The evolution of (un)fairness: the influence of inequity on cooperation

The evolution of (un)fairness: the influence of inequity on cooperation


Okay, thank you guys. So we’re gonna get started
with the second session or the second panel of today’s event. Thank you for
coming back in from coffee and treats. So moving away or moving in sort
of a slightly different direction, the second panel is focusing on
the primate origins of human decisions. And brings together three
scientists working on quite different aspects of decision
making in non-human primates as well as a little bit of work in human primates. So our first speaker
today is Sarah Brosnan. Sarah Brosnin is, is an associate
professor of psychology, philosophy and neuroscience at Georgia State University. And her work explores mechanisms,
the mechanisms that underlie cooperation, reciprocity, inequity and a whole suite
of other kinds of economic decisions in nonhuman and now increasingly
starting to work in human primates and she does all of this from
an evolutionary perspective. And one of the things that I really like
about Sarah’s work is the degree to which it really highlights the interdisciplinary
that we see in the social sciences. So, from her affiliations you can
see how interdisciplinary she is. But it really also comes out both in where
she publishes, the service she does as a, as editors at journals like,
the Proceedings of The Royal Society. Biological sciences, animal cognition,
evolutionary psychology, science as well as journals like social justice research
and frontiers in decision neuroscience. In addition to a whole suite of accolades,
Sarah also has the perhaps, well, it’s a distinction of having
what I think is actually the, the best video of a scientific
result ever put on the Internet. Which you guys are all gonna have
the chance to see today, and which, as of, as of this morning when I checked had
nearly 750,000 individual hits so, Sarah we’re looking forward to your talk.>>[APPLAUSE]
>>Thank you, Meg. I’m really excited to be here and to get the chance to tell everybody
what I’ve been working on. I think an institute on social sciences
in an absolutely fantastic idea. Obviously I fit right in to, to the idea of doing these as a group
rather than in individual disciplines. And today what I’m gonna talk about is some of the work
I’ve done on decision making. And in particular, I’m interested in
social influences on decision making. So how do the individuals
that are surrounding you influence how you perceive what you have
and what decisions you make about it? And I’m particularly in
the concept of fairness, in part because it’s something
that people care largely about. I probably don’t have
to explain this to you, this is a some what more dated
photo of Occupy Wall Street but people really deeply care about
the equity of distributions. And they not only care when
they’re the ones getting less but they also care when they’re
the one’s getting more. And a question that arises again and again
is, why should people care about this? After all,
if you’re getting enough to survive and reproduce, why should you care if
somebody else is getting more? And yet, we know people do. But of course,
an important question is why is this? Is this some sort of cultural phenomenon
in which case it’s the way in which we’re raising our offspring or
does it go more deeply than that? Given that I studied on human primates you
can probably guess that I think it goes more deeply than that. And if it does, we should be able
to see some evidence of something related to a sense of fairness
by looking at other species. Now, of course, they aren’t going to have
a sense of fairness in the same way we do. Quite obviously, I mean they can’t talk so
they can’t sit around and talk about, you know, what Rousseau said or
anything along those lines. But we can look at some of
these basic behaviors that may tie into the human sense of fairness. But of course, there’s a problem there
because fairness is a social ideal, so even in humans how do you really study it? It’s not particularly tractable to
break down for empirical research and I’m sure that there are many
ways that you could do this. But the way that we got into it all starts
with a monkey named Ozzie, who was, he was Capuchin monkey,
he still is a Capuchin monkey. He was, [LAUGH] he’s the alpha male or was was the alpha male of
the group I was working in. And I was following my graduate adviser’s
advice to sit out and watch the animals. And let them tell me what was important
about their lives, to give me ideas about what was the most, what were the most
interesting things to study. And one afternoon when I was sitting
out there, feeding them peanuts, which they like quite a lot. Ozzie was coming over and
trying to get more and more peanuts but of course,
as the alpha male he typically could but I was trying to give peanuts
to other individuals. And eventually he ran away and
he started coming back with things and trying to trade them with me in
order to get some of the peanuts. So he brought monkey chow,
then he brought an orange peel. And that was interesting because he seemed
to be increasing in value, but, of course, all of these were things that were much
less valuable to a Capuchin monkey than to a peanut. Then he came back with
a whole section of orange. It was about a quarter of an orange and
