Daniel Kahneman interview (Rutgers, Foundations of Probability)

okay we're going to go ahead and get started thanks for coming this is the those you haven't been here before welcome to the foundations of probability seminar this is a joint seminar of the statistics and philosophy departments my name is Harry crane I'm from the statistics department I'm one of the organizers of the seminar along with Barry lower of philosophy and Glenn Schaffer from the business school so we started the seminar last year in the fall and since then we've been pleased very fortunate to have speakers from all different areas and we're very interested very interesting talks and discussions about probability and any different disciplines and today in keeping with that theme we're glad to have Daniel Kahneman coming to us from Princeton so as I'm sure many of you already know Daniel is a world renowned psychologist well known for his work in behavioral economics and rationality among other things and I'm sure we'll have plenty of opportunity to learn more about Daniel's work and its life throughout the next two hours I'm told in today so today's seminar will be in the form of a of an interview conducted by Glenn Shafer Glen aside from being one of the regular participants of the seminar is border governance professor and former dean of the Rutgers Business School in Newark and I also I'd like to thank Len for organizing this event and for preparing quite diligently over the past couple of weeks preparing for the interview so I know that Glenn has a lot of questions that he'd like to ask and probably not enough time to ask them so I think that it's probably best that we go ahead and get started so Len why don't you take it Thank You Harry I would also secondary and welcoming danny kahneman to rutgers again he's been here it's not the first time I've had the opportunity to talk with him but I think this will be the most intense conversation that that we have had and I hope and I have gathered I invited our audience in advance to suggest some questions and so some of the questions I have come from them I since it is a philosophy department I'd like to start with a philosophical question philosophers debate about many things but one thing they debate about is what we know and about the meaning of knowledge for example at least in one time there's a lot of debate in philosophy about what about whether knowledge means justified true belief the social psychologists have shown us that people often understand a lot less than they a lot less about the past and know a lot less about the future than they claim and I'd like to ask Danny if what the social psychologists have learned on this can inform the philosophical debate well you know to a psychologist knowledge is a state of mind and it's it's a state of mind in which in some sense you believe that you have a justified belief that is true but whether or not it it is true is is ineffective relevant psychologically it's the same thing whether it's true or not it is important I think is that the view of knowledge that I think philosophers have that is widely shared among scientists is really psychologically and and I would save in socially very narrow there are other ways of knowing I mean we think of knowledge as you know beliefs that are true and and when we mean true we mean justified by evidence and and we're anchored in science but most people who feel they know things know them differently not by evidence not by they for example people who are religious they they know the truth and it's not that we know more certainly than they know they know the truth their basis for knowledge is is different it's you know revelation and faithful transmission for us it's actually it's a belief what is common to religious feeling of knowing and to most of them our feeling of knowing a secular feeling of knowing I think what is common is that it's really much more socially dependent then we tend to think in that what we call knowing and the knowledge that we have acquired and the beliefs that we have a very largely anchored in our beliefs in people I mean I know that this is how I believe in in global warming and you know I think I know there is global warming but the reason I think I know that that I believe in the National Academy of Sciences and and and their committees that have declared that there is natural that and we it really I think is very important to think of knowledge more broadly and you know as justified true belief for me psychologically you feel you know things when you can't imagine alternatives so when when you have a belief and alternatives seem unimaginable then you feel you know but it's it's relatively easy to get into that state without evidence well let me young let me ask you in the same interdisciplinary spirit let me ask you about economics I know in your recent book Thinking Fast and Slow the most cited author after a mr. risky is Richard Thaler and of course the Nobel Prize Committee agrees with you about his importance but I've heard other people other psychologists in fact that that say well what did the you know Danny Kahneman and Amos Tversky and all those other social psychologists have learned all this already what did the behavioral economist added so I guess I have that question for you what well you know I happen to think that Richard failure is a genius and quite and it's not a word I throw around lightly and wildly underestimated there's a genius because he is very funny and people laugh a lot and make other people laugh than to be underestimated you know in the genius category he a Richard Thaler wrote a paper in 1980 which I think everybody should read because it's readable by everybody it's called food a positive theory of the of consumer choice I believe and it's a set of ironic observations about how people make choices and about consumer choices including some ironic observations that are pointed and directed at at his professors like you know they in the attitude though the one of his observations is about a wine collector who was actually I think his mentor a wine collector who has a bottle in his cellar that he wouldn't sell for $1,000 but he would not pay a hundred dollars to get an equivalent that is economically absurd but it's a well-known fact now we'd have the name it's called the endowment effect invented that name in 1980 so in many ways Richard Thaler I mean you know the sense of precedence of risky Kahneman and Richard Thaler is to some extent reversed Richard had all these observations independently then you know telling a story that he tells better in a recent book called misbehaving which I also recommend because it's a good read he had these observations as a graduate student and as a very junior faculty with highly uncertain tenure prospects because he was an economist who didn't do maths and who was just observing human nature he had these observations independently and then a student of ours and he met a student of ours who showed him an early version of prospect theory which was the theory of choice that Amos and I were about to publish but hadn't published and the version he saw was an early version and he saw in prospect theory a way to explain some of the phenomena notably the endowment effect in that losing and winning do not have the same way and what he discovered in our work generally was what he viewed as a fundamental challenge to economic theory in economic theory the theory of the rational agent no economists no serious economists believe that people are truly rational all the time they make errors but the errors are sort of assumed to be random and insignificant and what Richard Thaler learned from our work in the existence of systematic and predictable errors and that he says was fundamental I mean his is a very emotional description of reading our paper on on air is actually our papers of judgment and that that had been published a few years earlier so he is really an independent thinker we learned a lot from him and and he has continued to make one breakthrough after another all of them look absolutely obvious after the fact so and that's a sense you know anybody could do this anybody could see this people don't see it people don't see the obvious and you know the way I think about it is that you know the child who's who noted that the Emperor is wearing no clothes it really deserves a lot of credit because the people who saw the Emperor they didn't really they were not really sure he was naked you know they saw clothes there and the child seemed the obvious