In a paper published in Nature in 2003, Sarah Brosnan and Frans de Waal provided what was, arguably, the first evidence for inequity avoidance in a non-human animal. Brosnan and de Waal found that brown capuchin monkeys, refused to participate in trials if they observed conspecifics receiving a more attractive award for performing the same task. Thirteen years after the paper was published, I spoke to Sarah Brosnan about the making of the study, her memories of the experiments and what we have learnt since about inequity aversion in non-human animals.
Citation: Brosnan, S. F., & De Waal, F. B. (2003). Monkeys reject unequal pay. Nature, 425(6955), 297-299.
Date of interview: 11th October 2016 (via Skype)
Hari Sridhar: I want to start by asking you about your motivation to do the work presented in this paper. By looking at your publication profile, I came to know that this formed part of your PhD work, and so I wanted to ask how this fitted in with the rest of the work you did in your PhD.
Sarah Brosnan: It’s an interesting story. When I first started graduate school, my graduate advisor, Frans de Waal, told us that the best way to learn about the animals was to spend lots of time watching them. The best questions derive from the animals’ natural behaviours. One day I was outside doing exactly that, as well as feeding the monkeys. When you feed a monkey group, the dominant monkey will try to grab everything. To avoid that, you hold a treat in one hand, off to the side, and try to tempt the dominant monkey over there, while not letting him have it and simultaneously feeding the other monkeys with the other hand. I was doing that with the peanuts and the dominant male, Ozzie, finally got frustrated with this. He ran back to the inside area of their enclosure and came back with a piece of monkey chow. He pushed it through the fence at me and tried to get the peanut. When I didn’t give it to him, he went back inside, where they had just gotten their fruit and vegetable trays, and returned with an orange peel. Then he did the same thing, pushing the orange peel through the fence at me. Again, I didn’t give him the peanut. Then he went inside again and came out with a whole quarter of an orange, which is quite large, of course, and he pushed that through the fence to me. I finally gave him the peanut, but it got me thinking, because I was relatively certain that if I walked up to Ozzie and offered him a choice between a single peanut and quarter of an orange, he would chose the quarter of the orange. If so, it was possible that he wanted the peanut because everybody else was getting one. Now, oranges weren’t a terribly limited resource, because they had just gotten their fruit and vegetable tray, so he could have gone inside and gotten another orange. But it was a very interesting interaction, and I really wanted to know whether my thought was right.
My dissertation project was about value perception in primates, determining whether they could assign values to tokens and then use those tokens to trade for the associated foods. I realized that I could test the hypothesis that Ozzie cared about the peanut because the other monkeys had gotten it by putting two monkeys next to one another and having them trade a single token with me, but for different food rewards. I was particularly interested in how they responded to a food that normally they were happy to eat when their partner got the same thing, versus when their partner got something more preferred. However, it didn’t end up being part of my dissertation, in part because I had proposed so much and in part because some members of my committee thought it was obvious that they would care and some thought it was obvious that they wouldn’t.
HS: When you made this observation on Ozzie, was this the first time you had any sort of inkling that monkeys were paying attention to what others were getting and making this comparison?
SB: Well, we knew they socially learned, which meant that they had to be paying attention to each other, but this was the first time that I had seen anything that suggested they were explicitly comparing what they got to others’ outcomes and judging their own accordingly. But of course, a single observation doesn’t prove anything, so then we had to design a study to look at it empirically.
HS: Do you remember when you made this first observation?
SB: It would have been my first year or so in graduate school, so long time ago – somewhere between the fall of 98 and the spring of 2000.
HS: Just to clarify – did you say the first time the monkey went in and brought out something called monkey puzzle?
SB: Sorry, monkey chow. A lot of people call them biscuits. It looks like dry dog food. Although they largely eat fruits and vegetables, they need the chow in addition because it provides the extra nutrition that they need, particularly protein for capuchins.
HS: Stepping back a bit, could you also tell us how you first got interested in primates and primate behaviour, and how you decided to do your PhD on this topic?
