WEBVTT - Advancing Behavioral Economics with Colin Camerer 0:00:02.400 --> 0:00:11.119 Bloomberg Audio Studios, Podcasts, radio news. This is Master's in 0:00:11.200 --> 0:00:16.000 Business with Barry Ridholds on Bloomberg Radio. 0:00:16.320 --> 0:00:20.479 This week on the podcast finally I get Colin Camera 0:00:20.680 --> 0:00:26.040 in the studio to talk about neuroeconomics, behavioral finance, and 0:00:26.640 --> 0:00:29.840 really all the fascinating things he's been doing at Caltech 0:00:30.240 --> 0:00:34.480 for the past. Gee's been there for almost thirty years. 0:00:34.560 --> 0:00:37.199 Is that about right? He's really an interesting guy, not 0:00:37.400 --> 0:00:42.240 just because he has the mathematical and behavioral finance background, 0:00:42.640 --> 0:00:47.120 but because he essentially asked the question, what's going on 0:00:47.200 --> 0:00:51.040 inside our brains when we make decisions? What's happening before 0:00:51.040 --> 0:00:55.320 we even have a degree of awareness of our own decisions. 0:00:56.440 --> 0:01:00.600 I just find what he does fascinating, not just fmyes, 0:01:00.640 --> 0:01:05.920 but eye tracking and eg and galvanomic responses of the skin, 0:01:06.000 --> 0:01:08.160 and just on and on, all these different ways to 0:01:08.240 --> 0:01:11.280 measure what's going on with your hormones, what's going on 0:01:11.319 --> 0:01:18.600 pharmacologically within your body. It's both fascinating and terrifying because 0:01:19.200 --> 0:01:23.240 you come to realize what you think is a decision 0:01:23.319 --> 0:01:27.120 you're making, very often is a decision your brain is 0:01:27.160 --> 0:01:30.640 making with or without you. I found our conversation to 0:01:30.640 --> 0:01:34.280 be absolutely fascinating, and I think you will also with 0:01:34.480 --> 0:01:40.120 no further ado my sit down with Caltech Colin camera. 0:01:40.400 --> 0:01:41.800 Thanks for having me so. 0:01:41.880 --> 0:01:44.600 I've been looking forward to having this conversation with you 0:01:45.120 --> 0:01:48.960 for a long time, not just because of my interest 0:01:49.040 --> 0:01:53.760 in behavioral finance, but because of the space you occupy 0:01:54.040 --> 0:01:57.760 in neuroeconomics. We'll talk a little bit about that in 0:01:57.840 --> 0:02:01.360 a bit, but let's start with your back, which is 0:02:01.480 --> 0:02:04.600 kind of astonishing. You get a bachelor's in quantitative studies 0:02:05.000 --> 0:02:09.360 from john Hopkins at seventeen, and then an MBA in 0:02:09.480 --> 0:02:13.399 finance and a PhD in decision theory from the University 0:02:13.440 --> 0:02:16.920 of Chicago at twenty one. That's a lot of school 0:02:17.000 --> 0:02:20.799 really quickly. What were the career plans? Were you thinking academia? 0:02:20.880 --> 0:02:21.959 Were you thinking finance? 0:02:22.600 --> 0:02:25.320 I was actually kind of not quite sure, So I 0:02:25.360 --> 0:02:28.520 got in. I went to Chicago Grad School for PhD 0:02:29.720 --> 0:02:33.480 in the now Booth School of Business. Because I had 0:02:33.680 --> 0:02:36.519 learned a little bit about finance. I took an independent 0:02:36.560 --> 0:02:40.360 study from Carl Christ who's a famous econometrician at Johns 0:02:40.360 --> 0:02:46.080 Hopkins when Gene Fama's book Foundations of Finance had just 0:02:46.160 --> 0:02:48.560 come out. In fact, I literally worked in the college 0:02:48.560 --> 0:02:51.679 bookstore part time, and I remember unpacking the box to 0:02:51.800 --> 0:02:54.320 have this Fama book, and so I immediately bought one, 0:02:54.440 --> 0:02:56.160 and you know, I was going to do this independent 0:02:56.200 --> 0:02:58.400 study and read through. And by the way, it really 0:02:58.480 --> 0:03:01.760 is some books often called Foundations of Blank. It really 0:03:01.800 --> 0:03:04.800 was Foundations, right, you know, it was the It was 0:03:04.800 --> 0:03:07.560 a summary in the nineteen seventy six, right, very early days. 0:03:09.600 --> 0:03:11.920 And so Carl christ had said, well, you should think 0:03:11.919 --> 0:03:16.520 about Chicago. That's a powerhouse place for finance. And so 0:03:16.600 --> 0:03:19.640 I started studying finance there and passed the prelam which 0:03:19.680 --> 0:03:23.280 is no which is no small feat that's very selective. 