which is what dick Thaler has been doing surprising people by telling them obvious things this is a major contribution so I think the Nobel Prizes overdue and certainly well deserved by the way not an accident that he got it around you know this is a significant fact in you know the sort of the noble story background the another question [Music] actually I feel like I should follow up on this question because what you raises the question of the impact of this work on economics do you think that sailors work and usually you've only mentioned famous work but there seem to be a lot of people doing behavioral economics now are they really having an impact on the way economists thing well the difference are we still looking at the emperor and seeing clothes on him well in the first place indeed there are many behavior economists and and the fact is that you know Nobel Prizes not because of the quality of the work but because of its impact so it's a measure of the impact of the work now in terms of economic theory there has been very little impact in terms of economic practice and in several fields of economics especially finance especially actually where there are a lot of data I can finance in savings in in responses to taxes and responses to incentives where there are lots of data this is becoming standard and and very large impact so many people there was a recent poll I think among major economists in their forties on members and and they you know the range is between agreeing that behavior economics is important and strongly agreeing that it's important there that is generally accepted the impact on economic theory is negligible thing and that's because I think it's because it's virtually impossible to create a tractable theory and very complicated assumptions and you know when you try to be realistic about the assumptions about the human agents you end up not being able to predict all that much and so there has been a there's a lot of theory in behavior economics a lot of theory being done but that theory all had the character of let's assume a rational agent except for but for something then all believes in they're all small numbers he has lost his loss averse or some properties and then and then you see you look for market implications of a local and particular specific deviations from rationality but in fact it appears to be impossible at least in the present to do serious large-scale economic theory without assuming mostly assuming rationality I should add that I'm not an economist and they don't know what I'm talking about but the but but that observation is an observation you can make from the outside you know without actually knowing the theory yeah we ask a slightly different question yes about decision analysis I think I heard you MIT say once said and I would agree that it's proven less useful than many of us expected 40 years ago and I wondered if you could say something about why that might be true I know in your and your book you wrote about the focusing illusion and I think we could argue maybe that decision it was a decision analysis just a bad idea no I I think I mean you know it's it's it's sort of a hunch I mean it's not what I have about what happened to decision theory I mean 40 years ago a mr. C and I believe that this would conquer the world you know it looked like no it looked so obviously obviously the right thing to do to measure beliefs to measure to measure values and then to combine them you know that seemed to be the way to go and and there was a major figures in the field howard raiffa Ron Howard I mean Ron Howard wrote a famous paper which was our introduction to in the 1970s about the decision to seed hurricanes and it appeared in science and it was a sort of breakthrough for many of us we were enormously impressed by it and he was willing to sue the government regardless of whether they decide to seed or not to seed unless they had done it was a proper decision analysis that's how certain people were at the time now when you look around you today decision analysis have saw the taught at business school but it's it's you know largely a curiosity as an applied field it exists mainly I think in Chevron where there are 300 decision analysts and it's it's not an accident that they it exists about oil drilling decisions the problem with decision analysis and we're going to have a similar problem I think with artificial intelligence is that it is very threatening to leaders of organizations it is in the first place leaders of many organizations do not want explicit values which is the major assumption of behavioral economics I sort of stumbled on that in the late 70s and I knew then there was probably a problem with the surge analysis when I was told you know the goal established in France la camisa we had it plumb sort of central planning authority that was there you know when the goal was in power and his successor was your risk of Easton and and Viscountess that was told somewhere that had been an analysis and there was a trade-off he was told that there is a paper that described the trade-off between inflation and unemployment and I was told visiting Francis the way to study the related question that risk Alistair said I do not want that number to exist and and that is not a trivial observation I mean the people do not want to be explicit on values and so that's one reason and now if you took it seriously and that's going to be the problem with artificial intelligence if you took it seriously the problem with decision analysis is that almost anybody can do it that is you know you have the probabilities you have it the utilities you can compute the decision but leaders of large organizations don't want their decisions computed for them they want to make them and so this is now why Chevron why is it worthwhile for Chevron to a 300 decision our analysis so I was told it's because all drilling decisions are large highly consequential but there are many of them and they can be delegated that is the the prestige of the CEO does not depend on oil drilling decisions so they can be delegated and decision analysis is useful for them let it go ahead but when it began when it comes to strategic decisions to the decision that that may drawn how it believed that decision analysis was important in the 1970s it's really not applicable and it will be resisted and it was soundly rejected ara will be more complicated you know more difficult to reject but it would present the same general kind of challenge to two leaders that you know decisions can be computed they don't want that anyway a leader is making decisions those of us who are have less Authority how are we going to resist it well no no I mean I think of AI in in the same way that you now have AI that is a better diagnostician than physicians you know this is happening there there is going to be you know I don't know when but there is going to be an artificial intelligence that that will be better at business decisions then then CEOs you know it's just a matter of accumulating very a lot of data enough data enough range of data and and analyses can be done that that are beyond human the capacity it's a question of time and how that will work out you know we can imagine physicians being displaced but how about CEOs when there are programs that anybody can run that will compute that that will compute decisions better than they can oh that blows the mind I have no idea how that will work out it's really an interesting question let me turn me turn to some decisions that we make organizations like universities I fifty years ago over 50 years ago Paul Neil wrote his began as work on disadvantages of holistic judgments and interviews as compared to more quantitative methods but and I know in Thinking Fast and Slow you've talked about trying to discipline interviews but my own experience is that in recruitment personality and our organization like Rutgers University that we still holistic judgment and refree ranging debate that we really don't discipline it in a way that Paul me Owen done there's some social psychologists are recommended is there any hope of making of getting us to do a better job in this area well you know progress in that is very very slow and that metruff that cycles back I think to the first question to the sense of certainty that you get you know when you hear a job interview and you know you like it or you don't like it or you're a very impressive you're not so impressed and and that feels like the truth about that human being about that candidate which is in a way