0:00:03.393,0:00:04.698
Science,

0:00:04.698,0:00:07.958
science has allowed us to know so much

0:00:07.958,0:00:11.108
about the far reaches of the universe,

0:00:11.108,0:00:14.303
which is at the same time tremendously important

0:00:14.303,0:00:16.307
and extremely remote,

0:00:16.307,0:00:18.828
and yet much, much closer,

0:00:18.828,0:00:20.919
much more directly related to us.

0:00:20.919,0:00:23.463
There are many things we don't really understand.

0:00:23.463,0:00:25.531
And one of them is the extraordinary

0:00:25.531,0:00:28.918
social complexity of the animals around us,

0:00:28.918,0:00:30.842
and today I want to tell you a few stories

0:00:30.842,0:00:32.866
of animal complexity.

0:00:32.866,0:00:36.354
But first, what do we call complexity?

0:00:36.354,0:00:37.656
What is complex?

0:00:37.656,0:00:41.222
Well, complex is not complicated.

0:00:41.222,0:00:44.608
Something complicated comprises many small parts,

0:00:44.608,0:00:47.005
all different, and each of them

0:00:47.005,0:00:50.265
has its own precise role in the machinery.

0:00:50.265,0:00:52.922
On the opposite, a complex system

0:00:52.922,0:00:55.655
is made of many, many similar parts,

0:00:55.655,0:00:57.525
and it is their interaction

0:00:57.525,0:01:00.999
that produces a globally cohering behavior.

0:01:00.999,0:01:04.758
Complex systems have many interacting parts

0:01:04.758,0:01:08.245
which behave according to simple, individual rules,

0:01:08.245,0:01:11.564
and this results in emergent properties.

0:01:11.564,0:01:13.467
The behavior of the system as a whole

0:01:13.467,0:01:15.052
cannot be predicted

0:01:15.052,0:01:17.426
from the individual rules only.

0:01:17.426,0:01:18.989
As Aristotle wrote,

0:01:18.989,0:01:22.219
the whole is greater than the sum of its parts.

0:01:22.219,0:01:24.896
But from Aristotle, let's move onto a more

0:01:24.896,0:01:28.249
concrete example of complex systems.

0:01:28.249,0:01:30.219
These are Scottish terriers.

0:01:30.219,0:01:34.054
In the beginning, the system is disorganized.

0:01:34.054,0:01:37.863
Then comes a perturbation: milk.

0:01:37.863,0:01:41.091
Every individual starts pushing in one direction

0:01:41.091,0:01:42.036
— (Laughter) —

0:01:42.036,0:01:45.073
and this is what happens.

0:01:45.073,0:01:47.710
The pinwheel is an emergent property

0:01:47.710,0:01:49.951
of the interactions between puppies

0:01:49.951,0:01:53.815
whose only rule is to try to keep access to the milk

0:01:53.815,0:01:57.314
and therefore to push in a random direction.

0:01:57.314,0:02:01.049
So it's all about finding the simple rules

0:02:01.049,0:02:04.109
from which complexity emerges.

0:02:04.109,0:02:06.895
I call this simplifying complexity,

0:02:06.895,0:02:09.046
and it's what we do the Chair of Systems Design

0:02:09.046,0:02:10.961
at ETH Zurich.

0:02:10.961,0:02:14.527
We collect data on animal populations,

0:02:14.527,0:02:18.370
analyze complex patterns, try to explain them.

0:02:18.370,0:02:21.142
It requires physicists who work with biologists,

0:02:21.142,0:02:23.957
with mathematicians and computer scientists,

0:02:23.957,0:02:26.470
and it is their interaction that produces

0:02:26.470,0:02:28.368
cross-boundary competence

0:02:28.368,0:02:29.977
to solve these problems.

0:02:29.977,0:02:32.204
So again, the whole is greater

0:02:32.204,0:02:33.588
than the sum of its parts.

0:02:33.588,0:02:35.815
In a way, collaboration

0:02:35.815,0:02:39.522
is another example of complex systems.

0:02:39.522,0:02:41.151
And you may be asking yourself

0:02:41.151,0:02:43.983
which side I'm on, biology or physics.

