1
00:00:03,393 --> 00:00:04,698
Science,

2
00:00:04,698 --> 00:00:07,958
science has allowed us to know so much

3
00:00:07,958 --> 00:00:11,108
about the far reaches of the universe,

4
00:00:11,108 --> 00:00:14,303
which is at the same time tremendously important

5
00:00:14,303 --> 00:00:16,307
and extremely remote,

6
00:00:16,307 --> 00:00:18,828
and yet much, much closer,

7
00:00:18,828 --> 00:00:20,919
much more directly related to us.

8
00:00:20,919 --> 00:00:23,463
There are many things we don't really understand.

9
00:00:23,463 --> 00:00:25,531
And one of them is the extraordinary

10
00:00:25,531 --> 00:00:28,918
social complexity of the animals around us,

11
00:00:28,918 --> 00:00:30,842
and today I want to tell you a few stories

12
00:00:30,842 --> 00:00:32,866
of animal complexity.

13
00:00:32,866 --> 00:00:36,354
But first, what do we call complexity?

14
00:00:36,354 --> 00:00:37,656
What is complex?

15
00:00:37,656 --> 00:00:41,222
Well, complex is not complicated.

16
00:00:41,222 --> 00:00:44,608
Something complicated comprises many small parts,

17
00:00:44,608 --> 00:00:47,005
all different, and each of them

18
00:00:47,005 --> 00:00:50,265
has its own precise role in the machinery.

19
00:00:50,265 --> 00:00:52,922
On the opposite, a complex system

20
00:00:52,922 --> 00:00:55,655
is made of many, many similar parts,

21
00:00:55,655 --> 00:00:57,525
and it is their interaction

22
00:00:57,525 --> 00:01:00,999
that produces a globally cohering behavior.

23
00:01:00,999 --> 00:01:04,758
Complex systems have many interacting parts

24
00:01:04,758 --> 00:01:08,245
which behave according to simple, individual rules,

25
00:01:08,245 --> 00:01:11,564
and this results in emergent properties.

26
00:01:11,564 --> 00:01:13,467
The behavior of the system as a whole

27
00:01:13,467 --> 00:01:15,052
cannot be predicted

28
00:01:15,052 --> 00:01:17,426
from the individual rules only.

29
00:01:17,426 --> 00:01:18,989
As Aristotle wrote,

30
00:01:18,989 --> 00:01:22,219
the whole is greater than the sum of its parts.

31
00:01:22,219 --> 00:01:24,896
But from Aristotle, let's move onto a more

32
00:01:24,896 --> 00:01:28,249
concrete example of complex systems.

33
00:01:28,249 --> 00:01:30,219
These are Scottish terriers.

34
00:01:30,219 --> 00:01:34,054
In the beginning, the system is disorganized.

35
00:01:34,054 --> 00:01:37,863
Then comes a perturbation: milk.

36
00:01:37,863 --> 00:01:41,091
Every individual starts pushing in one direction

37
00:01:41,091 --> 00:01:42,036
— (Laughter) —

38
00:01:42,036 --> 00:01:45,073
and this is what happens.

39
00:01:45,073 --> 00:01:47,710
The pinwheel is an emergent property

40
00:01:47,710 --> 00:01:49,951
of the interactions between puppies

41
00:01:49,951 --> 00:01:53,815
whose only rule is to try to keep access to the milk

42
00:01:53,815 --> 00:01:57,314
and therefore to push in a random direction.

43
00:01:57,314 --> 00:02:01,049
So it's all about finding the simple rules

44
00:02:01,049 --> 00:02:04,109
from which complexity emerges.

45
00:02:04,109 --> 00:02:06,895
I call this simplifying complexity,

46
00:02:06,895 --> 00:02:09,046
and it's what we do the Chair of Systems Design

47
00:02:09,046 --> 00:02:10,961
at ETH Zurich.

48
00:02:10,961 --> 00:02:14,527
We collect data on animal populations,

49
00:02:14,527 --> 00:02:18,370
analyze complex patterns, try to explain them.

