Games aren’t life and death – they’re much more important than that …
I’ve just discovered Herb Gintis. I suppose I should have noticed his name before now, given the long time that I’ve been staring glazed-eyed at the torrent of stuff that comes out on the evol-psych mailing list, to which he frequently contibutes. He’s Emeritus Professor of Economics at the University of Massachusetts and is at the Santa Fe Institute. His rich Website has cartloads of academic material on game theory, experimental economics, and suchlike voguish keys to the universe.
I know some instructors will be offended by my harsh treatment of neoclassical economics, classical game theory, and other traditional topics. My defense is that this stance counters the more common bias of professors, which is to hold received wisdom in excessively high regard, doubtless recalling the pain and suffering they incurred in acquiring it and the misgivings they were obliged to swallow before believing it.
…Experimental game theory has become increasingly influential in affecting research priorities. I have thus included references to tests of the descriptive accuracy of particular models throughout the book. Ironically, game theory is often hoist on its own petard: many of its most fundamental predictions—predictions that would have been too vague to test with any confidence in the pre-game-theoretic era—are decisively and repeatedly disconfirmed, in laboratory settings, with substantial agreement among experimenters, regardless of their theoretical priors.
Chapter 11 reviews this material at some length, concluding that these predictive failures are due to game theory's adopting Homo economicus from neoclassical economics. Homo economicus is great when people are faced with anonymous marketlike conditions, but not when engaged in strategic interaction.
…Economists are fond of using the Folk Theorem of repeated games and the Tit-for-Tat simulations to argue that human cooperation can be understood in terms of long-run, enlightened self-interest, but we will argue in chapter 11 that this view is profoundly incorrect. There are two major problems with the idea that cooperation can be understood in terms of long-run self-interest (charitably interpreted to include regard for kin). The first is that self-interest results in cooperation only when agents are sufficiently future-oriented (i.e., the discount rate is very low); but in situations where a social system is threatened and likely to be destroyed, cooperation is most central to survival and agents are likely to be very present-oriented, since the probability of future interactions is low. Therefore, societies in which cooperation is based on long-run self-interest will invariably collapse when seriously threatened. The second problem is that there is sizable evidence that we are considerably more prosocial than is predicted by the long-run self-interest models.
Except in the context of anonymous market interactions, the idea that human beings are self-interested is particularly implausible. Indeed, some of the major predictive failures of game theory stem from not recognizing the positive and negative aspects of preference and welfare interdependence. Homo economicus might be reasonably described as a sociopath if he were to be set loose in society.
…In short, evolutionary game theory replaces the idea that games have "solutions" that agents "learn," with the idea that games are embedded in natural and social processes that produce agents who play effectively.
Dispensing with the rationality postulate does not imply that people are irrational (whatever that means). The point is that the concept of "rationality" does not help us understand the world.
There are some course notes elsewhere that drive home part of this:
Strong Reciprocity and Human Sociality
•Human groups are highly social despite a low level of relatedness.
•There is an empirically identifiable form of prosocial behavior in humans, which
I will call strong reciprocity, that in part explains human sociality.
•A strong reciprocator (a) is predisposed to cooperate with others; and (b) punishes
non-cooperators even when this behavior cannot be justified in terms of extended
kinship, reciprocal altruism, or what is the same thing, long term self interest.
Libertarians should pay attention, regardless of ‘theoretical priors’: our rhetoric, still under the shadow of Ayn Rand, makes too much play with the notions of rationality and self-interest. (Rand thought her mother – (sister?) – was beyond hope of reclaiming for rationality when she visited Ayn in the States, tasted the delights of American civilization for weeks, and still pined for Russia. A concept of human excellence that has no place for love of homeland – the place in which you were born, not the place you choose to adopt – is a monstrous concept.)
Perhaps I should have heard of Gintis from a different direction – decades ago he coauthored a book called Schooling in Capitalist America, and he seems to have been - still is ? - as much of an egalitarian as that title portends. Pouf! I care not a fig. This is powerful science. And not for a moment will I be persuaded that its findings should give aid and comfort to the statists.