SB: Originally I didn’t study primates. My undergraduate work all involved ecology, focusing on stream ecology and small mammals. For instance, I did fieldwork looking at how small mammals used a field that a farmer had mowed smaller amounts of each year as he had gotten older. This resulted in a really nice succession that went from mowed annually to what was starting to become pretty established forest. We were tracking small mammals, to see how they used the habitat. I also worked in a prairie vole lab. Prairies voles are generally monogamous but would sometimes communally nest. I was interested in this because I was – and still am – fascinated by how individuals make decisions about cooperation. How do they decide who to cooperate with and when to quit cooperating? Or when it’s time to go find someone new?
When I was looking at graduate programs, I focused on the question, cooperation, rather than any particular species, and so applied to programs working with prairie voles, birds, amphibians and, of course, primates. I decided to come to Emory and work with Frans de Waal because working with him offered the most opportunities to explore really flexible decision making. Primates can think about each decision and flexibly adjust as needed.
HS: Could you tell us a little more about your collaboration with Frans – both the work for this paper and for the rest of the PhD? Was he involved in doing the experiments, or was his contribution mainly in terms of discussions and writing up of the papers?
SB: Frans was my graduate adviser, so, of course, he helped me immensely with discussing the topics that interested me, figuring out what would be a good way to test the questions that I had, and then designing the nitty-gritty details of the experiment that I needed for the test. After I ran the experiments and looked at the data, we would discuss the findings. I would write a first draft of the manuscript, and then we would work collaboratively on it. We continue to work together today, and continue to work very collaboratively.
HS: Could you give us a sense of what your daily routine was like when you were doing these experiments?
SB: One challenge to working with primates is that you only have a limited amount of time that you can work with them each day. They need to eat, rest, and socialize, there are husbandry needs, and there are usually several people working with them who are sharing time. In a typical morning, we’d come in and feed them first. The monkeys lived in a large social group in large indoor-outdoor enclosures (they did then, and still do at Georgia State, where I am now), so we’d call them inside and offer whoever we were working with the opportunity to come into the testing chamber, which is attached to the indoor area in their home enclosure. They only participate if they choose to, so if they came in, we’d run the study and then let them back outside to join the rest of their group. Usually, my afternoons were spent doing writing, data analysis, coding of video tapes, and other non-data collection tasks that needed completing. Finally, we’d give them their evening meal before leaving.
HS: Tell me a little bit about the history of these monkeys.
SB: Frans had established this colony of monkeys back in, I think, the late seventies or early eighties at the University of Wisconsin, Madison. When he moved to Emory he brought them with him. By the time I started working with these monkeys, they had been living in their groups for close to 20 years. We keep them in groups that are as close to species-typical as possible. Capuchins live in matrilines and females are philopatric (meaning they stay in their natal group their whole lives), so we keep offspring with their mothers and don’t move females around, although we occasionally bring in new males, because males move among groups.
HS: Did you continue to use the individuals used in this experiment, for later experiments?
SB: Yes, as a matter of fact some of them are still with me. The first capuchins came to Georgia State about the same time that I did, and then a few years ago, when Frans was preparing for retirement, we took one of the groups and the other went to the San Diego zoo.
HS: So you still use some of these individuals in experiments today?
SB: Yes, that’s one of the interesting features of primate research. We re-use the same individuals, which has benefits and drawbacks. The benefit is that I know their histories, including what studies they’ve done, their social relationships, and so forth, which gives us the opportunity to go back and correlate responses across different studies and look at changes over time. But of course the drawback is that these animals are highly experienced, and we have evidence that just being experienced with testing changes their behaviour. For instance, we recently got a social group from another facility that was ending their research program, and one of my students found that despite the fact that these new monkeys had some experience with behavioural and cognitive testing, they did far worse on a change blindness study. Since we can’t use naïve subjects for each new experiment, we do the best we can to account for previous experience that might have influenced their behaviour.
HS: Did you choose to work on capuchins because of that initial observation on Ozzie, or were always planning to work on capuchins?