nonsense and it's a small work sample that's all it is but the feeling of certainty that you get after a job talk and I you know I've had I've heard many of them and I've been very certain after every one of them although I knew better and so it's not you know it's very hard to master and to overcome hey you know eventually very slowly this might happen I mean I think I think all this issue disciplining intuition replacing intuition this is happening it's just happening very very slowly but it is happening and one wait is happening of course is by replacing humans all to get by algorithms so in many areas and this is happening structured interviews are gaining ground I mean it's it's unbelievable the amount of evidence there is that structured interviews which where you separately assess different traits the evidence for that I mean I did study when I was a lieutenant in the Israeli army 60 years ago I set up an interviewing system and structured interviews that was clearly superior to anything else that was that it competed with and there's still many interviews that are not structured I mean I was following meal directly so it it succeeds meals book is 1954 so but it's gaining ground structured interviews and and I'm thinking I've just had a paper rejected actually in which I was trying to argue that that structured interview ought to be a model for all decisions for that is that what we learn from structured interview is the virtue of breaking up problems and assessing their features independently before recombining and and you know the in it was rejected and that gives us hope it was rejected on the ground that it wasn't sufficiently novel so that's something is happening this is becoming accepted structured interviews are gaining ground well you see that the model organizations and you know where that and there is a huge variation actually across organizations in in how systematic they are in dealing with personnel and both with the hiring and with the evaluation which are very large problems in organizations there are some organizations and I think many people who agree that Google is is a model and and Google really applies those principles it's it's structured and it's independent and there and there is a sample of independent individuals making separate ratings and recombining them you know that's and this is being I think widely copied so I'm I'm fairly optimistic about that but but in many places it's not the case and oddly enough the more important the positions are the less well we do I mean you know that's no no I don't know that will you know we are too sure of ourselves yeah no I mean you know it's not a new idea it's just what we're discussing is in the social spread of this which is remarkably slow but I think it is happening well I can't I do have to ask if your what you learned as a social psychologist about how people could feel better with with paper rejections wives let me have some experience and Thinking Fast and Slow you wrote that you and Amos Tversky had retained utility theory is a logical rational choice but abandoned the idea that people are perfectly rational do you still see utility theory as a logic of rational choice and is it the only logical rational choice well it certainly is a logic of rational choice I mean there's no question about that that's that's been established whether it's the only logic of rational choice I mean that's that's a much deeper question of course and here I'm you know I'm uncomfortable because I'm in a philosophy department and what I'm going to say is is well but I what I am going to say that all the major ideas like competing utility theories and and many philosophical doctrines are based on our very path strongly-held intuitions that people have about something being true or something having to be true and we have those intuitions and you know then then it had that phrase that I really liked a lot about intuition pumps they're building on those feelings of certainty that you have except what do you do those the set of intuitions that we have on which we can potentially draw is not internally consistent and that I think is a psychological fact so it's a second so you can build a theory that will be intuitively compelling and built uncommon to ative ly compelling axioms and that it's not going to be unique because there are inconsistent basic intuitions that can be looking service sort of intuition pumps or the basis for developing theories so I believe yes it is a logic of choice and I certainly don't believe it's the only one it's what I should add and that is something that Amos was a lot more interested in than I was he was the formal theorist in in our collaboration but what Emma subjected to very strongly to a particular kind to a particular set of theories of choice theories of rational choice that with tweaks on utility theory that were designed to rationalize an existing paradox so there are major paradoxes of choice like the LA paradox the iceberg paradox um and there are a few others and theories have been built to rationalize them so that you train the axiom and and Amos I think was used calden that those the people who do that philosophers the decisions economists who do that kind of thing he called them lawyers for fools I mean that's the that they tried to justify something that we need you know you shouldn't try to justify you were less sure that no it's not I was I was left sure but everything that that had to do with Furioso so train from the normative to the descriptive you've written about blind spots and sort of are descriptive theories in fact some blind spots that you that prospect theory and utility of in common such as not allowing for disappointment or for regret on the other hand if I understand correctly you think that both of these theories your prospect theory and the utility theory are useful because they make some true predictions can you help us sort that out the status and values scientific theories that make predictions when summer Truman's and are false well you know I can try but I'm not going to succeed one observation I made in Thinking Fast and Slow that I talked about here for a minute was something that I call theory induced blindness that is that when when you have a theory it makes certain things easy to see and other things virtually invisible and and when when there is a shift when you know that all of a sudden it appears that people are truly blind that they couldn't see things and that was by the way the acceleration that Amos Tversky and I felt you know when we were building prospect theory well that we were discovering things that were completely obvious you know that there is you know that there is loss aversion so it's about it's about change the reason that's that cliche is true that it's the alternative theory that that is important it's not the counter examples but counter example there are many and it's remarkable that they draw very little attention and they're hard to see I have one in Thinking Fast and Slow that I'm very pleased with except that they can't recall it now but there was well it's the effect in general of things that didn't happen on your evaluation what didn't happen or it's the effect of things of choices you didn't make on your evaluation of the outcome of the choice you didn't make in prospect theory as in utility theory neither of those has anything but it's not difficult to construct examples with where they have major effects well you know what but if you choose one thing rather than another and just that the thing you didn't choose has no effect on you know the experience or the utility of what you do choose you know it's obviously wrong but it's an assumption in prospect theory just as in utility theory so yeah I have some good examples but there is there in the book I can't reconstruct them why would I ask about something else that's in the book which is a distinction between system one and system two and so one question is which of those systems is more like I think there is what the mean you know we shouldn't assume that people know what just the moaning system to our so I distinguished in Thinking Fast and Slow and it's not my distinction it's it's been suggested I adopted it and the terms from other people just develop them differently but system one is sort of what comes to mind immediately the way you recognize things how you perceive things the thought that occur so shi'ite of lee without much reasoning and system two is the slower more effortful kind of thinking and reasoning we tend to think that