he shoved it through the mesh at me and reached out his hand like
he wanted the peanut. And that caught my attention, because if you had asked me prior to that
whether Ozzie would choose a peanut or a quarter of an orange,
I would have said a quarter of an orange. I see you nodding over
there you agree with me. So this didn’t really make sense, why was
he willing to give up something that I thought he actually preferred
in order to get a peanut. Well, one possibility of course was that
he wanted the peanut because everyone else was getting it and so that got at
the idea we had for looking at this. Possibly by looking at how individuals
respond to unequal outcomes we could find an empirical way to test whether or
not they had this basic level of fairness. So, obviously just being upset when you
get something less is not the same as a sense of fairness but
it gets us started in that direction. And conveniently at the time from my
dissertation I was doing a project with Capuchin monkeys where we had them
trading tokens back and forth for differently valued foods. And what we were interested
in was whether or not they could learn that different
tokens had different value, which is completely unrelated to
what I’m gonna talk about today. Except for the fact that if you’ve already got
a monkey in here trading a token with you. Why not put another one next to them? Have them trade a token and give them
a different food reward for it, so that’s what we decided to do. So just to give you
an overview of the test and for those of you who’ve read any of this, there are probably close to a dozen
different versions of this we’ve done. But I’m just gonna focus on a few because
I don’t have too much time today. First you need to find out of course,
how the animals respond when their partner gets the same thing
as them for doing the same test? So this is what we call the equity
control, how do I respond when I exchange and get a cucumber after watching you
exchange and get a cucumber as well. And I should say that they all
like cucumbers, if you walk in and hand then 10 or 12 pieces of cucumber
they happily eat all of them. And that’s important because we’re not
interested in how they respond when they get something they don’t like and
somebody else gets something they like. We’re interested in how they respond
when somebody gets something better. So of course, we compare this to
an inequity condition, where I trade and get my cucumber after watching
you trade and get a grape. So now, I’m getting the same cucumber but you got something better and
we’re interested to see whether there’s a difference in my response to this
cucumber across these two conditions. But we really need one more
condition because there are actually two things
that are different here. One is that you’re getting a grape and
the other one is that my attention’s being drawn to the grape,
irrespective of who’s getting it. Now, when we do these tests,
we always have a bowl of cucumber and a bowl of grapes sitting out front so they
see grapes even in the equity condition but their attention’s
not being drawn to it. And there’s a long literature
going back to Hernstein and psychology showing that
there are contrast effects. Individuals compare what they’ve got
to other things that they’ve had in the past or other things that are out
there and that’s not social necessarily. So we include what we call
an equity plus grape condition. In that case, we wave the grape in front
of them until they gesture towards it which lets us know they saw it. Drop it back in the bucket, do the trade
and give them the cucumber and then we do the same thing for the partner. So we’re drawing their attention to
the grape actually twice as often, so if anything we’ve biased
it against our finding. But then everybody only gets a cucumber so
it’s still equitable. So by looking at these three conditions we
can try to tease apart how any changes in response here are due specifically
to the partner getting the grape. So, I’m going back to that,
famous video clip. I’m gonna show you here,
what the inequity condition looks like. So, this is a monkey named Lance
right here, and he’s gonna trade and get a cucumber. And then he’s gonna watch his partner,
his name is Winter, trade and get a grape. Where’d my mouse go? There it is. So, you can listen to it with sound on
the video if you, in YouTube, if you want. I like to talk over it. So, he ate his cucumber and
now he’s watching Winter. Winter trades. Winter gets a grape. And you can tell that he sees that. So Lance, come backs,
comes back and trades. He gets a cucumber. Watch what he does? He takes a bite, so he accepts it.>>[LAUGH]
>>And then he makes his preferences known.>>[LAUGH]>>It is important to point out, he’s the only capuchin I know who
can do aimed overhand throwing.>>[LAUGH]
>>But, this response is not atypical.>>[LAUGH]
>>If you look at the actual empirical data,
rather than just how they’re responding, you see that they refuse the cucumber
twice as often in the inequity condition, as they do in the other two conditions. And this has held true
across multiple studies now. So, they seem to care about
what their partner got. And it’s not just because their
attention’s being drawn to that grate, which actually,
I find interesting in and of itself. They put up with us
tricking them basically. But, they get really upset when
their partner gets something better. We’re actually going back now, and trying