0:03:23.960 --> 0:03:29.800 And then I got interested in behavioral science because finance 0:03:29.919 --> 0:03:33.520 was really obsessed with market efficiency and you know, there 0:03:33.560 --> 0:03:37.040 was no behavioral science, behavioral finance in sight at that time. 0:03:37.040 --> 0:03:38.840 But there were other folks at at Chicago. 0:03:39.120 --> 0:03:42.840 Well, if I recall correctly, Dick Taylor was there early 0:03:42.920 --> 0:03:47.680 in the behavioral finance or did he end up there later, Yeah, 0:03:47.720 --> 0:03:48.360 he came later. 0:03:48.480 --> 0:03:51.200 He came later. So when I came in the late seventies, 0:03:53.200 --> 0:03:56.960 a lot of Nobel Prize winners were their Fama, Miller Shoals. 0:03:57.520 --> 0:04:00.320 I think Fisher Black might have just left for MIT 0:04:00.720 --> 0:04:05.280 and when I came, but it was pre Andre Schleifer 0:04:05.360 --> 0:04:08.720 and Ravishni who did a lot of interesting behavioral finance, 0:04:08.720 --> 0:04:11.840 and then Dick Taylor came, I think around nineteen ninety five, 0:04:11.920 --> 0:04:12.440 nineteen to. 0:04:12.320 --> 0:04:15.480 Six, and you were at cal Tech by then, right, 0:04:15.640 --> 0:04:16.200 just correct? 0:04:16.200 --> 0:04:18.320 So yeah, so Dick and I had just passed like 0:04:18.360 --> 0:04:21.839 ships in the night, and I regard that sometimes not 0:04:21.920 --> 0:04:23.960 having just stayed and you know it's been part of 0:04:23.960 --> 0:04:25.039 a new vanguard. 0:04:26.320 --> 0:04:29.160 Well, but you actually are part of a new vanguard 0:04:29.160 --> 0:04:31.760 because the work you do in neuroeconomics, which we're going 0:04:31.839 --> 0:04:35.920 to get into, especially fMRIs and all the other things 0:04:35.920 --> 0:04:40.600 you've done more or less created that space. I mean, 0:04:40.600 --> 0:04:45.000 that's pretty foundational. Behavioral finance has a number of fathers, 0:04:45.040 --> 0:04:50.920 including Dick Taylor and Danny Kahneman. So well, let's circle 0:04:51.040 --> 0:04:53.960 back to the neuroeconomics in a little bit. But I 0:04:54.000 --> 0:04:58.160 want to ask what led you into decision making research? 0:04:58.200 --> 0:05:01.440 How did you find yourself taking the background you had 0:05:02.600 --> 0:05:08.479 in quantitative studies and your PhD and m b a 0:05:08.720 --> 0:05:10.200 and and go into decision making. 0:05:11.200 --> 0:05:13.880 So I some of it was when I was in 0:05:13.920 --> 0:05:16.719 college at Johns Hopkins. I studied physics and math that 0:05:16.800 --> 0:05:20.320 was too abstract, and number theory was just too mind blowing, 0:05:20.640 --> 0:05:22.200 you know for me, like I'm just not going to 0:05:22.560 --> 0:05:25.760 work at that level. And then I studied psychology and 0:05:25.800 --> 0:05:28.440 that seemed like just kind of a list of things 0:05:28.480 --> 0:05:31.880 that happened to people, but there was no unifying wish squishy. 0:05:32.520 --> 0:05:35.840 And then economics, which I really only took a little 0:05:35.839 --> 0:05:38.120 bit of a lot fewer than my peers I later 0:05:38.160 --> 0:05:41.040 competed with in grad school, was kind of in between, 0:05:41.080 --> 0:05:42.800 like the Three Little Bears, you know, there was I 0:05:42.839 --> 0:05:45.479 love that, and there was people right you know, physics 0:05:45.480 --> 0:05:47.479 didn't have people, psychology didn't have math. 0:05:47.600 --> 0:05:49.279 Economics was kind of the right mix. 0:05:49.160 --> 0:05:51.920 Exactly exactly. And I think a lot of a lot 0:05:51.920 --> 0:05:54.520 of social scientists may feel that way, and the people 0:05:54.520 --> 0:05:57.000 who like math lest stay in psychology or go to 0:05:57.040 --> 0:06:00.520 sociology or something where the mathematical structure isn't You found 0:06:00.520 --> 0:06:02.000 the canon and the foundation. 