SB: Once I decided to work with primates for graduate school, I was always planning to work with capuchins. They’re very interesting monkeys, highly social and cooperative, which makes them a great species for my research interests. In the wild they’re known to reciprocally groom, share food, and even occasionally allo-nurse, including among unrelated individuals. There are also reports of cooperative defence –for instance, mobbing of boa constrictors. They also are very intelligent, and have a large brain-to-body ratio, roughly equivalent to that of the great apes, and they do some of the things apes do, such as nut cracking and group hunting. I’ve also studied numerous other species, primate and non-primate, but I am particularly fond of capuchins.
HS: Did you have names for the monkeys?
SB: Oh yes, they all have names! Each monkey gets a name that starts with the same first letters as its mother, so it’s easy to remember kin relationships. For instance, one of my favorite monkeys is Wilma, whose daughter Wren now has a daughter named Widget.
HS: In the paper you mention that “Data were collected during testing by a second experimenter, who remained the same throughout the study”. Could you tell us who that was?
SB: It was Laura Antonucci, who was an undergraduate who worked in the lab for a couple of years. Undergraduates typically help us collect data and sometimes become very involved in the research. They often also help with inter-observer reliability, which is when someone who is blind to the hypotheses and conditions codes at least a subset of the data to help determine if there is bias in how the original data was recorded (by an experimenter who obviously knew the hypotheses and conditions).
HS: Could we also go over the other names in the Acknowledgements, to get a sense of who they were and how they helped?
HS: K. Bouxsein
SB: Kelly Bouxsein. She was also an undergraduate in our lab who worked with me on the study and did an honors thesis with me.
HS: J. Davis
SB: Jason Davis. Jason was also an undergraduate at Emory who then became Frans’ capuchin lab coordinator and worked on several studies with Frans.
HS: C. Freeman
SB: Cassie Freeman. Cassie was also an undergraduate who worked in the lab for several years. She was very involved and co-authored a paper with me a few years later.
HS: A. Katz
SB: Hmm, I don’t remember.
HS: And then you thank L. Ruttan for comments on the manuscript
SB: Lore Ruttan was a professor over in the environmental studies program who I had worked with. She did her graduate work study in cooperation, so she read the paper and gave us feedback. She is now an academic illustrator.
HS: And finally you thank R. Earley for statistical advice
SB: Yes, Ryan Earley, who is a friend of mine that I had worked with at the University of Louisville, where he was a graduate student. He is now faculty at the University of Alabama, Tuscaloosa. He is a great statistician who we asked for advice on the paper.
HS: In the paper you mention that you made video recordings of the experiment. Do you still have these recordings? Are they archived somewhere?
SB: Yes, I still have them, although they belong to Yerkes so they are not archived publicly, although there are some examples online.
HS: Yes, I remember seeing a video of a monkey throwing pieces of cucumber at the experimenter!
SB: Yes! That was Lance. She had particularly good aim. Lance is at San Diego Zoo now, but two of her brothers – Liam and Logan – are at Georgia State and also have decent aim.
HS: Do you remember the names of the monkeys you used in this experiment?
SB: Yes. We used Star, Bias and Mango, who are with me at Georgia State, Georgia, who died a few years ago of old age, and Nancy, who is at the San Diego Zoo.
HS: I was also curious about the items you used in this experiment – a piece of granite for the tokens, and cucumber and grape. Were you following what Frans had been doing?
SB: For the token, we needed something that was small and easy to trade, and different from anything else that was being used in experiments. From my ongoing research, they were used to tokens having a particular value, usually based on their color, pattern, or shape, so I wanted something that didn’t look like anything else that I had used.
As for the cucumber and grape – it took us awhile to come up with an appropriate reward. It’s important that the lower value reward be something that they actually like. So in order to do that, we needed a food that they were willing to eat, but didn’t like as much as the better reward. To determine that they liked it, our criterion, which I still use, is that if we handed them 10 consecutive pieces of something, they will eat them all. The other food they need to prefer when given a choice between it and the lower value food. To determine this, we use a dichotomous choice task, in which we present two foods in different hands and let them choose one to eat. We counterbalance which side each food is presented on and consider it a preference if they choose one food over the other eight out of 10 times on two consecutive sessions. In addition, we used red grapes because we wanted it to be visually distinct from the cucumbers.
