Monday, November 28, 2011

The difference between rationality and rational self-interest

by Massimo Pigliucci
Far too often in the course of discussions on this blog or with friends I encounter a fundamental confusion: people using the word “rational” (or “irrational”) as if there were only one clear meaning of it, from which their own position follows logically (or, alternatively, other people’s position is therefore deemed illogical). Not so.

The ancient Greeks had already made a distinction between “theoria” (theoretical reason) and “praxis” (practical reason), the first one being the sort of reasoning that is supposed to reflect a point of view from nowhere, as the phrase goes, while the second one is a type of instrumental rationality deployed in the pursuit of specific ends. (Yes, modern philosophy and cognitive science have respectively argued and shown convincingly that human beings can hardly access a view from nowhere, but it is an ideal, meaning that in some cases we want to transcend as much as possible our foibles and specific issues to look at the world as broadly as is feasible for us.)

While all systems of logic do need starting assumptions (or axioms, as they are called in math), and no interestingly complex individual system can be completely justified from within itself, the aims and assumptions of theoretical and practical reason are nonetheless clearly distinct. Theoretical reason is what we deploy when we wish to arrive at general principles of logic that apply regardless of the circumstances (e.g., the principle of non-contradiction), while the practical reason tells us what we need to do in order to maximize a specified utility function.

Of course, much of the confusion I referred to above comes in when I have discussions with libertarian-oriented people, who never fail to bring in rational self-interest, or the principle of rational egoism, as a trump card. We don’t need to get (again) into Rand-style pseudophilosophy here, as rational egoism was seriously discussed (and criticized) by Henry Sidgwick in his The Methods of Ethics already in 1907. One of the things that is often confused by libertarian/objectivists is that there is a fundamental distinction between rational egoism (a type of reasoning) and ethical egoism (an ethical position). One can readily agree that rational egoism is a particular kind of instrumental rationality (i.e., it isn’t “irrational”) without having to concede that it is ethical (I certainly don’t).

Indeed, this distinction between the (instrumental) logic and the ethics of one’s actions is one of the issues that makes economics such a fascinating field. Pace the most scientistically inclined of my readers and commenters, as this is one area where we clearly see the difference between facts and values. (Yes, I am aware of Quine’s denial of that distinction, I think that Quine is over-interpreted and over-rated in this respect. At best, he was able to argue that the difference isn’t always sharp, not that it doesn’t exist.)

I think of economics as a type of (soft) social science, not as quackery, despite the fact that much nonsense is being sold to us all in its name by a variety of pundits and experts. I mean that there are facts about economies that we can’t (or shouldn’t, at any rate) ignore, regardless of our values about fairness, distribution of wealth, etc. But it is also obvious that what is “rational” in economics does not depend only on those facts, it greatly depends on one’s values and how the latter determine one’s priorities. So, for instance, it may very well be that laissez-faire capitalism is the best way to maximize distribution of goods (I am not conceding the point, I am assuming it for the sake of argument), but even so we could (should, really) decide that other criteria need to be counted into the equation too, and sometimes may even override the goal of maximization of distribution (or even, gulp!, of maximization of shareholders’ profits). Criteria such as fair access to resources (like education and health care), safety, environmental impact, and so forth are among those that might be considered.

So, one cannot simply assume without argument that rational self-interest is equivalent to rationality tout court. Firstly, because it is a particular type of instrumental rationality whose assumptions may (should) be questioned. Secondly, because there are other uses of instrumental rationality available (such as maximizing societal benefit, enhancement of human flourishing, etc.). Thirdly, because it isn’t the type of general rationality that seeks principles independent of particular points of view (i.e., it is a type of praxis, not theoria).

Let me take up another example, which concerns issues that have arisen during the (ongoing) preparation of a collection of essays on the philosophy of pseudoscience that Maarten Boudry and I are putting together for the University of Chicago Press. It hinges on the following question: can we reasonably (ah!) say that it is rational to align one’s beliefs with the available evidence (a la Bayes theorem and Hume’s dictum)? Most skeptics would say hell yes, and repeat the Hume-inspired Sagan mantra: extraordinary claims require extraordinary evidence.

