Primate Origins of Human Decisions


I’m opening the floor for questions.>>[LAUGH]
>>Curtis.>>So the, the cool [INAUDIBLE], the
network, the joined network [INAUDIBLE]. So these things [INAUDIBLE]
>>No, sir.>>I mean I yell pretty loud, so I don’t know if a microphone,
a microphone is really required. So it looked like you were adding
these constrains sequentially. And it seemed like that
sequential order was arbitrary. Now I might be wrong on
both of those points. So I guess really my question is,
are these constraint functions and the difference in square that you reported
on that graph added sequentially? And if so, how do you determine
the order that they’re added?>>That’s a good question. I would.
>>Can I, can I contextualize it a little bit maybe? Cuz one of your claims was that the dual status constraint functions
started to matter more in in unstable groups, especially comp- or
the dual aggression, sorry network, started to matter more in unstable groups,
especially compared to the dual status. But, if you added those, if you look
at that graph you have a big drop off in dual from, from independence. To dual aggression and
almost nothing from status. So if those are added sequentially, then we can’t actually make claims
about the status network because it, it’s not independent of the aggression
network to get, you know what I’m saying? Anyway so, yeah.>>I’m.>>I’m not,
I’m not exactly sure what you’re asking so maybe we could talk after?>>So
do you have to sequentially [INAUDIBLE]>>They did add up, yes, they do add up sequentially. We work with the department of sta, st, statistics here a Davis pushing
[INAUDIBLE] and his, and his students. They do get added sequentially.>>And how do you determine that order?>>I think it was pretty
arbitrary in doing so. So I’d be happy to talk to you more
about your particular question.>>So I have two questions, one for
Sarah and the other for Brenda. So, Sarah I’m curious that
you are your studies, how’s the value of the items
evaluated by these individuals, and are they evaluated as the other value
in terms of how much do they prefer it, or is it actually always evaluated as
a relative value with respect to what the other the pattern that are
getting though, this is my question for you, Sarah, and my question for
Brenda is in the social network you were studying seems like there’s no
immigration or emigration allowed. So, in that case, if immigration or
emigration are, are allowed, do you see that will improve
the stability of the social network, and then reduce the probability of
the social collapse, society collapse. Thank you. So, great question. And I didn’t get into it because
of the time constraints but what’s really important of course is that
if you’re studying these relative values, you have to be able to say with some, with some consistency that everybody
has the same set of values. So we’re looking at how they prefer
the foods within themselves. So it’s a within subject’s evaluation and we also do it for each individual
separately in the social group. So, we don’t use the the lumped values of, oh on average the individuals
prefer one thing over the other. So, what we do is the least preferred food
is a food that they will accept if we walk in when nothing else is going on and
hand them 10 separate pieces of the food, they’ll eat all 10 of them
without pausing or hesitating. A food that is more preferred
than that on two separate ways. We do dichotomous choice tests
between the low preferred food and what we think will be
the medium preferred food. And we alternate back and
forth to account for side biases. And they have to prefer the medium
preferred over the low preferred at least 8 out of 10 times on both days. To make sure that it’s not only preferred
but it’s consistently preferred and same for between medium and high. And we have done studies, these guys show very consistent
preferences over the long term. So if prefer grapes to cucumbers
today I’m going to in a week and going to in 10 years and the other thing
we do is one, I always say cucumber and grape in the talks to keep it simple. But we don’t actually already
always use cucumber and grapes. We choose foods based on what every
single individual in the group prefers. So if we have one individual who doesn’t
show a strong enough preference for a purported median value food
then we keep going until we find a food where everyone does. Now of course what that means
is that I can’t tell you for certain that the difference in
preference value between the low and the medium and the medium and
the high is the same. And we’ve chosen to do it this way in
order to make it, at least we know for certain that there is
a consistent rank order. We’ve tried doing quantity assessments,
so 1 versus 3 versus 5, or something along those lines,
which would make them more consistent. Interestingly they don’t show us
an effect when you do quantities, at least in the few studies we’ve done. We think part of the problem especially
with some of these group studies is their looking from
across a great distance. So it may be very difficult for them to
tell whether I’m holding up 3 grapes or 5 grapes, but they can certainly tell the
difference between a grape and a cucumber. So I don’t actually think the fact that
they aren’t responding to quantity is particularly meaningful, because I think it’s confounded
with all these other issues. But that’s how we do it. And again if we have any one individual
that doesn’t it then we go back and re-test. So that’s our standardized definition.>>So you’re absolutely
correct that we don’t have.>>Emigration into these groups. We do have emigration because these
are animals in research projects, since we used those groups as a breeding
