English report - Noise. Why We Make Bad Decisions and How to Avoid Them | Lecture by psychologist Daniel Kahneman
We constantly make decisions and judgements in our everyday lives. How much we tip the waiter, whether we should accept a job offer, or if we like the person we just met for the first time. What we are not aware of is that ‘wherever there is judgement, there is noise.’ The temperature, the time of the day, whether one’s favourite football team has won or lost the match, or being hungry: our decisions and judgements are constantly influenced by those seemingly minor things called noise. During this event organized by Radboud Reflects at de Lindenberg in Nijmegen and livestreamed online, speaker Daniel Kahneman – professor of psychology and international affairs, and Nobel Prize winner – provided insight into the concept of noise to correct the imbalance in treatment of bias and noise, the former of which is commonly seen as a central problem, whereas the latter is mostly ignored. In a first part, Kahneman – who was present via Zoom – provided an introductory lecture on the concept of noise, the various forms it can take, and the difference between noise and bias. The second part consisted of a discussion between Kahneman, Noelle Aarts, and Daniël Wigboldus, focusing on the dangers of noise and how to avoid them. Noelle Aarts is a professor of socio-ecological interaction at Radboud University and interested in inter-human processes and communication. Daniël Wigboldus is chair of the Executive Board of Radboud University and a professor of social and cultural psychology. The discussion was moderated by Liesbeth Jansen – programme manager of Radboud Reflects.
What is noise?
In his lecture, Kahneman defined noise as ‘unwanted variability of errors’ and ‘a lack of consistency in judgement.’ He contrasted it with bias for which he provided the definition of ‘average error.’ Although both bias and noise can occur together, bias is characterised by its particular direction, whereas noise is scattered. Surprisingly, although it is commonly believed that bias is a central problem and noise is mostly ignored, both have equal weights in diminishing accuracy.
To further illustrate the three different forms of noise, he introduced the example of judges passing sentences. The first form is Level Noise: There are judges that are severe and others that are more lenient. Secondly, there is Occasion Noise: if the same judge is presented with the same case at two different occasions, they will come up with two different judgements which are influenced by random factors such as the weather or that their favourite football team has lost. Lastly, there is Pattern Noise, which refers to stable personal feelings and different tastes of people. For example, one judge might be more severe when the victim was old, whereas another is stricter when the crime was particularly violent. Kahneman argued that ‘Pattern Noise tends to be the biggest source of noise.’ Importantly, he made the point that system noise as we encounter it in for example the justice system or the medical system is produced by all of these three forms.
The Dangers of Noise
What are the dangers of noise? ‘Disagreement can be very interesting and useful,’ Kahneman explained, but ‘I think that many people would agree that you would want the judgement system to speak in one voice, that having a lottery that determines how many years one person will spend in prison is not a good thing.’ Indeed, Kahneman focused his analysis on the operation of noise in different systems, where it can lead to lottery-like judgements, such as the justice and medical system, but also insurance companies. As Kahneman explained, the problem is that – besides being costly – ‘noise creates unfairness and inefficiency.’ For example, noise leads to different sentences being issued for the same crime, and different treatments being prescribed for the same medical problem. In other words, fair and efficient systems require consistency in judgement, and it is precisely this that noise endangers.
How to Deal with Noise and to Improve Judgments
When asked for solutions for noise, Kahneman argued that disagreement can lead to innovation and progress, but the problem is that ‘there is no mechanism in the justice system […] to turn those disagreements and differences into a learning experience.’ Although not specifically emphasised in his book, in the discussion with Aarts and Wigboldus he admitted that what is important after measuring the amount of noise is to let the people involved discuss where the noise comes from in order to improve judgements and to reduce noise. In other words, one way to deal with noise is to turn it into a learning experience in order to improve the judgements in a system.
Furthermore, Kahneman introduced the notion of decision hygiene as a form of prevention of noise, which can be implemented by organisations. Decision hygiene consists of a ‘set of procedures for making judgements, rules for the way they conduct meetings, and the way they resolve disagreements.’
As a third option of how to deal with noise, Kahneman introduced the idea of breaking down problems. Here, one should ‘slow down’ and analyse merely the six to seven most important aspects of a problem in order to form a judgement without getting lost in detail.
Finally, when asked about whether in the future machines will take over the jobs of judges, Kahneman admitted that this would indeed be an improvement of judgements. ‘If noise is recognised as a problem, then there is pressure to formulate principles more clearly, and with that AI becomes a candidate for applying principles because it will apply them in a noise-free way.’
Luisa Koch, Research Master student Philosophy at Radboud University