we are like system two because no that's what we're conscious of so but in fact I think you know what governs us is mostly system one and you know described system to a sort of a minor actor who thinks that he is the center of the piece you know that's now the associative machinery is I think more Bayesian than our reasoning ability that is its Bayesian in the sense that priors matter enormously so when when you resolve an ambiguity and there are many examples but you know the one that comes to my mind always is she approached the bank and now what happened when I mentioned Bank now it could be a financial institution or it could be the bank of a river and you chose you chose one of the two and and most of you will have thought of a financial institution because we think more of this kind of bank than at the other but if the story had been you know the same sentence had been in the context of fishing you wouldn't have thought of the financial institution at all you would have made a different choice so it's context-dependent but it's also basically a dependent on prior experience a nun relative frequency of prior experience except that the context dependency allows us to address very quickly adjust our probability very quickly because they are dependent they change with the context of the probability of one interpretation of bank changes radically depending on the previous sentence and that you know I see you know I think is some is more Bayesian than a lot of what we see in people's reasoning ok I just to follow up on the question we were just I'm just asking about system 1 and system 2 I wanted to reference the work in perceptual psychology over the last few decades about like Geisler and Kristen's paper on illusions perceptions and Bayes for example it just asked whether this is changed the way you thought with us what work has changed anything about where you thought others something no no not really I think the let me pass on the move on to a different question you you might work that we have greater regret when we deviate from default choices so why are we like that well it's a long story but the basic idea is that as we go through life and we form we encounter events events evoke their own alternatives so that we don't go through life for example expecting many there are many things that we don't expect and yet they don't surprise us I mean there's a whole range of them that feel normal that there are many things that could feel normal in any one situation and and what feels abnormal is when and and we had that sense of normality that is associated with an event so if an event deviates from normality and is surprising that it is compared to alternatives there was an alternative that was more likely or in that context so it's not an expectation that you had ahead of time it is a computation that is made after the event itself as it evokes it's alternatives and behaving in a routine way or anything that is routine after the fact that is the alternative that will come most readily to mind so if and if I behaved in an irregular way and something happens usually the salient thing is when a bad thing happens but I behaved in an irregular way there will be a very salient thought that I could have behaved in the regular way whereas if you behaved in the regular way the thought that you could have behaved in the irregular way and avoided that bad consequence is much less compelling so it's that asymmetry between the associations to surprising events at one surprising event that is the root of that and I think the cognitive psychology is reasonably well understood I apologize for a voice that is not as strong as it used to be so thank you so in thinking fast and slow you talk about evidence that then we have distinct mental accounts for our money money isn't fungible across what we spend for different parts of our lives do we have different accounts for managing our time I should say by the way that the idea of mental accounts was one of the ideas one of the many ideas that make failure had in his 1980 papers you might I think it's a different name for it they mr. ski and I proposed the name mental accounting I forget what his name was but but he had the concept and and the concept is indeed that when we get money for saving then it's me we don't spend it as readily as we spend windfalls law so and this idea you know many people have developed similar ideas in this theology of money viviana Zelizer at Princeton has written books and many articles on different meanings of of money and different accounts that we have now whether we have that for time certainly I mean you know we allocate time we have a budget of time in just the same way as we have a budget of money and with different accounts and so when I have free time I feel very differently about an interruption then you know if it's there is an interruption when it's not free time so the similarity is very close one of my younger colleagues asked me to pose a question about that he has on his mind about his research agenda I think which was the question is whether you know he's thinking he's asking himself whether he should be more deliberate about that and he wonders whether your own progress in research whether the topics you follow just naturally progressed and imposed themselves and whether there were times when you step back and said I have to change direction and deliberate deliberated on that and if you dis is great on that was it really your free choice I mean you want to ask about freewill as well as as well as about the research I don't think I ever had a grand plan you know it was it's one damn thing after another I mean you know you find something and and then you follow it through and and and a very important kind of decision that you know that's my main advice to young researchers is to know when to quit and to know when this idea is just not going anywhere in particular so instead of forcing myself you know I I hated reading the discussion section of dissertations especially when the experiment has failed and you have a discussion section that is trying to make the failure interesting don't bother you know just do something else so that's and and I certainly had you know I think I can't remember the blind alley is that the funny thing so in retrospect my career makes a lot of sense you know in retrospect I'm now very interested in ideas like making decisions by breaking things up that I was interested in when I was 22 years old so you know you see those continuities and and there are I suppose continuities but you're not conscious of them it's not a plan it's just you have a few ideas and they guide you a speaking of big failure was much on my mind these days that 1980 article which I freely recommend has all his career in it just just about every major point that he made in his career is anticipated in the 1980 paper which he wrote when he was just out of graduates those things up but so let me read a quote from Thinking Fast and Slow he wrote and I'll try to speak up I see some people straining the ears no and Thinking Fast and Slow you wrote as follows I have yet to meet a successful scientist who lacks the ability to exaggerate the importance of what he or she is doing and I believe that someone who lacks a delusional sense of significance will wilt in the face and repeated experiences of multiple small failures and rare successes the fate of most researchers so I wondered how this view of success how that informs your work with doctoral students and if people don't have this delusional sense can you do anything about it you know I really don't think that you can induce that that delusional sense of importance and some very smart people don't have it I mean I I used that phrase to talk my son-in-law I mean you know it's that it didn't depend in his academic hurry and become a successful high tech person because I told him that but but he he would describe his research and said you know this is you know this is really a tiny little thing a tiny little problem and you know I thought this is the wrong way to think about it you've got to you know you've got to make a mountain out of that molehill otherwise you'll never get anywhere so yeah I think it is very important especially for young people it's it's a less desirable trait for older people the ability to exaggerate their own significance but among young people not exaggerate their personal significance but think of the problem that they're dealing with is really worth dealing with and that that isn't I think an essential aspect