If anyone would have just cause to complain about their childhood, it would be Daniel Kahneman. He was born in Tel Aviv, the son of Lithuanian Jews, but he spent his childhood in France, which was occupied by German fascists when he was barely six years old. The first blow was the internment of his father. Although after some time his father managed to escape from the internment camp, the persecution of the Jews meant that the Kahneman family were forced to flee to Juan-les-Pins on the Côte d’Azur. They didn’t settle there for long. When the Allies took North Africa, the Germans withdrew their forces to the south of France, forcing the Kahnemans to flee once again, moving from village to village. Eventually they ended up living in a chicken shed adapted for human habitation. Kahneman says of those times that he felt like a hunted animal—“We had the mentality of rabbits” (Shariatmadari 2015). Young Daniel concluded that God, to whom he prayed, must have been exceptionally busy in those times and unable to deal with major requests, so he thought that the most effective form of prayer would be to ask God for just one more day. Therefore, every day he prayed for just one more day of life ….
However, Kahneman does not complain about those times, nor does he seek self-justification in his experiences. Quite the opposite—he denies that the experience was traumatic. Perhaps it is precisely because of his attitude of not looking back that he was able to embark on an intellectual journey, guided by his unbounded curiosity. This journey eventually led him in 2002 to the Swedish Academy of Science, who awarded him the Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel. Today, he is one of the world’s most influential psychologists.
Shortly after the liberation of France, Kahneman moved to Israel where he decided to study psychology. He got his bachelor of science degree with a major in psychology, and a minor in mathematics from the Hebrew University in Jerusalem. In 1958, after completing military service in the Israel Defence Forces, he went to the USA to study for his Ph.D. in psychology from the University of California, Berkeley. In the spring of 1961, he wrote his dissertation on a statistical and experimental analysis of the relations between adjectives in the semantic differential. After completing his Ph.D. studies Kahneman went back to Israel and began his academic career as a lecturer in psychology at the Hebrew University of Jerusalem in 1961. His early work focused on visual perception and attention.
However, Kahneman began the greatest adventure of his life at the end of the 1960s, when he began co-operating with Amos Tversky, a younger colleague from Jerusalem. He speaks of this meeting as a “magical experience.” Together, Kahneman and Tversky formed that rare kind of team, where opposites complement each other perfectly, and they published a series of seminal articles in the general field of judgment and decision making, culminating in the publication of their prospect theory in 1979 (Kahneman and Tversky 1979). It is chiefly for these works that he was awarded the Nobel Memorial Prize in Economic Sciences. Kahneman and Tversky would continue to publish together until the end of Tversky’s life, but the period when they published almost exclusively together ended in 1983.
The culmination of his work on decision-making processes is his bestseller Thinking Fast and Slow, published in 2011. In it, Kahneman explains how the brain works, by describing two processes which together decide our way of thinking. The first is fast, intuitive and emotional, whereas the second is slower and operates in a more considered and logical way. Although the author shows these two brain mechanisms to be lie uneasy together (because they are counterintuitive), the book has enjoyed enormous popularity among millions of readers and has received many awards.
One of Kahneman’s rarely mentioned (not to say forgotten) achievements is in developing a procedure of so-called adversarial collaboration, which is an answer to the absurd (in his opinion) method of conducting scientific debate, where an author replies to his critics in kind, which, instead of leading to consensus, simply further polarizes the position of the adversaries. Adversarial collaboration depends on attempting to resolve differences of opinion through jointly conducted research. In this method the adversaries together develop a research procedure which they think will resolve the dispute. Kahneman has often expressed the hope that this method will become part of his legacy (Kahneman 2007).
During the 1990s Kahneman’s interests gradually shifted from decision making and economic psychology in the direction of hedonic psychology, a field which deals with research into the factors which govern whether our life experiences are pleasant or unpleasant. We question ourselves about the nature of pleasure and pain, boredom and interest, joy and sadness, satisfaction and dissatisfaction. The research results of hedonic psychology also show the whole range of circumstances, from the biological to the societal, that cause suffering and enjoyment.
