We want to hear from you! If you are enjoying these podcasts, please let us know. Email our editor with your comments, suggestions, recommendations, and thoughts about the discussion.
In this episode of The Decision Corner, Brooke Struck invites Erez Yoeli to share his insights on how people tick. Dr. Erez Yoeli is a research associate at MIT’s Sloan School of Management and co-director of the Applied Cooperation Team (ACT). ACT is a team of researchers that applies insights from the social sciences towards increasing contributions to real-world public goods.
Erez designs and tests large-scale interventions to promote altruistic behaviors such as charitable donations, volunteering, resource conservation, and medication adherence. He has worked as a researcher at Harvard University’s Program for Evolutionary Dynamics and Yale University’s Human Cooperation Laboratory. He holds a Ph.D. in Economics and an MBA from the Booth School of Business at the University of Chicago. In this episode, we discuss:
- What motivates people to do the right thing
- How group behavior works
- The difference between reputation and identity
- High and low tech solutions to problems of collective action
- The importance of speaking to communities in their own language (sometimes literally)
- What sets our current pandemic response apart from those of the past
- The unique power of social norms
- The meaning of community, and its unique role in mediating motivation
Human Motivation and Social Consequences
“Humans are particularly good at figuring out if they’re expected to do something. And if they don’t, they’ll suffer from social sanctions. And they’re also very good at socially sanctioning. They’re good at keeping track of who deserves to be sanctioned. They’re good at even keeping track of who didn’t sanction when they were supposed to sanction. It seems to be something that’s an innate feature of our psychology. And it’s one that actually differentiates us from other mammals and other primates. It’s one that makes us unique, and it makes, a lot of folks have argued that it is one of the things that undergirds our ability to cooperate at such a large scale, and such a larger scale than you see amongst other mammals and especially other primates.”
On Social Norms
“So it’s this norm enforcement psychology, which consists of the idea of, the key features of which are figuring out which behaviors are expected of you, figuring out who should be sanctioned, and figuring out even keeping track of those who sanction and those who don’t sanction, and sanctioning them also. That’s what I mean when I talk about norms.”
The Key Driver of Human Behavior
“Now if you ask me what I actually believe is going on, I actually believe that the key thing driving a lot of our being good is this reputation stuff. There are some places in which that worldview would deviate from a worldview where that’s a little bit more there are platonic goods and people are coded with some, or however it is that you might view the world that reputations are not the only thing that’s going on.”
The Importance of Community
“Shared bonds are usually useful beachheads for building meaningful relationships. At the end of the day, you build meaningful relationships based on the current conditions. Those beachheads help. They give people a little bit of traction. They give people something to start talking about. They often make it more intuitive to be closer to someone, and they’re important. But if you don’t have a real reason to be good to each other, then the beachheads won’t end up helping that much. I think the way that I view it is how important is this? Well, it can help a lot. It can really help smooth things over, but on its own it’s not enough.”
How to Make Effective Change
“When we were asking people to wear masks, we did it in a way that wasn’t culturally appropriate and sensitive to their sets of values and their priorities. That backfired on us. In the project that I told you about earlier where we were working with folks in Sub-Saharan Africa to promote medication adherence for tuberculosis. That’s another case where a bunch of outsiders trying to help folks who we don’t really know that well. And we knew immediately we needed to be teaming up with a local team to actually be able to make any progress.”
Brooke Struck: Hello everyone. And welcome to the podcast of The Decision Lab, a socially conscious applied research firm that uses behavioral science to improve outcomes for all of society. My name is Brooke Struck, research director at TDL, and I’ll be your host for the discussion. My guest today is Erez Yoeli, researcher at the MIT Sloan School of Management. In today’s episode, we’ll be talking about reputation, doing the right thing, and motivating collective action. Erez, thanks for joining us.
Erez Yoeli: Of course. Thanks for having me.
Brooke Struck: So collective action is always a challenge. It’s especially pressing these days in contexts such as this pandemic that we’re all living because everybody’s health is so dependent on the actions of people around us. Why is collective action such a critical thing, even beyond these contexts of pandemics?
Erez Yoeli: I think it’s fair to say that basically in every domain that you look, collective action arises to some extent. I mean, today I spent the morning working on a sustainable investment project. So if you’re thinking about how people invest their money, they could be investing their money in things that are more or less good for society. They could be attending to the fact that the investments they’re making are harming society, or they could be not attending to that. Earlier today, I was working on a different project related to medication adherence, which is a totally different domain of public health closer to the pandemic, where we were talking about people taking the medication so that others don’t get sick. We have projects in philanthropy and in volunteering, which are obviously pro-social. We have projects related to food sharing and reducing food waste or reducing waste in water.
Erez Yoeli: We have a really interesting project with Connecticut, where they have a bunch of fairly wealthy towns where people water their lawns too much. And that creates all sorts of problems and they can’t really get through to them because they’re a bunch of rich hedge fund owners whose lawns look like golf courses, and they don’t care that their water bill is 10,000 bucks. How do we get them to actually pay attention so that they stop ruining the waterways in Connecticut inadvertently, when what they really think that they’re doing is making their yard look beautiful for their neighbors, right?
Erez Yoeli: It just seems to show up in so many different domains. It’s small stuff like in the office, getting people to clean up after themselves at the office sink and big stuff like some of the problems I just mentioned or the pandemic. It’s all around us.
