Diego Gambetta on Signaling Theory

[Music] This is Social Science Bites with me, David Edmonds. Social Science Bites is a series of interviews with leading social scientists and is made in association with sage publishing. What do New York taxi drivers and the Italian Mafia have in common? Well, they've both been studied by the Italian social scientist Diego Gambetta. He's well known for his ingenious use of certain economic theories for the study of social phenomena. They include his use of what's called signalling theory. Diego Gambetta, welcome to Social Science Bites. Good morning, Dave. Thank you for having me. We're talking today about signalling theory. What is signalling theory? The theory we have to try and establish when truth can be communicated even in conditions that are difficult in which we expect the interest of the parties communicating to diverge or not to completely overlap. So is there a way or is it impossible to achieve that? And the theory describes what the conditions to do that are. Pat, you could just give me a simple example. The theory works both in conflict situation and incorporating situation. In a conflict, I may try to persuade you that I am really, really tough. If this is true and you believe me, then we may sort the conflict out cheaply for both of us because we don't enter into a damaging fight. So if I persuade you that I am tougher than you, but if I just say that I am tougher than you, you may suspect that I may not. So you may be looking for signs and signals about my being. So in a cooperating situation, I can tell you, look, Dave, I just need 500 pounds to settle something. I give them back to you next week. And you say, "Well, yeah, I appreciate that, but how can I be sure that you will return the money to me?" These are a typical situation of conflict or cooperation in which we have reasons to doubt that the intentions or the qualities of another person are what she says they are. Can you tell me something about the origins of signaling theory? Is it used only by economists? Or is it relevant for other disciplines? Yes, it was formalized first by economists, in particular by a man called Michael Spence, who received a Nobel Prize for that. And the issue in his case was, how can an employer establish whether a potential recruit is of the high productive or low productive end? Incidentally, the signaling theory shouldn't be seen just from the point of view of whether a signal believes a receiver, but also from the receiver because if you are a potential recruit and you want to persuade a BBC that you are highly productive, then you also ask this question, "What am I going to do? What can I do to persuade a BBC that they should recruit me?" So this was the first domain in which the theory emerged. More or less at the same time, there was intuitive expression of the theory by an Israeli animal behaviorist called Amos Zahavi, who had a very three-dimensional sense of animal communication. And here I've more or less at the same conclusion that Michael Spence and that thing happened almost simultaneously unbeknownst to each other, I believe. Fascinating. So let's return to what signaling theory is all about. Crucial to signaling theory then is the link between what we know and what we don't know, that there's a level of uncertainty out there and that there's a nice symmetry of information. Yes, it has to do with when I would like to know about you a lot more than is apparent. Because qualities that we are interested in are not written on our foreheads. So we wonder, is this person honest or not truthful or not strong in fights? And is she what she says? She is. Because in many circumstances, especially in the digital age, we may even doubt the identity of the person. So the theory has to do with the unobservable qualities that are of interest. For me to signal for you to receive. The question is, well, how much scoped do we have to con each other, to cheat each other? Right, so it's about trust. I wondered what the implications are for dating between partners. Because when we're dating somebody, we need to know whether they're going to be reliable, whether they're going to be faithful. It would seem to have implications for the dating scene. It very probably does. People when evaluate each other as potential partners, they must take a whole menu of unobservable qualities into account. Is she fertile? Is she going to be a good mother? The woman may wonder, is this man going to be loyal to look after the children? Yes, and people have studied the marriage market also in terms of signaling. One thing you could do is just say, well, trust me, I am faithful. You just have to believe me. I guess that's what they call cheap talk. Is that a tactic that might work? Well, we would be too lucky if it worked. I suspect. It should, it doesn't. I don't think so, but in order to explain what this hypothetical man could do to be a little bit more persuasive than just say that he is a faithful chap, is to invest in time and resources that he displays to the potential partner. Traditionally, the engagement ring was an investment. The receiver, the potential spouse may have to calibrate it with the income, right? So if big gates gives an engagement