50
00:02:18,370 --> 00:02:21,142
It requires physicists who work with biologists,

51
00:02:21,142 --> 00:02:23,957
with mathematicians and computer scientists,

52
00:02:23,957 --> 00:02:26,470
and it is their interaction that produces

53
00:02:26,470 --> 00:02:28,368
cross-boundary competence

54
00:02:28,368 --> 00:02:29,977
to solve these problems.

55
00:02:29,977 --> 00:02:32,204
So again, the whole is greater

56
00:02:32,204 --> 00:02:33,588
than the sum of its parts.

57
00:02:33,588 --> 00:02:35,815
In a way, collaboration

58
00:02:35,815 --> 00:02:39,522
is another example of complex systems.

59
00:02:39,522 --> 00:02:41,151
And you may be asking yourself

60
00:02:41,151 --> 00:02:43,983
which side I'm on, biology or physics.

61
00:02:43,983 --> 00:02:46,294
In fact, it's a little different,

62
00:02:46,294 --> 00:02:47,561
and to explain, I need to tell you

63
00:02:47,561 --> 00:02:50,258
a short story about myself.

64
00:02:50,258 --> 00:02:51,829
When I was a child,

65
00:02:51,829 --> 00:02:55,815
I loved to build stuff, to
create complicated machines.

66
00:02:55,815 --> 00:02:58,414
So I set out to study electrical engineering

67
00:02:58,414 --> 00:03:00,165
and robotics,

68
00:03:00,165 --> 00:03:02,032
and my end-of-studies project

69
00:03:02,032 --> 00:03:05,002
was about building a robot called ER-1

70
00:03:05,002 --> 00:03:06,899
—it looked like this—

71
00:03:06,899 --> 00:03:09,486
that would collect information from its environment

72
00:03:09,486 --> 00:03:12,968
and proceed to follow a white line on the ground.

73
00:03:12,968 --> 00:03:15,271
It was very, very complicated,

74
00:03:15,271 --> 00:03:18,331
but it worked beautifully in our test room,

75
00:03:18,331 --> 00:03:20,615
and on demo day, professors had assembled

76
00:03:20,615 --> 00:03:21,977
to grade to the project.

77
00:03:21,977 --> 00:03:24,756
So we took ER-1 to the evaluation room.

78
00:03:24,756 --> 00:03:27,079
It turned out, the light in that room

79
00:03:27,079 --> 00:03:28,824
was slightly different.

80
00:03:28,824 --> 00:03:31,123
The robot's vision system got confused.

81
00:03:31,123 --> 00:03:32,869
At the first bend in the line,

82
00:03:32,869 --> 00:03:36,532
it left its course, and crashed into a wall.

83
00:03:36,532 --> 00:03:38,695
We had spent weeks building it,

84
00:03:38,695 --> 00:03:40,337
and all it took to destroy it

85
00:03:40,337 --> 00:03:42,948
was a subtle change in the color of the light

86
00:03:42,948 --> 00:03:44,635
in the room.

87
00:03:44,635 --> 00:03:46,075
That's when I realized that

88
00:03:46,075 --> 00:03:48,201
the more complicated you make a machine,

89
00:03:48,201 --> 00:03:50,440
the more likely that it will fail

90
00:03:50,440 --> 00:03:53,088
due to something absolutely unexpected.

91
00:03:53,088 --> 00:03:54,757
And I decided that, in fact,

92
00:03:54,757 --> 00:03:57,768
I didn't really want to create complicated stuff.

93
00:03:57,768 --> 00:04:00,710
I wanted to understand complexity,

94
00:04:00,710 --> 00:04:02,698
complexity of the world around us

95
00:04:02,698 --> 00:04:05,396
and especially in the animal kingdom.

96
00:04:05,396 --> 00:04:08,287
Which brings us to bats.

97
00:04:08,287 --> 00:04:11,628
Bechstein's bats are a common
species of European bats.

98
00:04:11,628 --> 00:04:13,035
They are very social animals.

99
00:04:13,035 --> 00:04:15,427
Mostly they roost or sleep together.