0:06:03.440 --> 0:06:06.159 So what led you into game theory? You end up 0:06:06.200 --> 0:06:08.880 writing a book Behavioral game Theory that was published in 0:06:09.160 --> 0:06:13.320 three How does that relate to economics and decision making 0:06:13.600 --> 0:06:14.520 and investing? 0:06:15.440 --> 0:06:18.960 So in graduate school when I pivoted away from finance, 0:06:19.560 --> 0:06:24.599 there was a couple of psychologist Hilly Einhern and Robin Hogarth, 0:06:24.600 --> 0:06:27.320 who were interested in judgment decision making. They were doing 0:06:27.400 --> 0:06:30.120 things very similar to konomen and diversity. It was sort 0:06:30.120 --> 0:06:34.200 of somewhat mathematical attempts to understand actual human decision making, 0:06:34.240 --> 0:06:38.200 not really stylized like Bay's rule and optimization. You know, 0:06:38.200 --> 0:06:40.279 those are good things to know, but they were interested 0:06:40.279 --> 0:06:43.520 in deviations from those and what that might tell us 0:06:43.520 --> 0:06:45.640 and what the practical value. So that's what I ended 0:06:45.680 --> 0:06:48.200 up doing in grad school. Game theory came a little 0:06:48.200 --> 0:06:52.920 bit later because at Chicago at that time, in the 0:06:53.040 --> 0:06:55.719 late seventies, there was hardly any interest in game theory 0:06:55.760 --> 0:06:59.520 for peculiar reasons. They were, you know, the economic world 0:06:59.600 --> 0:07:02.799 was dominated by price theory supplying demand like Gary Becker, 0:07:02.839 --> 0:07:04.640 you know, there was a lot going on. Game theory 0:07:04.720 --> 0:07:07.840 just was not flourishing there. But my first job was 0:07:07.839 --> 0:07:11.440 as an assistant professor in Northwestern and that happened to be, 0:07:12.280 --> 0:07:16.600 through just historical coincidence, a hotbed of great game theory. 0:07:16.640 --> 0:07:19.960 Paul Milgram was there, Banked Holmestrom was there, Robert Weber, 0:07:20.000 --> 0:07:25.000 who worked on lots of things on auction theory, Dave Barren, 0:07:25.000 --> 0:07:28.120 who was interested in political economy, and you know, political 0:07:28.160 --> 0:07:32.680 systems as games. So Milgrim and Holstrom went on to 0:07:32.720 --> 0:07:35.760 win Nobel prizes and went to other places. So it 0:07:35.800 --> 0:07:39.440 was sort of this incubator place that then, you know, 0:07:40.080 --> 0:07:44.280 like a incubator like Hewlett Packard and things like that, 0:07:44.320 --> 0:07:46.800 where people then went off to do other stuff. And 0:07:46.880 --> 0:07:49.400 so I basically learned game theory in my first job 0:07:49.440 --> 0:07:53.720 as AISTM professor, and that game theory is similar to 0:07:54.120 --> 0:07:57.800 behavor economics. The standard theory that everyone teaches in every 0:07:57.800 --> 0:08:03.200 introductory course is people arend and make the best choices 0:08:03.240 --> 0:08:05.920 given what they think others will do, and they're correct 0:08:06.040 --> 0:08:08.600 guessing about what others do. Like a bunch of people 0:08:08.600 --> 0:08:10.760 who played poker with each other, you know, every Friday 0:08:10.840 --> 0:08:13.640 night for decades, they kind of know what the tells are. 0:08:14.520 --> 0:08:17.280 But we were interested in what happens before you get 0:08:17.280 --> 0:08:19.760 to this kind of what's called Nash equilibrium, you know, 0:08:19.760 --> 0:08:21.880 where everyone is guessed correctly what everyone's going to do. 0:08:23.080 --> 0:08:25.000 And so to me, there was a huge room for 0:08:26.200 --> 0:08:29.760 understanding the psychology of strategic thinking in game theory. 0:08:29.880 --> 0:08:33.880 So that's really interesting to me. I always found the 0:08:33.920 --> 0:08:40.400 traditional economic homo economists of humans as rational calculating profit 0:08:40.520 --> 0:08:46.480 maximizing actors is just complete contradiction of real life experience. 