Have I been wrong for years? What I mean is that I have often attributed ideas I believe to be mistaken to cognitive biases. But as Dan Kahan points out, this might not be the case. Thinking, he says. is not always a means of reaching accurate perceptions of the world. It can also be a way of protecting our status and (self) image - of expressing who we are.

Take for example a classic experiment (pdf) by Lord Ross and Lepper in the late 70s. They showed a group of people with strong views for and against capital punishment some empirical studies of the issue. They found that this led to greater polarization. This is inconsistent with conventional Bayesian reasoning. But it is entirely consistent with the theory that people form beliefs to maintain a coherent and consistent self-image. As Sherman and Cohen say (pdf):

People defensively distort, deny, and misrepresent reality in a manner that protects self-integrity.

Professor Kahan says this is not necessarily irrational. For most people, there is no cost to having erroneous beliefs about political issues, simply because their individual vote doesn't affect the outcome of elections. But there is a cost to being "good" Bayesians: you'll feel uncomfortable if you come to doubt your long-held beliefs, and you might even lose friends if you don't fit in. It can, therefore, be rational to hold views which are mistaken from a Bayesian perspective*.

One fact is consistent with this. Professor Kahan points out that people with greater cognitive skills tend to be more partisan than others. They have the skill to be what Glaeser and Sunstein call asymmetric Bayesians: they are capable of discrediting evidence against their beliefs whilst seeing the scientific merits of corroborative evidence.

Perhaps, then, I have often been guilty of a category error. Beliefs that I have attributed to irrational biases might instead be the product of a different form of rationality - expressive rationality.

One should ask of any theory in the social sciences: how widely does it apply? Many things are true of some people, but not of all. For example, I don't think expressive rationality is very relevant in my day job. When retail investors' make losing investments, it is often because of cold cognitive errors rather than expressive rationality**. There are, though, domains where it might be applicable:

- Stephan Lewandowsky says climate change denialism is motivated by "identity-protective cognition" - a desire of free-marketeers to reject interventions in the market.

- "Very Serious People" believe things which uphold their self-image as sober and serious people, regardless of their empirical validity. This, I suspect, is why professional investors prefer futurological judgments to the much more empirically grounded "sell on May Day, buy on Halloween rule".

- "No platforming" and "safe spaces" are absurd from a Bayesian point of view; they entail being wilfully blind to dissonant ideas. But they make sense from the point of view of a desire to maintain a coherent self-image. When rationalists such as Nick Cohen and Richard Dawkins criticise such policies, they might be committing the same sort of category error of which I might have often been guilty.

- Professional trolls such as Toby Young and Katie Hopkins hold beliefs which are wrong by our standards. But they are rational from their point of view, because they have based a career upon them. Newspaper and TV editors don't want columnists and guests who say "we don't know: the evidence is missing and ambiguous and the world's a complex place", even though this is often true.

You'll notice that these examples come from across the political spectrum. Insofar as expressive rationality has eclipsed Bayesian rationality, it is not a partisan phenomenon.

* Of course, if everyone is expressively rational there's a massive cost in terms of bad policy and terrible political discourse. This is another example of how individual rationality can conflict with collective well-being.

** Herding is an ambiguous case. People can jump onto bandwagons and buy at the peak of bubbles because of cognitive error - a mistaken belief in the heuristic of social proof - or because of a desire to be part of a group, to do what others are doing.