100
00:04:15,427 --> 00:04:17,996
And they live in maternity colonies,

101
00:04:17,996 --> 00:04:19,582
which means that every spring,

102
00:04:19,582 --> 00:04:22,655
the females meet after the winter hybernation,

103
00:04:22,655 --> 00:04:24,914
and they stay together for about six months

104
00:04:24,914 --> 00:04:26,940
to rear their young,

105
00:04:26,940 --> 00:04:30,079
and they all carry a very small chip,

106
00:04:30,079 --> 00:04:31,845
which means that every time one of them

107
00:04:31,845 --> 00:04:35,164
enters one of these specially equipped bat boxes,

108
00:04:35,164 --> 00:04:36,807
we know where she is,

109
00:04:36,807 --> 00:04:38,090
and more importantly,

110
00:04:38,090 --> 00:04:40,508
we know with whom she is.

111
00:04:40,508 --> 00:04:44,187
So I study roosting associations in bats,

112
00:04:44,187 --> 00:04:46,832
and this is what it looks like.

113
00:04:46,832 --> 00:04:48,919
During the day, the bats roost

114
00:04:48,919 --> 00:04:51,517
in a number of sub-groups in different boxes.

115
00:04:51,517 --> 00:04:53,063
It could be that on one day,

116
00:04:53,063 --> 00:04:55,235
the colony is split between two boxes,

117
00:04:55,235 --> 00:04:56,934
but on another day,

118
00:04:56,934 --> 00:04:59,252
it could be together in a single box,

119
00:04:59,252 --> 00:05:01,399
or split between three or more boxes,

120
00:05:01,399 --> 00:05:04,449
and that all seems rather erratic, really.

121
00:05:04,449 --> 00:05:07,529
It's called fission-fusion dynamics,

122
00:05:07,529 --> 00:05:09,242
the property for an animal group

123
00:05:09,242 --> 00:05:11,481
of regularly splitting and merging

124
00:05:11,481 --> 00:05:13,204
into different subgroups.

125
00:05:13,204 --> 00:05:15,689
So what we do is take all these data

126
00:05:15,689 --> 00:05:17,320
from all these different days

127
00:05:17,320 --> 00:05:18,839
and pool them together

128
00:05:18,839 --> 00:05:21,442
to extract a long-term association pattern

129
00:05:21,442 --> 00:05:23,823
by playing techniques with network analysis

130
00:05:23,823 --> 00:05:25,521
to get a complete picture

131
00:05:25,521 --> 00:05:28,072
of the social structure of the colony.

132
00:05:28,072 --> 00:05:32,263
Okay? So that's what this picture looks like.

133
00:05:32,263 --> 00:05:34,826
In this network, all the circles

134
00:05:34,826 --> 00:05:37,202
are nodes, individual bats,

135
00:05:37,202 --> 00:05:38,938
and the lines between them

136
00:05:38,938 --> 00:05:42,695
are social bonds, associations between individuals.

137
00:05:42,695 --> 00:05:45,373
It turns out this is a very interesting picture.

138
00:05:45,373 --> 00:05:47,479
This bat colony is organized

139
00:05:47,479 --> 00:05:49,254
in two different communities

140
00:05:49,254 --> 00:05:51,054
which cannot be predicted

141
00:05:51,054 --> 00:05:53,372
from the daily fission-future dynamics.

142
00:05:53,372 --> 00:05:57,000
We call them cryptic social units.

143
00:05:57,000 --> 00:05:58,615
Even more interesting, in fact:

144
00:05:58,615 --> 00:06:00,842
every year, around October,

145
00:06:00,842 --> 00:06:02,494
the colony splits up,

146
00:06:02,494 --> 00:06:05,050
and all bats hibernate separately,

147
00:06:05,050 --> 00:06:06,684
but year after year,

148
00:06:06,684 --> 00:06:09,573
when the bats come together again in the spring,

149
00:06:09,573 --> 00:06:12,394
the communities stay the same.