HS: Do you remember roughly how long it took you to write the paper and when and where you live most of that I think.
SB: The manuscript was written relatively rapidly, because it’s a pretty straightforward result. We wrote it up in the winter or early spring of 2003 and had a few weeks for revisions in the early summer. At that time, both of us were traveling, Frans in Europe and me in California, so we spent a week trying to find a time when we could talk about the revisions from half the world away.
HS: Why did you get only 2 weeks?
SB: It was Nature’s policy, but the revisions were very straightforward.
HS: How did you communicate with Frans at that time? I’m guessing there was no Skype then.
SB: No, there wasn’t! We used email, but remember that no one expected to be able to talk from afar at the time, so it probably sounds surprising today but certainly wasn’t unusual then.
HS: Since the revisions were minor, the paper didn’t change much from the first draft you submitted?
SB: The revision was mostly in terms of how we framed the study. The reviewers made some really nice suggestions about other papers we could look into, particularly in the economics literature, so they were extremely helpful.
HS: What kind of attention did this paper receive, in the popular press and academia, when it was published?
SB: There was an enormous amount of media coverage, which I suspect was partly because it was a very slow news week. I did something like 60 interviews over the week or so after the study came out. Talking to the media was new to me, but Frans, who was used to talking to the media, gave me great advice. It was also very important to him that I be the one who spoke to the media, so I did almost all of the interviews. It was interesting because it was the first time I had personal experience with science reporting in the popular press. The main thing is that the story line gets simplified because it has to be – most people don’t care about all the details that those of us closest to the research think are important – so learning how to do that is important. We got a lot of push-back, in part because a lot of people interpreted our result as monkeys having a sense of fairness, which never claimed, but of course the news media framed it that way, which probably biased how people interpreted it.
HS: Did the push back come from people in academia?
SB: Mostly other people within academia, because of course scientists are supposed to question new results. The standard response from people outside of academia was, more or less: “Yes, but of course we already knew this! You know, my two dogs get upset when I give one of them a bigger bone than the other, or, my cats get upset when I give one more attention than the other.” The push-back from academia was in part the concern about whether this was truly a sense of fairness, and in part the assumption that this resulted from a very higher order cognitive process that there was no evidence for. For instance, the reaction could result from the animal thinking about what they have, what their partner got, and comparing the two to decide whether to be upset, or it could be an emotional reaction- they just get frustrated and respond, and maybe throw something! I think it’s the latter, but back then emotion wasn’t discussed much in relation to animals. We still don’t know for sure; we need more work to figure out exactly what is going on.
HS: Were there also formal published responses to the paper?
SB: There were two of them. One was by Joe Henrich, whose major point was that you couldn’t call this inequity aversion in the economic sense, because the animals were only showing one side of inequity aversion, responding when they got less, and not when they got more. Of course, this was true, and indeed, our study wasn’t designed in a way that we could really test that. Since then, we have looked for this other side of inequity, and we see some evidence for it in chimpanzees, but not capuchins.
The second comment was from Clive Wynne, who argued that they were just frustrated by the different values of the rewards that they saw, but weren’t paying any attention to what their partner got. That is, it was a contrast effect, based on frustration because they saw better rewards that they did not get, rather than a social effect based on what their partner got. Although clearly they care when there are better rewards they can’t get, we also saw a different pattern of responses when they saw better rewards versus when their partner got them. Since then, we and others have done numerous other controls that show that capuchins do respond differently when their partner gets that better reward. But I think that the underlying mechanism to both inequity and contrast effects is the same – individuals get frustrated when they see something better, either because it’s available or because their partner got it, and they respond; the main difference is what the referent is, something in the environment or what someone else got.
HS: This partially answers what I wanted to ask you next, which is, how did this paper influence the future course of your research. I notice you have a paper on inequity aversion as recently as 2016.