Brooke Struck: So the conditions in which we find these problems are ones where generally we’re talking about a communal good, right? Something that we all get to enjoy collectively, but where the actions of plus or minus one person participating in supporting that communal good or maintaining it doesn’t tend to make too much of a difference. So it’s kind of easy and tempting to cheat because if you don’t have to do your part to do the maintenance, you kind of get to avoid the cost, but enjoy the benefit anyway. Are there any more criteria that we should be looking for in trying to identify the circumstances in which these collective action problems arise?
Erez Yoeli: The example you just gave is one particular kind of collective action problem that arises, and to make this fairly stark distinction between ones that have the features that you’re describing and maybe others that don’t. If you look at other fields, that distinction ends up being somewhat less important. And what they focus on instead, and to be honest, this is more of the way I’ve come to think of things, they tend to focus on whether the action you’re taking is somewhat costly to you and somewhat beneficial to society or to others around you in some way. So it’s actually a fairly broadly defined thing. It could be that it has this communal good aspect like you’re describing, where the costs and benefits are both relatively low. But it could be something else as well. It could be a host of other kinds of behaviors. As long as they have the feature that they’re costly to me, beneficial to others, then they kind of end up falling in this set of behaviors we think of as being cooperative or candidates for using norms to promote them, which is the kinds of stuff I work on.
Brooke Struck: And in your work, one of the other features that you talked about, just so we get that out on the table and we can discuss sort of where it falls in our conversation as we move forward, is the idea that lots of people need to participate to reach a threshold for the good to kind of be delivered. Do you want to say a little bit more about that before we launch any further?
Erez Yoeli: I wouldn’t say that every problem I work on has that feature. So take for instance philanthropy. There are some philanthropic projects where if you don’t get past the threshold, there’s no point in doing the project. You just don’t have enough funds to go build a building or whatever. But there are some philanthropic projects where that’s not the case. So suppose to what you’re doing is instead of building a building, you’re providing deworming medication? Well, deworming medication if you have $10, you can buy a certain amount and if you have $20, you can buy twice that amount. So in those cases, you’re still doing this costly thing that’s benefiting others. You’re giving money for the benefit of somebody else. It’s still a behavior that we can think of as being a pro-social behavior and can think of as the kind of behavior that we want to encourage. But it doesn’t necessarily have this key feature of a threshold.
Erez Yoeli: The threshold things can matter, right? It can be important that this is something that you kind of need to get a lot of people on board for, and that can make it a harder behavior to push. It’s not necessary for it to be a behavior we focus on.
Brooke Struck: So some will be characterized by this non-linearity, but not necessarily all?
Erez Yoeli: Yeah. Yeah, definitely.
Brooke Struck: You’ve done a little bit of work on this I hear. And in fact, I gather that you’ve put together a bit of a model for how it is that we can more effectively promote effective collective action. Can you tell us a little bit about the model that you’ve put together?
Erez Yoeli: Yeah. The way I think about this is, and I admit it’s not very original. I’m definitely standing on the shoulders of a lot of folks is what I describe here. But I think of this as a problem related to norms. And that word is one that has lots of different meanings. So let me explain what I mean.
Erez Yoeli: Humans are particularly good at figuring out if they’re expected to do something. And if they don’t, they’ll suffer from social sanctions. And they’re also very good at socially sanctioning. They’re good at keeping track of who deserves to be sanctioned. They’re good at even keeping track of who didn’t sanction when they were supposed to sanction. It seems to be something that’s an innate feature of our psychology. And it’s one that actually differentiates us from other mammals and other primates. It’s one that makes us unique, and it makes, a lot of folks have argued that it is one of the things that undergirds our ability to cooperate at such a large scale, and such a larger scale than you see amongst other mammals and especially other primates.
Erez Yoeli: So it’s this norm enforcement psychology, which is comprised of the idea of, the key features of which are figuring out which behaviors are expected of you, figuring out who should be sanctioned, and figuring out even keeping track of those who sanction and those who don’t sanction, and sanctioning them also. That’s what I mean when I talk about norms. I talk about that psychology.
Erez Yoeli: What we believe is that a huge chunk of good behaviors are motivated by this kind of psychology. The idea that what I want is to be seen as a good person to meet those kinds of expectations. And that’s not something that necessarily I’m thinking about consciously each time I do something good. This whole subconscious thing going on of trying to do this, just kind of call reputation management. I start to believe in doing things that are good for the environment. I start to think that littering is really immoral and things like that.
Erez Yoeli: And the reason that these things arise in my conscious mind, that’s what I focus on. Littering is actually aversive to me. I would never litter. That’s actually gross to behave that way. I don’t think about reputation or norms when I think about littering, it’s just a gut reaction almost. But that gut reaction came from somewhere, and from a scientific standpoint, what we think is that it came from this concern for reputation, this subconscious concern for reputation. And that’s the model that we use once we have that perspective, that model spits out a bunch of really useful prescriptions basically. It gives you some guidance given that you think that this is a big part of why people are doing these good things. Okay, well then maybe what we can do is ensure that the correct features of the decision environment are in place in order to encourage those good things.
Brooke Struck: Norms are kind of what it’s all about, but reputation is what helps the rubber to meet the road there. So what we need to do is the kind of appropriate signaling around norms to get the reputation wheels turning. And then we’re all hamsters on our wheels and everyone’s off to the races.