The result of these interests was the development, along with David Schkade, the concept of “focusing illusion” (Schkade and Kahneman 1998) which explains the mistakes people make when estimating the effects of different scenarios on their future happiness. The illusion occurs when people consider the impact of one specific factor on their overall happiness, they tend to greatly exaggerate the importance of that factor, while overlooking the numerous other factors that would in most cases have a greater impact. Life satisfaction, in Kahneman’s opinion, is to a significant degree the result of how far we fulfill our expectations and achieve our life goals, and not, as we might mistakenly believe, due to the circumstances in which we find ourselves. The actual result of the research on happiness was also a method for collecting data on emotional well-being and time-use, called the day-reconstruction method (Kahneman et al. 2004).
At present he is studying “noise” in decision making and how to address it. Noise is an unsystematic error and it is the complement of bias, which is systematic error. The concept of noise is a useful way of thinking about decision making. One of the main reasons why simple algorithms are typically superior to human decision makers is not because of bias or systematic errors by the humans, but rather the inconsistency of human judgment (Kahneman et al. 2021).
Ever since the start of his career, Kahneman has scrupulously taken care of the methodological propriety of his research by repeating it many times. Disturbed by the quality of research conducted by social psychologists, as well as low replicability and in particular priming research, he addressed an open letter to them in 2012, which was extensively commented upon in the media. The most widely quoted of Kahneman’s lines was “I see a train wreck looming …” (Kahneman 2012). Without too much exaggeration it could be said that these very words were the first shot in the methodological revolution intended to restore credibility to the results of psychological research of which we a currently witness.
Currently, he is a senior scholar and faculty member emeritus at Princeton University’s Department of Psychology and Woodrow Wilson School of Public and International Affairs.
Professor Kahnemann, your research work is a stream of successes for which many of us admire you. We can read about them in many publications, which is why I’d like to start our conversation from a somewhat different angle. What has been the greatest disappointment with psychology that you have experienced in your life?
I think to some extent the replication crisis…
If so, please allow me to discuss the replication crisis a little, a subject that has been a focus of attention for several years, partially because of your famous letter in which you stated, “I see a train wreck looming.” However, the trouble with replication was noticed by some a long time ago. You yourself wrote about it in your autobiography: as early as the 1960s, in order not to pollute the literature, you wanted to report only findings that you had replicated in detail at least once. Nobody was talking then about a replication crisis. How did you arrive at the conviction that the replications which nobody else really treated seriously at that time were, in fact, so important?
I think it must have been because I tried once to replicate the same result and it didn’t work, so I didn’t trust my results of my work unless they were replicated. I don’t remember the details, but this is what must have happened. I was working in parallel on vision and problems of acuity, where all my results were absolutely replicable, and I was trying to do things in the development of psychology, and not replicating. I was replicating my work in vision, but not replicating work that I was doing on individual differences between children, I tried to do work like Walter Mischel was doing at that time, because I was very impressed with what he was doing, and I tried to do work on that, but I couldn’t get the same results twice. I was probably expecting too much. That work actually led to the very first publication that I co-authored with Amos Tversky, which was basically about replication.
How many scientists do you know that have embraced standards as high as yours in regard to the replication of their research?
I didn’t know anyone. It was just a decision that I made for myself, because I didn’t like the idea of achieving a result and trying to write it up, and then finding out it wasn’t true. But I did entirely for myself—I wasn’t part of a movement or anything.
Why in your opinion did others ignore the necessity of replication for so long?
I think that by and large people did not know that they have a problem, and this was to a large extent because of a self-delusion that every investigator thought that his procedures in doing science were adequate, and didn’t realize the extent of self-deception that was involved in collecting results. I think that the most basic finding was about the p-hacking, the extent to which investigators quite honestly take steps that cause the results to be unreliable.
In recent years there has been much said about replication projects, and indeed quite a bit is happening in that area. Some are even talking about an age of great replication in psychology. According to the PsycARTICLES database, in the last three years less than 1% of all published articles contained the word “replication” in the title. Is this a lot, or not much? How do you assess the efforts currently being made to overcome the replication crisis?
