Kahneman’s Fast and Slow Talk
Daniel Kahneman Fast and Slow Thinking, Google Talks (book promotion), 2011
A 45 minute talk with 15 minutes of questions. This covers much the same ground as his Nobel speech and papers, but is more accessible. I generally have difficulty with Kahneman that seems to be because he talks an alien language, but I felt that I had less difficulty here. I comment on the talk out of order.
Kahneman cites medicine where there is a clash between trusting experts and evidence-based approaches. This clearly an important area where Kahneman’s work could help.
Aside: In many areas of the UK, particularly those that are governmental or regulated by the government ‘evidence based’ often tends to mean ‘objective’ evidence, to the exclusion of expert judgement that may be hunch-like. The choice is thus presented as being between expert judgment and ‘evidence’, discounting expert judgment. But an experienced drive may ‘feel’ that it is a good thing to slow down without necessarily being able to articulate why. It is probably a good thing if they slow down. On the other hand if a driver gets into smoother car than he is used to, it is probably best o watch the speedometer. A general evidence-based would take account of all sources of evidence and give hunches some weighting, appropriate to the experience n the circumstances.
Kahneman talks about his notion of intuiting and reasoning aspects of thought, which he calls ‘system 1’ and ‘system 2’. System 1 relies on ‘reinforced practice’.
“the mind is so set that if there are rules in the environment and we are exposed to them for a long time and we get [fairly] immediate feedback on what is right or wrong, we will acquire those rules”
System 1 thus recognises a situation and applies the appropriate ‘rules’ to generate a ‘story’ or explanation. Contrary to what I thought he said in his Nobel speech, it seems that this has an associated ‘degree of confidence’ which (if accepted) become probabilities. These degrees are dependent on the ease with which the story is constructed and its coherence. Thus confidence can be unreasonably high when there is little data to go on. We should take account of the environment and our experience in that environment.
There is a discussion at the end about how to remedy typical failings.
Kahneman’s biases have influenced behavioural economics, but the above quote seems much more pertinent. As Keynes observe, some processes, such as finance, have a reflexive feedback between subjective judgements and reality. For example, the more people expect house-prices to go up the more they are prepared to pay, and hence the more prices tend to go up. Such processes can appear to be stable for long periods so that, as in the quote above, people come to rely on the rules. But such processes have the tendency to go into reverse when some part hits a limit, thus causing surprise and possibly a crisis.
Here the bias is a tendency to unprincipled induction. The remedy might be to develop our understanding of these kinds of process and to the traditional role of academics and others in challenging the rule-based groupthink.
Other work by Kahneman.