Skip to main content
Loading Explore guides…
Bounded Rationality & Policy · Acedra
Home
Explore
Economics
Bounded Rationality & Policy
Sign in to save
Chat
If YouTube asks you to sign in, open the video on YouTube.
Watch on YouTube
Reload player
Transcript
82 segments
Copy
Hide video
0:00
Good afternoon everyone. It's wonderful to see you virtually. I can't really see you, but it's wonderful that you're here. Welcome to the Harvard Law School Library Faculty Book Talk series. This talk is being recorded as you heard as you entered the room and we'll post it to our YouTube channel in the next week or so. I want to thank Dean Manning for his generous support of these talks and a huge thanks to our team, Meyer Burger Moscow and Martin Rachel Parker,
0:26
Debbie Ginsburg, and Theresa Knapp for helping us put together this series. We welcome your questions for our authors at the end of our talk, which you can use the Q&A feature throughout, and we'll pick those up toward the end of the conversation. Please visit your local library or your local bookseller, including the Harvard Coop for a copy of today's book, bounded rationality, heuristics judgment, and public policy. It's my pleasure to introduce today's authors.
0:53
Sanjit Dhami, is Professor of Economics at the University of Leicester, and Cass Sunstein is the Robert Walmsley University professor at Harvard and here at the law school. I'm going to turn it over to you in the interest of time so that we can hear all about your new book. Thanks for being here. Thank you. I think the run of show will be, I'm going to talk for about two minutes and then we'll go to Sanjit and then I'll say a little bit on policy. This book has a very unusual origin.
1:25
A number of years ago, Sanjit sent me a note he wrote basically in response to some concerns about bounded rationality, about maybe 10 or 12 pages he can correct and it was a brilliant note that at one Scott a foundational debate between neoclassical economists and behavioral economists that's been going on for decades, and Adam or recent debate about bounded rationality and the role of heuristics. A debate that is very lively, it has been like for about 8-10 years, a little more than that.
2:04
The note had such content and thickness and brilliance really in it that I thought or he thought or the thought some things should come event that would be thicker than this amazing note. We worked together for several years what became an academic article on bounded rationality. But the foundational nature of the issues discussed in the article on the original note is basically it's an enduring set of questions that are arguably the most basic and all of social science about
2:43
our species and our minds and that led to what is not a short book, and our goal really is both to capture what we know now about bounded rationality and to set a lot of confusion straight, that's a goal, and Sanjit certainly succeeded in setting straight a lot of my computer. But also to set an agenda for, this is ambitious, but this is our goal for people who are exploring these issues now and we'll be exploring them or what we hope will be a long time to come.
3:25
I'm very grateful to my coauthor as well as those of you who are listening to this, and Sanjit over to you. Thanks Cass. I have a set of slides if you don't mind. I'm going to try to share my screen put them on if I can.
3:51
That didn't go through. I'm going to try again. Yeah, there. Can you see my screen now? Yes. Brilliant. Thanks very much for inviting me, and Cass is a truly wonderful co-author, and I learned much from him as the book progressed. Thank you Cass for your very kind words. The plan is I'll go for about 20 minutes. But please, if I run over or if you want to stop earlier, just give me a hint and I'll move it across two Cass. I'm sorry, this has gone all the way. Let's try again.
4:45
The dominant framework in economics, known as neoclassical economics uses something known as the Bayesian rationality approach, which I'll abbreviate as BRA. Anybody who goes to graduate school in economics is really grilled in this approach. It's been widely adopted in political science, for instance, in the rational actor model, in law, in management, in finance, even in studying the behavior of neurons in the human brain believe it or not. In its actual practice,
5:21
it is essentially a set of core assumptions, and these core assumptions impose a certain model of human behavior. They ask us to believe that humans behave in a particular manner and then various conclusions are drawn from these core assumptions and some of these core assumptions are unlimited computation and cognitive abilities that humans possess, or supposed to possess. That humans make considered choices that respect certain mathematical axioms of rationality,
5:57
such as completeness and transitivity. That humans have perfect memory, they have perfect attention. That humans know all the math and stats that there is to know, including research that has just been published in math's journals and even research that is at a revise and resubmit stage. Now, what it typically ignores is the role of emotions and morality in humans. Humans are essentially immoral in this framework.