of success so Thomas Schelling was well known for arguing that our moral intuitions depend on how alternatives are framed and I believe in in your and Thinking Fast and Slow you have some examples of how how we think about fairness of tax rates that if we talk about talk about taxing people we'll have one intuition if we talk about giving them exemptions or something well offset to intuition so what does this say about about the validity of our moral intentions can we really make more illuminations if this if we're so if we're so malleable and but the example you mentioned is my favorite example of framing and it is indeed Tom shillings example and and in his example I think this is the way it goes you ask a bunch of people consider you know the tax table and think of the difference between the tax rate on people our trial s and people have two children now should the rich and the poor have the same child exemption and and there is a very strong there is a very strong intuition I think among many people that well actually it's not be now I'm now I'm getting it backwards the other formulation is think of the normative family is a family with two children and now we're going to have a surcharge for a childless family should that surcharge be the same for direction for the poor and of course we say no the tribes cannot be the same the rich should have a higher surcharge then then the poor but of course the surcharge and the exemptions are just two ways of framing the difference between being tried less and having two children and we have contradictory intuitions and they are very powerful so this is an example where we have moral intuitions and and they're wrong I mean they're inconsistent now about the same you know objective reality of the difference between now the interesting question which I think shelling an interesting question that trailing doesn't raise its okay so what's the correct answer you know it shouldn't be huh and I what you would want to design is a tax table that is not framed in terms of exemptions or surcharges it is framed in terms of final assets it is framed in terms of somebody was that income and that many children they should have they should paid that amount of tax that's the way if you had such a representation in terms of final assets you wouldn't run into the framing problem you would have with Emma so she and I called economic representation and its unique and then it felt the problem is at that level we have no intuitions so the level at which intuitions are needed moral intuitions are needed they're not there Amos and I had a related example on life saves and lives lost and so you know in in that example you ask people well it's a complicated scenario but the disease that we call and and you can you can either save two hundred people for sure or save six hundred people the six hundred people are threatened with a probability one-third and what you choose now you can frame that as you can depending on your choice with probability 1/3 600 people will die or with probability one yeah or 2/3 600 people will die or two hundred people will die for sure and the intuitions are reversed you you're risk seeking in one situation and risk-averse in the other and again the remarkable thing is okay so you have to make a decision about a vaccine and it's going to have that consequences in terms of the number of people will be alive at the end of the exercise there are no intuitions I mean you do you really have a preference you know and when about how many million people will be million plus six hundred million plus two hundred nothing so the intuitions we do have tend to be the in this case they're clearly about changes and moral intuitions and and it's from the little I know well you know from studying fairness of course you have a theory you have a theory of distributive justice about how to distribute you know what kind of distribution of wealth we would we would like to have but then you also have intuitions about what changes in wealth we can impose by taxation of a redistribution the intuitions we have about changes do not lead to a favored distribution they're just two sets of intuitions and so it's dependent it's depending on framing and you know in a major way I mean it's not and and I think it's interesting that we have conflicting intuitions in some cases and I think it's particularly interesting and as you have zero intuitions about the economical problem when we have very powerful intuitions about different ways of framing the same canonical problem that that is partly this might not be slightly a related question you distinguish between in your book I may be elsewhere you distinguish between life satisfaction and experienced well-being I wondered if you could explain that difference and I also want to ask you which one do you value the most which one would you want for your grandchildren yeah well this was this was actually an exception I think to what I was describing earlier in terms of sort of blundering from one problem to another I was very interested when Amos and I were working on prospect theory I was really fascinated by the question of what should it be you know what what's real about those intuitions we have and I'll give you an example that sort of accused me of a lot of illusions about theory and it's it's a puzzle I posed to to a masseuse to remember the conversation about it so you have somebody who is going to is facing the prospect of having a series of injections in the but I remember I mean that's that's how we talk and and they're going to be once a day and they're going to be equally painful now what intuitions do you have about how valuable it is to have 10 of those you know negative value of having 10 of those or 20 of those and if there are any justification at all for deviating from linearity when I'm telling you that those are experiences that are going to be equally painful so that's that's a real question about and then if if you think that that there is no justification and yet we have powerful intuitions that 20 is much less than twice 10 in terms of action how bad it is that means that our intuitions which we think are about the utility of injections are really about numbers they're not about the injections at all you know that's that that were the puzzle and and Amos and I never got around to doing it together so we I worked on that on a related problem by myself it would have been much better if we had done it together but I studied situations in which people have an experience and you can measure it you know how many of these experiences they have well at the same time recording like in the injections how much pain or how much pleasure they're experiencing so that you can actually do the calculus of experience and it turns out that choices do not follow the experiences so I'll give you an example that you you put people in a situation where they have to hold their hand for 60 seconds in water at 14 degrees Celsius this is called me and sort of thankfully cold I mean nobody really likes it it's now so people had that experience 60 seconds at 14 degrees Celsius oh they have another experience which lasts 90 seconds the first 60 seconds are the same 14 degrees Celsius and then unbeknownst to them for the next 30 seconds the temperature the water rises gradually to 15 degrees Celsius which is not pleasant but distinctly less unpleasant than 14 degrees Celsius now people the two experiences in the 9th of all of three minutes something like that one with the left hand when with the right hand and then they are told you signed on for three of these experiences which of those would you like to repeat and you get a significant majority preferring the 90 seconds over the 60 seconds and that's because it ends better and and the end has a disproportionate weight on the evaluation of the experience in fact there is another phenomena the duration of the experience doesn't matter at all so we ran that's not an experiment that actually there was an experiment to we we studied people's experiences of a colonoscopy which is an experience for middle aged and and over and it's unpleasant and we don't need to go into details and and it turns out and so you monitor every 60 seconds how much pain the how much discomfort the the patient is experiencing and then you ask overall you know how bad was it there is essentially zero correlation and colonoscopies vary in duration from six minutes to an hour a minutes the highly variable there is no correlation between the duration of the colonoscopy and the evaluation