150
00:06:12,394 --> 00:06:14,851
So these bats remember their friends

151
00:06:14,851 --> 00:06:16,928
for a really long time.

152
00:06:16,928 --> 00:06:19,034
With a brain the size of a peanut,

153
00:06:19,034 --> 00:06:21,220
they maintain individualized,

154
00:06:21,220 --> 00:06:23,546
long-term social bonds,

155
00:06:23,546 --> 00:06:25,502
We didn't know that was possible.

156
00:06:25,502 --> 00:06:26,814
We knew that primates

157
00:06:26,814 --> 00:06:29,382
and elephants and dolphins could do that,

158
00:06:29,382 --> 00:06:32,195
but compared to bats, they have huge brains.

159
00:06:32,195 --> 00:06:34,471
So how could it be

160
00:06:34,471 --> 00:06:36,760
that the bats maintain this complex,

161
00:06:36,760 --> 00:06:38,339
stable social structure

162
00:06:38,339 --> 00:06:41,841
with such limited cognitive abilities?

163
00:06:41,841 --> 00:06:44,593
And this is where complexity brings an answer.

164
00:06:44,593 --> 00:06:46,379
To understand this system,

165
00:06:46,379 --> 00:06:49,498
we built a computer model of roosting,

166
00:06:49,498 --> 00:06:51,516
based on simple, individual rules,

167
00:06:51,516 --> 00:06:54,049
and simulated thousands and thousands of days

168
00:06:54,049 --> 00:06:55,956
in the virtual bat colony.

169
00:06:55,956 --> 00:06:57,712
It's a mathematical model,

170
00:06:57,712 --> 00:07:00,279
but it's not complicated.

171
00:07:00,279 --> 00:07:02,903
What the model told us is that, in a nutshell,

172
00:07:02,903 --> 00:07:06,409
each bat knows a few other colony members

173
00:07:06,409 --> 00:07:09,206
as her friends, and is just slightly more likely

174
00:07:09,206 --> 00:07:11,530
to roost in a box with them.

175
00:07:11,530 --> 00:07:13,928
Simple, individual rules.

176
00:07:13,928 --> 00:07:15,610
This is all it takes to explain

177
00:07:15,610 --> 00:07:18,215
the social complexity of these bats.

178
00:07:18,215 --> 00:07:19,947
But it gets better.

179
00:07:19,947 --> 00:07:22,350
Between 2010 and 2011,

180
00:07:22,350 --> 00:07:26,011
the colony lost more than two thirds of its members,

181
00:07:26,011 --> 00:07:28,880
probably due to the very cold winter.

182
00:07:28,880 --> 00:07:32,098
The next spring, it didn't form two communities

183
00:07:32,098 --> 00:07:33,819
like every year,

184
00:07:33,819 --> 00:07:35,653
which may have led the whole colony to die

185
00:07:35,653 --> 00:07:37,824
because it had become too small.

186
00:07:37,824 --> 00:07:42,952
Instead, it formed a single, cohesive social unit,

187
00:07:42,952 --> 00:07:45,868
which allowed the colony to survive that season

188
00:07:45,868 --> 00:07:49,187
and thrive again in the next two years.

189
00:07:49,187 --> 00:07:50,796
What we know is that the bats

190
00:07:50,796 --> 00:07:53,657
are not aware that their colony is doing this.

191
00:07:53,657 --> 00:07:57,051
All they do is follow simple association rules,

192
00:07:57,051 --> 00:07:58,659
and from this simplicity

193
00:07:58,659 --> 00:08:00,962
emerges social complexity

194
00:08:00,962 --> 00:08:03,756
which allows the colony to be resilient

195
00:08:03,756 --> 00:08:07,043
against dramatic changes
in the population structure.

196
00:08:07,043 --> 00:08:09,708
And I find this incredible.

197
00:08:09,708 --> 00:08:11,530
Now I want to tell you another story,

198
00:08:11,530 --> 00:08:13,208
but from this we have to travel from Europe

199
00:08:13,208 --> 00:08:16,303
to the Kalahari Desert in South Africa.

200
00:08:16,303 --> 00:08:18,222
This is where meerkats live.