6:29
Always argue these two are not a part of the rationality requirements in this framework. The good thing about this approach is that it makes very precise, very testable predictions. The problem is, several of the core features are empirically rejected. From around the second half of the 20th century, this framework has gradually purged insights from the other social and behavioral sciences, making it into an extremely insular discipline. In addition over time, the rationality requirements on
7:13
individual behavior have almost grown exponentially, become extremely stringent. At its heart, it is motivated by models in theoretical physics that deal with inanimate particles. The hope very much was that the mathematical equations that work with individual practicals might also work with humans. Rather than borrowing from evolutionary biology, this approach borrowed from theoretical physics. However, it does not adopt stringent standards of testing and physics.
7:52
In physics, you test your theories very stringently. That bit is missing in this approach. On the other hand, it claims a very distinct status for economics, which is designed to protect it from reputation. The central feature of this approach is mathematical optimization. It is assumed that the man in the street either maximizes some well-defined objective function or minimizes some well-defined objective function. For example, firms might wish to minimize costs,
8:26
individuals might wish to maximize the satisfaction that they get from consuming goods and services. This maximization and minimization is done in an engineering sense. In the same way that the engineer would measure the top of a hill, you maximize stuff. Or in the same way that an engineer would measure the bottom of an ocean, you minimize stuff [NOISE] In contrast to this approach, the evidence shows that people have limited attention, limited computational abilities, limited willpower.
9:02
People often make choices they regret. People have limited financial literacy and other kinds of abilities. Imperfect and motivated memory. We don't recall events as accurately as we would like to. We have motivated recall and we're often overconfident. Humans are also emotional, moral and other regarding.
9:30
Although these two being emotional, moral and other regarding is not related to the rationality requirements in the Bayesian rationality approach, and that's an important point to make and we do make that point. Let me talk a little bit about mathematical optimization. It imposes very unreasonable cognitive and computational requirements on humans. Consider a very simple dynamic programming problem. This was a technique that was invented around 1950 by mathematicians.
10:05
Supposing you have 30 time periods and individual lives over 30 years let's say. In every time period, there is a 50, 50 chance of income being either high or low. We don't know whether it will be high or low in the future. All we know is a 50, 50 chance. You make a consumption stroke savings choice for each level of income. In order to solve a dynamic programming problem, which is the bread and butter solution method in economics.
10:37
An individual would need to make two raised to power 30 calculations. That's about a trillion calculations in their heads, and we show this formula in the book. Most problems in economics are in fact more complex than this. But the man on the street is assumed in this framework to be able to solve these problems in their head in an instant. Now this would be true of undergraduate and graduate courses in economics in an investor in university. That's what the framework would ask students
11:14
to believe and that's reflected in research as well. Then our mounting empirical reputations of the Bayesian rationality approach, hence we need concepts and bounded rationality. But why has there been an endurance to the Bayesian rationality approach for so long?
11:36
In the book, we identify the reason in terms of homegrown methodological positions within economics. The centerpiece of this methodological position in economics is the as if argument due to the Nobel Prize winner Milton Friedman. Friedman argued that people may not literally obey empirically rejected assumptions, even assumptions that are patently false and obviously false. He said people might not literally obey them, yet the predictions of such models might be accurate.
12:11
Hence he said people act as if they follow the Bayesian rationality approach. However, when these predictions are tested, they often fail. This was the beauty of the work that Kahneman and Tversky did in several seminal papers in the 1970s. They showed that the predictions of the Bayesian rationality approach do not even hold in an as if sense, that the evidence at the moment on that has grown exponentially and is extremely strong.
