08 November 2022

Kahneman: pseudoscience on a grander scale

• I thought it would be interesting to alternate our reading of Paul Collier’s The Future of Capitalism with a book by another highly decorated economist: Daniel Kahneman. Professor Kahneman is a well-known name among economics students. Research carried out by him and Amos Tversky in the 1970s highlighted some of the limitations of conventional economic analysis, by showing that choices made by the average person often fail to conform to what economic theory predicts. But in Kahneman's book Thinking, Fast and Slow, this awareness of theoretical limitations is inverted, and spun into a grand narrative about human rationality.
   The work has become hugely popular with intellectuals. "Daniel Kahneman has done us a great service" is a typical comment by a reviewer working in the humanities. Why has the book struck such a chord with intellectuals? Various explanations are possible, including that proposed by another reviewer, claiming that Kahneman is on a par with greats like Freud in advancing understanding of human psychology.
   I suspect one of the principal reasons the book has proved popular is its central thesis, according to which research shows that humans are irrational. Why does this thesis appeal to intellectuals? Because it provides ammunition for the interventionist-paternalist programme, which tacitly assumes that intellectuals should rule society (in the sense of controlling, among other things, education, medicine and cultural output — supposedly in everyone's best interests) rather than leaving things to the decisions of individuals and the markets.
   It's ironic that, having carried out research which usefully demonstrated that some of the assumptions of economic theory about how humans behave were plain wrong, Kahneman’s book assumes another theoretical model of rationality, and in effect says that where theory and practice differ with regard to behaviour, it is practice which is wrong!

• The issue hinges on the concept of rationality, and whether it is possible to define it objectively. The short answer is: no, it's not. There is no behaviour, or belief, about which it is possible to assert irrefutably "this is irrational".
   Take for example a textbook illustration from economics: you are bargaining with a buyer, who could be an employer, for the sale of an object or your own labour, and the buyer offers a choice between you getting £1000 and £1100, all other things being equal. Some would argue that you are definitely irrational if you strongly prefer the £1000 option — after all, you could (they would say) dispose of the extra £100 easily enough. But you may well have reasons for making that choice which cannot simply be dismissed. You may not even be aware of what the reasons are, but it would be impossible to disprove the proposition that ultimately, in some sense, this choice is in your interests. (All sorts of effects could be present here to complicate the picture but being left out of the equation; some of them known about, such as reputational effects; others not known about.)
   Or take a belief in something supernatural, for which (a sceptic would say) there is no good evidence. How about belief in the existence of God? Richard Dawkins has argued this belief is irrational, but that would make a lot of clever people from history irrational. In any case, the concept of God is too ill-defined to say what would constitute evidence. How conclusive would the evidence have to be? The evidence for global warming, or the carcinogenicity of tobacco, is strong, but not completely conclusive. At what level of evidence does a belief stop being irrational, and start to be rational?
   The point is: the question of what is rational is ultimately subjective. Kahneman, and the psychologists he cites, may have done experiments which comply rigorously with scientific standards and which generate interesting results, but such experiments are — and arguably always will be — incapable of yielding the sorts of conclusion that Kahneman draws.
   Conclusions such as the following; Kahneman is here referring to an experiment in which subjects are asked to express a preference between two types of experience involving mild pain (my italics):
An objective observer making the choice [on behalf of an individual] would undoubtedly choose [differently from the individual].

... The choices that people made on their own behalf are fairly described as mistakes.
Again, there's an irony in the fact that Kahneman at other points in the book criticises evaluations made on a gut basis, in ignorance of reality being more complex, yet is here guilty of the same thing. He appears to think we can obviously dismiss some judgments as being irrational or inadequately thought out, and that some preferences are just wrong. "This person says she prefers strawberry jam because it leaves a nice aftertaste, but she ought to prefer blueberry jam because it is more satisfying while it is being consumed" is a statement Kahneman does not make — but it's analogous to some of the things he does say.

• Some beliefs or preferences may strike a high percentage of ordinary people as bizarre or unjustifiable. Others may strike an even higher percentage of intellectuals as ridiculous. No doubt some intellectuals would like to have conclusive scientific support for rejecting certain beliefs or preferences. Attempts to use science to justify the decisive rejection of one preference over another, however, inevitably involve the abuse of science.
   I don't wish to de-legitimise the concept of 'irrational' as used in an everyday context, but we have to recognise that judgments about rationality are judgments, not scientific findings, and are ultimately not capable of being given irrefutable justification.
   Whether something is true or not may seem simple in some cases (is London the capital of the UK or not?) but most questions do not have easy yes-or-no answers, meaning there is little conclusive basis for assigning irrationality to one answer rather than another.
   With regard to preferences, there is certainly no adequate justification for intruding on individual choices to argue: your preference for A is wrong, our data shows you should be preferring B [*said in severe tone, by figure in lab coat carrying clipboard*]. Believing such intrusions are justified by science is not only wrong, it is dangerous.

• In the next instalment we'll take a look at the experiments on which Kahneman bases his conclusions.

Daniel Kahneman, Thinking, Fast and Slow, Farrar Straus & Giroux, 2011. Quotes are from p.409.