0:02:43.983,0:02:46.294
In fact, it's a little different,

0:02:46.294,0:02:47.561
and to explain, I need to tell you

0:02:47.561,0:02:50.258
a short story about myself.

0:02:50.258,0:02:51.829
When I was a child,

0:02:51.829,0:02:55.815
I loved to build stuff, to[br]create complicated machines.

0:02:55.815,0:02:58.414
So I set out to study electrical engineering

0:02:58.414,0:03:00.165
and robotics,

0:03:00.165,0:03:02.032
and my end-of-studies project

0:03:02.032,0:03:05.002
was about building a robot called ER-1

0:03:05.002,0:03:06.899
—it looked like this—

0:03:06.899,0:03:09.486
that would collect information from its environment

0:03:09.486,0:03:12.968
and proceed to follow a white line on the ground.

0:03:12.968,0:03:15.271
It was very, very complicated,

0:03:15.271,0:03:18.331
but it worked beautifully in our test room,

0:03:18.331,0:03:20.615
and on demo day, professors had assembled

0:03:20.615,0:03:21.977
to grade to the project.

0:03:21.977,0:03:24.756
So we took ER-1 to the evaluation room.

0:03:24.756,0:03:27.079
It turned out, the light in that room

0:03:27.079,0:03:28.824
was slightly different.

0:03:28.824,0:03:31.123
The robot's vision system got confused.

0:03:31.123,0:03:32.869
At the first bend in the line,

0:03:32.869,0:03:36.532
it left its course, and crashed into a wall.

0:03:36.532,0:03:38.695
We had spent weeks building it,

0:03:38.695,0:03:40.337
and all it took to destroy it

0:03:40.337,0:03:42.948
was a subtle change in the color of the light

0:03:42.948,0:03:44.635
in the room.

0:03:44.635,0:03:46.075
That's when I realized that

0:03:46.075,0:03:48.201
the more complicated you make a machine,

0:03:48.201,0:03:50.440
the more likely that it will fail

0:03:50.440,0:03:53.088
due to something absolutely unexpected.

0:03:53.088,0:03:54.757
And I decided that, in fact,

0:03:54.757,0:03:57.768
I didn't really want to create complicated stuff.

0:03:57.768,0:04:00.710
I wanted to understand complexity,

0:04:00.710,0:04:02.698
complexity of the world around us

0:04:02.698,0:04:05.396
and especially in the animal kingdom.

0:04:05.396,0:04:08.287
Which brings us to bats.

0:04:08.287,0:04:11.628
Bechstein's bats are a common[br]species of European bats.

0:04:11.628,0:04:13.035
They are very social animals.

0:04:13.035,0:04:15.427
Mostly they roost or sleep together.

0:04:15.427,0:04:17.996
And they live in maternity colonies,

0:04:17.996,0:04:19.582
which means that every spring,

0:04:19.582,0:04:22.655
the females meet after the winter hybernation,

0:04:22.655,0:04:24.914
and they stay together for about six months

0:04:24.914,0:04:26.940
to rear their young,

0:04:26.940,0:04:30.079
and they all carry a very small chip,

0:04:30.079,0:04:31.845
which means that every time one of them

0:04:31.845,0:04:35.164
enters one of these specially equipped bat boxes,

0:04:35.164,0:04:36.807
we know where she is,

0:04:36.807,0:04:38.090
and more importantly,

0:04:38.090,0:04:40.508
we know with whom she is.

0:04:40.508,0:04:44.187
So I study roosting associations in bats,

0:04:44.187,0:04:46.832
and this is what it looks like.

0:04:46.832,0:04:48.919
During the day, the bats roost

0:04:48.919,0:04:51.517
in a number of sub-groups in different boxes.

0:04:51.517,0:04:53.063
It could be that on one day,

0:04:53.063,0:04:55.235
the colony is split between two boxes,

0:04:55.235,0:04:56.934
but on another day,

0:04:56.934,0:04:59.252
it could be together in a single box,

0:04:59.252,0:05:01.399
or split between three or more boxes,

0:05:01.399,0:05:04.449
and that all seems rather erratic, really.