I think there really has been considerable progress. Everybody is aware of it. People are running larger samples, are more careful, are registering studies, and pre-registration is becoming standard, and even people who don’t preregister think in those ways. So I think we have had a huge change, in psychology in recent years, all for the better. Many areas of psychology have changed, and it clearly is a better science than it was a decade ago because of the replication crisis.
About one year before our conversation Many Labs 2 results have been published. Half of the 28 influential psychological researches were replicated successfully. Among the latter was also the study by you and Tversky. Is 50% much or few? Is the glass half full or half empty?
Well, 50% is not enough, but some of the difficulty, I think, lies not only in the original papers, but also with the replication that they were far from perfect. Some of the failures are due to the replicating teams.
Some scientists I know take a relatively blasé attitude to the replication crisis, claiming that in other sciences, such as medicine, the percentage of successful study replications is as low or even lower. Does the fact that it can be equally bad in other sciences relieve us of the burden to care about higher rates of replicability?
I think that’s ridiculous. Medicine has a problem, and they should solve that problem. We should solve ours.
What in your view should we do first to overcome the replication crisis?
I think that everything that is being done now is good. I think people should use large samples, and pre-registration of studies is a good idea. But even if you don’t preregister, you are thinking as if we were. We are thinking of how we will analyze the data, and we want to plan it all ahead, so that we don’t have the option of deluding ourselves about the results. This is happening now. Psychology is improving very rapidly, and I don’t think the problem will go away, but it certainly is diminishing. It’s been a big success.
So it’s a kind of internal control system now somehow?
Yes. The culture has shifted, and even the social psychologists, who still don’t believe in the replication movement, and are strongly opposed to it, have changed their procedures to get more replicable experiments. Basically, this movement has been successful.
Low reproducibility of psychologists analyses and especially social psychological research is not the only problem contemporary science is facing. For over half a century, first Jacob Cohen (1962), and later you and Tversky (1971) along with others, have pointed out that studies with low power should not be treated seriously. Has this issue been solved in contemporary studies?
It is being solved. People are now running larger samples than ever before, and are very conscious of sample quality, so in mainstream science this problem is being solved.
In your autobiography you mentioned reading Clinical Versus Statistical Prediction, in which the author, Paul Meehl (1954), demonstrated the greater accuracy of actuarial forecasts based on statistical data as compared to those made by experienced clinicians making use of their experiences and intuition. When you attempted to implement those recommendations into the practice of selecting officers in the army, you ran into a wall of resistance. Even greater resistance was shown to Meehl himself and his assistants. What do you think now about that issue? Have clinicians and other specialists come around to adopting the tools offered by Meehl, significantly improved thanks to modern technology and the potentials for processing big data, or do they remain stuck in the belief that machines, never mind tables, are not capable of replacing them?
There is absolutely no question that Meehl was right, and Dawes went even further in showing that any sensible combination of variables could do as well or better than humans. So this result is now completely accepted, and I don’t think there is a serious debate about the Meehl results anymore.
And now all these methods Meehl was talking about can be improved by artificial intelligence algorithms and computers.
When we started with artificial intelligence, multiple regression was better than people, then it was found that even simple models, unit weight models were better than people, and now artificial intelligence is clearly better than regression. The artificial intelligence picks up non-linear relationships in data, so that’s the new standard. And by the way, the reason that people are so inferior is something that I am studying now, and this is something that we call noise and unreliability. It is because of noise more than anything else that people are inferior to simple rules. Meehl’s findings were primarily a result of noise, not of bias.
Once we started talking about AI and before we move on to your current research interests please tell me what is your opinion about artificial intelligence? How important it is in our field? Because scientists are rather divided. Some of them say it will play a crucial role, and some of them say it’s won’t, maybe it will be useful only in improving our research methods.
