Saturday, February 18, 2006

Pinker Peeves, Final Edition

[I posted this message twice before, but Blogger problems caused the posts and comments not to work properly. The problem has allegedly been fixed; we shall see.]

This will be my last post on The Blank Slate. Lest it seem I’m hostile to the book, I actually thought it was excellent. But as I said in my original post, points of disagreement are more exciting.

If I had to make just one complaint about the book, it would be this: Pinker has an annoying tendency to construct straw men, sometimes by mischaracterizing a person or position. His unjust vilification of Spencer, noted in my first post, is one example. As someone observed in the comments to that post, Pinker’s motivation is most likely the desire to distance himself and his thesis from ideologies that some readers will find offensive. The motive is understandable, but it’s frustrating to those who know better – and who are willing to take the science where it goes, instead of condemning it just because it might give succor to the wrong sort of people.

It’s also frustrating when a misrepresented position is, um, yours. Here’s a passage that irked me:
Steven Rose and the sociologist Hilary Rose, for instance, call evolutionary psychology a “right-wing libertarian attack on collectivity.” But the accusation is factually incorrect ... and it is conceptually incorrect. The real alternative to romantic collectivism is not “right-wing libertarianism” but a recognition that social generosity comes from a complex suite of thoughts and emotions rooted in the logic of reciprocity. (p. 255)
Riiiight – because libertarians don’t believe in reciprocity? Nobody who understands libertarian thought at all would think that. In fact, reciprocity is a key feature of the libertarian conception of good social order; it is the leading alternative to coercion. Pinker assuredly understands this; he’s just trying to distance himself from libertarianism because some people don’t like it.

Just a page later, Pinker has this to say about economic models of human behavior:
Studies of altruism by behavioral economists have thrown a spotlight on this sword of Damocles [the threat of reprisal against those who break the rules of reciprocity] by showing that people are neither the amoral egoists of classical economic theory nor the all-for-one-and-one-for-all communalists of utopian fantasies. (p. 256)
Now, you don’t have to try hard to convince me that the neoclassical model of homo economicus leaves something to be desired. It doesn’t capture everything interesting about human behavior, and sometimes it leads us astray. But what evidence does Pinker adduce to show an alleged flaw in the neoclassical model? The results of game theory experiments:
In the Ultimatum Game, for example, one participant gets a large sum of money to divide between himself and another participant, and the second one can take it or leave it. If he leaves it, neither side gets anything. A selfish proposer would keep the lion’s share; a selfish respondent would accept the remaining crumbs, no matter how small... In reality the proposer tends to offer almost half of the total sum, and the respondent doesn’t settle for much less than half... We know that the proposer’s generosity is driven by fear of a spiteful response because of the outcome of two variants of the experiment. In the Dictator game, the proposer simply divides the sum between the two players and there is nothing the respondent can do about it. With no fear of reprisal, the proposer makes a far stingier offer. The offer still tends to be more generous than it has to be, because the proposer worries about getting a reputation for stinginess that could come back to bite him in the long run. We know this because of the outcome of the Double-Blind Dictator game, where proposals from many players are sealed and neither the respondent nor the experimenter knows who offered how much. In this variant, generosity plummets; a majority of the proposers keep everything for themselves.
These results are hardly a threat to the neoclassical model. The behavior of the proposers is perfectly consistent with that of selfish rational agents seeking to advance their long-run interest. The behavior of the respondents is a little tougher to explain, but the key is seeing their behavior in the context of a more comprehensive game in which agents have to signal that they won’t let themselves be taken advantage of. This is, again, a matter of seeking one’s long-run interest. Here’s my question (to which I don’t know the answer): what do the respondents in the Ultimatum Game do if they know their own identities are concealed? Rational selfish agents would accept the most paltry offers, since they’d have no need to protect their reputations. If they still rejected lousy offers, that would be evidence against the standard neoclassical model.

Again, I’m not trying to defend the neoclassical model in all respects. I suspect what’s really going on is that rational long-run strategies of reciprocity and reprisal for non-cooperation are, to some extent at least, built into us by evolution (Pinker’s thesis) rather than deliberately chosen. Clearly the result is mediated by conscious thought, though, because otherwise knowledge of anonymity wouldn’t affect the proposers’ behavior in the Double-Blind Dictator game. In any case, the neoclassical model effectively predicts much actual behavior in these situations, so why choose this particular context to cast aspersions on the neoclassical model? Again, I think it’s Pinker’s attempt to placate a specific audience – this time people in non-economic social sciences who dislike the economic approach.

1 comment:

Mike Giberson said...

You may find useful the discussion in Vernon Smith's collection Bargaining and Market Behavior.

I don't find an explicit answer to your question about respondents in double-blind Ultimatum games, but his introduction to Part II of the book and the following chapters address some of the issues you raise. Smith suggests that, contrary to the usual game theory experimentalists' assumptions, subjects act as if the laboratory game is part of a larger social exchange. Smith and co-authors interpret their results in the paper in chapter 5 as indicating that subjects are not acting so much from concern for others as from concern over what others may know and think about the subject.

On the question of "defend[ing] the neoclassical model," I think issues are raised worth thinking about. But to really get anywhere here I think it pays to specify just what you mean by neoclassical model. The world of "neoclassical economics" is just too diverse and adaptable.

I'm not sure you can give the neoclassical approach much credit for predicting "much actual behavior" however. Experimental work has demonstrated that, for example, awarding the Proposer role to the high-scorer in a Trivia quiz systematically reduces the offers to Responders. I don't think that neoclassical economics in its standard simiplified version says anything about this or other systematic effects that can be induced in laboratory games. Rather, for one-shot games simple economic theory predicts the proposer will make the smallest possible positive offer and the responder will take it. A richer understanding of human choice is required to make sense of the behavior.

As for Pinker's attempt to placate some part of his audience, well I hate to see someone as smart as he is just playing to the prejudices of the crowd. I wouldn't take it as a serious challenge to economics, but still pushing back -- like you do here -- is a good thing.