Erez Yoeli: It’s just like three or four. But yeah. It’s definitely the right metaphors and the right description of what’s going on. I mean, basically we have to set expectations. We have to say this is something that’s going to be norm enforced. That’s going to be something that’s important. And then in addition to that, we have to make sure that it’s something that can be norm enforced. So we have to figure out what to ask. That’s actually not trivial because it turns out that the way social sanctions work, they involve some coordination. Anytime there’s coordination, that means that you have to basically generate consensus when somebody does the right thing versus not. Otherwise, somebody doesn’t want to sanction them when they think that they did the wrong thing because they’re not sure other people agree with them. So that creates this mess. So you have to be careful about what you ask for. And then you also have to make sure that people can see the desired behavior so that they can socially sanction or socially reward the desired behavior and so on and so forth. But you can see how that all pops out of this idea that these norms, and abiding by norms, and maintaining your reputation is really important.
Brooke Struck: So let’s dig into the three features that you’ve identified for what work we need to do around norms to really be able to leverage this reputational power. So from what I understand, the three key features are observability, eliminating excuses, and communicating expectations. Can you walk us through those three, one by one?
Erez Yoeli: So what I’ll do is I’ll tie it to the discussion we’ve been having. And I’ll give you some examples of each. So observability just means people should be able to find out about others’ good deeds. And the more likely they are to find out, the easier it is for them to socially sanction or socially reward. That’s the one that we just covered a moment ago. You can’t really have reputations without observability. So a major predictor here is that observability should have a reasonably large impact. And it does in practice. We’ve found that it’s sort of the lowest hanging fruit for encouraging prosocial behavior. And we always kind of start with observability and check to see whether we can increase observability in the setting that we’re working. What I think are illustrative examples.
Erez Yoeli: We worked with a power company, and we were recruiting for a program that helps prevent blackouts. You might even be familiar with these kinds of programs. Basically there’s a risk on hot days that our company will run out of power as people turn on all the ACs. When that happens, historically what power companies did was do rolling blackouts and rolling brownouts, which are terrible. Nowadays, we have better solutions. We have these radio switches where, if the power company is running out of power, they just hit the switch. The radio switches get the signal, and they know to turn the AC compressor on and off less often than it otherwise would’ve been. And each individual person basically just feels a one to three degree difference if they notice it at all. That ends up adding up if there’s tens or hundreds of thousands of these installed. So it’s a small cooperative action that you can take. Most of the cost comes from the fact that you are actually attending to the request from the utility to participate in the program and letting the installer come to your house to install it. The actual cost of participation is trivial once that’s done. So a nice prosocial behavior to look at.
Erez Yoeli: We started working with the power company and we found that they were recruiting for this program by sending out letters and having interested people call into a hotline. And our immediate reaction to that was, “Whoa, whoa, whoa. Where’s the observability?” These are people who are making a decision in private about something nobody will ever see. There’s zero kudos for doing a good deed here that’s helping to prevent a blackout. How can we change that? And what we ultimately came upon was a pretty low tech solution. We ended up asking people to sign up on signup sheets that we would post near the mailboxes in their buildings. And now you can see who signed up. You can see if your neighbor signed up. Your neighbor will be to see if you’ve signed up. That ended up tripling participation in this program. It just was much more effective. When we actually have a fairly, this was an experiment that we ran. So we have a fairly clean control where we have anonymous sign up sheets. So it’s the same method of signing up. People were much less likely to sign up on those. So that’s an example.
Erez Yoeli: Other examples are if you reward people for giving blood by listing their names in the local newsletter, they become more likely to give blood. If you move the donation box for a national park so that it’s more in view of the rangers, then people give more to the national park. If you post the names of people who are late in paying their taxes, they pay their taxes more quickly. There’s just another example that we did was we actually sent out a letter to people encouraging them to vote and told them, “Oh by the way, we might call you afterwards to find out how the experience was.” So again, sort of making it feel like somebody is actually attending to whether or not you vote. These are some of the examples that I like to give, but there’s actually dozens of examples like this in the field and in the laboratory. There’s even more examples where observability really has a very large impact on desired prosocial behaviors.
Brooke Struck: Okay. And moving on from observability, the next one is eliminating excuses. This is one that I’m very curious to hear about because humans of course are very ingenious in their ways of finding a way out.
Erez Yoeli: They sure are. There’s a great paper that illustrates this beautifully. There’s a bunch of guys that team up. Actually, excuse me, there’s two guys and a female researcher as well. So I shouldn’t say a bunch of guys. They team up. And they team up in particular with a Salvation Army outfit here in Boston. And they went to a supermarket, volunteers do what Salvation Army volunteers always do. Which is stand in front of the doors, ring their little bells, ding, ding, ding, ding, ding, and then solicit donations.
Erez Yoeli: What the researchers decided was that they would test different strategies for soliciting the donations and see which strategies encourage more donations versus less donations. They had some of the research teams sit in cars in the parking lot, and they would count the number of people who gave donations and the number of people who were going in and out of the grocery store. And one of the things that they varied amongst a handful of things was whether the volunteers stood in front of one door or two doors. And they found that when somebody was standing in front of just one door, then the customers of the supermarket would just go out the other door so they wouldn’t have to engage with the volunteer. So there was this nice excuse there of, “Oh no, I just wanted to pick up the chocolate from over there. They’ve got my favorite magazine over there. I wasn’t thinking about it. It’s closer to my car,” whatever. There’s tons of these excuses that people could give, but really what they were doing was avoiding the volunteers.