12:47
Just to give you an insight into what some of the leaders think about this approach, and by it, there is such strong adherence to the Bayesian rationality approach. This is one of the leaders in economic theory. It's Gilboa. He says, ''Economic models are expected to convey a message much more than to describe a well-defined reality. The economic theorist is typically not required to specify where his model might be applicable and how. This is Ariel Rubinstein,
13:21
another very respected leading theorists in the world. He says, '' As in the case of fables, models in economic theory are not meant to be testable.'' Yes, this is an article that appeared in print. ''A good model can have an enormous influence, not by predicting the future, but rather by influencing culture. Yes, I do think we are simply the tellers of fables, but is that not wonderful?''
13:54
In our book, we serve at the literature on bounded rationality and consider the implications for welfare and for public policy. WE have some challenges speculate on which theories might work better and offer a research agenda for modern bounded rationality. We start by defining what rationality means in its various contexts, and then provide empirical evidence on the underlying axioms of rationality. We show that the evidence rejects in fact, many of those axioms.
14:30
We also argue that the evidence is much more supportive of behavioral economics, which is a more interdisciplinary approach that is strongly grounded in the social and behavioral sciences. In fact, it's fair to say that [inaudible] and I, for many years now have been using insights in behavioral economics and developing the underlying framework in various areas. We do not shy away from challenging topics in this book. We've designed this book for a very wide audience,
15:04
so the temptation was to dump it down and not include challenging topics. We resisted that, and we included these topics, and here's the taster list; evolutionary game theory, stochastic social dynamics, formal dynamic epidemiological models of the spreads of epidemics and narratives, dynamic programming, axiomatic foundations of utility functions, behavioral decision theory, complexity and chaos, and the bias-variance dilemma in machine learning.
15:43
Once you decide to go towards bounded rationality, then you have to decide and you do it. We identify two parts. There's an incremental program, which essentially argues if fixed all the features of the Bayesian rationality approach, relax one feature at a time and see how it goes. For example you might relax unlimited attention or unlimited memory to limited attention and limited membrane. Keep everything else fixed in the Bayesian rationality approach, see how it goes.
16:17
Relax one assumption at a time and see what happens over time. That's the approach. But the end result might be quite influenced by the order in which you relax these assumptions. There's no guarantee that relaxing assumptions in an arbitrary manner will lead to the same end result. A second possibility is the big push program. This essentially argues [NOISE] that you give up most of the empirically rejected features of the Bayesian rationality approach in one fell swoop.
16:57
Then go on make empirically founded assumptions and see what's the best you can do over time. Most economists have preferred the incremental program. This is a largely as if approach, and it's based on substantive rationality. By that is meant you look at how good the predictions of the model are, the assumptions might not be empirically correct.
17:25
We make a case in this book for taking more seriously the big push program as well. With a particular emphasis on replacing mathematical optimization with judgment heuristics, I'll speak a little bit more about them in a moment. A central feature of this big push program is procedural rationality. You look at the quality of decisions and the process by which decisions are made. Famous example of that is the aspiration adaptation model, which we described.
17:57
[NOISE] The Kahneman-Tversky research program is quite an important part of the book. Let's first think about heuristics as giving us simple rules of thumb that are fast in terms of the time it takes to reach a decision and also frugal in the use of information that is required in order to reach a decision, and they allow cognitive simplifications in making decisions. In pathbreaking work in the 1970s that I just spoke about a little while ago,
18:37
Kahneman and Tversky showed that instead of using mathematical optimization, people seem to be using very simple heuristics in making choices, and they do not follow the laws of classical statistics. Bayesian rationality approach does not hold, in fact, even in an as if sense in this framework. We survey and formalize a relevant heuristics, we evaluate the criticisms, and empirical evidence for them as well. Some of the heuristics that we examine include representativeness,
19:12
anchoring, availability, confirmation bias, and hindsight bias. We show applications in almost all areas of social science. We also try to resolve some really important and fundamental confusion in the literature on heuristics. We also examined other related approach by Gerd Gigerenzer and colleagues at the Max Planck Institute, and argument is that these two frameworks are not necessarily adversarial.