Yes, but that is true from the point of view of theoretical reason. Cognitive science has begun to show that human beings apparently need to believe, at least some of the time, in things for which there is no evidence (or which are, in fact, contradicted by the evidence). Optimists — which can be defined as people who disregard some of the negative facts about their own abilities or the state of the world — apparently live longer and happier lives than pessimists (who actually tend to be realists, i.e., they come closer to aligning their beliefs to the available evidence). So from the point of view of practical reason, it may be argued that it is actually rational for people not to behave too strictly as Bayesian estimators, theoria be damned.

The upshot is that we skeptics are perfectly justified in telling someone who believes in, say, astrology, that he is theoretically irrational, but not necessarily instrumentally so. Then again, things get complicated by the fact that if the person in question starts acting on the basis of his false beliefs in situations where his health is at risk (say, by going for homeopathic “remedies” instead of proven medical procedures) then we can justifiably say that he is behaving irrationally both from the point of view of theoria and of praxis — at least assuming that he actually values his life over his temporary psychological satisfaction.

So, the next time you are about to hurl the “that’s irrational” judgment at someone, stop and think in what sense you mean it, and consider whether it actually applies to the situation at hand. It may save you some grief and frustration, and who knows, you may even get your point of view across in a less threatening manner. Assuming that’s of concern to you, of course.


  1. Excellent post.
    Some remarks:
    1. Economic reasoning (even the neoclassical sort) does NOT require that individual agents behave rationally, even if that hypothesis is used to construct theories. For instance, Gary Becker showed (in his 1962 paper "Irrational behavior and economic theory", The Journal of Political Economy, 70(1): 1-13) that postulating a population of irrational actors such as consumers or firms (with two kinds of "irrationality": reacting at random or persisting stubbornly in the same behavior, in the face of a change in, say, relative prices) generate the same conclusions that can be generated by postulating a set of perfectly "rational" (utility maximizing) consumers or firms; e.g. the same supply and demand curves.
    2. Amartya Sen has shown that rational individual preferences need not be "selfish": utility can be derived from the welfare of others as well as from self regard. He has proposed a number of economic models reflecting these kinds of preferences.
    3. Practical rules of behavior can be derived from false scientific theories; e.g. a sailor can find his way in high seas by the apparent position of stars, using the Ptolemaic theory of a fixed Earth surrounded by rotating celestial spheres, without needing Copernican, Newtonian or Einsteinian cosmology. Assuming the Earth is immobile and stars rotate around gets you to port just as well (within certain bounds of imprecision of course). You may even predict eclipses, lunar phases and the position of planets (albeit with some degree of imprecision, some nasty "anomalies" and plenty of epicycles) based on Ptolemaic cosmology. Being practically rational (in the sense of using means conducive to your goals), and even being practically right in making predictions, does not imply being scientifically right.

  2. Massimo: ...we could (should, really) decide that other criteria need to be counted into the equation too, and sometimes may even override the goal of maximization of distribution...

    I hope that you don't mind a lengthy quote, but the following interview with Herman E. Daly lends support to what you're saying, only from a source within the field of economics.


    ...I think economics has been very successful in one very important area: allocative efficiency. So that [regarding] the efficiency of allocation of scarce resources among competing ends, economists have preached the importance of decentralized decision-making coordinated by markets and the price system. This has historically dramatically proven to be much better than central planning in this collapse of the former Soviet Union and so forth, [and] is something that needs to be recognized and taken seriously. As far as market control of allocation of resources [goes], I think economics has provided a whole lot. And good reasons were given for why this is so.

    Now, my problem is that allocation is only one fundamental economic problem. It's important, but it's only one. There are two others, which I mentioned briefly before: there's distribution and scale. So allocation is about how resources get divided up among different users - how much goes to produce bicycles, how much to cars, how much to houses. You know, is that efficient given what people want and their ability to pay. You end up giving people the most that you can get of what they want with the resources available. That's a question of efficiency - are allocations efficient or inefficient? The distribution question is a question of justice. Who gets all the stuff that was produced. Does it go to you or me. And that's a question of justice - is it fair, is it a fair distribution? And then the third question of scale - the total amount that gets produced in all of the resources and the depletion and pollution generated by the use of those resources. Is that at a total scale which is within the absorptive and regenerative capacity of the ecosystem? Or are you destroying natural capital at a rate which is too great? The welfare effects of destroying natural capital may be greater than the welfare benefits of what you produced. And whose products required the destruction of that national capital? So just from a purely anthropocentric view, not giving any value to other species or nature intrinsically, just as an instrument for human betterment, you still run into this limit of scale.