Erez Yoeli: So then the researcher said okay, fine. So when people are standing in front of both doors, it’s better. Right? And sure enough, that’s what they found. But they also found this other really weird result. That when they had people in front of both doors, folks stopped coming out of the supermarket at all. At first we were kind of puzzled by this, they’re just congregating in the back by the meat aisle, what’s going on? And then they eventually realized no, that’s not it. There’s a third door we didn’t know about. It’s this utility door on the side of the building. Most supermarket doors are these nice wide double doors. And this supermarket had those. But this supermarket also had one of these tiny little metal doors where you went to throw out the trash and the recycling. And now the customers were leaving from that door because they didn’t want to have to say no to the Salvation Army volunteers directly and wanted to have some sort of more plausible excuse. So yes, you’re absolutely right. Humans are super good at finding any possible excuse to leave on the table. And this study I think illustrates that beautifully.
Erez Yoeli: In general, what we find is that the more you lock down these excuses, the more likely you are to get the desired behavior. That if you give people lots of outs, what’s going on there to tie this back to the reputations thing that we were talking about earlier, what’s going on there is that when there are lots of plausible excuses, then the link between the action and the reputation is weaker. Because if you don’t do the desired action, there’s lots of plausible excuses you could give for that. Then you don’t take as much of a reputational hit.
Erez Yoeli: And this is coming from that thing that I said earlier about how social sanctions involve an element of coordination. What a plausible excuse really is, is an excuse where even I as somebody who might sanction, I might know that that excuse is BS. That really the person, they meant to avoid this behavior, they meant to avoid the Salvation Army volunteer. But I don’t know that others can tell that. There’s too much context that you need in order to know that. And I can’t communicate that context in a practical way. So it undermines my ability to sanction because while I may have wanted to sanction, then I’ll look like a jerk for sanctioning wrongly and I’ll get sanctioned. So the thing that we say is you have to generate consensus when the person fails to do the good deed. And that’s equivalent to saying you have to take away all of those possible excuses. You have to eliminate that third door that was there in the supermarket. You have to stand in front of all of the doors and so on. And I’m happy to give you more examples of how we do this in practice so far. I’ve focused on a couple of illustrative ones, but you tell me-
Brooke Struck: Yeah I was going to ask you, are there certain categories of outs that we should be keeping in mind when we’re trying to seal those third doors?
Erez Yoeli: Yeah, that’s a good question. It’s funny. I think the answer might be no. At least the way that I think about it is it’s very context specific. So I’ll give you another example to sort of illustrate that.
Erez Yoeli: We were working on this digital health platform that I alluded to at the very beginning of our discussion. It’s a platform that’s designed to encourage adherence to tuberculosis medication. And TB is a super tough disease to beat. You spend about six months on treatment for some of the easiest treatments, and then there are tougher treatments from there. You have to take a very strong medication every single day, that medication typically will make you feel sick. There’s a big stigma. You have to go back to the clinic all the time. You want to avoid that because you don’t want to be seen if somebody could go into the clinic. It’s a mess. And people are all the time, you see people being like, “I give up, it’s too tough.” And that’s a pretty reasonable thing to do.
Erez Yoeli: But what you find is, as a consequence of this, there’s basically this huge epidemic that as soon as folks go off their medication regimen, they become infectious again. And that really helps to feed this epidemic. And you might not know this, but TB is actually the world’s deadliest infectious disease. It kills more people than HIV and malaria combined.
Erez Yoeli: So it’s a pretty big problem. And it’s coming largely from the fact that there’s this desired behavior, staying on your medication regimen that isn’t happening. It’s not really a medical problem. We have the cure. It’s just people aren’t sticking with it.
Erez Yoeli: So that’s the context here. When we started working with a digital health startup called Kaheela, which is based out of Israel. They’re building very simple dumb phone platforms to provide people with better support and better motivation. Our job was to help build in that motivation piece. And there’s a whole bunch of things in this platform, and I’m happy to go into more details if you’re interested. But I want to focus in on the key thing we did to eliminate excuses.
Erez Yoeli: So one of the things we wanted because of this observability thing we were just talking about was for people to respond to us after we send them a reminder to take their medication, telling us that they indeed took the medication, verifying that they’ve taken their medication. And that we realized, there were a lot of excuses that one could give. If they got a text message reminder, then they could always say, “I didn’t see the text.” They could always say their phone died, that their mother had borrowed their phone for the day or whatever. So we knew we couldn’t just do a text message reminder. That left too many excuses on the table. So we needed to start figuring out a way to lock those down.
Erez Yoeli: So we started asking okay, well what if we ask them to verify? Well, that’s good. It both increases observability, as we just said, and it also starts to eliminate the excuses. Because now they know we’re expecting a response. But that’s not enough. What if we then text them an hour later and say, “Hey, have you verified?” What if we text them another hour after that, “Hey, have you verified?” And what if they know that if they don’t respond to the third reminder, that there’s a team of folks, of real humans who will follow up with them if they don’t verify?
Erez Yoeli: Well now they know it doesn’t matter whether your phone died. It doesn’t matter whether your mom has your phone. None of that matters. You have to tell us whether you’ve taken your medication for that day. We’ve just locked down all of those excuses. That was the process we went through when we were thinking about excuses in this particular context of medication adherence in Sub-Saharan Africa.