to code these behavioral responses and see if we can figure out whether or
not there are differences in behavioral response, in behaviors, in the 80% of
the cases here where they aren’t refusing. Because refusing something that
you actually like is very costly, so one might anticipate that there
are other sorts of agitation or stress responses that we are gonna get,
even when they are willing to accept it. So next we tested this with chimpanzees, these are photos from a long time ago
taken at the Yerkes field station. There actually are chimps down here in the
shade, it gets hot in Atlanta too, and so we tested these two groups in
particular,because this group was an established social group, that had
been together for close to 30 years, so all the individuals in the group
had grown up together. Whereas in this group, the individuals
had been put together as adults, about six years prior to the study. Which means they had a good relationship,
but it’s not the same as
having grown up together. Chimps typically would have a much
more long-term relationship. But otherwise,
these groups were pretty similar. They had the same number of animals,
same basic demographics, and so forth. And we also looked at some
pair housed individuals, which is an exceedingly atypical
social situation for a chimpanzee, because we wanted to see how social
housing influenced their responses. What we found was interesting, looking
just at the short term group in orange, and the pair housed group in pink. You can see that they responded
more strongly to inequity, although interestingly, it was a much stronger effect compared
to the equity plus grape condition. With the pair housed individuals who
don’t have any other social outlets, than the short term group, but the sample size is small enough it’s
hard to make much out of that right now. What’s really interesting is
the long-term group’s response. They didn’t respond to anything. These are standard error bars right here,
and there were ten individuals in that sample. So that shows you that was very,
very low variability. So, we got to wondering, if humans whom
we know that in responses to inequity are much exacerbated for individuals with
whom you don’t have close relationships. So maybe, these guys here have these
well established relationships and it’s not worth screwing them up
over a cucumber versus a grape. But, the problem is that these
guys relationships were confounded with how long they’ve been living
together six years, versus 30 years. So, we tested this again at
MD Anderson Cancer Center in Bastrop, Texas who has all these social groups, who
have lived together in this similar sort of outdoor corral system
since the late 70s. And so, these guys had been together for
close to 30 years, so we could disentangle any
differences in social interactions, from how long they’ve
been living together. So, we ran the study again,
we had 16 new chimpanzees, which for those of you who work with humans, is
a really big sample size for us, trust me. And we got something totally unexpected. There was no influence of groups,
but we found a sex effect. Males are responding more strongly
to inequity than either equity, or the equity plus grape condition. Interestingly, this difference is bigger. Whereas females didn’t
care about inequity, but they cared when we tricked them. They cared about this contrast condition,
and going back and looking at the results
from Yerkes, none of the males or female, none of the females
showed this pattern at Yerkes. So this wasn’t some sort of sex difference
that had been hidden by a couple of outliers at Yerkes. It was really different. So it became clear to us that whatever
was underpinning the variability in the inequity response, was a lot more
complicated than are you male, female, are you from this social group or
that social group. So next, we took four social groups
of chimpanzees from MD Anderson, and we were interested in how they
responded with different individuals? We thought this was relationship based,
we still, I still think it probably is, but we were convinced
that what was going on, was that individuals were having different
responses with individuals that they had close relationships with
versus those that didn’t. And that our differences between the two
different facilities were due to the fact that we were pairing individuals,
and we had tried to do it randomly, but probably inadvertently
had introduced bias. So for each one of these groups, we
paired every individual with every other individual in the social group
with whom they would pair. And then did this same with this
individual, and this individual, and so forth. And chimps won’t pair with everybody. Really high ranking individuals,
we’ve got lots of the lower ranking individuals won’t pair with them, and
some of them don’t like each other, but we got slightly more than half
of the possible pairs to form. It took us four years to do this. And we were interested in comparing their
inequity responses in each of these possible pairings, to their relationship
quality based on a separately collected, set of data on how often they were, how much time they were spending
grooming one another and so forth. And we also used the time
they lived in the same group as a separate measure of
relationship quality. So we had two measures
of relationship quality, that we could tease apart statistically