0:08:47.240 --> 0:08:50.559 How did you go from your initial interest in behavioral 0:08:50.600 --> 0:08:57.400 finance into neuroeconomics, where you're looking at the biological underpinnings 0:08:57.559 --> 0:08:59.960 of what happens as people make decisions. 0:09:00.360 --> 0:09:02.439 Yeah, So the neuroeconomics to me was sort of a 0:09:02.520 --> 0:09:06.000 natural extension of behavior economics, which was We're going to 0:09:06.360 --> 0:09:09.200 grab from any interesting data and different ways of thinking 0:09:09.200 --> 0:09:12.600 about humans outside of standard economics and kind of pull 0:09:12.600 --> 0:09:14.640 it in and try to, you know, generate a kind 0:09:14.640 --> 0:09:16.959 of hybrid. It was almost like an import export business. 0:09:16.960 --> 0:09:19.520 And I'm going to import some psychology or dicktale or 0:09:19.559 --> 0:09:22.240 imported from konomon and what is this going to tell 0:09:22.280 --> 0:09:25.320 us about fairness and reference points and loss aversion what 0:09:25.440 --> 0:09:27.680 have you? And neur economics seem to me like just 0:09:27.720 --> 0:09:30.720 another thing to do. Part of it is my personality 0:09:30.760 --> 0:09:33.840 is kind of like intellectual entrepreneurship. So I liked you know, 0:09:33.880 --> 0:09:36.160 doing different things. You know, over the years, I've worked 0:09:36.200 --> 0:09:38.800 on lots of different methods and with different groups of people, 0:09:38.840 --> 0:09:41.720 and neureconomics was just a chance to do something even 0:09:41.760 --> 0:09:43.559 more dramatic. 0:09:43.440 --> 0:09:48.160 And tell us about your patent on active learning decision engines. 0:09:48.600 --> 0:09:49.960 What on earth is that? 0:09:50.200 --> 0:09:53.280 So active learning is the commuter scientist term is sometimes 0:09:53.280 --> 0:09:57.120 called dynamic adaptive learning. For basically, like if I was 0:09:57.160 --> 0:10:00.440 going to try to figure out how much you like risk, 0:10:00.600 --> 0:10:03.160 like you were a client, and if a financial advisor 0:10:03.240 --> 0:10:06.080 is asking you know, I might start by saying, well, 0:10:06.440 --> 0:10:09.200 here's a portfolio. Is this too risky or not risky enough? 0:10:09.720 --> 0:10:12.040 And if you say, nah, that's not risky enough, you know, 0:10:12.080 --> 0:10:14.960 I'd rather go for more, and then I would give 0:10:15.000 --> 0:10:16.680 you a better one that's a little has a little 0:10:16.720 --> 0:10:19.960 more risk in it. And in chemistry it's called titration. 0:10:20.320 --> 0:10:22.760 You know, you kind of change the mixture of the chemicals. 0:10:23.280 --> 0:10:25.840 And so for each person, you're asking them a dynamic, 0:10:26.760 --> 0:10:29.360 customized set of questions to get to the best answer 0:10:29.400 --> 0:10:33.200 as quickly as possible, and that's called active learning. So 0:10:33.360 --> 0:10:36.560 one of my colleagues at Caltech at that time, Andreas Krauss, 0:10:37.160 --> 0:10:39.920 was studying he was a Gonna Better scientist. So they're 0:10:39.960 --> 0:10:43.160 always on the frontier of how to get the truth 0:10:43.480 --> 0:10:47.600 faster and subject to computational constraints, like you know, because 0:10:47.600 --> 0:10:50.360 sometimes it's not just a question. I'm getting there, but 0:10:50.480 --> 0:10:52.200 can you do it in real time so you don't 0:10:52.240 --> 0:10:55.040 have to wait half an hour, you know, to ask 0:10:55.080 --> 0:10:59.320 the next highly unformative question. And so the patent was 0:10:59.360 --> 0:11:02.800 just a a method that Andreas and another guy who 0:11:02.920 --> 0:11:05.480 now works like Google, I believe Daniel Goldman and me 0:11:05.800 --> 0:11:09.520 had worked on to apply this in a particular way. 0:11:09.600 --> 0:11:11.480 And so it was basically a software pattern. There was 0:11:11.480 --> 0:11:13.080 an Advazon pattern on an algorithm. 