Brooke Struck: And what about the third one? Communicating the norms that you’re both expecting people to adhere to, that you will sanction them for breaking, that you will reward them for following. And that you want to observe people either following or breaking.
Erez Yoeli: That one comes from the fact that in any norm enforcement environment, basically if we do a game theoretic analysis of these environments, this is a game which is a repeated game. You have these social sanctions happening, and people interacting over and over again. And in these kinds of repeated game environments, there’s always an equilibrium where people don’t do the desired behavior and their reputation isn’t really affected by it. You’re not expected to sanction that. You don’t get sanctioned for not sanctioning it, etc. You don’t do it. You don’t get sanctioned. You don’t sanction for not sanctioning. That equilibrium in, for instance, the simplest of these games, the repeated prisoner’s dilemma is called the all-D equilibrium, always defect. And all of these games, all of these environments have this equilibrium. So anytime we’re talking about an environment with these sort of social reputational things, there’s always this one equilibrium where you just don’t care. And that translates to in practice, people being very, very sensitive. Because there’s multiple possibilities. There’s a possibility that your behavior will affect your reputation. And there’s a possibility that it won’t. That people being very sensitive to whether their reputation will be affected by their behavior in this particular context, whether they are expected to do the good thing in this case.
Erez Yoeli: And lots of different ways of communicating to them that expectation exists. There’s no one silver bullet. It kind of depends on your context. And behavioral economists have a field day with these. They have descriptive norms, and injunctive norms, and identity frames. And oh man, now there’s dynamic norms are the new big thing that they like to talk about. But what all of these things at the end of the day are doing is telling you “hey, this is a behavior we’re expecting you to do.” That’s what we mean when we say communicate expectations. And that’s where it’s coming from.
Brooke Struck: Are there some more and less effective ways of signaling those norms, of communicating those expectations?
Erez Yoeli: Yeah, I think so. And I think the behavioral literature has done a pretty good job of exploring that space. So for instance, one of the famous things that people always come back to in this space is when you’re using a descriptive norm, a descriptive norm is telling people “hey, lots of other people are doing this thing.” Descriptive norms are really effective. They’re one of the oldest nudges in the nudger’s toolkit. They’re one of the most famous ones. They’re often found to be effective. They’re often found to have lasting effects. It’s a good one to go to for communicating this. And you can understand why if you’re thinking about expectations. If you see that lots of other people are doing this thing, it becomes very clear that you’re expected to do the thing. Obviously that’s the takeaway that you take from it.
Erez Yoeli: And one of the downsides of descriptive norms is that if you say hey, 20% of people are doing this. Then the person takes away that they don’t need to do it. Or if the person went into the situation thinking 90% of people did it and you told them 75% did it, then they think it’s a little bit lower than I thought before. So that’s also a potential problem for you and you might get some sort of boomerang effect. So that’s the kind of problem that we see.
Erez Yoeli: The problem that you face is that if this is a behavior that’s not very popular, or if people’s expectations is that it’s more popular than you end up telling them, you’ll get a rebound effect or a boomerang effect is what they’re called. And people have documented this. So if that’s a risk, you don’t want to use a descriptive norm. And people have come up with other ways of communicating expectations in those contexts. You can use injunctive norms, which is to say you should do this. And you can use dynamic norms, which is “this thing is growing very rapidly.” You can use the other kinds of tools like identity frames. You can use folks who are figures of authority. And you can communicate expectations that way. Celebrities and other people who are well regarded in society who are good at sort of generating these kinds of, this impression that people are expected to give. All of those different ways will work if you can’t use a descriptive norm, and that’s fine.
Brooke Struck: I want to overlay this three-piece or this three-part approach onto another model that exists in the same ecosystem of behavioral approaches to promote collective action. And that’s the panic model. So P-A-N-I-C. P is for personal, A is for accountable, N is for normative, I is for identity relevant, and C is for connected. Now if I overlay that onto yours, it sounds like two of them from the PANIC models, especially around accountability and normativity line up really well with observability and eliminating excuses. That people are going to see you do it, and people are going to hold you accountable for what you do as they see you. That seems to line up super clearly. The personal and connected, the P and the C of the PANIC model, I think resonate really strongly with that last point you were making that the norm being communicated has to somehow have traction with the individual in question, right? It has to not just be a norm that’s out there. It has to be a norm that’s relevant to me. And one of the ways to kind of promote the traction of that norm with the individual is to have it delivered by people that they recognize as like them. So either friends or family directly within their network, or colleagues within their professional network. People of high repute, recognizable figures, this kind of thing.
Brooke Struck: But the one area where things seem to diverge a little bit, and this is where I want to take this conversation now, is about identity. And I want to talk about the relationship between identity and reputation. So first of all, what’s your thought on the relationship between identity and reputation before I tip my hand?
Erez Yoeli: The way that I think of identity here is that it’s another tool like personal, like normative, it’s another tool for generating expectations. Basically what you’re selling somebody is, “Hey, if you’re this kind of person, well this kind of person does this kind of thing.” We used an identity frame at the start of one of our get out the vote letters, and it started “you’re a voter.” And that’s because we had the person’s voting records. So we could tell they were indeed a voter. And what do voters do? They vote. And we were advocating for them to vote.
Erez Yoeli: So you’re basically pulling on this consistency motive. And in addition to pulling on that, you’re also setting that expectation. An identity free frame seems to me like a combination of those two things. That’s how I typically think of it.