and we also included their personality. It’s controversial to some degree what
personalities actually measuring in non-human animals, but what is clear
is that personality is measuring very highly replicable long-term
individual differences in behavior. And then we had those data from a
separately collected study that one of my post-docs had done,
when she was a graduate student. So, we got all these data,
put em together, and it turns out, much to Sam Gosling,
who was participating in this delight. It was personality that was driving the
responses, not their relationship quality. So chimps like Keno here, who have
really good relationships with lots of individuals, they’re always grooming,
they’re outgoing, they spent lots of time
interacting with others. Were very responsive to inequity,
but not so much to contrast. Whereas chimps that spent lots and
lots of time interacting with humans and were really good at training, and things like that,
tended to be more responsive to contrast. Which actually makes you wonder
what they were responding to. Do they see us as the target, or was it really the fact that
it was just the contrast? But, that’s a separate question. So again, it does seem like it’s
these stable individual differences, that are driving responses. But of course, even in this case, where we
are trying to pair everyone with everyone else, we’re still missing one of the key
features of chimpanzee social life, which is you don’t usually
see chimps like this. You see them like this. In a big cluster. All together,
all interacting with one another. That’s how they,
that’s how they’re typically living, but that’s not how we’re testing them. So what we’re doing now in these
data are still in progress, so I don’t have any
actual data to show you. But both with chimpanzees at MD Anderson
with Katie Hall who is my post-doc, and with capuchin monkeys at Georgia State
with Julia Watsec who is one of my graduate students,
we’re doing the same thing, but we are doing it in a group setting. So I’ll show you a video
slip of how this looks, but we’ve got all the chimps who were outside,
there’s a bucket here of. And we’re actually not using cucumber and
grapes for every one of them, but I’m gonna keep calling it that. You’ve got the grapes in one,
you’ve got the cucumbers in the other, and any individual can come up and exchange. And we actually have
three foods in this case. So what we’re interested in is,
how much does chimp A, say Joey, here. How much does Joey exchange for
the medium value food, when one or more individuals are getting
a better food than him, and when one or more individuals
are getting a less good food than him? So we had to introduce three foods since
we obviously can’t control how many exchanges they’re getting and so forth. So if you look at this video, there is
my cursor, Joey’s been getting the lower value food, while his partner here has
been getting the higher value food. And he’s having a little
chimp temper tantrum. [LAUGH] So, we seem to be getting nice
data showing that they do care, and they’re paying attention to one another. It’s hard to see but
there’s another chimp back there too. Just to give you another example of what
we’ve been getting, these are two chimps who’ve known each other for
a very long time, named Mary and Martha. So Martha here is trading and
getting orange, and Mary is trading and getting celery, and I don’t know if
you noticed, but Martha traded and got her orange and ate it. Mary’s celery is piling up here. She’s not eating it. [LAUGH] And she also, she, watch this. Mary just stole Martha’s orange, [LAUGH]
which is something that we didn’t see in the paired test, even though
they were sharing an enclosure and next to each other in just the same way. So we’re hoping that the social dynamics
that we’re gonna get from this will show us more about what’s going on, and
possibly how relationship quality may be interacting with some of the personality
results we’re finding and so forth. So all of this raises
an interesting question. We’ve got clear evidence that they
respond to these differences, but what is the function of this behavior? Why did it evolve in the first place? I can give a whole one hour lecture on
that, so I’m gonna summarize in one slide at the risk of getting rid of
a lot of the critical detail, you can come find me afterwards. We’ve done a, there’s in the literature,
especially in economics, it’s been argued that it’s
related to cooperation. Having a sense of inequity allows you to
identify good cooperative partners, and continue to interact with those partners
and leave partners who are not so good. So we’ve done lots of studies looking
at how inequity influences cooperative ability in these guys and
we find that it does. But we’ve also been doing
a comparative approach, looking at how all these different
species respond to inequity, and comparing it to the tendency they have to
cooperate in their natural environment. And what we find is that
there’s a pretty good match. So, individuals that cooperate with
non-kin, so group hunts, coalitions and alliances, either to attain rank, or
to keep out individuals from other social groups and so forth,
tend to respond very strongly to inequity, whereas those that don’t have such
strong tendencies to co-operate don’t. And it seems to correlate with that
better than things like brain size, or size of the social group. Now, admittedly we only have nine