colony, and so we do have that occurring. But it is actually a really good
question and we do see these kinds of social collapses in more semi-free ranging
places as well, maybe not quite as often. It probably does have something
to do with being in captivity and having four, you know, walls around you. But it’s a really, it’s a really
interesting question because what we’re doing now is experimental work where we
are doing knockouts of these animals, permanent and temporary knockouts and
one of the things that we have found we hypothesize that these natal males
that are not leaving these groups, high ranking natal males are problematic
and can help, help cause social collapse. And we have done some
experimental work showing that these high ranking natal
males actually gain rank through their relationships with their
high ranked sisters and mothers. Sisters and mothers. And so they’re not gaining
rank by the normal way. An adult male would gain rank in a social
rank by, by, through the use of power and, and that sort of thing. And they actually develop
their own SBT networks. Separately from the high ranking alpha,
you know the alpha beta males, slowly. Not all do, but many do. The ones that we think
are causing problems and so, one of the management recommendations
we now have is to remove high ranking beta males from these groups, and
to bring unrelated males at some. You know, some rate in order to keep
that stability going cause it turns out the unrelated males are the ones that
are actually policing the conflict in these groups cause they
have no skin in the game.>>Sounds like the demedicheese.>>[LAUGH]
>>So.>>Okay, this is interdisciplinary so
I can say this is, outside my field, but I was really struck by the,
relationship between the, signalling and
decision making that was talking about. And some of the themes
in Brian’s talk earlier. And then also,
it was interesting that the, to me in the. What’s that? Er, oh sorry. And then it was interesting to me in the
first, in the first talk, how the response of people around you, or primates
around you in terms of splitting or not splitting really influence the reward
or penalty that you felt like you were receiving and so therefore should
also influence signaling in the or the outcome as Brian was
talking about this morning. And so I, I’m not sure if anyone
has thoughts on that theme but>>Eliza Barach just briefly, addressing our two talks. So, obviously we’ve both been
talking about stochastic and probabilistic mechanisms of,
of decision making. Just from,
from different viewpoints, and, and what you’ll basically find is
in the end the mathematics that you need to describe these, these
different processes are, are obviously. Conversion right, regardless of whether
you’re looking at the behavior or the underlying neural mechanisms, so yeah.>>And I’ll just add one thing. I don’t study signaling but one of the
things that we’re really interested in in these large group experiments
is what are we seeing. And I think the video clip I showed where
Joey was pitching the temperature tantrum That looks like signaling to me. [LAUGH] So that’s the sort of
thing we’d like to get into more.>>Hi, I have a question for
Sarah you alluded to a little bit that kin matters when it comes to sort
of protester responses to inequity. Have you tried with sort of mother and
offspring? Because it seems, at least in human
species, we’re willing to sacrifice and give and make it so our offspring have
way more than we, like that’s sort of the American dream or a dream in a lot of
different cultures to give more to your children to yourself, so that you would
expect that the protestors are signalling to an equity in an offspring
situation would be very different. Yeah.
I agree. We don’t have much data on kin versus
non kin pairs because we’ve been trying explicitly to work on non kin pairs. And to avoid that compound. I do have one study I didn’t talk about
where we looked at the willingness of individuals to work together to pull
in a heavy bark pull in order to get food rewards. And sometimes the food rewards were the
same, and sometimes they were different. And, what we found was that whether or
not the food re, whether the foods were the same or not, made no impact on
the frequency with which a pair polled, what mattered was whether or not they were
sharing when the rewards were unequal. So if Jochen and I are working together,
if I’m dominating the better reward in that unequal case,
we quit cooperating in all conditions. But, if I’m only taking it half the time,
then we cooperated a very high rate in all conditions, whether the foods
are equitable or inequitable. And interestingly in that case,
we did non kin pairs and kin pairs and kin pars from mother offspring pairs. And we saw the same pattern in both pairs. Kin cooperated at the higher
rate across all conditions. But in some kin pairs,
we saw inequitable relationships and the decrease in cooperation across
all conditions and in some pairs, we saw the cooperative ones develop. And we didn’t have very many non kin,
or kin pairs, so we can’t say too much. But in some cases where they, we only
have three pairs that were kin pairs and they weren’t sharing. And I can’t remember which order it was
but in two cases, I think it was the mom who was taking more and in one case
it was the kid who was taking more. So we’re not seeing
a consistent pattern there. But obviously it’s something
we’d like to look at further.>>I’m gonna ask, I’m gonna ask you to
repeat the question into the microphone, so we have it for the tape.>>In,
in these highly hierarchical groups, how does that impact resource sharing? There, the, the, that impacts it and that
there’s not a lot of resource sharing by, I mean particularly in there’s