I belong in the first camp. I think it’s an enormously important development, and I think it’s going to have huge consequences, although as I said, I am not attempting to forecast the details, but I am certainly among those who believe in it. Our model of artificial intelligence is only 10 years old, so I do not take very seriously people who say what it cannot do, because here are going to be major breakthroughs. Currently, artificial intelligence is not intelligent, so there are going to have to be big developments in that field. There is a huge amount of talent in the field at the moment, a huge amount of interest, and there is a lot of money, and so I think great things are going to happen, but I also think that some very dangerous things are going to happen. Socially, AI could have very negative consequences, but that belongs to the world of forecasting.
What kind of negative consequences do you mean?
I mean that entire professions are going to be wiped out, and the relationship between people and tools is going to change. And basically it’s going to turn out that artificial drivers are better than artificial human drivers, artificial doctors are safer than real doctors, and so on. I think this is inevitable. At some stage we are going to get there, and at some stage developments in robotics are also going to make a lot of difference. It will be a very long time before you get a haircut from a robot, but I think really nobody is entirely safe from AI, and my guess is that very significant opposition is going to arise.
The future will probably show. Let’s return to psychology in the meantime. In your opinion, is there any chance in the near future of the unifying theory in psychology that would reduce the proliferative character of our field a little bit? Do we need such a theory or perhaps we aren’t ready for one yet?
Well, the facts speak for themselves. At the moment the field is very fragmented and there is no theory. The understanding of the brain and the mind will eventually help to organize psychology, but we are very, very far from it. So at the moment it’s staying fragmented, and I’m making no predictions when it’s going to change.
What are your thoughts on the views of Robyn Dawes (1994), whom you have also met in person, and whose opinions about psychology were very radical?
Yes, his opinions about clinical psychology were very radical. He just said that basically they were statistically uninformed, and he was obviously correct. At that time.
How are his views applied to contemporary clinical psychology?
I cannot evaluate contemporary clinical psychology. I know that there are some parts of clinical psychology, like cognitive behavior therapy, which are evidence based, and good science is being done. How much of clinical psychology has this character, I do not know.
Alongside many of your well-known achievements that can be read about in numerous places, you are also the creator (or re-creator as you mentioned in your autobiography) of a method designed to replace traditional scientific polemics, which you name adversarial collaboration. In a database of scientific articles I was able to locate just a few empirical works carried out in accordance with the recommendations of that method, and a few dozen publications involving discussion of adversarial collaboration. As the creator of that method, do you consider that it has achieved sufficient recognition among researchers, or do they rather prefer to engage in traditional polemics with one another?
Obviously it hasn’t been a very successful idea. My own experiences have been interesting in this regard, and I’ve had several adversarial collaborations, some of them experimental, and all of those failed in the sense that we ended up with series of experiments, we understood and liked each other better, but we did not accomplish anything scientifically to clarify the field. That happened to me several times. The only two successful adversarial collaborations that I’ve had have been in writing papers together. I had one, that was very adversarial with people with whom I was quite angry, and who were quite angry with me (Mellers et al. 2001), I had another one with Gary Klein (Kahneman and Klein 2009), in which we became friends while carrying out that collaboration.
Do you see any chance for polemics among scholars to take on a more constructive character?
I think that this is actually likely to happen to some extent, and it’s part of a new standard. The fact that people have to be much more open about their techniques and methods is going to make things much easier.
What other threads to psychology do you perceive which we haven’t discussed yet?
I don’t think much along those lines. Maybe I don’t believe in my ability to forecast and I do not really know what is happening today. I think old people generally don’t know what is happening today. What is most important is what young people going to graduate school decide what to study, and it’s those decisions that make the future for the next 20 or 30 years, and I’m not following that closely.
What would you recommend for these young people trying to select a specialization at the outset of their career now?
I have no idea. I personally would go into AI, because that’s a thing that I find the most thrilling, but I wouldn’t give advice to anyone.
This book’s subtitle is Perspectives on Legacy Controversy and the Future of the Field, but we have spent almost all this time discussing the controversies. Lest we upset our readers even more, let’s talk about the accomplishments in our field. Which of your achievements do you feel have made the most important impact on science?
I think there is no question that prospect theory has had the most impact.
Have you made any discoveries that you yourself consider to be of importance, but which have gone unnoticed by others and not given consideration in other studies?