0:05:04.449,0:05:07.529
It's called fission-fusion dynamics,

0:05:07.529,0:05:09.242
the property for an animal group

0:05:09.242,0:05:11.481
of regularly splitting and merging

0:05:11.481,0:05:13.204
into different subgroups.

0:05:13.204,0:05:15.689
So what we do is take all these data

0:05:15.689,0:05:17.320
from all these different days

0:05:17.320,0:05:18.839
and pool them together

0:05:18.839,0:05:21.442
to extract a long-term association pattern

0:05:21.442,0:05:23.823
by playing techniques with network analysis

0:05:23.823,0:05:25.521
to get a complete picture

0:05:25.521,0:05:28.072
of the social structure of the colony.

0:05:28.072,0:05:32.263
Okay? So that's what this picture looks like.

0:05:32.263,0:05:34.826
In this network, all the circles

0:05:34.826,0:05:37.202
are nodes, individual bats,

0:05:37.202,0:05:38.938
and the lines between them

0:05:38.938,0:05:42.695
are social bonds, associations between individuals.

0:05:42.695,0:05:45.373
It turns out this is a very interesting picture.

0:05:45.373,0:05:47.479
This bat colony is organized

0:05:47.479,0:05:49.254
in two different communities

0:05:49.254,0:05:51.054
which cannot be predicted

0:05:51.054,0:05:53.372
from the daily fission-future dynamics.

0:05:53.372,0:05:57.000
We call them cryptic social units.

0:05:57.000,0:05:58.615
Even more interesting, in fact:

0:05:58.615,0:06:00.842
every year, around October,

0:06:00.842,0:06:02.494
the colony splits up,

0:06:02.494,0:06:05.050
and all bats hibernate separately,

0:06:05.050,0:06:06.684
but year after year,

0:06:06.684,0:06:09.573
when the bats come together again in the spring,

0:06:09.573,0:06:12.394
the communities stay the same.

0:06:12.394,0:06:14.851
So these bats remember their friends

0:06:14.851,0:06:16.928
for a really long time.

0:06:16.928,0:06:19.034
With a brain the size of a peanut,

0:06:19.034,0:06:21.220
they maintain individualized,

0:06:21.220,0:06:23.546
long-term social bonds,

0:06:23.546,0:06:25.502
We didn't know that was possible.

0:06:25.502,0:06:26.814
We knew that primates

0:06:26.814,0:06:29.382
and elephants and dolphins could do that,

0:06:29.382,0:06:32.195
but compared to bats, they have huge brains.

0:06:32.195,0:06:34.471
So how could it be

0:06:34.471,0:06:36.760
that the bats maintain this complex,

0:06:36.760,0:06:38.339
stable social structure

0:06:38.339,0:06:41.841
with such limited cognitive abilities?

0:06:41.841,0:06:44.593
And this is where complexity brings an answer.

0:06:44.593,0:06:46.379
To understand this system,

0:06:46.379,0:06:49.498
we built a computer model of roosting,

0:06:49.498,0:06:51.516
based on simple, individual rules,

0:06:51.516,0:06:54.049
and simulated thousands and thousands of days

0:06:54.049,0:06:55.956
in the virtual bat colony.

0:06:55.956,0:06:57.712
It's a mathematical model,

0:06:57.712,0:07:00.279
but it's not complicated.

0:07:00.279,0:07:02.903
What the model told us is that, in a nutshell,

0:07:02.903,0:07:06.409
each bat knows a few other colony members

0:07:06.409,0:07:09.206
as her friends, and is just slightly more likely

0:07:09.206,0:07:11.530
to roost in a box with them.

0:07:11.530,0:07:13.928
Simple, individual rules.

0:07:13.928,0:07:15.610
This is all it takes to explain

0:07:15.610,0:07:18.215
the social complexity of these bats.

0:07:18.215,0:07:19.947
But it gets better.

0:07:19.947,0:07:22.350
Between 2010 and 2011,

0:07:22.350,0:07:26.011
the colony lost more than two thirds of its members,

0:07:26.011,0:07:28.880
probably due to the very cold winter.