  3. PS: Just to put a finer point on my Daly quote, I should add: "...while economics has done a great deal in the matter of economic efficiency, it has been negligent regarding distributive justice and extremely negligent regarding optimal scale. So those two things are where the effort should go."

    But what Daly does not mention is the degree to which this professional negligence among economists is non-accidental; e.g. as a predictable consequence of prior ideological bias and/or personal conflict of interest (as per the academic portion of the documentary "Inside Job"). In such cases, the pursuit of rational self-interest is perhaps central to the problem.

  4. Random bits: Oh ancient Greeks, what didn't you know? Does Godel's incompleteness theorem require "interesting complexity"? Does "pace" and "contra" mean the same thing, and what is the antonym of both?

    The fact that nonsense is being sold in its name does not make a something a pseudoscience. Economics is a social science, but I wouldn't modify it with soft. I think of "soft" modifying "science", where softer means indirectness of measurement. So something like fMRI data for neuroscience could be softer than a company's data for microeconomics. (And if you just use it to rank sciences on a scale, psychology and political science would be softer than econ.)

    The terminology in economics is positive, mathematical formula for maximization or dealing with large data sets, and normative, the values chosen to pursue.

    What does "maximize distribution of goods" mean? Equal distribution? Maximize the number of goods? Economic efficiency? I'm afraid your economic thinking is muddled.

    Mark Erickson

  5. If, as I believe, moderate optimists are on average more successful than realistic pessimists, then it's rational to self deceive with regard to one's prospects in taking economic risks that require a modicum of "luck." Feeling lucky brings you luck in other words. So that it's theoretically rational to be practically so as well.

  6. Mark,

    > Economics is a social science, but I wouldn't modify it with soft. I think of "soft" modifying "science", where softer means indirectness of measurement. <

    In Nonsense on Stilts I talked about soft sciences being sciences that have a good amount of empirical evidence but not much theory, or a good amount of theory but not well connected empirical evidence. Economics falls within that group for reasons similar to sociology and psychology. It certainly is no pseudoscience.

    > What does "maximize distribution of goods" mean? Equal distribution? Maximize the number of goods? Economic efficiency? I'm afraid your economic thinking is muddled. <

    I don't think it is, clearly we are talking about maximizing distribution of goods in terms of economic efficiency. At any rate, I don't see why this was confusing, since the contrast was with criteria that are inherently non economical, like fairness, wealth distribution, environmental impact, etc.

    Oh, and someone already suggested that you consider changing your nickname. I don't know if you realize it, but in the US it has acquired horrible connotations after the recent shootings in Norway. The first 9 out of 10 google hits for that phrase refer to the tragedy (yes, one does refer to your blog).

  7. Economics has plenty of theory and empirical evidence. I don't understand your framework. What is the prototypical all theory-no evidence science and vice versa?

    Fairness is purely normative, but wealth distribution and environmental impact are not. You can (and I'm sure someone has) compare the economic performance of different countries controlling for all but wealth distribution. Externalities, or social costs, are basic Econ 101 concepts. An example would be to measure the effect of pollution on productivity.

    While on principle I don't feel I should give up my 20-year old nickname - it will stay on my blog - I'll change my display name for Google. I'd given up pure anonymity with all my other accounts a while ago, but it just didn't occur to me to change Google.

  8. Mark,

    > Economics has plenty of theory and empirical evidence. <

    Right, the problem is that, by and large, they don't match. Unlike, say in physics (a hard science).

    > Externalities, or social costs, are basic Econ 101 concepts. <

    Yes, I'm aware of that. But they are, ahem, external to economics per se. It's a matter of quantifying our values and insert them as "externalities" into the system. A free market economy by itself takes them in no account whatsoever.