different groups up here, so there’s, there’s certainly some room for
error, but it looks promising. One of the most interesting details
actually is the fact that the cooperative breeders we’ve tested are not responding
to inequity, but we think that’s because in all three studies that have been done,
we’ve used mated pairs. And invaded pairs, it’s very
expensive to go find a new partner. So losing your partner and
your entire reproductive output for the year over a cucumber
doesn’t seem to make sense. [LAUGH] so. So we think that the two
of them are related. However, everything I’ve talked about thus
far is how you respond when you get less, but fairness is not just responding
when you get less than someone else, it’s also responding when you get more. In humans, if I went around and got upset
every time I found out I was making less money than somebody but didn’t care when I
found somebody who is more deserving than me who is making less than me,
you wouldn’t say I was fair. You might say I was a jerk, or
that I was self-interested, but that’s not the same thing. And so we’ve also been looking at how, or whether,
individuals respond when they get more. Of course it’s, the negative responses to
getting less are related to fairness but they’re, they’re just one step. One other thing. So essentially what we’re interested in, in these original studies, is, how does
this guy respond to getting the grape? How in that video you saw, how does
[INAUDIBLE] respond to the fact that she’s continuing to get the grape
while her partner gets the cucumber? So one of the other conditions we’ve done,
is a high value test where both of them exchange and get a grape, and
using the same logic as the inequity test, if Winter responds differently to this
grape when her partner also gets a grape, than to this grape when her
partner gets a cucumber, then we might argue that
she notices this inequity. Capuchins don’t seem to. But if you look at chimpanzees, we do see evidence that the male chimps
in that study from M.D. Anderson, who if you recall were the only ones who
responded to getting less than a partner, also refuse m-, somewhat more often,
and it is statistically significant. They refuse about twice as often when
their partner gets the cucumber and they get the grape,
than when they both get the grape. So this seems to indicate
that the chimps notice this. Now to put this in perspective. First of all this doesn’t
mean that they’re altruistic, it just means that they
notice when they get more. Could be that they’re afraid of
repercussions later in the day. The other thing is, if you look at their
responses to inequity when they’re getting more compared to the responses to inequity
I’ve showed you when they’re getting less, they’re clearly much more
upset by getting less. I would argue that that’s
true of humans too. Not everybody agrees with me. But there is, there is a change here. But the important thing is, yes. We do have evidence that they’re
noticing when they get more. We wanted to test this in a little more
detail using a, a paradigm that was more designed to do this using
a version of the ultimatum game. I’m gonna skim over this, cuz I’m
betting many of you’ve heard of it. But essentially, there are two players,
a proposer, and a responder. A repose, proposer splits a cash
endowment any way that they want. If the responder accepts the division, then they get the money as
the proposer suggested. If the responder refuses it,
nobody gets anything. And in human societies, generally
speaking, when people offer 40 to 50% and most people won’t accept it
if it’s less than 30 or 40%. There is a lot of cultural variation and
variation across game types, but people tend to offer more than
the minimum, and then seem to want to, seem to expect more than the minimum. And we can’t do this with chimpanzees of
course because we can’t explain to them how they’re supposed to take
these grapes or whatnot and divvy them up between individuals, but we can do a limited form game, where
the proposer was trained to choose tokens, one of which represented a 5:1 split and
one of which represented a 3:3 split, and then the responder can decide whether or
not to accept the reward. The way we did this, for those of you
who aren’t used to working with chimps, is we first trained the proposer
that the blue token gave them five, five of the rewards, and the red
token gave them three of the rewards. Then we put them in
a preference test condition, where they were sitting next to a social
partner who was from their social group. They were separated by mesh so that
they couldn’t steal each other’s food. And we had two conditions. One was a preference condition. And this was a, now we did this to establish a baseline
preference between the options. The proposer chose a token,
returned it to the human experimenter, who then divided the six
actually it was banana slices. Six banana slices that were out
there between the two animals, pushed them forward and gave them
the food the way it had been determined. We compared the proposer’s choices for
the blue token versus the red token, to the ultimatum game condition where
after the proposer made a choice, they had to pass the token through
the mesh to their partner, who then had to quick return
it to the experimenter. And if they returned it to the
experimenter, the experimenter divided up the tokens the banana slices