Brooke Struck: So maybe that’s a really incisive way to get into the kernel of this question. When we talk about the motive for consistency for instance, who is it that we’re trying to convince? Are we ultimately trying to convince others about the consistency of our character and our adherence to the appropriate norms? Or are we trying to convince ourselves that we have consistent characters, that we are good people, that we have the identities that we say we have and that we want to have?
Erez Yoeli: The answer is both. But they’re kind of happening at different levels. There’s a guy named Bob Trivers. He’s a super famous biologist actually, an evolutionary biologist who came out with a book a few years back called The Folly of Fools. And the core research that this book is based on is actually a collaboration with a psychologist named Bill von Hippel who’s in Australia, Melbourne. Excuse me, in Brisbane, not Melbourne. Bill and Bob have this very nice paper showing that basically a major part of why we convince ourselves of things is so that we’re better at convincing others of it. So when you ask me the question are we doing it so that we can convince ourselves or so that we can convince others, the answer is yes, we’re doing it for both reasons. But one of those reasons really exists for the other in and of itself.
Erez Yoeli: The important thing isn’t so much that we argue over whether the right way to think about it is this way or that way or whatever. The important thing is to recognize that these things are kind of pushing in the same direction. They’re just different levels of analysis. One is what a psychologist would call, or a biologist, they would call them the proximate level of analysis. And that’s the level of analysis where you’re looking at the actual processes, what’s actually going through somebody’s mind. What are they thinking about? If somebody is thinking about their identity, we would call that a proximate analysis. If you’re trying to trigger this proximate psychology, we would call that a proximate analysis. And I think when somebody is talking about identity, the clearest way to think about identity is at a proximate level of analysis.
Erez Yoeli: When I talk about reputation, as we sort of mentioned earlier, a lot of this is going on subconsciously. It would be kind of weird to think that people are consciously thinking about what others think of them and whether others catch them doing the thing, or see them doing the thing. I mean, sometimes I think about that, sure. But a lot of the times that’s not what’s going on through their minds. So basically what I’m saying is there’s another level of analysis that you can do that kind of explains where the proximate psychology came from and why it’s so sensitive to plausible excuses, and why it’s so sensitive to observability, why it’s so sensitive to communicating expectations.
Erez Yoeli: When I talk about reputations at that level, and when people talk about identity, they talk about it at the proximate level. And I think probably what’s going on is that reputations and identity are very tightly linked. One is more proximate and one is more ultimate. And identity also tends to have this other feature that we were talking about, which is about this consistency piece. Where you really want to behave consistently with past behaviors that helped you form this identity, which is a very particular kind of reputational dynamic that’s unique to identity and maybe not as important if we’re not using an identity frame and if we’re using say descriptive norms and like that.
Brooke Struck: The reason I push on this of course is because as soon as we come down on one side of that fence about whether it’s reputation or identity, what we’re going to be saying next right after that is, “Well, humans are inherently good people. They want to have this strong sense of identity. That’s going to be the biggest motivator. No no, humans are terrible individuals. We all interact with each other for strictly instrumental purposes. That’s why reputation is the way to go.” I agree that a more nuanced understanding is perhaps more helpful. So we can approach this from a different angle.
Brooke Struck: So first of all, I think it’s really valuable to acknowledge that there are massive overlaps between these things. That actually, regardless of which side of that fence you come down on or whether you decide you’re not interested in the conversation about whether humans are inherently good or inherently terrible, a lot of the kind of experimental and practical results that would follow actually converge. That, whether you think identity and inherent goodness or reputation and inherent instrumentalism is really the primary factor, the motive force in this ecosystem, you’ll reach the same conclusions in a lot of cases.
Brooke Struck: So for instance, clear communication of norms is I think an important one here. Perhaps making excuses is the better way to go. It’s just as important for us to, if we want to be promoting collective action. It’s just as important for us to seal off those third doors for people to make excuses for others or to others as it is for us to seal off those third doors for us to make excuses to ourselves. I also will rationalize my own behavior. Like yeah, I really did go out that fire door that I didn’t even know existed until the Salvation Army was here. Because of this completely unrelated factor. It’s not that I’m a bad person and I want to avoid the Salvation Army and their bells. It really is that yeah, my car is closer to the store. I can tell myself that story just as easily as I could tell that story to somebody else. The plausible deniability is just as important in both realms.
Brooke Struck: So if there are these kinds of strong convergences between the two approaches, one of the questions we might ask is where might we see divergences between the kinds of predictions they would make about types of interventions that are effective or the level of effectiveness of interventions? And one of the things that you noted earlier is that observability is a really, really powerful factor in this. So maybe I can invite you to talk a little bit about what observability means in relation to reputation versus in relation to identity. Do you see important differences there? Or is this not the kind of acid-test that we might hope it could be?
Erez Yoeli: I’m not sure it is, because I think that you could argue, in fact I’ve heard this argued, you simultaneously have this explanation that people want to convince themselves that they’re good people. They also want to convince others that they’re good people. So observability will of course work.
Erez Yoeli: That’s not how I would try to argue. I wouldn’t try to argue against that explanation in that way, because I think it’s true that if both explanations are going on, observability will matter. And in fact, the way I often get out of this argument, you’re not allowing me my usual plausible excuse, is that basically I say, “Look, I don’t care if you think that reputations are the only thing that’s going on. I just care that you think that it’s an important thing that’s going on.” And then the three things on that checklist that we’ve been discussing. Observability, eliminating excuses, and communicating expectations are important. So here’s a useful tool kit. You can also believe all these other things are going on, but at least now I’ve armed you with a useful toolkit.