not a lot of resource sharing. [LAUGH] They’re kind of known for
being very despotic. A highly aggressive, and, and not very cooperative in terms
of those kinds of things, although they are very, very strong and,
and complex grooming networks, as well. So, m, but yeah, resource sharing is, they do a lot of thought fighting over
resources and the, and the obviously, the higher ranking animals are more
likely to partake in those resources.>>I have a question. Well actually one for Sarah, regarding, do they ever kind of,
come up with teamwork approaches to you know,
if, if, if one person, or if one monkey is consistently
getting the lower ranked food. Did they ever think of handing
the token to the other one, so that they can get the high
rank food instead. And my question for Sarah is, there’s a, you, clearly really value
group cohesion over collapse. But then you’re kinda
suggesting that maybe some of the other monkeys might
prefer actually you know, we could reframe collapse as
reconfiguration and then so you know, as clearly a lot of congressmen choose to,
you know, so it’s the idea that we wanna prevent collapse by telling people
it’s gonna happen might be the opposite. They might say great, that’s our aim.>>So no, we haven’t seen any evidence
that they are trying to really work hard to rectify these situations. We have, in all of the studies we’ve done,
I think we have three cases of chimps where the subordinate had a grape and
let their partner have it. But in every case it wasn’t clear. Chimps hands are configured
very differently from ours. So, picking up something like
a grape is very difficult. They can’t do a pencil grip. So, it wasn’t clear that
they didn’t just drop it and decide to not fight the dominate over it. But of course, this isn’t a great
situation for sharing, because we’re giving them a, a single grape,
which you really can’t share very well. And all of the good studies of sh, food sharing have involved foods
that are more inherently sharable. Let’s give you a few apple slices so
you can break something off and share it. One of the things we’re also trying to
do with these group studies is set up these situations where, multiple
individuals can work together and for all sorts of things where maybe
one gets a better reward and one gets a less good reward and maybe
in that case they would share it, okay. You know, you pulled in for
that crummy piece of celery but over here, I got half an apple sliced up in pieces
and I’ll let you have some of it. We do have some evidence
from Capuchin monkeys that they are sharing more in these cases
where the foods are more divisible. And it looks like chimps in
a similar sort of groups, groups cooperation setting
are doing the same thing. So I think we might find it. It’s probably not all that common. But the studies that we’re doing
are not well designed to look for it because they’re
inherently non-shareable.>>And you’re absolutely right, that there are many individuals in those
groups that are very happy about the idea of a collapse occurring,
in fact they are negotiating that. One of the things that we’ve found
is that when we temporarily remove high ranking animals, animals do try to
vie for increasing their social status. And so presumably, these social
collapses are recurring as a result of, of subgroups of animals that are sort of
being opportunistic and in trying to, in trying to increase
their social status and therefore their ability to
partake in the resources. So you’re that’s what’s happening. It’s kind of a constant process in
these groups that these, there’s this, it’s monkey politics is what it is. Yeah [SOUND]
>>[INAUDIBLE] Could you say a little bit more about how you
get the neurons doing log likelihood? You know, you said, and
some plausible assumptions. I mean it’s very interesting. I’d like to hear a little
bit more if you could.>>Yeah, yeah.
Yeah, so, so, to, to make it short, there’s basically two assumptions that,
that you have to make there. One of them is what the actual
distribution of the activity of those neurons is, when, when, the underlying
value that they are representing changes. And for example, one of the assumptions
that you can make there is a pause order distribution which is often used for
describing the activity of, of cortical neurons, and when you make that assumption
actually turns out it, it works. The second assumption that you have
to make is the idea of basically, a neuron anti neuron pair. So you basically need two
groups of neurons that have exactly opposite properties. So basically in our case,
right with the, with the left, right, as one of the groups increases its
activity when one direction is stronger, the other one has to decrease accordingly. And just based on those two assumptions,
just when you write down the equations, you’ll see in the end that the log
likelihood will be proportional to that, to that difference.>>Hi, so I, I came kind of late, so I don’t remember where like this exact
study is, but I feel like I have these general questions for
interdisciplinary study in social science. So like people tend to use monkey or
like animal behaviors, who like cannot study our own society, so
I wonder if there is like, so for you, for like the scholars do you think that your
research has some kind of restrictions in terms of external validity, when you
are like applying what you learned from the monkeys to human beings
>>Okay. [LAUGH]
>>I can, I can start so for the, the kinds of,
of tasks that we are studying we are actually comparing other behavior
between humans and non-human primates and we are basically making sure that we are
only looking at types of decisions that, that have similar behavior to basically
make sure that there is the validity. Right?
