about how money and trade work. But there is no work comparable to The Wealth of Nations before The Wealth of Nations.
The more one learns about the history of economic thought, the shallower seem the roots of many apparently fixed features of the contemporary economic landscape. GDP? No one seriously bothered to measure national income at all, apparently, until after the Great Depression. The idea that the sole responsibility of a corporation is to increase profits for the benefit of shareholders? Apparently this theory didn’t really catch on until the 1980s. The private limited liability corporation itself? An invention of the nineteenth century. The now-sacred idea of a 2% inflation target for central banks? Made up off the cuff by some guys in New Zealand around 1990.
One of the values of studying history is coming to understand how different things have been and could be — learning that what may seem a natural feature of the landscape, fixed and unavoidable, is in fact the contingent outcome of contingent forces and events, and remains susceptible to change. This was, in any case, the effect that studying the history of economic thought had for me: it strengthened my sense that the current orthodoxies of economics and other rational-choice-driven social sciences could be displaced.
It was also encouraging to learn about the kinds of thinkers who once wrote on economics, before it became a largely mathematical discipline. Smith was a philosopher of moral sentiments before he was a theorist of national wealth. Keynes was famously a member of the Bloomsbury Group. Hayek and Schumpeter were political philosophers as much as technical analysts of the economy.
Yet another consideration that made me skeptical of rational choice theory, and curious about the possibility of an alternative, was my sense of a parallel between the current states of academic economics and of academic philosophy. In general, I happily defer to technical experts and tend to be very skeptical of outsiders voicing marginalized perspectives on technical issues — like anti-vaxxers, or conspiracy theorists opining on the technical aspects of a plane crash, or the Danish “environmentalist” who made a name for himself by downplaying the threat of climate change.
But my experience with academic philosophy had illustrated that it is sometimes possible for a highly questionable methodology to maintain a nearly exclusive dominance in an academic field based not on any publicly defensible reasons, but simply based on sociological factors (as I have considered here and there on this blog). It seemed plausible that something similar might be happening in academic economics — and for partly similar reasons, including a misplaced, scientistic desire to ape the deductive rigor of the natural sciences.
Finally, my interest in an alternative to rational choice theory was heightened by developments in behavioral economics. By drawing empirically demonstrated insights about human behavior from psychology and applying them to economics, behavioral economics seemed to be moving in the right direction — away from rational choice theory and toward models of human behavior that remain highly tractable but achieve increased predictive power by incorporating simple refinements to the homo economicus model. The fact that behavioral economics was making inroads into mainstream academic economic thought, rather than being rejected and ridiculed like nearly all heterodox economics, also seemed like a promising sign.
On the other hand, behavioral economics was not itself the alternative to rational choice theory that I was hoping to find. The refinements to rational choice theory offered by behavioral economics seemed both lacking in any kind of methodological unity and at the same time needlessly modest — in one sense, too randomly scattered, and in another, scattered across too small a space. Behavioral economics seemed to be based on a grab-bag of relatively obvious psychological findings, with most of the findings related to inaccurate risk perception and a few other cognitive biases. The identification of those biases is of course valuable — and appears to be proceeding at a rapid pace — but such biases don’t seem up to the task of modeling what Keynes called “animal spirits,” the visceral forces that drive, for example, speculative booms and busts.
Nor do the small-bore, “thinking fast“-type cognitive biases of behavioral economics seem capable of incorporating Thucydides’ plausible claims regarding fear and honor. Both fear and honor are deeply rooted and persistent. They are not knee-jerk reactions that are overcome once we “think slow” and engage in rational cost-benefit analysis.
The psychological phenomena used in behavioral economics may be the most useful phenomena for predicting how consumers will shop, or how workers will invest for their retirements — in other words, for predicting paradigmatically economic behavior — but may be less useful in social science contexts further from economics, it seemed to me, such as predicting political behavior, or the drive to war.
When it comes to politics, war, family, and countless other deeply significant realms of human behavior, I suspected that the most useful models might draw less on the one-dimensional rationalism of homo economicus, even if modified by various cognitive biases, and more on the basic categories of human experience analyzed by, say, anthropologists: tribalism, the sacred, kinship, fairness, etc. This was the kind of thing I had in mind in the posts on “tribal cognition,” as well as the post on a possible hierarchy of political needs.
Interlude: A Workshop
As I was thinking about the issues above, it occurred to me that it could be interesting to teach a year-long multidisciplinary class, structured like a workshop, that went through three phases: First, in the literary phase, the students would read one or more works of realist literature, works that are known for their psychological perceptiveness, such as Anna Karenina — or maybe Macbeth. The students would attempt to draw from the work various generalizable observations about human behavior. Second, in the psychological phase, the students would treat the observations from the first phase as hypotheses. They would design simple laboratory experiments to test whether the observations from the first phase were in fact accurate generalizations about human behavior, and if so, to what extent and in what respect. Third, in the behavioral economics phase, the students would use an experimentally supported conclusion from the second phase as a basis for developing a simple (ideally formal, mathematical) model about some aspect of economic behavior.
At the end of the course, the students would have a better idea about how to read literature closely, how to develop and conduct a psychological experiment, and how to develop a model in economics.