0:11:13.240 --> 0:11:17.440 So you're asking people questions, how do you know they're 0:11:17.480 --> 0:11:21.600 giving you honest answers? And I asked that question for 0:11:21.720 --> 0:11:24.680 very specific reasons that will be evident in a moment. 0:11:25.080 --> 0:11:27.679 How do you know the answers are legitimate? 0:11:27.760 --> 0:11:31.800 Okay, So in experiment economics, one of the main rules, 0:11:31.840 --> 0:11:35.120 like a commandment, is we almost always pay people unless 0:11:35.160 --> 0:11:37.400 we can't, like with children sometimes or what have you. 0:11:38.040 --> 0:11:40.360 We almost always pay people money or something we know 0:11:40.400 --> 0:11:43.440 they value based on the decisions they made. So when 0:11:43.480 --> 0:11:46.640 we do these kind of risk assessments, again not with clients, 0:11:46.640 --> 0:11:48.760 but say in a simple experiment for modest amounts of 0:11:48.760 --> 0:11:51.280 money twenty bucks, fifty bucks, what we'll do is we 0:11:51.360 --> 0:11:52.719 say at the end, we're going to pick one of 0:11:52.760 --> 0:11:54.280 the things you said you wanted, and we're going to 0:11:54.320 --> 0:11:57.840 actually play that for money. And if you if you 0:11:57.840 --> 0:11:59.280 don't tell us what you really wanted, you're gonna get 0:11:59.320 --> 0:11:59.880 stuck with something. 0:12:00.600 --> 0:12:04.280 So you're creating an incentive for them to be somewhat honest. 0:12:04.520 --> 0:12:04.800 Correct. 0:12:05.040 --> 0:12:08.160 The reason I ask we're recording this about two weeks 0:12:08.200 --> 0:12:12.000 before the twenty twenty four presidential election. I wrote something 0:12:12.000 --> 0:12:16.000 a month ago about why polling errors are really a 0:12:16.040 --> 0:12:20.080 behavioral problem because when you ask people who you're going 0:12:20.160 --> 0:12:22.640 to vote for, what you're really asking is not just 0:12:22.720 --> 0:12:25.560 their preference, but hey, you're gonna get your lazy butt 0:12:26.000 --> 0:12:27.920 off the couch and go to the library and vote. 0:12:28.280 --> 0:12:31.520 And I assumed, hey, there's an era of five, six 0:12:31.640 --> 0:12:34.360 seven percent built into that, and that's why polls are 0:12:34.360 --> 0:12:39.679 so bad. Researching your work about hypothetical bias, I was 0:12:39.800 --> 0:12:43.160 shocked the data that you came is when you ask 0:12:43.200 --> 0:12:45.600 people if they're going to vote, about seventy percent say 0:12:45.640 --> 0:12:49.360 they will, In reality, just forty five percent of them do. 0:12:49.840 --> 0:12:54.000 That's a massive error of twenty five percent. What value 0:12:54.040 --> 0:12:56.520 is there in polls when people have no idea what 0:12:56.559 --> 0:12:57.360 they're really going to do? 0:12:57.960 --> 0:13:00.360 Yeah, So I mean I think the best post are 0:13:01.280 --> 0:13:03.480 know that, and so they try to phrase the question 0:13:03.640 --> 0:13:06.560 or gather some other data. But this is often called 0:13:06.559 --> 0:13:09.760 acquiescence or yes bias. Right, so you say, people, are 0:13:09.800 --> 0:13:11.840 you planning to vote? Oh, yeah, I'm planning to vote, Well, 0:13:11.880 --> 0:13:13.320 you're going to Are you going to not vote because 0:13:13.320 --> 0:13:14.840 it's too Yeah, I may not vote. 0:13:14.880 --> 0:13:16.880 What happens if it rains, what happens if you're busy. 0:13:17.520 --> 0:13:19.040 So you can often get numbers in it up to 0:13:19.040 --> 0:13:21.199 more than one hundred percent. Are you having to vote? No, 0:13:21.360 --> 0:13:24.240 you have seventy percent. Yeah, I probably won't vote fifty 0:13:24.240 --> 0:13:26.280 five percent. That's one hundred and twenty five percent current. 