0:07:28.880,0:07:32.098
The next spring, it didn't form two communities

0:07:32.098,0:07:33.819
like every year,

0:07:33.819,0:07:35.653
which may have led the whole colony to die

0:07:35.653,0:07:37.824
because it had become too small.

0:07:37.824,0:07:42.952
Instead, it formed a single, cohesive social unit,

0:07:42.952,0:07:45.868
which allowed the colony to survive that season

0:07:45.868,0:07:49.187
and thrive again in the next two years.

0:07:49.187,0:07:50.796
What we know is that the bats

0:07:50.796,0:07:53.657
are not aware that their colony is doing this.

0:07:53.657,0:07:57.051
All they do is follow simple association rules,

0:07:57.051,0:07:58.659
and from this simplicity

0:07:58.659,0:08:00.962
emerges social complexity

0:08:00.962,0:08:03.756
which allows the colony to be resilient

0:08:03.756,0:08:07.043
against dramatic changes[br]in the population structure.

0:08:07.043,0:08:09.708
And I find this incredible.

0:08:09.708,0:08:11.530
Now I want to tell you another story,

0:08:11.530,0:08:13.208
but from this we have to travel from Europe

0:08:13.208,0:08:16.303
to the Kalahari Desert in South Africa.

0:08:16.303,0:08:18.222
This is where meerkats live.

0:08:18.222,0:08:19.799
I'm sure you know meerkats.

0:08:19.799,0:08:21.397
They're fascinating creatures.

0:08:21.397,0:08:24.817
They live in groups with a[br]very strict social hierarchy.

0:08:24.817,0:08:26.306
There is one dominant pair,

0:08:26.306,0:08:27.596
and many subordinates,

0:08:27.596,0:08:29.418
some acting as sentinels,

0:08:29.418,0:08:30.724
some acting as babysitters,

0:08:30.724,0:08:32.560
some teaching pups, and so on.

0:08:32.560,0:08:35.850
What we do is put very small GPS collars

0:08:35.850,0:08:37.575
on these animals

0:08:37.575,0:08:39.450
to study how they move together,

0:08:39.450,0:08:42.998
and what this has to do with their social structure.

0:08:42.998,0:08:44.565
And there's a very interesting example

0:08:44.565,0:08:47.127
of collective movement in meerkats.

0:08:47.127,0:08:49.571
In the middle of the reserve which they live in

0:08:49.571,0:08:50.795
lies a road.

0:08:50.795,0:08:54.136
On this road there are cars, so it's dangerous.

0:08:54.136,0:08:56.062
But the meerkats have to cross it

0:08:56.062,0:08:58.794
to get from one feeding place to another.

0:08:58.794,0:09:03.530
So we asked, how exactly do they do this?

0:09:03.530,0:09:05.488
We found that the dominant female

0:09:05.488,0:09:08.110
is mostly the one who leads the group to the road,

0:09:08.110,0:09:11.166
but when it comes to crossing it, crossing the road,

0:09:11.166,0:09:13.610
she gives way to the subordinates,

0:09:13.610,0:09:15.074
a manner of saying, you know,

0:09:15.074,0:09:18.639
"Go ahead, tell me if it's safe."

0:09:18.639,0:09:19.963
But what I didn't know, in fact,

0:09:19.963,0:09:22.690
was what rules in their behavior the meerkats follow

0:09:22.690,0:09:25.615
for this change at the edge of the group to happen

0:09:25.615,0:09:29.865
and if simple rules were sufficient to explain it.

0:09:29.865,0:09:33.503
So I built a model, a model of simulating meerkats

0:09:33.503,0:09:35.876
crossing a simulated road.

0:09:35.876,0:09:37.687
It's a simplistic model.

0:09:37.687,0:09:40.388
Moving meerkats are like random particles

0:09:40.388,0:09:42.672
whose unique rule is one of alignment.

0:09:42.672,0:09:45.170
They simply move together.

0:09:45.170,0:09:47.970
When these particles get to the road,

0:09:47.970,0:09:50.022
they sense some kind of obstacle,

0:09:50.022,0:09:52.303
and they bounce against it.