201
00:08:18,222 --> 00:08:19,799
I'm sure you know meerkats.

202
00:08:19,799 --> 00:08:21,397
They're fascinating creatures.

203
00:08:21,397 --> 00:08:24,817
They live in groups with a
very strict social hierarchy.

204
00:08:24,817 --> 00:08:26,306
There is one dominant pair,

205
00:08:26,306 --> 00:08:27,596
and many subordinates,

206
00:08:27,596 --> 00:08:29,418
some acting as sentinels,

207
00:08:29,418 --> 00:08:30,724
some acting as babysitters,

208
00:08:30,724 --> 00:08:32,560
some teaching pups, and so on.

209
00:08:32,560 --> 00:08:35,850
What we do is put very small GPS collars

210
00:08:35,850 --> 00:08:37,575
on these animals

211
00:08:37,575 --> 00:08:39,450
to study how they move together,

212
00:08:39,450 --> 00:08:42,998
and what this has to do with their social structure.

213
00:08:42,998 --> 00:08:44,565
And there's a very interesting example

214
00:08:44,565 --> 00:08:47,127
of collective movement in meerkats.

215
00:08:47,127 --> 00:08:49,571
In the middle of the reserve which they live in

216
00:08:49,571 --> 00:08:50,795
lies a road.

217
00:08:50,795 --> 00:08:54,136
On this road there are cars, so it's dangerous.

218
00:08:54,136 --> 00:08:56,062
But the meerkats have to cross it

219
00:08:56,062 --> 00:08:58,794
to get from one feeding place to another.

220
00:08:58,794 --> 00:09:03,530
So we asked, how exactly do they do this?

221
00:09:03,530 --> 00:09:05,488
We found that the dominant female

222
00:09:05,488 --> 00:09:08,110
is mostly the one who leads the group to the road,

223
00:09:08,110 --> 00:09:11,166
but when it comes to crossing it, crossing the road,

224
00:09:11,166 --> 00:09:13,610
she gives way to the subordinates,

225
00:09:13,610 --> 00:09:15,074
a manner of saying, you know,

226
00:09:15,074 --> 00:09:18,639
"Go ahead, tell me if it's safe."

227
00:09:18,639 --> 00:09:19,963
But what I didn't know, in fact,

228
00:09:19,963 --> 00:09:22,690
was what rules in their behavior the meerkats follow

229
00:09:22,690 --> 00:09:25,615
for this change at the edge of the group to happen

230
00:09:25,615 --> 00:09:29,865
and if simple rules were sufficient to explain it.

231
00:09:29,865 --> 00:09:33,503
So I built a model, a model of simulating meerkats

232
00:09:33,503 --> 00:09:35,876
crossing a simulated road.

233
00:09:35,876 --> 00:09:37,687
It's a simplistic model.

234
00:09:37,687 --> 00:09:40,388
Moving meerkats are like random particles

235
00:09:40,388 --> 00:09:42,672
whose unique rule is one of alignment.

236
00:09:42,672 --> 00:09:45,170
They simply move together.

237
00:09:45,170 --> 00:09:47,970
When these particles get to the road,

238
00:09:47,970 --> 00:09:50,022
they sense some kind of obstacle,

239
00:09:50,022 --> 00:09:52,303
and they bounce against it.

240
00:09:52,303 --> 00:09:53,383
The only difference

241
00:09:53,383 --> 00:09:55,440
between the dominant female, here in red,

242
00:09:55,440 --> 00:09:56,925
and the other individuals,

243
00:09:56,925 --> 00:09:59,556
is that for her, the height of the obstacle,

244
00:09:59,556 --> 00:10:01,999
which is in fact the risk perceived from the road,

245
00:10:01,999 --> 00:10:03,979
is just slightly higher,

246
00:10:03,979 --> 00:10:05,717
and this tiny difference

247
00:10:05,717 --> 00:10:07,401
in the individual's rule of movement

248
00:10:07,401 --> 00:10:09,770
is sufficient to explain what we observe,

249
00:10:09,770 --> 00:10:12,499
that the dominant female

250
00:10:12,499 --> 00:10:13,995
leads her group to the road

251
00:10:13,995 --> 00:10:15,558
and then gives way to the others

252
00:10:15,558 --> 00:10:18,235
for them to cross first.