0:13:26.440 --> 0:13:29.880 The math doesn't math, and you see it. Particularly. One 0:13:29.880 --> 0:13:32.640 of the things we study was product purchases. So when 0:13:32.640 --> 0:13:34.599 you show people new products and say, you know, you 0:13:34.600 --> 0:13:37.000 think you'd be interested in this, you get way too 0:13:37.040 --> 0:13:40.240 many yes's. And that's one recent new products fail. It 0:13:40.320 --> 0:13:43.559 is because somebody who's the product champion inside the firm, 0:13:43.720 --> 0:13:45.960 like in a consumer products company, looks at this polling 0:13:46.040 --> 0:13:48.440 date and says, see, see you know, give me money 0:13:48.480 --> 0:13:51.960 to roll this out in a test market. So what 0:13:52.040 --> 0:13:53.680 one of the things we have done is to try 0:13:53.720 --> 0:13:56.200 to see if we didn't, we'd wro'te a few papers 0:13:56.240 --> 0:13:58.480 on this, but I don't feel like we exactly crack 0:13:58.559 --> 0:14:01.920 the nut was to see if a combination of what 0:14:02.000 --> 0:14:04.600 people look at if you measure where their eyes are looking, 0:14:05.320 --> 0:14:08.080 like how often they look back and forth between a 0:14:08.120 --> 0:14:12.160 price and a product, and maybe brain signals can help 0:14:12.240 --> 0:14:14.720 us predict when they say, yeah, I'm going to vote, 0:14:14.760 --> 0:14:15.800 are they really going to vote or not? 0:14:16.160 --> 0:14:20.800 And neuroeconomics, as as I've learned about it through you 0:14:21.480 --> 0:14:24.080 is you're putting people in a functional MRI machine. You're 0:14:24.120 --> 0:14:27.360 asking them a series of questions, and you're identifying what 0:14:27.560 --> 0:14:29.800 parts of the brain are actually lighting up. 0:14:29.920 --> 0:14:32.480 Correct exactly so that so and and by the way, 0:14:32.480 --> 0:14:37.480 ephraimri is glamorous and fantastic, but there's lots of other 0:14:37.480 --> 0:14:39.640 methode that they are used as well. You know, it's 0:14:39.720 --> 0:14:42.760 unnatural because people are in this tube. It's very loud. 0:14:43.120 --> 0:14:44.840 You know, if you want to study a phobic, if 0:14:44.840 --> 0:14:46.880 you want to study close to probi, you cannot, you know, 0:14:46.920 --> 0:14:50.240 because the closer roobics won't go in there. But it 0:14:50.280 --> 0:14:52.240 does give you a picture of the whole brain. And 0:14:52.280 --> 0:14:55.560 in the in the case of the we need some 0:14:55.600 --> 0:14:58.320 experiments where we show people to consumer good and in 0:14:58.360 --> 0:15:00.920 one condition. The first part of the experiment we say, 0:15:01.040 --> 0:15:02.880 you don't have to actually buy this, but just tell us, 0:15:02.920 --> 0:15:04.640 you know if it was on sale for this price, 0:15:04.800 --> 0:15:08.560 like yes, no, strong, yes, weak yess. So we get 0:15:08.560 --> 0:15:11.560 a four point scale and then we surprise them and say, 0:15:11.760 --> 0:15:13.640 now we're going to show you some different products and 0:15:13.720 --> 0:15:15.840 these are going to actually buy. So if you say 0:15:15.920 --> 0:15:18.560 yes and we choose that one out of this. 0:15:18.760 --> 0:15:19.720 Bin, you get it. 0:15:19.960 --> 0:15:21.960 You have you have to buy it. Well, we give 0:15:22.000 --> 0:15:24.040 you some money and we're going to take the price 0:15:24.080 --> 0:15:27.080 out and give you the residual money and the product, 0:15:27.120 --> 0:15:29.160 and you're going to leave here with this product or 0:15:29.240 --> 0:15:30.920 I think some of them we have we mail it 0:15:30.960 --> 0:15:33.600 to them at Amazon is something we actually had products 0:15:33.640 --> 0:15:37.680 there in a box. And so the question is what's 0:15:37.720 --> 0:15:40.240 going on in the brain when they're seriously thinking about 0:15:40.280 --> 0:15:44.240 buying something for real versus hypothetical, which is like a survey, right, 0:15:45.400 --> 0:15:47.520 And what we found was the tricky part is to 0:15:48.280 --> 0:15:53.080 predict when people say yes, hypothetical, but the brain says no, 0:15:53.880 --> 0:15:55.360 you know, can you can you see a brain? 0:15:55.600 --> 0:15:56.520 Can you identify that? 0:15:56.920 --> 0:15:58.840 Uh? Modestly well? 0:15:59.160 --> 0:15:59.280 Right? 0:15:59.520 --> 0:16:02.920 And it turns out the most. There's two interesting markers. 