I think nobody ever has quite enough, and so I can think of pieces of mine that could have become more famous than they did, but I really cannot complain about anything.
Sometimes your studies are criticized for being mainly based on hypothetical questions. Critics say that the process of making a decision on how to spend or invest one’s own hard-earned money is significantly different from the answers we select in a questionnaire that doesn’t cost us anything. The same is true in relation to other decisions. What is your response to such criticism?
The responses could be that the work that was done in the laboratory has been extended largely through behavior economics into the market, then into the field, and it works quite well. So I think those criticisms are passé, and they aren’t currently relevant. There is much successful applied work in the spirit of this research. It’s successfully applied work in finance, and in social influence, and so on. I don’t think that many people still talk about this area of research as disconnected from reality. This is nonsense.
In the 1990s you engaged in research on the feeling of happiness. Do you feel that these studies brought a breakthroughs of a similar caliber to those on thought processes and decision making?
My work on happiness actually was influential, but in ways that are widely known and had some influence on the design of the Gallup World Poll. The poll recognizes the distinction between life evaluation and the experience of living. The idea existed before my work, but I helped in making this acceptable.
And what is your judgment on positive psychology as a whole?
I have never been entirely sure of positive psychology. I have a temperamental opposition to a certain kind of humanism, so I haven’t been very sympathetic to positive psychology. But that’s entirely my character, I am a little bit cynical. There is a lot of talk about virtue and meaning, and so on, which I am usually skeptical of, but I recognize that there have been significant achievements. Some of my best friends are doing important work in that area, and in particular the work in the United Kingdom of which Richard Layard has been at the forefront is extremely useful. One of the problems that I have with the positive psychology movement is that I firmly believe that reducing misery is a more important objective for society than enhancing happiness, and this should have implications for research. That’s just my own philosophical bent.
Some people call positive psychology as a psychology of feeling good. Would you agree with this label?
No, I don’t think that I would use single labels. Positive psychology is many things. I don’t know them all, but I think that the psychology of good feelings is not enough. It’s not a friendly definition and it’s not what I would choose.
What research areas are you currently focusing on? What questions are you trying to find answers to these days?
I am writing a book with two collaborators on noise in judgment and decision making. And noise is unsystematic error and it is the complement of bias, which is systematic error. Most of my career I have studied bias, and now I am very impressed by the amount of noise there is. And there are two kinds of noise. There is variability within individuals on different occasions. That’s one kind of noise. And there is another kind of noise which is individual differences that exist, but shouldn’t exist, like differences in judgments between judges, differences in underwriting decisions, differences in patent decisions, differences in medical decisions and so on. We are studying the variability that shouldn’t exist in judgment and decision making. This is my topic today.
Many people who know your earlier books are eagerly awaiting Noise. I think that this will be another contribution to understanding how far off we are in our judgments about reality. Without too much exaggeration, we can say that you have dedicated nearly your entire life to showing us this. In a certain sense, you and Amos Tversky, along with your collaborators, created a new field of knowledge—about the meanderings of our minds. Did psychologists—both research and clinical—learn something from your lessons and apply that knowledge in practice?
Many people are aware of our work, and I hope its impact is positive.
What burning questions still demand answers in psychology?
I am going to disappoint you, because I really don’t think that way. You know, it’s not that there are questions that demand answers, it’s what are the questions that are in the mind of scholars these days. I don’t want to presume that I know what’s going on. Clearly, there has been a trend over the last 10 or 15 years that is turning psychology to brain science. This is the biggest thing that is happening in psychology over the last several decades. Brain research is considered superior, and is taking over. That’s a big development, but I have no insights about the future.
I think that a person who has participated in the most critical breakthroughs in the history of our field can afford to share his vision of a future with others. How do you imagine our field of science over the next two or three decades?
I do not believe in forecasting, because it just doesn’t work and it’s not the sort of exercise I engage in. I really think that the future is completely unpredictable and we have very short horizon. There are people with long term plans, and there are trends I would like to follow if I could live long enough. I am very interested in what’s going to happen with artificial intelligence, more interested than in psychological thinking. The history of forecasting is not promising.