19:43
To give you a flavor of the Kahneman and Tversky finding, let me give you some empirical evidence that was with students from Harvard actually. In coming to any situation, we have a set of prior beliefs. Then new information arrives or a new signal is observed, and then we form stereo beliefs by updating our prior beliefs. How do we update our updates? In the Bayesian rationality approach, the answer is very simple, we use the laws of classical statistics, which take conditional probabilities,
20:19
and it's Bayes law, which is a relevant way of updating from prior beliefs to posterior beliefs. But empirical evidence typically rejects Bayes law. To consider this problem given to Harvard Medical School Students 1978 paper, suppose that a police breathalyzer test discovers false drunkenness in five percent of the cases when the driver is sober. Even if a driver is not drunk in five percent of the cases, you get a false positive.
20:53
But the breathalyzers always detect a truly drunk percent with 100 percent chance. If you're drunk is going to catch you out anyways. In the general population, one out of every thousand drivers engage in drinking whilst drive. Now this is known as the base rate, the number of drivers per thousand who actually drink and drive. Driver is now checked at random, just turns out to be positive, what probability do you assign to the driver being drunk?
21:29
This is a case of updating prior beliefs to posterior beliefs. The modal response amongst Harvard Medical School students was the probability is 95 percent. The mean response was 56 percent. The correct Bayesian response is just two percent. Its because lot of people ignore the base rate, they ignored the fact that only one out of 1,000 drivers is driving whilst drunk, they take the fact that if it discovers false drunkenness in five percent of the cases,
22:07
so they subtract it from 100 and get 95, and that's the answer. Compare 95 percent with two percent, huge difference. [NOISE] Another theme in the book is True Uncertainty. This is a quote from Donald Rumsfeld, former Defense Secretary, who I believe was both the youngest Defense Secretary and the oldest Defense Secretary because he had to tenures in office. This is a very well-known quote due to him. As we know, there are known knowns. These are things we know, we know.
22:42
We also know there are known unknowns. That is to say, there are some things we do not know. But there are also unknown, unknowns. The ones we don't know, we don't know. If one looks throughout history, it is a latter category that tends to be the difficult ones. Now, these three things are at the heart of the book in some sense, in at least one or two parts of the book. Known knowns in economics is known as Risk. Known unknowns is known as Subjective Uncertainty.
23:18
Unknown unknowns is known as True Uncertainty. Economics unfortunately, makes little in the way of predictions, when you have true uncertainty. Mathematical optimization completely fails in this case, because you simply don't have the data that is required to run mathematical optimization. It's unknown unknowns. You don't know what data to put in. Yet people do make decisions under true uncertainty. We propose in this book that heuristics and social norms,
23:51
and we do a lot of stuff on social norms in this book as well, because we offer cognitive simplifications, may offer a powerful tool for dealing with true uncertainty. The last slide before the conclusions, just a quick word on private rationality and social rationality. Modern game theory relies on private rationality. The prisoner's dilemma, game is a leading metaphor for human cooperation, and that's the one I'm going to use now. Two prisoners caught for a minor misdemeanor,
24:27
held in separate cells by the police and are being interrogated. Police suspect they were both involved in the crime. Each prisoner does not know what the other is talking to the police. They simultaneously choose either to cooperate, i.e, not snitch against each other, or to defect, which is to snitch against each other. Now, you can think of C and D, Cooperate and Defect as the first prisoners' choice's. Operate and defect along the columns as the second prisoners' choices.
25:03
If both cooperate, they both don't snitch against each other. The police lets them go. They both get two each. This is in some units.
25:15
If they both snitch against each other, D, D, they both get lower payoffs, 1,1. If one cooperates, doesn't snitch, but the other snitches against him, then the snitch gets a high payoff, three. He gets off and the police can come for that. But the one who doesn't snitch is put in jail. The payoff of the one who doesn't snitch is zero. Similarly, this is the case where the Prisoner 1 snitches, Prisoner 2 does not. The snitch again gets three and the one who doesn't snitch gets zero.