Erez Yoeli: Now if you ask me what I actually believe is going on, I actually believe that the key thing driving a lot of our being good is this reputation stuff. There are some places in which that worldview would deviate from a worldview where that’s a little bit more there are platonic goods and people are coded with some, or however it is that you might view the world that reputations are not the only thing that’s going on. One of them for instance with regards to identity is that identity might be a little bit more stable and resilient in a world where it is itself a driver than in a world where it’s basically there in order to help you. It’s a press secretary that’s there to help you with your reputational endeavors more generally.
Erez Yoeli: The litmus test that I would try to use, and I don’t think the literature out there really convincingly rules out one argument or the other yet. It is an interesting thing to keep an eye on to see how things develop. But the litmus test I would use is to try to figure out whether identity really is a little bit more, whether identity tends to move into directions that make it seem like it’s particularly well suited to maintaining people’s reputation. Whether people’s identity is more important to them when the chances of observability are higher, for instance. Whether their identity is sensitive to cues about how others think of those identities. Or whether it tends to be quite resilient to that, resistant to that. Those are the kinds of things that I would look for. I would look to see whether identity seems suspiciously well-designed to be a press secretary or not. And I think it’s a really good question how to design good experiments to do that in and of itself. And then what kind of evidence, we’ll have to wait and see.
Brooke Struck: He’ll arm you with another out in future, or for the future I should say. And that is this. You can push back on identity and reputation being two decoupled things. And one of the things that you can appeal to there is community. That both identity and reputation are really, really strongly tied up with a notion of community. Who I am is strongly connected to who my people are. And reputation is also extremely strongly connected to who my people are, because those are the people whom I want to be impressing. Those are the people with whom I want to have a positive reputation. Those are the people whose sanctions I care about. If we kind of follow that turn and say okay, well community is an important ingredient that we need to be keeping in mind here. To echo an idea that I mentioned earlier, I don’t think that that is inconsistent with the view that you’re putting forward and with the toolkit that you arm us with. If community is what’s at issue, then we need to be really cognizant of who people’s communities are and how community membership is shifting and evolving. Because that’s going to be crucial in order for us to be able to identify which norms people will consider relevant, to identify which people’s observations they will consider relevant, and this kind of thing.
Brooke Struck: So now to lay another heavy at your feet, social solidarity and sense of community is extremely strained right now in some places around the world. Some places that aren’t even that far away. On the one hand, do you see risks associated with that for the viability of applying the tools that we’ve been talking about? And on the other hand, do you see an opportunity to leverage reputation and identity, and collective action to start kind of remediating the situation, to start walking back towards the sense of stronger bonds between us?
Erez Yoeli: The easy part is the impediments. Yeah. There’s definitely impediments. Look, no matter how observable, how few excuses you have, and how much you’ve communicated around masks. If there are some communities where that behavior is not viewed as normative, you first have to do the work of turning it into a normative behavior. It’s not going to help you to just ask for it. That’s a major problem because what we saw was that because of the political rifts in the country, it was a behavior that was ripe for politicization. And now we have communities where not wearing a mask is a signal that you don’t trust the establishment and that you aren’t liberal and so on. And ultimately, that’s going to be something that hurts those particular communities. It’s their old and sick folks who are put at risk. It’s their hospitals that are going to be overwhelmed. It’s not ours here in Boston, right? Like if they’re in Indianapolis or Florida, I know there’s some viral video going around today of some anti-maskers invading Target in Florida. It’s the Florida hospital that’s going to be impacted by behaviors like that.
Erez Yoeli: It is definitely a problem, and it makes our job way harder. It’s much easier to start with a behavior that everybody agrees is a good thing and say okay, well how do I get people to do it now? It’s much harder to first get people to believe the thing is a good thing and sort of everyone be on board, which is what I was saying earlier, which I called earlier making the behavior normative or establishing a new norm. Much harder to do that.
Erez Yoeli: That said, I’ll say two things. Number one is the same perspective on thinking about how norms work, how these social sanctions work, what the importance is of reputation. The same perspective is useful for thinking about how to establish norms in the same way that it’s useful for thinking about how to enforce them once they exist. So you don’t get exactly the same observability, plausible deniability, expectations checklist. You get a different checklist, which is more relevant to the establishment of norms. But you still can kind of pull together a checklist for that. That’s what we’ve been doing.
Erez Yoeli: So when COVID hit, the first thing we did was put together documents for folks in marketing departments of their organizations. In mayor’s offices etc. on how do you write a messaging campaign? What are the things you say in your signs? What are the things you say in your radio announcement or the prerecorded call that I get from the mayor of Somerville? How should he talk? We used that same perspective when we came up with a different list of things based on that perspective for these messaging campaigns. So that’s one thing I’ll say, and I’m happy to walk you through some of the details of that.