at the end and that is a very deep phenomenon because it is not isolated actually the way that people evaluate an experience retroactively Lee is as a weighted average not as an integral and that is the way that categories are represented in the mind they are represented by exemplars by typical prototypical exemplars this is true in many domains and it explains violations of dominance this is violation of dominance of course it explained violations of dominance in multiple domains but it extend there is what I call you know the complete insensitivity to extension so it's all about the categories are represented by their intention and the extension seems to be essentially immaterial I'll give you just another example to to highlight this just so that you can see the similarity between two topics that look radically different you ask people how much they would be willing to pay to prevent various disasters or you know in bad events in nature and how much would you be willing to pay to say it's a two thousand birds from dying in an oil spill and how many how much would you be willing to pay to save two hundred thousand birds from drowning in an oil spill turns out to be the same amount and I couldn't be but it is and the reason is that you are really not saving two thousand birds you're selling one bird you know you have the image of the drowning bird of the bird running in oil you want to prevent that the extension is essentially irrelevant so that's this is this is the way that categories are represented and experiences are categories now so I read those experiments and from those experiments I got to the question of how do we measure well-being because and and well-being is typically measured by asking people satisfied are you with your life that's the basic measure and that is a retrospective measure it's like asking people after the 90 seconds and cold water you know how which of these do you like better how do you like your life and so I this is wrong and we shouldn't be doing this because in the case of the two experiences of cold water we know which is the correct answer I mean you know you should sixty seconds is just better so maybe I thought at the time I was young and naive you know I was in my late sixties and then and then I change was that it became clear to me that what people want this life satisfaction because they don't really the decisions they make are included the remembering self as they want memories and people evaluate their life in terms of their memories and and where people want is sort of a good story and which is not the same thing as as you know prolonged set of experiences and then it seemed to me that it doesn't make sense to have a theory of well-being that ignores what people want I mean you know that's that's a bit too too authoritarian even for me and so I have to give that up and I'm completely undecided about you know hero2 I have strong intuitions that this is important that is important which is more important what do I want for my grandchildren I have no idea I really I find it very difficult to decide I want both you know I want both and if I am forced to choose I'm going to have a lot of difficulty thank you I I watched your Nobel Prize interview which I enjoyed one thing you said in that was that sometimes a scholar it takes a long time for a scholar to understand what they said in their youth I wonder if you could give it back well my favorite example is actually not my experience that's it's my wife's experience my wife is an treatment was a famous psychologist then and she she had an insight and a theory around 1975 which took her several years to develop experimentally she was experimental psychologist was about attention and perception and she published a very important article in 1980 that explained that theory and then somebody told her but you know you had published that idea you have that idea a long time and she said she denied it but in fact black-on-white she had published that idea in 1969 and so the idea was there and you know actually formulated but she hadn't she didn't know she had it it had to mature in some way and a lot of the work I think that Amos Tversky and I did had in some ways the same character that we would have an insight or an idea something that we thought was an insight and and it would take a long time before we worked it out so there is a difference between formulating the idea that there is an asymmetry between gains and losses and really understanding the asymmetry between gains and losses and in in some cases in this case this is the case where you're conscious that that there is an important idea there but sometimes you're not conscious that the ideas and interests emerges over time that among the many thoughts that you have there is that one that you have been following and that was worse [Music] another comment you made in that interview was that in your youth you thought people could be educated to think better and you're not so sure could you elaborate on what you think education can actually accomplish well education clearly accomplishes a lot I mean there are major differences between you know the quality of people thinking depending on their educational attainment so it's that would be silly to to say that people cannot be educated of course they can be educated the idea that you can be educated to the point of being a rational agent that I think is absurd I mean you know then we don't even know as I pointed out earlier what a rational agent would be but but you know even a crude definition of a rational agent and that is because of something that I mentioned earlier I'm I'm reasonably convinced that system one you know the associative way of thinking has a much greater influence on the way we think and the conclusions we we have then we're aware of and I don't think that system one is easily editable and I will elaborate on that a bit because it's it's an idea that that I don't think people have followed enough we think of reasoning and this is by we I mean lay people thinking of their own thinking and probably philosophers and most psychologists we think about thinking as proceeding from assumptions and arguments to conclusions the psychological analysis suggests that this is just not the way it happens it could happen all at once or very likely the conclusions come first and and then there are stories to rationalize the computer it's not logical it's psychological I mean that's the way it's just the one I mean this is but this is the way we think I mean it's not there is an experimental demonstration of that which is due to a Jonathan Evans an english psychologist and it's people are asked to evaluate the validity of syllogisms and and the syllogism is some flowers fade quickly all roses or flowers therefore some roses fade quickly yes I mean this this gets about 80 to 90% endorsement as a valid syllogism among university students so education was not sufficient to eradicate this but notice what this is you believe in the conclusion and the argument seems valid clearly the conclusion came first now you present that when when the conclusion is false people trivially you know find it trivial when it's with X's and Y's people find this trivial you know they think Venn diagrams but when you have a conclusion that is true valid is so close to true that you don't really know the difference when you are asked how valid it is the fact that it's true is so appealing so that was the essential view of thinking I think that Amos and I developed of heuristics substituting one question for another that's an example of but this really if you accept this view of thinking you know us as in the Trump case you know if we think of conclusions first and have reasons later then the world is really different from you know the way the world would look if conclusion if conclusions came last and I believe that psychologically psychology provides a lot of evidence that in many cases conclusions come first you use the word heuristics which of course was very important and your earliest work I didn't see that word as much in Thinking Fast and Slow so I wanted to raise the issue you know heuristics feels like something simple and easy are you risking easy simple frugal well when when imps and I started our work together we we did not have a clear distinction between system one and system two between intuitive thinking and you know the the more and and so we never made a decision we use the term heuristic which we borrowed from Herbert Simon and you know others had used that before and