22 July 2007

The bias of cognitive bias theory



This article strikes me as a textbook illustration of the fact that much of post-war academic research in the humanities has a tendentious character. That is to say, it appears to be motivated by ideological considerations. And not in the sense which most humanities academics mean when they talk about "ideology". (They confine their use of this term to criticisms of capitalism.)

What I mean is that such research, although not incorrect, and sometimes even illuminating, appears to be motivated by a desire to make anti-individualist points. And is duly used for this purpose. Sometimes, as in this case, the use may be inappropriate, but it doesn't really matter. The object — to create a different worldview in which the old concept of 'the individual' is undermined — has been achieved.

The article by Professor Norman Siebrasse, on the Oxford-University-associated blog Overcoming Bias, invokes two classic examples of such anti-individual research: (i) the concept of "cognitive bias" and (ii) the Prisoner's Dilemma model.

The idea of cognitive bias is fashionable because it allows society to question the decisions and preferences of the individual. You may observe that it is particular popular with commentators of a leftist persuasion. By contrast, the idea that the state is biased by the motivations of its agents has hardly been explored. To the extent it has, by people like James Buchanan and Eric Nordlinger, the research in question is distinctly unfashionable. Enthusiasm for cognitive bias theory in its usual form could therefore itself be regarded as a form of cognitive bias.

The Prisoner's Dilemma, as I explained here, has received the massive exposure it has — not, as Adam Curtis of The Trap ludicrously suggested, because it promotes the idea that people behave mechanically (although he is right that this mechanistic ideology has been widely peddled by academia) — but because it is one of the few basic models of microeconomics which suggests, at least at first sight, that markets could fail.

Siebrasse invokes these two concepts in critically considering the right to privacy. He complains that "in much of the debate as reported in the media no argument at all is made in favour of privacy — it is just accepted as presumptively good", and that he has "never come across a sound policy argument that justifies a general presumption in favour of privacy".

But to put the argument in this form (i.e. "please give a social justification") already predetermines the type of answer you are going to get. The point is that everyone wants the right to privacy, for themselves, other things being equal. It is a freedom to choose. If you have the right, you can if you wish expose yourself exhibitionistically on the internet (for example). Without the right, it is other people who will determine if you have privacy in practice.

Like any other liberty, privacy should be regarded as a basic right which does not need justification. It may then need to compete against other claims, e.g. the need to detect crime. Demanding that there be utilitarian justifications for the basic right is pointless. If you must pursue that line, take it back to the question of whether there are good enough arguments for liberty per se. Although I happen to think doing so is to be equally avoided, since that too is a type of question the mere posing of which — "can we justify this in terms of the good to society?" — to a large extent shapes the answer.

Having decided that privacy doesn't make a lot of sense, Siebrasse reckons it must be a cognitive bias. This is rather a popular line among contemporary intellectuals. We don't quite approve of what people seem to want? It doesn't fit with normative ideals as formulated by humanities professors? Must be a cognitive bias.

Of the two possible cognitive biases which Siebrasse considers as explanations of the preference for privacy, he leans towards the hypothesis that we have been hardwired for a type of Prisoner's Dilemma.

Suppose that free flow of information is in fact that the best social policy. This would set up a classic prisoners’ dilemma: the best case overall is if no one keeps information private, but the best case for me is that I keep my information private and everyone else reveals theirs. Since everyone has the same reasoning, everyone elects to keep their information private, even though free flow of information would be substantively desirable.
Now labelling the preference as a Prisoner's Dilemma sounds good, because it implies a type of tragedy of the commons in which everyone is worse off, but could be made better off by means of intervention by an outside agency (i.e. the state). But I find the use of PD here somewhat implausible. In PD-type situations, individuals are typically aware of what they are missing out on. Those overgrazing a piece of land would surely say, "yes, it's a shame we can't cooperate about this". Or they would create a framework for cooperating; that is after all the purpose of many aspects of civil legislation. But I see no reason to think that most members of the public would regard current information-sharing between them as seriously suboptimal. Not pre-9/11, and not even now. Suggestions that it would be in our interest to reduce privacy seem to come largely from government representatives or their mouthpieces.

There may of course be a demand from one section of the population that more information about members of another section (e.g. the rich and famous) be made available. That, however, is not a Prisoner's Dilemma.