Erez Yoeli: The second thing I’ll say is this might be a strange time, but you have to marvel at the fact that we responded to the pandemic as effectively as we did at all. If you look at responses to pandemics in the past, they were way worse. If you look at the 1918 influenza, basically we killed like 4% or 5% of Europe because we just ignored its existence and shipped our troops over there. Then got it back and completely failed to be on top of it. We held parades. We did all sorts of disastrous stuff, and we failed to contain it completely. If you look further back at the bubonic plague, I mean the MO during the bubonic plague was “leave the poor in the cities and literally lock the gates so that they can’t leave.” And the rich would go out to their estates and try not to engage with anybody until the thing blew over. That was the way we responded to pandemics in the 1500s and 1600s. Okay. I think we’re doing better. Most of the world has actually responded fairly effectively to the pandemic. Some extraordinarily effectively, countries like South Korea or Japan are managing the pandemic in ways that are somewhat mind-boggling. There’s others whose response is perhaps less effective, but still very effective like Germany. And then the United States’ response has been so-so, quite so-so.
Erez Yoeli: But still by historic standards, it’s still better than better than locking the poor in and waiting for them all to die so that you can come back. We’ve moved up one notch on the ladder from that. So I think there is a little bit of room to be optimistic. We are seeing cooperation at a greater scale than we’ve ever seen before in a pandemic I think. That is somewhat exciting to see.
Erez Yoeli: I remember being quite blown away when finally the first shutdowns hit in the United States and then people actually did them. I was like yes, amazing. I think that’s something to be happy or excited about. It’s a little bit of a silver lining in this tough situation.
Brooke Struck: Okay. So if we are kind of scoring some, not scoring some points. If we are achieving some victories here, if we are making progress, how important is it that we find a way to talk about this in the future that allows us to kind of have this common cultural touchstone that we can look back at this as a shared struggle that we all came through together? Ultimately those narratives are really, really important in sewing or weaving the bonds of community, which we’re talking about now as really important for being able to create and use social norms in the future.
Erez Yoeli: The way that I think of that kind of stuff is those kinds of shared bonds are usually useful beachheads for building meaningful relationships. At the end of the day, you build meaningful relationships based on the current conditions. Those beachheads help. They give people a little bit of traction. They give people something to start talking about. They often make it more intuitive to be closer to someone, and they’re important. But if you don’t have a real reason to be good to each other, then the beachheads won’t end up helping that much. I think the way that I view it is how important is this? Well, it can help a lot. It can really help smooth things over, but on its own it’s not enough.
Brooke Struck: What’s missing? What’s the missing ingredient? And I promise I’ll stop pressing you after this one.
Erez Yoeli: Well, you really want to build a world where if you’re trying to encourage people to be good to each other, you want to build a world where they’re legitimately not in conflict. And there’s always conflicts that arise. That’s a difficult thing to build. And then you want to build a world where you have a set, if you are asking for things, there are always going to be these things that require some elements of self-sacrifice for the benefit of others. When that’s the world that you’re trying to build, then you want to be thinking about the decision making environment that people are making those decisions and then try to optimize that to get them to do the right thing more often. And that’s what the last hour has been about. Regardless of whether you are thinking about it through reputations or using a more proximate approach like the PANIC model, either of those approaches are useful for trying to get more people to do more of the right things.
Brooke Struck: All right. So if I can try to put together, try to distill one summative insight that takes all of what we’ve discussed which has covered a lot of territory, and pull it down into one very concrete actionable item that we can come away with. It’s that a model like the PANIC model or the three-part model that you’re advocating, there’s one concrete step that needs to be taken before, at least needs to be taken before, probably needs to be kept on the radar along the way, which is you need to pay attention to which communities people believe that they’re in, in order to know which kinds of identity or reputation levers will actually be effective with them. That if you don’t have your eye on community membership, then you’re much more likely to step into a hole, no matter which of these models you choose.
Erez Yoeli: I agree with that. When we were asking people to wear masks, we did it in a way that wasn’t culturally appropriate and sensitive to their sets of values and their priorities. That backfired on us. In the project that I told you about earlier where we were working with folks in Sub-Saharan Africa to promote medication adherence for tuberculosis. That’s another case where a bunch of outsiders trying to help folks who we don’t really know that well. And we knew immediately we needed to be teaming up with a local team to actually be able to make any progress. And every single word on that platform was not worded by us. We came to the team and we said, “These are the principles that we think we should use to design this platform, but we need you guys to actually design it.” And that ended up being very powerful. I mean, everything from the slogan that the company uses, which is “health hero shujaa afya, which in Swahili, shujaa is a warrior/protector of the community. It’s not a word that doesn’t translate very well, but it has this almost nationalistic connotation. And afya means health. So he’s a warrior of health or a hero of health. That’s language we wouldn’t have known to use. And it’s language that we know really resonates with folks. And we give folks a wristband when they participate that says it. They wear it, and they’re super proud of it.
Erez Yoeli: That kind of stuff is really, really important. Being the difference between your interventions resonating with the population that you’re trying to reach and not resonating at all, and just totally falling flat. It requires a lot of legwork to do it right, to engage the community well and understand the community well. But I think you’re right, that regardless of whether you’re using a PANIC model or the checklist, the reputations based checklist, it doesn’t matter. You’re going to fall flat. And that’s definitely true.
Brooke Struck: All right. Erez Yoeli, thank you very much for being with us today. This was a great discussion.
Erez Yoeli: Thank you very much.
Brooke Struck: If you’d like to learn more about applied behavioral insights, you can find plenty of materials on our website, thedecisionlab.com. There, you’ll also be able to find our newsletter, which features the latest and greatest developments in the field, including these podcasts. As well as great public content about biases, interventions, and our project work.