253
00:10:18,235 --> 00:10:21,886
George Box, who was an English statistician,

254
00:10:21,886 --> 00:10:24,848
once wrote, "All models are false,

255
00:10:24,848 --> 00:10:27,054
but some models are useful."

256
00:10:27,054 --> 00:10:30,166
And in fact, this model is obviously false,

257
00:10:30,166 --> 00:10:34,319
because in reality, meerkats are
anything but random particles.

258
00:10:34,319 --> 00:10:35,863
But it's also useful,

259
00:10:35,863 --> 00:10:38,690
because it tells us that extreme simplicity

260
00:10:38,690 --> 00:10:41,679
in movement rules at the individual level

261
00:10:41,679 --> 00:10:44,167
can result in a great deal of complexity

262
00:10:44,167 --> 00:10:46,367
at the level of the group.

263
00:10:46,367 --> 00:10:50,486
So again, that's simplifying complexity.

264
00:10:50,486 --> 00:10:51,671
Now, I would like to conclude

265
00:10:51,671 --> 00:10:54,212
on what this means for the whole species.

266
00:10:54,212 --> 00:10:55,967
When the dominant female

267
00:10:55,967 --> 00:10:57,749
gives way to a subordinate,

268
00:10:57,749 --> 00:11:00,082
it's not out of courtesy.

269
00:11:00,082 --> 00:11:01,361
In fact, the dominant female

270
00:11:01,361 --> 00:11:04,292
is extremely important for the cohesion of the group.

271
00:11:04,292 --> 00:11:07,496
If she dies on the road, the whole group is at risk.

272
00:11:07,496 --> 00:11:09,778
So this behavior of risk avoidance

273
00:11:09,778 --> 00:11:12,411
is a very old evolutionary response.

274
00:11:12,411 --> 00:11:16,371
These meerkats are replicating an evolved tactic

275
00:11:16,371 --> 00:11:18,558
that is thousands of generations old,

276
00:11:18,558 --> 00:11:20,849
and they're adapting it to a modern risk,

277
00:11:20,849 --> 00:11:24,236
in this case a road built by humans.

278
00:11:24,236 --> 00:11:26,723
They adapt very simple rules,

279
00:11:26,723 --> 00:11:28,966
and the resulting complex behavior

280
00:11:28,966 --> 00:11:31,876
allows them to resist human encroachment

281
00:11:31,876 --> 00:11:34,648
into their natural habitat.

282
00:11:34,648 --> 00:11:36,265
In the end,

283
00:11:36,265 --> 00:11:38,887
it may be bats which change their social structure

284
00:11:38,887 --> 00:11:41,044
in response to a population crush,

285
00:11:41,044 --> 00:11:42,778
or it may be meerkats

286
00:11:42,778 --> 00:11:45,673
who show a novel adaptation to a human road,

287
00:11:45,673 --> 00:11:48,220
or it may be another species.

288
00:11:48,220 --> 00:11:51,304
My message here, and it's not a complicated one

289
00:11:51,304 --> 00:11:54,206
but a simple one of wonder and hope,

290
00:11:54,206 --> 00:11:57,050
my message here is that animals

291
00:11:57,050 --> 00:11:59,754
show extraordinary social complexity,

292
00:11:59,754 --> 00:12:02,149
and this allows them to adapt

293
00:12:02,149 --> 00:12:05,615
and respond to changes in their environment.

294
00:12:05,615 --> 00:12:08,337
In three words, in the animal kingdom,

295
00:12:08,337 --> 00:12:11,234
simplicity leads to complexity

296
00:12:11,234 --> 00:12:12,902
which leads to resilience.

297
00:12:12,902 --> 00:12:14,754
Thank you.

298
00:12:14,754 --> 00:12:21,434
(Applause)