the way that it was indicated. What we expected to see was responders
refusing to return the token. That’s not what we actually did see,
what we saw was perhaps more interesting. We only did 12 trials, which is a very,
very small number of trials, but we didn’t want to get them, we didn’t want
to see too much in learning affects, and what we found was that the chimps showed
no expect, as we expected a very high preference for the five one token when
it was just the preference condition. But that flipped in the ultimatum game
condition, and they prefered the token that gave them only three, and gave their
partner three, about 65% of the time. And we also tested this in young
children in a daycare setting who who were tested with kids
from their daycare class so we could also sort of control for
the level of a relationship. And we used stickers instead
of banana slices and we got the same thing with the kids. The kids also never ever refused and
they flipped their preferences. So we would argue that what we’re seeing
here is, that in a non-anonymous, non-stranger, non-one-shot situation
with someone you know you’re gonna act, interact with later today and
tomorrow and the day after that and so forth, this may be a more
rational response. So they aren’t refusing, but
we did see some communication. We saw chimps doing things
like banging on the mesh and spitting water on one another. Kids were saying things like,
why didn’t you pick the other one? That’s not fair which seems to
be rather obvious communication. So it might have been
a subtle form of punishment. You didn’t pick the one I liked. If you don’t do this in the future, maybe
I’m not gonna reward you the way you want. And we would argue that this
is actually much cheaper and better, you’re better off in the long run
doing this when it’s an individual where you have a long term relationship. Because in this case,
then you can change their behavior and continue to get the better reward,
whereas if you just refuse and don’t play with them anymore, you lose
out on future opportunities as well. So taking all of these results together,
recently, Frans de Waal and I made an argument that the evolution,
we do see evidence for the evolution of fairness
in other species. We see good evidence for it. That doesn’t mean, again, that other
species have a sense of fairness. But what we propose is that
the first thing you would expect to emerge is what we call First-Order
Inequity Aversion or First-Order Fairness. Basically you’ve got two individuals,
one gets something more, one gets something less good. The one who gets something
less good protests. They throw their cucumber
at the experimenter. They get agitated and
pitch a temper tantrum. They steal their partner’s orange. They do something that,
that shows a protest. But individual A doesn’t
really do anything. This is not a sense of fairness in and
of itself, but this is a precursor. You’re noticing these different rewards
and you’re, you’re upset about it. Presumably that protest is trying
to rectify the situation, but that doesn’t need to be the case. In the second order, you have the same
initial situation, you get more, then I get less, I start to protest. But in this case, individual A tries to forestall B’s
protest by equalizing the outcomes. It doesn’t mean they
have to share everything. I intentionally put only a single
grape out of that bunch here. It could just be something small
to get across the point, you know, we’re trying to help. Yes, you know, your salary’s much less
than everybody else’s, but we’re gonna give you a small bonus this year and
hopefully, you know, do some cost sharing. Hopefully they’ll feel better about it. If that happens, then individual
B’s protest may go away and, importantly for both of them,
the relationship continues. So we think this has a very strong this
is probably very strongly selected, because individuals can maintain this
cooperative relationship and keep going. Whereas if there’s just protest and
individual A doesn’t do anything, the relationship may pall, fall apart. Which again, means you don’t get these
continued benefits from cooperation. Of course, the only evidence we’ve seen
for this, outside of humans right now, is in the chimps. I don’t think that’s the only place
we will see evidence for it, but I suspect that this requires
a high enough degree of planning, forethought, maybe ability to inhibit
because you have to give up something you really like in order to forestall this
protest that we might not see it expect, except in some really highly
encephalized species. Things like elephants, cetaceans,
probably the other apes, and so forth. And of course in humans we’ve
taken this much further. Humans have unparalleled
ability to plan and so forth. But by studying these other species,
we can learn about the roots of human fairness, which may help us to
better understand situations like Occupy Wall Street, because now we know
it’s not just a cultural phenomenon. But we can start thinking about
the underlying biological aspects of it. So this work takes an enormous amount of
time and effort and money, so I would like to thank all the people who’ve worked on
all these projects through the years, and thank you.>>[APPLAUSE]

One thought on “The evolution of (un)fairness: the influence of inequity on cooperation

Leave a Reply

Your email address will not be published. Required fields are marked *