Pollyanna and others and we followed Herbert Simon or so we thought and a heuristic as we used it initially was sort of a strategy to solve problems this is a way to solve problem so I'm asked about probability I'm going to use similarity as a strategy to solve the problem of the question or else I will count the instances that come to mind and that it wasn't clear to us we never really dwelt on it because the distinction was not very much in the air when we were whether this was indeed a deliberate strategy or something that just happened by the time I I got to write Thinking Fast and Slow which was you know more than 30 years later than almost 40 years later I had really changed my mind in very deep ways about heuristics of judgment so Amos and I had written about a finite number of heuristics Abreu's than a major paper on that topic and listed you know the three of them representativeness availability and anchoring which by the way is not the heuristic in my current thinking but but there is another way of thinking about the whole thing which is that it all happens in system one and what happens in system one is that you are asked a question but the answer that comes to mind is not to that question many answers come to mind too many related questions and the one that gets there first is the one that provides our answer and if you I call that the attribute substitution principle rule or mechanism and if we have that then the fact that we use similarity to drug probability is not a strategy something that happens I'd love to give you an example how heuristics work which is my favorite example so I'm going to tell you a story about about Julie and Julie is a graduating senior and I'm going to give you one fact about Julie that she read fluently when she was four years old what is her GPA now an interesting thing happens which is that probably all of you had an answer and and most of you had an answer that was probably over 3.6 how did that happen and and the answer came immediately to mind you know you didn't deliberate you didn't compute you know I asked you what's the GPA and a number came to mind now it's interesting because that is really a process that I think is well understood so what happens she read fluently at age four places Juli somewhere in a distribution that you feel how precocious how smart she was so you get that judgment which in effect you can you could easily represent as a percentile where does she fit in a distribution and what you do in answering the question of whether it's her GPA what you do what happens in your head is that there is also distribution of GPA in your head and you find the percentile of GPA that matches the percentile of speed of reading and then you read off the DPA that corresponds to that percentile and that's the answer that comes to mind now this seems like a complicated process we know I think just about every step of it is very likely to be true and that's a heuristic as I think of them now so it's not a deliberate process it's completely associative I mean they're all deliberate heuristics as well but the the heuristic that I'm most interested in I have that character something that just happens to you and that was an example Dulli example example for me the example that I have a purist ik thinking you were asked you were asked a question about her GPA but you answer the question the but how precocious she was and and you were not aware that you were answering the wrong question because it could not be you know this answer is wrong which I then I mean it will be obvious to you but this answer is wrong it's not regressive enough I mean it's completely non regressive the process that I described is a non be aggressive prediction you know it's an exact matching from percentile to percenter as if the correlation between reading precocity and GPA was one which clearly it isn't so that leads to non big dress illness which is the kind of you know one of the errors of reasoning that they must first see and I studied and and it's a completely automatic process I mean you want to drag me into you are going to ask me know you had a question that you know we have an enemy a masseuse key and I had an intellectual enemy who played an important role in our lives was fairly important his name is good gig arranger and he is a fairly eminent German psychologist who has a very different view of heuristics and he describes heuristics as fast and frugal and and is very very appropriate and ecologically valid and clearly that puts him in some opposition to us so the process I described for Julie is fast no question is it frugal I don't think I don't think that intuitive thinking is particularly frugal you would want to be you know Giga render the background is he thinks that we solve problems of prediction by using one cue when we have many cues we use only one in the Julie example i gave you one kill and you know and you used it but he thinks that what equals frugal is that we simplify problems by using one cue and i see that's not the way the mind works far as i'm concerned i think we we perceive the world in parallel and we see many things at once and we have many thoughts at once and so there is no need to be frugal so the system one is not no I mean system one you know the it's the multiplicity of possible associations it's a multiplicity of parallel processes that characterize the system one it's essentially month not frugal but fast one of my colleagues wondered to what extent do you would identify the idea of system one with the idea of the subconscious we have a better theory in the subconscious and Sigmund Freud had I mean you know what Freud's theory had in common with this that we're dealing with processes that are certainly not completely conscious we are not and in general but that that's a sort of trivial psychological statement that I'm making is we're conscious of the products of our mental processes when they are system one process when where reason deliberately then we're conscious of the process of reasoning because it has steps and we're conscious of when things just happen associated ly we're not conscious of the process then also so the the an important part of our mental life is not conscious that is common to you know this theory and and Freud's theory and and that's about all they have in common so another thing that struck me in your book was the concept of a pre-mortem i wonder if you could tell us about that and whether you actually yeah this is one of my favorite ideas and it is due to the psychologist Gary Klein with whom we became friends because we are really intellectual adversaries he hates everything I do and so we wrote a paper together trying to figure out under what conditions can intuitions be trusted because he likes intuitive thinking and I'm sort of skeptical of it but he came up with that idea quite independently of this work and the pre-mortem which strikes everybody not just you is that when you have a group that is say making an important decision and they are close to the final stage of agreeing on the decision you then convene a special meeting everybody has a sheet of paper and we're going to run an exercise and in that and the exercises follows we made that decision now two years have passed it was a disaster now please write the history of that disaster that's the exercise that's the Primo and it turns out it's a beautiful way of doing things too and it's quite effective I mean in a limited way it is very effective and it's a social process when a group is approaching a decision doubts are suppressed pessimists are redirected objections are sort of turned that and but the pre-mortem exercise does is it legitimizes all all these aspects of critical thinking that are suppressed by the process of group decision-making it's brilliant and it works so so I'm told I mean I haven't I haven't used it but you can see that it would work and you can also see that it would work in a limited way I really very much doubt whether any group would do a pre-mortem and decide not to do it but they can do if they will find loopholes they would find problems tweaks that they can make I don't really and and Gary has never told me of any instances that he knows of of a group doing a pre-mortem and giving up that's too much to expect too much rationality to expect but but it's a very good technique well you've been very generous with your time and I think we could continue a little longer if someone in the audience would like to I mean we've been going you know I mean you know you guys are very