0:16:02.960 --> 0:16:05.960 One is there's a very old area in the brain, 0:16:06.040 --> 0:16:10.440 old you know, evolutionary yes, called the mid brain, which 0:16:10.480 --> 0:16:14.280 is actually where all of the dopamina drid neurons live 0:16:14.600 --> 0:16:18.240 and then and then connect to middle areas of the 0:16:18.240 --> 0:16:21.440 brain called basoganglia that are kind of computing reward and value. 0:16:21.480 --> 0:16:24.200 And then frontal cortex, which is really putting together the 0:16:24.240 --> 0:16:27.560 modern the modern exactly like it's like a thinking cap 0:16:27.600 --> 0:16:30.960 on top of the monkey brain. And in the mid 0:16:31.000 --> 0:16:36.600 brain there's a stronger signal when they say yes and 0:16:36.640 --> 0:16:40.720 they actually do do yes hypothetical and it's a yes reel. 0:16:41.080 --> 0:16:43.800 There's a stronger signal than when they say yes hypothetical 0:16:44.080 --> 0:16:47.080 no real. So it's almost like way upstream in the 0:16:47.080 --> 0:16:52.680 brain if if if in that region they say yes, 0:16:52.720 --> 0:16:55.640 I'm gonna buy it hypothetically, there's enough activity they're gonna 0:16:55.640 --> 0:16:55.920 buy it. 0:16:56.400 --> 0:16:59.880 So my general sense of this, and I'm curious as 0:17:00.080 --> 0:17:03.240 to how you what the reality is, my sense of 0:17:03.280 --> 0:17:06.840 it is, on the one hand, people are social animals 0:17:06.840 --> 0:17:09.919 and they want to be agreeable and say yes to people. 0:17:10.680 --> 0:17:13.040 On the other hand, we really don't know what the 0:17:13.080 --> 0:17:16.359 hell we want, especially if you're talking about something six 0:17:16.440 --> 0:17:19.760 months from now. I guess the tricky part is how 0:17:19.760 --> 0:17:22.680 do you get people in MRI machines when you have 0:17:22.720 --> 0:17:24.639 a question for them. We can't even get people to 0:17:24.640 --> 0:17:28.240 pick up their phone to answer polls. How difficult is 0:17:28.280 --> 0:17:31.399 it to get subjects to go through this process? Or 0:17:31.400 --> 0:17:34.840 are these all mostly undergraduates and you know their lab rats? 0:17:34.880 --> 0:17:35.680 You can do whatever you want. 0:17:35.760 --> 0:17:39.600 Some of them are undergraduates, although in Caltech they're very 0:17:39.680 --> 0:17:43.640 unusual human beings because they're actually useful. They're very useful 0:17:44.119 --> 0:17:47.120 labrats who pay for economics because the media and matthe 0:17:47.240 --> 0:17:51.639 Is et Is eight hundred, they're the most mathematically skilled except. 0:17:51.400 --> 0:17:52.879 For that's a perfect score, isn't it? 0:17:52.920 --> 0:17:55.480 Like exactly, that's the perfect score, like Harvey mud Mit. 0:17:55.640 --> 0:17:58.800 There are other places that have, you know, similarly hyper 0:17:58.800 --> 0:18:02.479 analytical kids. So if like if they can't do something 0:18:02.600 --> 0:18:07.359 like a computation easily, nobody can. So it's very useful 0:18:07.520 --> 0:18:10.720 establishing like bounds on rationality. You know that people. We 0:18:10.760 --> 0:18:13.320 often get critiques like well, you wouldn't get bubbles if 0:18:13.359 --> 0:18:16.240 people were smart enough, Like, well, we have the smartest 0:18:16.240 --> 0:18:17.320 people and you get bubbles. 0:18:18.520 --> 0:18:20.879 It's got less to do with the frontal cortex and 0:18:20.920 --> 0:18:24.199 intelligence and everything with that, the limbic system and the 0:18:24.240 --> 0:18:25.000 lizard brain back. 0:18:25.080 --> 0:18:27.120 Yes, exactly, so they have they have all the things 0:18:27.200 --> 0:18:29.000 in the brain they have, they have other skills that 0:18:29.040 --> 0:18:33.479 are cordically expressed. But so in a lot of these 0:18:33.600 --> 0:18:36.440 MRI studies we also use. We work pretty hard actually 0:18:36.440 --> 0:18:39.960 to get regular folks from the community who and who 0:18:40.200 --> 0:18:42.800 you know are different ages. You know, we we don't 