ring, the size of it needs to be as big as the rooms in which we are sitting. But if I give an engagement ring, then one that fits on a finger is probably sufficient, or imagine if you are a fan of opera and you know about Mozart's Don Giovanni, you can think about the difference between Don Giovanni and Don Octavio in Cording Don Arna. Don Giovanni is more the type you were describing, the guy that says, you know, I'm very nice, I make your life a delight and so on, come with me right now. And Don Octavio by contrast, who is truly in love with Don Arna, is the faithful kind, he spends more time in Cording, in serenading, in displaying that he is prepared to sacrifice. So what happens in this case, it means that the power of the signal to persuade is in the difference in cost between the phony shooter and the truly loving shooter. So if you're prepared to spend money, if you're prepared to invest, that exhibits a degree of trustworthiness. It exhibits that you are true to your word. It's not just money by the way, if you imagine me giving in cash the value of an engagement ring, it may not be so persuasive. You say, well, this is a crossway of proceeding. The theory in itself isn't often not complete, you need also a theory of conventions. So we have the conventions that in order to signal that you are a good shooter and a potentially reliable partner, you have to do it with a ring or you have to do it with a dance or you have to do it with some kind of thing that is shared in the community and understood. If I give you the cash, I don't think you would work. You could try. When one talks about signaling, it suggests that one is somehow doing it deliberately, that one wants to send a signal and I'm just wondering whether it matters whether or not the sending of the signal is at a conscious or at an unconscious level. It doesn't matter really because in the animal behavior world, signals are as far as we understand not very often intentional. They're part of the selection process. So things that persuade get positively selected because you attract mate, you discourage competitors, whatever you do stays around and is transmitted to your offspring. There are however in the animal world also intentional signals. So we know that certain species call each other by name. Dolphins have names that are a particular sequence of knots or frequency, so they can use it intentionally, they can call each other. In the human world as well, there is a mixture of signals. So there are things that we have developed at a greater cost and therefore signals that we have a better immune system or we have a better fertility, potential and so on. So for instance, symmetry or the ratio between hip and waist are also observed by potential partners to make decision and what matters is still the difference in cost. If you were not strong enough, you wouldn't have that level of symmetry in your facial feature or you wouldn't have that kind of high hip-waist ratio. However, we can tamper with it. we can manipulate that so we can use the makeup or we can use corsets, culture, intrusion in our world. And therefore, the wastey peritior that you observe at first blush may not be the necessarily accurate one, right? And that has a level of intentionality, which is there. I mean, people realize that certain features are more appealing and they may stress them. I want to talk a bit about your own empirical work. You seem to straddle the worlds of economics and political science. And you've done this fascinating research in a number of areas. Can we start with your work on the mafia, organized crime, and how that links to signalling theory? Well, in a sense, it links not in the way you would expect, in the sense that I was alerted to the importance of symbolic communication in the mafia by the way they communicated with each other. And also, on a couple of occasions, the way they communicated with researchers, myself included, that is, they display a subtlety in communication that surprised me. We tend to expect organized criminal to excel at brutality and intimidation, but we can hardly expect them to be a lot subtler than we are in communicating. There was an Canadian researcher that went to Palais, a famous scientist, and he didn't shy from saying that he was going to research the mafia. And apparently, he had his car robbed, and his laundry is dirty laundry taken away. And then, after a few days, the laundry came back, cleaned an iron with just a not saying goodbye. Right. So, violence is known to us because it leaves body on the ground because it's goryly tracks our attention. There is a subtlety in threatening, which is, you know, I have a whole list of these. So, they were signalling with the iron shirt. I've done your ironing for free, but I know how to get you. I know how to reach you. I can enter your room. This is my power. Yes, in a sense, it was the more enthusiastic rendition of, I know where you live, right? Is the kind of. That was the message. And the point is, had you been a phony mafioso, or which there are of you, you would not have been able to do that, so gingerly and effectively, right? So, it's not the fact that it's costly for you. It's the fact that it's too costly for a phony