patient but but we can bring it to an end any time yeah when you were talking about your adversary I was reminded of something that Jerry Fodor you still like to say and that was that every philosophers theory of mind is true of only himself yeah I was wondering whether your German adversary this theory was true the family no I mean you know it um I'm not sure you know I think I think he has an interesting theory as many interesting things to say and Amos and I always thought that he disliked us first and then he found reasons you know that was in that sense I think that back to the years back I didn't hit I'm hard of hearing at all no I am NOT and and that mapping with left-brain right-brain worries me because the left brain right brain is not is not current in in current neuroscience and so it would worry me if but but it's true that it sound left brain right brain sounds very much like system one system to verbal not verbal conscious not conscious etcetera but it can be pushed too far so just a quick question on your because you mentioned system one is not through little and knowing your work that kind of surprised me so are you saying that there are lots of things happening but at the same time it's still cognitively easier right because well I mean the way the essential characteristic of the essential distinction between system one and system two as I drew it whether system one is essentially effortless so things happen and you know so when you present a word you know like river or like doctor you don't have one Association yeah hundreds of them I mean hundreds of three associations many ideas are evoked don't become conscious necessarily most of them don't almost all of them don't but but that is the non frugal aspect that you don't pick one intuition you have many now we worked that may have surprised you in that we we focused unpredictable errors and you know in that sense it seemed as if it's a simple process but we just focused on the case where one of these processes is likely to be faster and stronger than its competitors and that's that's why twins one of the many project they were exciting for this century the intersection of philosophy psychology and this was the foundations of measurement project that firstly who's stupid disengaging I was wondering how this project has influenced your thinking and your work in terms of theoretical and experimental technology the answer I think is not at all and and you reminded me of a joke that that Amos like telling and it was one of his and this is that magnificent building that exists you know it's all there and and way down there in the cellar there are people working on the foundations and you know that was his view of the Foundation's work so he loved the precision of it he loved the formulas mother I I don't think it affected our joint work very much well that's you know what I drew on is sort of basic cycle psychological stuff it is really not not the mathematical foundations of measurement yeah okay not at all yeah that's this is an observation that there isn't enough of rhythms for it and I'm convinced that it's true but it's a hunch so but there is some evidence and it's about sandwiches actually so people have an endowment effect about sandwiches so they they require more to sell a sandwich they have even if they have been given it then they'd be willing to pay for it now there are one of the explanations it's not the only one is is the loss aversion but the interesting thing is that when you buy when you are making decisions on behalf of someone else there's no loss of food you buy and sell at the same price so this this is quite deep I think because it turns out that agents don't think in terms of gains and losses because they are not emotionally involved in what is happening they don't feel the gain and the loss they are thinking terms of final assets which makes them much more rational an interesting implication of that is in the social domain people who proposed reforms reforms are invariably much more expensive than anticipated and the reason is that the reformers view a final state that is highly desirable and clearly better than the current state but on the way to the final state you're going to create winners and losers and the losers are going to fight much harder than the vendor potential winners and they'll have to be compensated and it will turn out much more expensive than anticipated so this idea that we mail that sometimes we don't have loss aversion sometimes it ends in disaster as in many before on the other hand an answer question I was not asked because I liked the answer and and the question is who would you seek advice from and to that my answer is Richard Thaler for me but I will explain who Richard Thaler is in that case it's somebody who likes you who is very smart of course he likes you but he doesn't feel your pain he couldn't care less about my feelings which makes him a very good adviser because you know it's about consequences he views the big picture he doesn't view the immediate emotions that are keeping the occupied and so you know there are good consequences and bad consequences are the fact that agents or advisers or generally outsiders do not have loss aversion so done groundbreaking work doing purely behavioral research using that psychology departments worship going away from purely beginner research so if you think that this is a problem in itself no I mean it's I don't think it's a problem I mean you know if I had been 20 years or 30 years younger at beans I'd be doing neuroscience I mean you know I think it's I think the brain is fascinating it's another way of studying you know it it hasn't had major payoffs yet that you can you know pinpoint where our thinking psychological thinking broadly speaking has been affected by it but it will be and and I I'm not among those who deplore it there are some bullet proof I find it very exciting and this seems to be at odds with prospect theory because in fragile systems benefit from variations whereas the prospect theory and the value function of loss aversion it seems like variations may resulting negative values I wonder how you well I had never you know the the theories are at such different levels that I had never thought that that they were opposed or contradictory machine has developed but I think is a very interesting theory that pushes an idea he had had earlier about the importance of extreme events and so he's been fascinated by extreme events for 30 years or so and and now he's pushing you know what would we do if we take Siri if we take rare and extreme events seriously and then he gets into a building system that are in effect robust anti fragile is as more than robust that actually benefit from from being stressed and I think he pushes it to ridiculous extremes than in the sense that a society cannot really function along his principles but the idea is interesting and and certainly deserves to be taken very seriously I don't think that the conflict was prospect theory as would bother either Nassim or me we might have one more question as I guess it's similar to Andy's did you from your side where's mathematics always a way to express insights that you already have or did you ever have a moment where inside you arrived at insights well yeah there was a very big no-no here there was a very big difference between a masseuse key and myself he was a true mathematical social scientist and there was he was in the second generation there was a resurgence of mathematics in the social sciences foundation of measurement loose Soapy's and so on that happened I think in in late fifties early sixties and he was in the second generation profoundly influenced by it and he was the only one I think of the mathematical psychologist whose work I knew for whom mathematics was really a way to think clearly it really helped him think it didn't do that for me so I am I don't I've never really you know I I did some of the mathematics of prospect theory I could do that but it it always came later you know I didn't get my ideas from it I mean you know the money is gone mind I can tell you a dick Thaler you know who keeps popping up today but his answer to what he was going to do with the money whether he was going to spend it as irrationally as possible

Channel: Harry Crane

Views: 5,497

Rating: 4.8961039 out of 5

Keywords: daniel, kahneman, amos, tversky, richard, thaler, nassim, taleb, prospect theory, probability, bounded rationality

Id: XWWlcaHXxZs

Length: 109min 2sec (6542 seconds)

Published: Mon Oct 30 2017