0:09:52.303,0:09:53.383
The only difference

0:09:53.383,0:09:55.440
between the dominant female, here in red,

0:09:55.440,0:09:56.925
and the other individuals,

0:09:56.925,0:09:59.556
is that for her, the height of the obstacle,

0:09:59.556,0:10:01.999
which is in fact the risk perceived from the road,

0:10:01.999,0:10:03.979
is just slightly higher,

0:10:03.979,0:10:05.717
and this tiny difference

0:10:05.717,0:10:07.401
in the individual's rule of movement

0:10:07.401,0:10:09.770
is sufficient to explain what we observe,

0:10:09.770,0:10:12.499
that the dominant female

0:10:12.499,0:10:13.995
leads her group to the road

0:10:13.995,0:10:15.558
and then gives way to the others

0:10:15.558,0:10:18.235
for them to cross first.

0:10:18.235,0:10:21.886
George Box, who was an English statistician,

0:10:21.886,0:10:24.848
once wrote, "All models are false,

0:10:24.848,0:10:27.054
but some models are useful."

0:10:27.054,0:10:30.166
And in fact, this model is obviously false,

0:10:30.166,0:10:34.319
because in reality, meerkats are[br]anything but random particles.

0:10:34.319,0:10:35.863
But it's also useful,

0:10:35.863,0:10:38.690
because it tells us that extreme simplicity

0:10:38.690,0:10:41.679
in movement rules at the individual level

0:10:41.679,0:10:44.167
can result in a great deal of complexity

0:10:44.167,0:10:46.367
at the level of the group.

0:10:46.367,0:10:50.486
So again, that's simplifying complexity.

0:10:50.486,0:10:51.671
Now, I would like to conclude

0:10:51.671,0:10:54.212
on what this means for the whole species.

0:10:54.212,0:10:55.967
When the dominant female

0:10:55.967,0:10:57.749
gives way to a subordinate,

0:10:57.749,0:11:00.082
it's not out of courtesy.

0:11:00.082,0:11:01.361
In fact, the dominant female

0:11:01.361,0:11:04.292
is extremely important for the cohesion of the group.

0:11:04.292,0:11:07.496
If she dies on the road, the whole group is at risk.

0:11:07.496,0:11:09.778
So this behavior of risk avoidance

0:11:09.778,0:11:12.411
is a very old evolutionary response.

0:11:12.411,0:11:16.371
These meerkats are replicating an evolved tactic

0:11:16.371,0:11:18.558
that is thousands of generations old,

0:11:18.558,0:11:20.849
and they're adapting it to a modern risk,

0:11:20.849,0:11:24.236
in this case a road built by humans.

0:11:24.236,0:11:26.723
They adapt very simple rules,

0:11:26.723,0:11:28.966
and the resulting complex behavior

0:11:28.966,0:11:31.876
allows them to resist human encroachment

0:11:31.876,0:11:34.648
into their natural habitat.

0:11:34.648,0:11:36.265
In the end,

0:11:36.265,0:11:38.887
it may be bats which change their social structure

0:11:38.887,0:11:41.044
in response to a population crush,

0:11:41.044,0:11:42.778
or it may be meerkats

0:11:42.778,0:11:45.673
who show a novel adaptation to a human road,

0:11:45.673,0:11:48.220
or it may be another species.

0:11:48.220,0:11:51.304
My message here, and it's not a complicated one

0:11:51.304,0:11:54.206
but a simple one of wonder and hope,

0:11:54.206,0:11:57.050
my message here is that animals

0:11:57.050,0:11:59.754
show extraordinary social complexity,

0:11:59.754,0:12:02.149
and this allows them to adapt

0:12:02.149,0:12:05.615
and respond to changes in their environment.

0:12:05.615,0:12:08.337
In three words, in the animal kingdom,

0:12:08.337,0:12:11.234
simplicity leads to complexity

0:12:11.234,0:12:12.902
which leads to resilience.

0:12:12.902,0:12:14.754
Thank you.

0:12:14.754,0:12:21.434
(Applause)