that carries the message. Nobody else could fake this. This was clearly genuine, and this was a signal that you need to be scared. But then, one reason why the theory is so attractive and so underapplied, is that it's difficult because it depends on context. So, if I put a card of a jack of spade under your door in easily in town, you would think that some kid dropped it on the way to school. But if you received a jack of spade under your door, in Palermo, when you're researching the mafia, then that may acquire a different light. So, it means that the context in which a signal is produced may saw different effects. Right. The jack of spades in Islington in North London, obviously, might not be threatening, but the jack of spades in Palermo, anybody could put a jack of spades under the door. There's not a great deal of cost involved in shoving a card into somebody's room. No, in fact, that's why I'm saying that the context matters, because in Palermo, there is a kind of atmosphere of threat or there used to be. A threatening atmosphere makes people extremely careful about evaluating any noise in the environment. Anything can acquire a meaning. So, there is an obsessive search for meaning. And the mafia, they are particularly good at that. The thing I realize is that the stillness in the way they comport themselves, in which they behave, they find that they are very still, very impassive, they don't move very much, is very likely to be an attempt to silence misunderstandings. Or to silence wrong interpretation that you don't want to give out. So, it's a very highly controlled, so they are the opposite of your traditional stereotype Italians. Let me ask you a bit about how you carried out this research, because, as you say, the mafia don't really welcome people investigating their methods. Tell me a bit about how you set about doing it and whether you yourself felt under threat. If we want to stay on signal, let me tell you this. You asked me correctly, I mean, anybody could put a card on your door. The difference between Palermo and Islington is that, when you get a card on your door in Palermo, you take steps to check whether the sender is the real thing or not. So, there is an investment, also from the receiver point of view, saying, does this request of extortion mean that there is just some kid pretending to be tough? Or does it come from the source of the water, was the expression? So, that means you start asking around. The mafia said to you, ask Anno who is the boss here. And if you receive the same answer from random people, then only the real thing could achieve that. There was an entrepreneur who was asked for extortion money, and he said, "But how can I be sure that you're not a punk?" He says, "Ask Anno." That was the answer. And if the tobacco he says is that guy, if the carabiner, he says, "Yeah, they say is that guy." If the cleaning lady tells you, and they all say the same guy, then you go and see that guy. Did you ever get a jack of spades under your door? Yes, I did as a matter of fact, and I took it as if I was in Islington, yes. I didn't realize what he was. It's only later that some better knowledgeable, and near-polyton friends told me, "But look, the jack of spades, in certain underworld, are go." He's, you know, you're too nausea. Go away. You didn't understand the signal. I didn't understand the signal. In fact, so far we've been talking about the threat to signals that comes from misrepresentation, but there is a threat to signals that we encounter very often, that comes from misunderstanding. Let's talk a bit about your work on taxi drivers and signaling. That was conducted where? It was conducted badly, but fast, where my co-author, Councillor Romain, had very good understanding of the well there, and badly in greater New York, parts of New York in which yellow calves don't dare to drive. So we took two places that were not, to our usually, dangerous for taxi drivers. And the issue was which passengers taxi drivers would pick up, and how they would be able to predict that their passengers were safe and trustworthy. - Correct, is that? Yeah, that's exactly what we had in mind. We wanted something simple. If you think about it, taxi drivers come into contact with lots and lots of strangers. And like few other professions do, maybe prostitution, but not that many. So they have to have some way to decide quickly and safely whether the potential customers is somebody they can trust. Trust in a limited sense. Trust of saying, "I get in a taxi, I truly want to ride, "and at the end I pay." I don't ask you whether you are trustworthy in any other deeper sense. We wanted to find out what signals do they look for, they attend when they decide. - And what was the answer? What signals can passengers give off? So that taxi drivers know that they're going to be safe at the end of the journey and they will receive their money. - It's quite rich. You'll be surprisingly rich cognitive operation that you encounter in taxi drivers' decision. So in Belfast, things were so tense that the signals were very crude and very clear. That is, you come from a certain part of the city and not from another. You are known or you know your way around this place because