Monday, October 10, 2011

Intelligence is not Rationality.

Unlike most of the HBD crowd, I don't hold IQ in nearly the esteem that they do. Medicine attracts a lot of highly gifted people, people who still manage to do incredibly dumb things. Still, any fair observer of the literature out there cannot but agree that IQ is a strong correlate to worldly success.

It's therefore with some interest, that I've had the pleasure of  acquainting myself with the  works of Dr Keith Stanovich, Professor of Human Development and Applied Psychology of the University of Toronto, who has much to say on the subject of intelligence and rationality.

Stanovich is critical of the unquestioning acceptance of the IQ test. Unlike other detractors, he does not claim that the tests are not valid measures of intelligence, or that there are different types of intelligence, rather, Stanovich recognises that the IQ testing is valid for the determination of intelligence but misses the mark with regard to rationality. Rationality, according to Stanovich, is an all-together different thing to intelligence, and unlike most critics of IQ testing, is able to prove his case with some conviction.

Here is an easy-to-read article he wrote for Scientific American. 

There is a good YouTube video of him presenting his ideas. (Warning, it's about an hour and a half long). His use of George Bush in the early part of the video, as an example of the limitations of IQ testing--Bush was reputed to have an IQ in the 120-130 range--is reason enough to watch it.

He also has a home page where numerous papers of his are available for those who are interested.

One of the very surprising findings of his research is that high IQ, is in many instances, either weakly or not at all correlated to rationality, as it appears that high IQ individuals are just as able to irrationally "solve problems" as their low IQ peers.  The cause of rationality failures amongst the high IQ crowd seem to cluster around cognitive biases, information lack and "cognitive miserliness ". There also seems to be some evidence of cognitive limitations in rationalisation. All in all he provides a convincing, and more importantly, empirically justifiable argument.

There is also a kindle version of his book, What Intelligence Tests Miss. I'm off to read mine now.

Stenosophic Liberalism.

In my previous post, commentator KJJ made the following comment:
Have you considered Jonathan Haidt's examination of the moral reasoning of "liberals" and "conservatives"?

It dovetails with your concerns about overoptimization for one set of priniciples.

Whether by custom or nature, liberals tend to overoptimize for the "caring" and "fairness" virtues, while conservatives have a broader moral palette.

Haidt has noted that liberal populations tracks with major trade centers, especially longtime port cities.

This suggests to me that liberalism is the default governing morality for a diverse, commercial society in a time dominated by global trade.

And so our policies end up being set by specialists in equality and utilitarian analysis, with traditionalist concerns going by the wayside.
I'd like to thank KJJ for the link to Jonathan Haidt, whom I'd not heard of before. Haidt's specialty is the psychology of morality, particularly with regard to how people come to moral determination. He began by analysing the moral codes of many different cultures and was able to identify five common meta-themes. They are:

  1. Care for others, protecting them from harm. 
  2. Fairness, Justice, treating others equally.
  3. Loyalty to your group, family, nation. 
  4. Respect for tradition and legitimate authority. 
  5. Purity, avoiding disgusting things, foods, actions.
Haidt notes that the meta themes had different functions. Care and fairness seemed to related to our personal relationships with one another, whilst loyalty, respect and purity are group binding virtues.
Now, he is not the first person to have identified these themes. C.S. Lewis, in his book the Abolition of Man, recognised  this concordance amongst religions which he described as the  "Tao of Life". (He had a few more categories than Haidt). What Haidt notes is that loyalty, respect, and purity are virtues necessary for group binding. But what Haidt then proceeded to do is analyse how liberals differ from conservatives when it comes to moral determinations. The result is fascinating :

Haidt discovered when it comes to moral discernment, Liberals tend to weight Fairness and care above all else, whilst conservatives tend to weigh all elements equally. He described liberal moral reasoning as being a "two channel" [Ed: parameter] determination whilst conservatives were five. Conservatives take more factors into account at coming to moral determinations than liberals do: They are more multiparametric.

Now, Haidt doesn't ask why liberals are less "parametric" than conservatives, he simply registers the fact,  but it does demonstrate that there are different cognitive process which separate the two.  Now, it's my theory that multiparemetric analysis is computationally intensive and a high "broad" IQ is harder to achieve than than a "thin" high IQ. It would appear that there is now some evidence for this hypothesis in an experiment performed by Wright and Baril. In this experiment, conservatives defaulted to a liberal morality when they were intellectually distracted or cognitively exhausted. In other words, conservatism is a computationally intensive exercise. Liberalism is intellectually undemanding.
Previous research on moral intuitions has revealed that while both liberals and conservatives value the individualizing foundations, conservatives also value—while liberals discount—the binding foundations. Our control group displayed this same pattern of responses. This study examined two alternative hypotheses for this difference—the first that liberals cognitively override and, the second, that conservatives cognitively enhance, their binding foundation responses. We employed self-regulation depletion and cognitive load tasks, both of which have been shown to compromise people’s ability to effortfully monitor and regulate automatic responses. In particular, we were interested in determining whether, when the ability to monitor/regulate their automatic moral responses was compromised (either by exhausting their cognitive resources or by distracting them), liberals would give more moral weight to the binding foundations or conservatives would give less. What we found was support for the latter: When cognitive resources were compromised, only the individualizing foundations (harm/fairness) were strongly responded to by participants, with the binding foundations (authority/in-group/purity) being de-prioritized by both liberals and conservatives. In short, contrary to Joseph et al.’s contention that the “…automatic moral reactions of liberals could be similar to those of conservatives”, we found that the automatic moral reactions of conservatives turned out to be more like those of liberals.
Haidt gives a very good talk on the subject here.  It's well worth the look.