It occurred to me later, though, that the kinds of psychological insight we value in literature are often the kinds of deeply contextual, even person-specific observations that cannot be generalized. They’re valuable precisely because of their specificity, because they’re not the kinds of bland, generic observations that can be tested by putting college students in front of a computer game in a lab. This is one reason that psychology can never hope to displace literary study, and why so many students who plan to study psychology — because they think it will deepen their understanding of other people, of the complexities and subtleties of other people and of themselves — end up largely disappointed.
An example: when Tolstoy makes an observation about a character, or even simply shows a character doing something, we may be struck by the thought — yes, that’s exactly what Count Vronsky would do. These insights have value in part because we can also say — yes, that’s exactly what someone like Count Vronsky would do. We may even be able to realize: “My friend is like Count Vronsky in that respect,” or “I am like Count Vronsky in that respect.” But the most interesting insights into character that Tolstoy offers will probably not prompt the response — yes, that’s what human beings do, that’s a general fact about human beings that I hadn’t noticed before. Or am I wrong? My thoughts remain unsettled here…
Either way, I suppose I still think it would be fun to try to design and teach a workshop like this. Generalizable, testable observations about human behavior might not be the most valuable or distinctive thing that literature can offer. But if you’re looking for such observations, realist literature could offer a treasure trove.
(Incidentally, there is a defense of some kinds of literary study buried in this line of thought — a quasi-scientific defense of the humanities that I’m not sure I mentioned in earlier posts on that subject. Through reading a work like Anna Karenina, it is not inconceivable that we might become better able to predict the behavior of those around us, by drawing useful analogies between characters’ thoughts and behaviors and the thoughts and behaviors of real people. Once you realize that someone you know is like Count Vronsky in some specific respect, and you know that this aspect of Count Vronsky’s character led him to do something under some set of circumstances, you might predict that the person you know will do something similar under an analogous set of circumstances. You might not have come up with this prediction had you not read Anna Karenina. Is it implausible that readers might gain a better understanding of specific others through reading literature, and that this improved understanding might even result in an improved capacity for prediction — although one that it is hard to imagine testing? And because the process would involve drawing deeply contextual analogies between literary “thick descriptions” and one’s own experiences of other individuals, or of oneself, it is hard to imagine how any social science could offer the same kind of understanding.)
Conclusion: Pluralistic Disillusionment
You might have noticed that nearly all of the post up to this point has been phrased in the past tense. The idea of a grand alternative to rational choice theory appealed to me, etc. That’s because — and here this longest-ever post will end — I eventually came to feel that the social sciences simply do not need a grand unified theory.
For someone like me who looks to the sciences, including the social sciences, solely for increased predictive power — in fact, the ability to provide such power is basically how I would define a “science” — there is no need for all the predictive tools to rest on a single set of substantive or methodological foundations.
The more one studies actual contemporary economics, rather than the history of economic thought, the more one has the sense that the former consists of a hodge-podge of models for predicting various phenomena. When Paul Krugman uses an IS/LM model to predict the effects of fiscal and monetary policies in a liquidity trap, the heterodox economist Lars Syll dismisses the model as a “gadget” resting on unacceptable premises — but as I suggested in a recent post, if the gadget makes good predictions, why should we denigrate it?
Certainly, where a field of investigation rests on premises that appear unrealistic in ways that might affect the field’s ability to make useful predictions, the field could be ripe for a paradigm shift, and the shift might involve rethinking things based on simplified but in relevant respects more realistic assumptions. But the payoff — and the test of whether the rethinking was worthwhile — should still be improved predictions, in my view.
Sadly, it is unclear to what extent the use of more realistic models of human behavior by heterodox economists is currently resulting in more useful predictions than those offered by orthodox economists — especially now that many prominent mainstream economists on the left have shown themselves, as Dean Baker recently noted, willing to defend approaches and conclusions that were ridiculed as “off-the-wall” as recently as a decade ago.
There is already so much water under the bridge in economics, so many useful models for predicting various things, admittedly scattered among all the predictively useless exercises in mathematical cleverness. It’s hard to imagine a paradigm shift involving a new picture of human behavior coming along and improving on the predictive power of the existing models in a wholesale way. At most, one could imagine a new picture being used to generate predictively useful models for those aspects of the economy where existing models have shown the greatest weakness — such as the prediction of speculative manias, panics, and crashes (to echo Charles Kindleberger). Just as we continue to use Newtonian physics for predicting most phenomena that happen at an ordinary human scale, while we use general relativity for (among other things) very large dimensions and quantum mechanics for (among other things) very small dimensions, so one could imagine using rational-choice-based gadgets, modified where useful by behavioral economics, for a lot of familiar questions in economics, and some more realistic conception of human behavior for questions in certain special domains.
And if economics by itself needs no grand unified theory of human behavior, then the social sciences as a whole — political science, sociology, psychology, anthropology, etc. — need one even less. Wherever the social sciences fail to give us the predictive powers that we seek, it may be useful to skeptically rethink our substantive and methodological assumptions — and in doing so, it may be useful to consider models of human behavior that lie outside the modern social sciences, such as Thucydides’ tripartite scheme. But it doesn’t seem likely that the development of a general alternative to rational choice theory, unbound from any specific problem or subject of inquiry, would be a productive project.
Indeed, if such an abstract alternative could have any value, it would most likely not be in the social sciences, but in a more abstract, more philosophical, and less predictive field such as political theory.