Does This Pass the Smell Test?

I'm amazed how scientists can discover "truths" which end up being not just misleading but dangerous. Think of all the eugenicists at the turn of the 20th century who claimed scientific validation for the peril of "breeding with inferior types". I'm afraid that the following is an example of a scientist rushing over a cliff without taking time to really examine his premises and data.

This is a post by Satoshi Kanazawa on his Psychology Today blog:

In an earlier post, I explain why virtually all stereotypes are empirically true. Stereotypes come from the aggregation and generalization of the daily experiences of millions of people, so they cannot possibly fail to be true. Stereotypes are based on massive empirical data, and empirical data don't lie. However, I also explain that one of the very few stereotypes that are not empirically true is the aphorism "You can't judge a book by its cover." Contrary to popular belief, you can very accurately judge people's character just by looking at them.

There have been a large number of experiments conducted in recent years to show that altruists (people who tend to cooperate in situations where self-interested behavior might benefit them personally) and egoists (people who tend not to cooperate in such situations) genuinely look different, and people can tell them apart simply by looking at them. These studies show that people pay more attention to the faces of cheaters, and later remember their faces more accurately, even when they don't know who are cooperators and who are defectors. The latest in the series of such experiments has just been published in the September 2009 issue of the journal Human Nature, with the very descriptive title "Altruism Can Be Assessed Correctly Based on Impression."

The study, conducted by Ryo Oda of the Nagoya Institute of Technology and his colleagues, clearly demonstrates that you can judge a book by its cover. ... In their study, a large number of male undergraduate students complete a self-reported altruism scale. Those in the top 10% on the altruism scores are designated as "altruists," and those in the bottom 10% are designated as "egoists." These "altruists" and "egoists" are then individually videotaped during a normal conversation with a blind confederate in a closeup shot. The first 30 seconds of the videotaped conversation are then shown to different groups of students at a university more than 800 miles away (in order to eliminate the possibility that the perceivers may personally know the targets). The video clips of the targets are shown to the perceivers without sound, to prevent the targets from betraying their level of altruism by verbal cues.

Their study shows that the perceives, when asked to estimate the targets' levels of altruism, can accurately guess who are altruists and who are egoists. Interestingly, even though perceivers who are themselves altruists tend to think that others are more altruistic in general than perceivers who are themselves egoists, both types of perceives can nonetheless accurately judge who are altruists and who are egoists among the targets. Surprisingly, men and women in their study are equally good at estimating the altruism level of total strangers. (There are theoretical reasons to believe that women are better judges of character than men in general.) Further, altruists are judged to be significantly more active, more generous, more responsible, friendlier, kinder, more extroverted, and as giving better impression than are egoists, but altruists are not judged to be more discreet, more hurried or more intelligent. Further analyses of Oda et al.'s data show that the key to detecting altruists is genuine smile, which is under involuntary control and is therefore difficult to fake. Altruists genuinely smile more frequently than egoists during natural conversations.

Oda et al.'s study is only the latest in the series of experiments which demonstrate that we can indeed judge a book by its cover. Nice, altruistic, and cooperative people look nice, altruistic, and cooperative; nasty, egoistic and uncooperative people look nasty, egoistic, and uncooperative. And we (both altruists and egoists, both men and women) have the capacity to tell them apart, after looking at them for only 30 seconds without sound! In retrospect, this should come as no surprise. We have been dealing with potential cheaters throughout evolutionary history, and being duped and deceived by them has always carried tremendous costs. In other words, the presence of cheaters has exerted strong selection pressure on our ancestors. It would be a miracle if a capacity to judge people's character based on their appearance did not evolve sometime during the course of human evolution.

First of all, the necessary disclaimer: I am not a scientist and don't have background in the above fields. But the above doesn't pass the smell test.

If it is so obvious by just "looking" at a person, then how did Bernie Madoff manage to create so many victims? How does any con man succeed? They have an air about them, a look, a self composure, an assuring air that lures people into trusting them and getting duped. If the world were as simple as Satoshi Kanazawa presents it, then no con man could ever thrive.

There are well know experiments where various people are put on trial using the same "facts" but we get different results. The ugly, the different, the poor, and the outcast are more likely to be judged "guilty" while the pretty, the well-dressed, the successful, and the socially elite tend to be judged innocent: on the very same set of facts. We are more favourable to the beautiful than the ugly so we get miscarriages of justice. Here the stereotype simply ignores the facts.

I think an element of truth can be found in Kanazawa's claims. I am willing to accept that people with a lifetime experience of altruism and collaboration may develop traits that we can read. Similarly, those who have had a lifetime of failure and rejection (as well, maybe those who have spent a lifetime of manipulation and greed) have acquired traits -- mannerisms, presentation of self, etc. -- that we can sometimes read. But from here the leap to the simplistic generalization that you can "read a book by its cover" is unjustified.

I accept that stereotypes can be useful. But they are not always the "the aggregation and generalization of the daily experiences of millions of people". Stereotypes can be "taught" and that is exactly what racists do. So how do you distinguish between (a) an "unbiased" stereotype learned from examples and (b) a "biased" stereotype that is unfounded but which you were taught as if it were a valid generalization? I don't think many people are perceptive enough to clearly distinguish the two cases. Worse, I don't believe the world is black-or-white. All of our stereotypes are some kind of mixture, part honest generalization, part learned prejudice. The exact proportion varies both by the specific stereotype and by specific individual. Generalizations here can be dangerous.

In the late 19th century there were "scientists" like Kanazawa teaching the view that there was a criminal type, that you could spot a criminal from his features. We laugh at such views today. Similarly we should laugh at Kanazawa's views. Life is just too complicated for such simple "truisms" to be true.

I do believe that you can read some facts from a face. I accept the work of Paul Ekman. I believe that he has both a sound theory and the practical data to back up his views. Kanazawa is an armchair "sociobiologist" (or evolutionary psychologist) who is making claims with no data and which no sound theory could support.

At the very least, Kanazawa should tone done his generalization to match empirical reality: the skills by which people can "read" and distinguish altruists from cheaters may be statitistically significant, but it is not a 100% right-every-time skill and it most certainly is a "skill" that varies from individual to individual. So you can run a test and get the result that Kanazawa claims, but this would not lead to the conclusion that one person can "read" another. And it certainly doesn't lead to the idiotic generalization that you can "read a book by its cover".

Anybody who, like Kanazawa, says "stereotypes come from the aggregation and generalization of the daily experiences of millions of people, so they cannot possibly fail to be true" is telling you something that is untrue. That's my stereotype. So take that Kanazawa!

If you believe Kanazawa, I have a bridge in Brooklyn which I swear is a really, really good deal and I'm willing to part with it cheap! Just look at me... with a face like this, could I be lying to you? And if you don't trust me, my buddies Bernie Madoff and Ted Bundy are willing to vouch for my character!