SB: Definitely! The Nature paper got a lot of attention because it was the first paper to demonstrate inequity in a non-human species, but it actually proposed more questions than it answered: Why are they doing this? What is the underlying mechanism? Does this response occur across species? What happens if they get more than their partner instead of less than their partner? What is the relative impact of social inequity versus contrast effects? I’ve spent the better part of the last 15 years following up on these questions. Our 2014 review in Science provides a good summary up to that point. We’ve found, for instance, that it does not occur in all primates, and seems to occur most in species that routinely cooperate with non-kin. We also don’t see a response to getting more than a partner in most species, but do in chimpanzees. We also see a lot of individual variability, which is often hidden in our inferential statistics. One of my big questions is how differences in context or the individual that leads to different reactions. For instance, chimpanzees’ responses seem to be influenced by their rank, their personality, their social group, and how the experiment is set up. Whether capuchins cooperate depends on whether their partner dominates better outcomes or they get them sometimes, too. We’ve got ongoing studies looking at other species’ responses and how the social context influences their responses.
HS: You have already sort of answered this question, but I wanted to ask you , whether today, 13 years after the study was published, would you say that the main conclusions still hold true, within the limited scope of these experiments?
And the second part of my question is: If you were to redo these experiments today, would you change anything about them, given the availability of new technology, the theoretical advances and other development?
SB: I don’t think that I would. Of course it didn’t answer every question, but it needed to be a simple, straightforward experiment to see what was going on. Indeed, even today many of the studies looking at new species or new contexts use a very similar design including an inequity condition and controls for both baseline and contrast. We have run the contrast control in several different ways since then. In the original study, in the contrast condition the subject was alone and the better rewards accumulated in an empty enclosure next to them. We chose this as it seemed like the most conservative control, since by the end of the session, they had 25 grapes sitting right next to them that they couldn’t reach! However, these monkeys are social, and in the rest of the conditions they were next to a partner, so it’s possible that being by themselves influenced reactions. Since then we’ve used a social control condition, in which two monkeys are next to one another and both are shown the better food but get the less good one, and get the same result. That kind of convergence is good in science as it suggests that their response is relatively robust.
Another question we had was whether it made a difference if you show a reward to the subject before they do the trade, so they know what you are “offering”. We tested this and found no effect. One thing we have changed was that in the original study, we had both cucumbers and grapes in cups in view of the monkeys for three out of four conditions, including both the inequity and contrast conditions, but not the baseline control condition. In that one, both monkeys got cucumbers and so we only had cucumbers in view. Since then, we’ve always displayed all foods in all conditions, whether or not they are ever used in that condition, to control for what is visible.
HS: Your answer to the first part of my question is that the main conclusions of that paper still hold true, more or less?
SB: Yes. We and others have repeated this study in different forms on capuchins and continue to get the same results.
HS: Would you say the paper had an influence on your career – what you did immediately afterward, where you went for your post-doc etc?
SB: Absolutely, and in two ways. First, when a paper that gets that kind of attention you get a degree of visibility helps immensely for whatever the next step is, whether it’s a postdoc or a grant or a faculty position. Second, it gave me a foundation upon which to build a career. Of course, I have moved in many new directions over the years, but it has remained a topic that both gave me direction when I was establishing a career and has continued to influence how my work develops. I think for a young researcher, finding a research focus is probably the most important benefit.
HS: This paper has been cited over 1000 times. Do you have a sense of what it mostly gets cited for?
SB: You know, I actually don’t know. Obviously it’s typically cited in reports of inequity in other species, but that’s a small percentage of the citations. It has also gotten a lot of attention in other disciplines and the popular press in relation to the evolution of human behaviour. For instance, considering the latter, John Oliver did an episode several years ago on the gender pay gap in the US, and included the clip of Lance throwing the cucumber. Regarding the latter, there has been an increasing focus on economic inequality and the evolution of moral behaviour. Since fairness is one of the key human moral behaviours, and we see at least some aspects of it in other species, it is a topic that is covered a lot.