    As for your nickname, it was just a suggestion. As I said, it did start to jar me, which is why I did a google search and found what I reporte above.

  9. Massimo, wonderful post! This makes me think in a different way about that famous line from Hume: "'Tis not contrary to reason to prefer the destruction of the whole world to the scratching of my finger."

    (Following up on Alexander Kruel's comment -- It looks like the practical-reason and aristotle-noncontradiction links are also broken.)

  10. Are the Academic Skeptics putting Descartes before DeHorse? :)

  11. Massimo: An externality is a cost or benefit incurred by a party who is external to a transaction, but it's hardly "external to economics per se." Indeed, we can thank economists for acknowledging the problem (in its various forms) and for proposing solutions (usually involving some form of government action, such as Pigovian taxes).

  12. Good article.

    I'm curious to know what it would mean to optimize not for "rational self-interest" but rather "enhancement of human flourishing". Is it up to the individual to decide what it means for them to flourish? If so, it would seem to align with personal utility. If not, then who decides what it means to flourish? Is it the general consensus view of society, regardless of the view of that person?

  13. Scott,

    thanks, links fixed.


    yes, I'm aware of how economists use externalities. My point is that those are not inherently "economical," meaning that there is nothing at all in laissez-fair capitalism that would consider them. They are added in because they reflect society's values, which means that economics isn't just a matter of facts, values play a huge role.


    that's a long discussion about virtue ethics, see for instance

  14. Massimo: I agree that values (as in: morals) are required inputs in economics. Otherwise, who cares about externalities?

    But then your use of "economical" in this thread suggests that the term only belongs to laissez-fair capitalism (or what Quiggin calls "market fundamentalism"). I mean to suggest that it belongs equally well to other value-laden approaches (e.g. social democracy).

  15. Massimo, you know just enough about economics to be dangerous.

    On the more philosophical level, the discussion about hard and soft science: you are still confusing. "Don't match," is that the technical term? I don't see how evidence matching theory is different in any science. If it supports the theory, the theory is good for now. If it doesn't, then toss the theory and try again. Economics has plenty of theory that has plenty of matching evidence. Don't you think that economic research that wins a Nobel has lots of evidence that matches?

    So I'll take physics as the prototypical hard science, that wasn't hard, although I still don't like your definition. What is the prototypical soft science, or the softest?

  16. Massimo
    Just wanted to bring this to your attention. Could add to topic of either "free will" or "determinism"

  17. mufi,

    > your use of "economical" in this thread suggests that the term only belongs to laissez-fair capitalism <

    Yes, and that makes sense in the context of my discussion (I was, after all, referring to my chats with libertarians, aka economic fundamentalists...).


    > you know just enough about economics to be dangerous <

    So you say, but at the least I try.

    > "Don't match," is that the technical term? I don't see how evidence matching theory is different in any science. <

    Depends on how much the theory matches the evidence. In classical economics the match is close to zero, from what I understand. Hence the growing popularity of approaches like behavioral economics (or even, gasp! of neuro-economics, though I think that's just pandering to the neuro-band wagon).

    > What is the prototypical soft science, or the softest? <


  18. Hurrah, this argument needed to made again and it is argued well here! I will refrain from picking any nits.

  19. Well, one nit. I concede that I can see the point of the instrumental rationality of, say, believing in astrology if it "works for you" (as people invariably say). Yet this sort of activity makes me very, very nervous. If one gets practiced at anti-epistemology in a harmless context, then one may be more prepared to use it in a more serious one.

    It also occurs to me that although there might be some theoretically irrational behaviour that is instrumentally more rational than doing nothing, it's still hardly optimized. Sure, getting all excited about your horoscope may be better than lying on the couch picking lint out of your navel, but on the other hand... gah, for the love of Cthulhu, pick up a book!

    (By the way Massimo, I'm just arguing against this strain of idea, not any position you've staked out.)

  20. You do try, E for effort. And you can't be expected to know everything!

  21. Rand really bothers Massimo, doesn't she.


Note: Only a member of this blog may post a comment.