25:52
What's the prediction for this game? What should they do? Private rationality predicts they should just snitch on each other. You can see that very clearly here. Look at Prisoner 1, if he snitches, he gets a payoff of three greater than two, one greater than zero. Always pays to snitch. If you look at the column there, it's exactly the same situation. If he snitches, he gets three instead of two, or one which is greater than zero. It always pays both of them to snitch and
26:29
the prediction of game theory is snitch, snitch. Yet, when you look at the empirical evidence, 60 percent of the choices are actually cooperate, cooperate. These are the most common choices. The prisoners don't snitch mostly. In general, we observe more cooperation than can be accounted for by private rationality. Social rationality might explain the data. Prisoners might have a social norm or a code of conduct so that they do not snitch against each other.
26:59
They might defect to behave as if they are maximizing some sort of team pay off. In the book, we outlined several concepts in social rationality, excuse me, such as team reasoning, co-related equilibrium, social norms, and content rationality. In its current form, the Bayesian rationality approach is untenable. It has serious implications for the social sciences. The way forward is to develop models in bounded rationality. In the book, we outline what's been achieved, the likely challenges,
27:31
and the possible way forward. We do not have all the answers, but that is the nature of science. It also has critical and hugely important implications for welfare and public policy, which I believe Cass is going to talk next. Thank you very much. Thank you, Sanjit. That was admirably clear. The extent to which the argument you just heard is meant as a Declaration, capital D, with respect to a long period of work and economics and
28:10
an effort currently in psychology to rescue unbounded rationality, let's call it. We should see the capital D, Declaration. I'm going to talk a little bit about policy. I'm going to try to start with things that are very, very concrete. Think of these, if you would, in the context of questions about whether people are boundedly rational. It turns out that whether people get vaccinated is importantly affected by whether there is a participate in retail pharmacy vaccination site in the county.
28:49
There's a 26 percent increase in the per capita number of doses administered, indicating that proximity and familiarity play a really big role in determining vaccine take-up decisions, bracket for the moment the question whether that is a reflection of bounded rationality or reflection of perfect rationality. Here's a good predictor of whether people are going to be evicted from their homes. That's a high-stakes issue. Here's a good predictor of whether a tenant is going to be evicted,
29:22
whether the tenant lives close to or far from the courthouse. That's a startling very recent finding that if a tenant lives far from the courthouse, the likelihood that the tenant shows up in an eviction proceeding is substantially lower, which means the tenant will be evicted. Being evicted is a catastrophe. To avoid the risk of eviction, it's really good to show up. People don't show up if the place is far away. Is that a reflection of perfect rationality or bounded rationality?
29:56
That is a rhetorical question. Third of my four items is that in the District of Columbia, as is all over the world, there are opportunities that parents can take advantage of to give their kids better educational opportunities. It turns out in the District of Columbia, there's one that's really, really good. We know from the data that if the parents use the technology on every dimension, the kids do a lot better. If parents are asked, "You want to opt in?"
30:27
You go on the website and opt in, use the technology. The participation rate is one percent. If participation opportunity is simplified in the sense that you don't have to go to the website, you just click on the email, the participation goes up by, I'm not the mathematician of the team, but goes up by 800 percent, I believe, to eight percent from one percent. That's good. It's terrible. It's good, eight percent is a lot better than one percent. It's terrible, eight percent of parents means that
31:00
92 percent of parents aren't taking advantage of this opportunity. The third intervention is to say, the third way of doing it is say, you're automatically in parents, you want to opt out? As a result of that, over 95 percent of parents are in, their kids do much better in terms of their educational opportunity. We go from one percent to eight percent to upwards of 95 percent. The kicker, the punch and the teeth, I think in terms of rationality, is that experts on
31:35
educational take-up district leaders who've been working on this for decades, overestimate take-up under the standard condition by almost 40 percent, and underestimate the take-up under automatic enrollment by over 31 percentage point.