So, there are certainly limitations. So, there of course,
many kinds of decision behaviors that humans are able to do right that
you could never study in animals. So, we have certain limitations but we basically make sure that
the validity is there.>>That’s a great question, of course. And, one of the things,
we do the same things. So, for all of my behavioral experiments,
we test usually kids and adults if we can do it and we’re also trying to begin
testing people within subjects designs. And so for most of the work I’ve done
on experimental economics schemes we’ve tried to test humans and primates with individuals they know
on these repeated designs to increase the validity of comparing the non-human
primate paradigms to the human paradigms. Because for obvious reasons
that Jochen nicely touched on, in his talk, they’re different. I mean we can’t tell them what to do. But actually, I’d also like to point
out that one of the values of studying the non-human primates is, I can do what
I talked about here where we go from very controlled studies with dyads
that we pull out of the group and choose to studies where we try to compare
everyone with everyone else to these group level studies in a way that really
can’t feasibly be done with humans. I can’t grab the same group of people and bring them in over
the course of several years. I can’t control their social environment, I can’t decide necessarily who
they are going to be paired with. We can do this to some degree with
working with kids in day care so we can go in repeatedly,
but even then, you know, if their mom moves across the country
there’s nothing I can do about that. [LAUGH] So studying the non-human
primates gives us a way to ask these questions that
we can’t ask in humans. And also does a better job of
helping to disentangle where, where some of the biological constraints
on human behavior may come in. I’m not talking about any sort of
biological determinism here but understanding where, where they are,
these evolutionary restrictions. Now Connor nicely refers
to it as the tangled wing. You know, if your wing is
tangled in the bushes there’s so many ways you can get out. And we’re interested in trying to figure
out what are those constraints in humans. And the way to do that is
to look at other species.>>Yeah I agree, with,
with what Sara said completely. In fact, you know, we’re,
in fact we’re trying to use this, these models here at our primate center. Not only to understand group stability and
and, and increasing the welfare of these, these animals, but also as
a translational model for human health. So we have a whole health component
that we’re looking at and how networks are how structures and
networks and roles of networks have felt,
effect health outcomes. And, and frankly we couldn’t do this, the
kind of work that we’re doing with with human populations because these animals
are very well characterized behaviorally. We know everything about
their kinship patterns. We, we have data on
biobehavioral organization from physiological measures to personality and
temperament. So we’re able to put together a very
complex set of variables to look at how variation in those variables affects
health that we think will be applicable, even though not exactly the same
in terms of the types of networks will be applicable to human populations. So, if they offer us a wonderful model
we’ll be able to, to look at complexity in ways that, that we can’t do it
is easily in human populations. Particularly because of experimental
component and control of so, of a social environment.>>Well thank you all very much for,
for attending and I’m sure, oh is there one more question? Okay, like we, one more question. [LAUGH]
>>Okay>>I, my question is for Sarah. So I was wondering if you could comment
on whether there was any influence of dominance relationships
in response to inequity?>>Another great question. So we would assume that there would
be an influence of dominance, I mean obviously it permeates their
lives and influences almost everything. And we would expect that the relationship
would be more dominant individuals who’re used to getting more. So they would be more upset by
inequity than subordinate individuals. Now interestingly, we don’t see that
large of an infective dominance. When we do see it,
it’s always in the predicted direction. You always see dominants
are more upset than subordinates. But of course, one of the confounds
is we’ve been doing a lot of these on pairs that we pull in, and
they’re interacting with the human, so it may be that having the human there
influences their reactions to some degree. It may ameliorate, it may either
ameliorate the dominant response or maybe the dominant says well you know I’m,
yes I’m interacting with this subordinate over here but
they’re not the ones who caused it. That’s one reason we’ve tried
to go to these more group paradigms which helps with some
of this external validity and helps us look at behaviors that maybe, you
know, Joey had the temper tantrum right, right in front of us but
maybe he wanders off to do it. And we’re also trying to look at paradigms
where we can look more at the animals interacting with one
another to get at that. But where we do see dominant interactions,
they’re exactly as you’d predict.>>Okay.>>Yeah.
Thank you all for wonderful panel.>>[APPLAUSE]

Leave a Reply

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