the screening continues once they get into the car. So they can ask you what football team you may support or whether you know a certain pub. But they also look at something subtler, say. So if you sit in the back of the car and look around in a debon air kind of way, then they assume you're not interested in doing anything strange. So where you look becomes interesting or how you move. But if you establish that they support the wrong football team, by the time they're in the back of the cab, it's too late. I mean, you've got to take them to where you've promised to take them. - If they are of that kind, they know that you would ask that question and they wouldn't answer or they would answer in the way you want them to answer. The screening is quite rich because it's a strategic game between the passenger that has an interest to attack you. You know, many drivers were attacked in bedfast during the troubles. And others had their cars hijacked. - But to go back to that, isn't it too late to establish that you've got a dangerous passenger if they're already in the back of the cab? - It can be too late. However, you may take counter action. So you may discover that this person is suspicious and instead of driving to the destination that the passenger asks you to, you're driving in front of a police station. I'm citing events or I can stop and ask you to get out. And in New York? In New York things were a little bit looser and people were mostly receiving not a political terrorist type of attacks but robbery kind of attack by drag addicts and people like these. For instance, you would try to see where people were coming from. Are they coming out of a church? And then you can say, "Well, anyone can pretend to come out of a church." So you have to look, is this a afraid or lady or a brute pretending to be like in red-riding hood, afraid or lady? Right, so. But this type of reasoning that you and I are bouncing about is the type of reasoning that seems to be going on in the heads of the drivers trying to decide what to do. And you established that how? We use that every possible means. Mostly by asking, you know, we had a list of items that would be costly to fake. And our general prediction was, given the theory, you would expect drivers to look for signals that would be costly to fake for somebody with bad intentions. The simplest one in very stable environment like in Belfast would be who you are. You couldn't go out and fake to be you. So being known was the first thing they looked for. The other were gender and age. You mean the older and female, the less threatening you were thought to be. These were the kind of statistical discrimination, but at the same time, it's very difficult to fake. So it works because it's difficult to fake. What about race? Race was important in New York, even black African or black American drivers did not pick up black American youth. They just stay away because they considered them two dangers. And this race is an interesting problem for a signal interior. In any way that, suppose you are a good youth, black guy, or whister, or many in New York, how do you get attacked? And the question is, you probably can't. The cost of proving that you are bona fide and you are a real passenger despite your age and color is too costly, too complicated. One part of the theory which hasn't been developed very much, cost that you would want to have available, but they are not. So, signning theory so far has developed a diagnosis of the problem of racial profiling, but hasn't yet come up with a solution, but there are potential solutions out there. I don't know whether there are, but it's certainly worthwhile looking for them. So, for instance, joining association which can be credibly flagged through the internet, in which work is a accreditation of your bona fide could develop. And they may be developers as we speak in fact. It's like looking at your credit history. If you build up a history of having successful exchanges with taxi drivers and you have a way to convey your identity, it was really you. It wasn't somebody just looking like Dave, it was kind of a taxi driver. It was really you. If you can signal that identity credibly, then you can build up a reputation for goodness and you may sort the problems that black youth wanting a taxi encountered. Well, something like Uber is a potential solution to that because everybody has a personal profile. They also give a credit card. You're absolutely right. There is potential for technical solution to the problem of misrepresentation in exchange. Let me ask you just one final question. I was wondering whether your expertise on signaling has changed your own behavior in any way. Has it changed how you emit signals? You know, they say that people interested in rationality are usually very irrational, them says. I may become better at noticing things. You know, I have my antennas out for noticing things that other people do. But I don't know whether I'd become any better in my own behavior or any worse actually. Diego, can better. Thank you very much indeed. Thank you Dave. It was a pleasure. Social Science Bites is made in association with Sage Publishing. For more interviews, go to socialsciencebase.com.