31:55
After learning the actual take-up rates, there's a 140 percent increase in willingness to pay for the technology. The last of these is that as many of you know, cars in the United States now have cameras in them so that people can see in back. That's reducing accidents and deaths, it's also reducing inconvenience and difficulty of parking and driving. Have the camera, you can see in back. The market didn't produce cameras in cars. I want to suggest that all of these four cases,
32:34
they're all very simple, are very plausibly attribute to the power of bounded rationality and the opportunity for a policy response based on welfare analysis that helps. The stakes are really high. In the first case, it's about being alive or dead. In the second case it's about being evicted or not. In the third case is about their kids educational opportunities. Then the fourth case, it's about being in an accident or not hitting someone small or not, or knocking your car.
33:13
Again, something that costs a lot of money. The welfare analysis or the, what's the right word? The cost-benefit analysis or the choice theoretic analysis. I think those are the two ways to go, are pretty straightforward. The welfare analysis standardly would be a cost benefit calculus, and in the cases just given, we need to run the numbers to find out. But it's very plausibly the case that the benefits crush the cost in every one of the four cases.
33:51
That's true even though the fact is that consumers aren't making a choice that fits with the choice that regulatory policy generates. I've collected my own data on the cameras in the cars, and ex post at least people who had experience with the cars, with the cameras in them, would demand a lot of money to give up those cameras and the cars on the order of $300. That's a lot more than the cost of the camera and the car, which is on the order of $50.
34:23
At least with the experience of driving with the cars, the judgment is thank goodness for the regulatory in position that overcame my OPIA may be on the part of drivers, meaning present bias or maybe limited attention on the part of drivers at the stage of purchase or maybe simple lack of information. We don't know exactly the mechanism. The book does talk a lot about welfare analysis in the face of bounded rationality, and sees this as a world of opportunity for
34:59
private and public institutions to help people's lives go better. If we don't do it in a top-down technocratic way, there's a choice theoretic way which the book explores, which is a research agenda and I think a thrilling research agenda where we would think what would consumers who were not boundedly rational do? That's not a thought experiments, it's empirical agenda. It can be dis-aggregated, the question, by asking for example, what do inform choosers choose, do they choose cameras in cars?
35:38
Do they choose to get vaccinated even if it's quite a drive to get there? What do active choosers choose? Often it's the case that people are passive choosers. That is they are just going with the default. Maybe we can learn something by seeing that active choosers typically get that educational technology, whereas passive choosers get it if they're automatically enrolled in it, don't get it if they have to opt in to do it. In circumstances in which people's view screen is broad,
36:11
in which various things that affect their well-being are placed before their eyes, what do they choose that would overcome or reduce the least limited attention? What happens now? Finally, we can ask if people are helped to be, let's say, unconstrained by unrealistic optimism or present bias. What do they choose to have? That's not a science fiction tale, you can tell people that over the course of five years, if you got this product, you're going to save that amount of money
36:43
compared to if you've got another product, then what do you want? That would be a choice theoretic way of trying to figure out what people choose, it wouldn't be technocratic. Now, describe this as a thrilling research agenda because it's actually happening before our eyes. It's an application of, let's say, behaviorally informed welfare analysis, where we ask what to inform choosers, choose with respect to cars or educational technologies.
37:11
What do active choosers choose in circumstances in which people's view screen is broad because they're informed of a range of things, then what do they do? What do they do in which present bias and unrealistic optimism are met by something that gives people a sense of the statistical reality and the futures likely existence, even with a discount rate? We know from what Sanjit described, that there are ways in which people depart from perfect rationality that is trouble for all of us,
37:49
and the fact that people die prematurely or get sick when they needn't, or get crashed in a car, that in some ways attribute to bounded rationality. But we know from the four things with which I began, there are things we can do about it. Now, what exactly we should do shouldn't be unmoored from welfare analysis. Welfare analysis we suggest easily survives an encounter with bounded rationality. Still can do it. We might want to do it with analysis of the effects, negative and positive.
38:26
That's the technocratic approach, or we might want to do it with a choice theoretic approach with the four questions I described. In a way, this part of the book is, I'm not sure I'm going to pronounce this word right, and I'm not sure I know what it means. Prolegomena. That's a word. It means it's a prolegomena to, let's say, welfare analysis in the face of bounded rationality, animated by a world of optimism and hope. In this case, not unrealistic. Thanks.