0:18:42.800 --> 0:18:45.359 really have a representative sample, although you could, you could 0:18:45.359 --> 0:18:49.200 try to get pretty close in southern California, and then 0:18:49.240 --> 0:18:51.399 we we we almost always never do a study this 0:18:51.680 --> 0:18:54.600 just take outing undergrads because we worry about the robustness 0:18:54.600 --> 0:18:57.080 across right, It is true in the case of something 0:18:57.200 --> 0:18:59.960 like trying to get brain signals to break when people 0:19:00.000 --> 0:19:03.560 will actually buy products. The other type of study we've 0:19:03.640 --> 0:19:05.719 used to involves eye tracking and things like that, and 0:19:05.760 --> 0:19:09.480 it turns out that when when you ask people hypothetical questions, 0:19:09.520 --> 0:19:11.119 would you buy that? You don't really have to buy this, 0:19:11.240 --> 0:19:13.320 but would you, they just don't look at the price 0:19:13.359 --> 0:19:16.960 that much, and when they're really shopping, they really look 0:19:17.000 --> 0:19:19.840 at the price. So one way to tell whether people 0:19:19.880 --> 0:19:23.720 are being serious in expressing a genuine what I'm going 0:19:23.720 --> 0:19:26.320 to really do, it is just something like how much 0:19:26.359 --> 0:19:28.080 time they spend looking at the price and looking back 0:19:28.119 --> 0:19:31.520 and forth. And there may be other like if if 0:19:32.840 --> 0:19:35.280 if it was consumer products company was trying to use 0:19:35.560 --> 0:19:38.600 FROMRI or other methods. There are others that are much 0:19:38.640 --> 0:19:41.520 more portable, like EEG, and you can get a pair 0:19:41.520 --> 0:19:43.919 of glasses you walk around and it, you know, it 0:19:44.000 --> 0:19:46.159 records where your eyes looking. So there are there are 0:19:46.160 --> 0:19:49.119 things you can do outside of the confines of a 0:19:49.119 --> 0:19:52.320 campus lab. I think we would just look for things 0:19:52.359 --> 0:19:56.800 that are that are easy, easily seen biomarkers of this 0:19:56.960 --> 0:19:59.400 mid brain activity of f MRI, because we're never gonna 0:19:59.400 --> 0:20:01.320 be able to do that, you know, at scale in 0:20:01.400 --> 0:20:03.000 a shopping mall or something. 0:20:03.160 --> 0:20:04.919 So let's go through each of these. We know what 0:20:05.119 --> 0:20:10.280 fMRI is, right, you're in an MRI machine. EEG and SCR. 0:20:10.400 --> 0:20:11.359 Tell us what those do. 0:20:11.680 --> 0:20:15.280 So eg's electro and cephalography and it's basically all the 0:20:15.320 --> 0:20:19.760 little things on your electrodes. If your ball like me, 0:20:19.880 --> 0:20:23.359 that's good for seasons. You know, if you're a supermodel 0:20:23.359 --> 0:20:27.400 with big puffy Texas beauty pageant hair, then no good, 0:20:27.520 --> 0:20:27.880 no good. 0:20:28.520 --> 0:20:31.560 So you're measuring electrical activity in the brain, and you 0:20:31.600 --> 0:20:35.240 could really specify where it is by you know, just 0:20:35.640 --> 0:20:39.240 triangulating with all the different leads that you have spect. 0:20:39.200 --> 0:20:42.240 Exactly so the you know, you can put sixteen to 0:20:42.280 --> 0:20:45.280 one hundred and twenty eight different electrodes. The signals are 0:20:45.440 --> 0:20:48.840 very weak, but the advantage of EG is it's really fast. 0:20:49.280 --> 0:20:51.399 So if you want to study something like thinking fast 0:20:51.440 --> 0:20:53.359 and slow, you know, like if I show you a 0:20:53.359 --> 0:20:55.600 picture of a person, you have a snap reaction that 0:20:55.680 --> 0:20:57.800 they're scary or they're someone you want to vote for, 0:20:58.359 --> 0:21:01.359 then FRI is too slow because it measures these blood 0:21:01.359 --> 0:21:03.480 flow signals that take like one or two seconds to 0:21:03.520 --> 0:21:03.920 show up. 0:21:04.240 --> 0:21:08.400 But like one, one or two seconds is too slow for. 0:21:08.920 --> 0:21:11.959 You know, a lot is going on in the in 0:21:12.320 --> 0:21:14.040