HS: I want to now talk about a few specific sentences in the paper where you sort of suggest that there are things that were unknown at this point in time and that might be areas of future search. I want to know what’s happened since then.
The first is about the difference between females and males and how they responded to these tests. You say “Our limited sample size did not allow a conclusive comparison of the sexes, but independent evidence indicates that capuchin females pay closer attention than males to the value of exchanged goods and services.” Have you looked at the difference between females and males subsequently?
SB: Yes, and further research did not bear that out; male capuchins’ responses were no different than females’. In addition, some of our initial chimpanzee studies found sex differences, but interestingly they went in opposite directions(1, 2) and other work found no sex differences at all. Thus I suspect that there is some other factor or factors influencing responses and the sex difference was a spurious finding due to small sample sizes.
HS: The next is about the latency to exchange. You obtained some results which you say were not really clear because they are contradictory – the capuchins completed exchanges most quickly in inequality tests and more slowly in the control tests than inequality tests. You say this result is hard to interpret. Do you know more about latency to exchange now, and what influences it?
SB: We have continued to measure latency to exchange in every study and haven’t seen any consistent response, so at this point I would say that they change their willingness to refuse, but not their latency to exchange.
HS: Towards the end of the paper you talk about the motivations for these behaviours. You say: “Although our data cannot elucidate the precise motivations underlying these responses, one possibility is that monkeys, similarly to humans, are guided by social emotions”. Today, do we know more about the motivations for these behaviours?
SB: Not really. In particular, it’s really tricky to empirically study affective responses in other species. People are starting to look at hormonal changes in the context of inequity, and there is interest in studies of physiological changes, such as heart rate and pupillary dilation, and those will help. Our best evidence for any social emotions continues to be changes in their behaviour, such as the evidence from Friederike Range’s lab that dogs are less likely to interact with the experimenter and the partner dog after getting less than a partner in an experimental test.
HS: When you say you haven’t been able to do it so far, is it because of practical difficulties in doing the experiments.
SB: Yes. Our monkeys live in long term social groups, and we don’t do any invasive research, so there are some studies that we just can’t do. However, my post-doc, Marcela Benítez, is interested in hormones, which can easily be collected non-invasively, so we are beginning to look at whether there are hormonal correlates of some of their responses.
HS: In the 13 years since this paper was published, have you ever read the paper again?
SB: Oh, yes! I use it in class, and, of course, I re-read it carefully when Frans and I were writing our review that was in Science in 2014.
HS: If you compare this paper to the papers you write today, do you notice any striking differences in the way you write?
SB: I don’t think my writing has changed much. Frans is an outstanding writer, and I learned to write a lot like him and continue to use the same style. The thing that always stands out to me is how short it was. I generally dislike papers that short because it’s difficult to include all of the important details, but that paper described a very straightforward experiment so it was possible to provide sufficient information.
HS: Would you count this as one of your favourites among all the papers you’ve written?
SB: Yes and no. I like it both because it was such a foundational paper in my research and because it was so straightforward, however I generally prefer my studies that go into more depth. For instance, my 2010 paper in Animal Behaviour had an enormous number of conditions and answered a lot of really important questions about what was causing variation in inequity responses across chimpanzee groups, and the 2011 paper on the Assurance Game in PNAS distilled three years of data in three species and provided important insights into variation across the primates. But, I will always have a soft spot for the Nature paper.
HS: What would you say to a student who’s about to read this paper today? What would you say he or she should take away from it from this paper published 13 years ago? Would you add any caveats they should keep in mind when reading it?
SB: Reading a paper from the past and not reading the follow-up literature won’t give you a complete picture of the phenomenon. Subsequent studies have provided quite a bit more shape and nuance to our understanding of this behaviour and how it varies across species. That being said, the flip side is also true, that it’s important to read the foundational papers from a field to understand where the field came from and how it developed. I also think it’s important to recognize that this was a very simple experiment, and sometimes a simple study is what is needed. And lastly, I came up with this idea from watching my animals, and there is no substitute for knowing your species and your particular animals in recognizing what are the most interesting questions to ask.