39:04
Thank you both so much for sharing your insight. We'd like to take this time to move to our audience Q&A for our attendees joining us today, if you could use the Zoom Q&A function to submit questions to our speakers. We will read them aloud in the order that they are received, so it might take a minute for folks to formulate a question. I'd also like to invite the speakers if you have questions for each other. Around the research behind this book or co-writing together, we'd love to hear more.
39:35
I think I'd let the course take that one. The only question for Sanjit I have is, how did you produce so much so quickly? [LAUGHTER] Well, because I have great co-authors like you. I mean, it's just genuine academic curiosity. I mean, it's so exciting really see empirical evidence, which is contrary to what I learned in graduate school. That once you start thinking about it, it's difficult to think about anything else. I suppose my position is a bit like scientists when
40:18
the Cern Hadron Collider started just being functional. Physicists were excited that, now we're going to learn something really amazing about the empirical evidence on relativity. If it were to reject relativity, how exciting then we're going to really know what the way forward is. Now, with respect to behavioral economics, we have thousands of cern hadron colliders around in so many different areas. This is very palpitable sense of excitement
40:54
about the developments that are occurring in many different areas. Writing this book and the other books has given me an opportunity to come up to speed with large bits of the literature and figure out what's the way forward.
41:13
Thank you, professor Dhami. We do have an audience question. The question for both of our speakers, how is your research being received by those who still subscribe to the Bayesian model?
41:30
Cass you want to say something to begin with? Sure. Well, there are two ways to think of that, and they're both in the book. One is that the Bayesians are good Baysins, and they see that there's data that's inconsistent with what they think, and so they're updating in a way that is admirably consistent in their own performance with their theory of the case. Some people have read the book, it's hasn't been out for a long time. It's a pretty dense book. Some people have written me notes saying I
42:12
finally understand and I see the various problems. Good Bayesian updating. There are others who I think are, let's say, behaved or more behavioral than Bayesian in the sense that they have strong emotional commitment to what they used to believe, and they're fighting real hard against the arguments in the book. But some of the arguments, of course, I would think this, but I do think this sincerely. They're a little bit of an argument that a law professor named
42:43
Charles Black once described as one step ahead of the marshal arguments. There are arguments where you're running really fast to try to keep up. Not because you're not smart, but because the arguments are ahead of you and you're emotionally committed to them. We have not seen that.
43:10
Thank you. Next audience question. What is the relevance of reductionism in behavioral economics with respect to Neuroeconomics?
43:22
Again, I think Cass, this is probably more up your street as well. Reductionism.
43:29
We both learned from neuroscience. The book I think, shows that influence. I'm thinking that sometimes neuroscientists, a little like historians, give a thick account of a process which might may or may not lead to testable predictions. If I understand the question right, you might have an understanding of the amygdala and the prefrontal cortex and have a sense of what's happening in a choice that might not. It might, but it might not give you a prediction with
44:12
respect to whether people are going to be loss averse. It might show a neural marker of loss aversion where the amygdala is activated more by a loss and again and again. But some of the neuroscientific work is profoundly illuminating about the brain. But whether it is necessary to illuminate behavior remains to be seen, and it may be there's some reductionism and behavioral economics, that's okay or necessary for the purpose.
44:54
Looks like those are all the questions that we have. I want to thank both of you so much. I still don't really understand economics, but that's okay. I do feel like I have a better sense of what's going on, so thank you both for that. I just want to let our audience members know that our next book talk will be on November 8th at 12:30 on the book Constitutionality And A Right To Effective Government, edited by Vicki Jackson and Jasmine Dog Woods. I hope that folks will join us for our next one.
45:26
It will also be in-person and on Zoom and check out our Twitter feed at backslash HLS LIB. HLS LIB for more information. Again, Professor Dhami, Professor Sunstein, thank you so much for being with us today. It was a delight to hear from both of you. Thanks all. Thank you very much. Thanks all.
Here to help you learn
Sign in to keep a private conversation with your course tutor.
Ask me anything about the material…
Sign in to chat