1
00:00:05,120 --> 00:00:09,520
Speaker 1: What is intelligence And if we look hard, might we

2
00:00:09,760 --> 00:00:13,600
Speaker 1: find it in very weird places, not just in brains,

3
00:00:13,720 --> 00:00:17,080
Speaker 1: but in all kinds of structures in our universe? And

4
00:00:17,239 --> 00:00:20,360
Speaker 1: how would we even recognize it? And what does any

5
00:00:20,440 --> 00:00:23,240
Speaker 1: of this have to do with a dog born without

6
00:00:23,280 --> 00:00:27,920
Speaker 1: front legs learning how to walk bipedally, or making new

7
00:00:28,000 --> 00:00:33,120
Speaker 1: little organisms out of single cells, or how Wikipedia might

8
00:00:33,200 --> 00:00:37,159
Speaker 1: be like an axilautel and why we are so blind

9
00:00:37,400 --> 00:00:41,680
Speaker 1: to the vast variety of minds that might surround us.

10
00:00:44,440 --> 00:00:48,000
Speaker 1: Welcome to Intercosmos with me, David Eagleman. I'm a neuroscientist

11
00:00:48,000 --> 00:00:50,919
Speaker 1: and an author at Stanford, and in these episodes we

12
00:00:51,000 --> 00:00:55,440
Speaker 1: sail deeply into our three pound universe to understand some

13
00:00:55,520 --> 00:01:13,600
Speaker 1: of the most surprising aspects of the world around us.

14
00:01:15,200 --> 00:01:18,840
Speaker 1: Today's episode is about intelligence, not in the way that

15
00:01:18,880 --> 00:01:22,680
Speaker 1: I've talked about in earlier episodes, about how the structure

16
00:01:22,720 --> 00:01:25,520
Speaker 1: of the human brain gives rise to intelligence and how

17
00:01:25,520 --> 00:01:29,839
Speaker 1: we can measure whether AI's intelligence. Today's episode is way

18
00:01:29,880 --> 00:01:32,600
Speaker 1: beyond that. Today I'm going to talk with one of

19
00:01:32,640 --> 00:01:37,000
Speaker 1: my most brilliant and creative colleagues, Michael Levin, about how

20
00:01:37,040 --> 00:01:41,720
Speaker 1: we might find intelligence all around us in ways that

21
00:01:41,760 --> 00:01:45,240
Speaker 1: we don't typically into it. So let's start at the beginning.

22
00:01:45,840 --> 00:01:49,520
Speaker 1: What is intelligence. It's a word that we usually reserve

23
00:01:49,920 --> 00:01:54,480
Speaker 1: for something abstract and cerebral, something associated with problem solving

24
00:01:54,560 --> 00:01:58,680
Speaker 1: and planning and passing IQ tests. We tend to picture

25
00:01:58,760 --> 00:02:03,680
Speaker 1: intelligence as a property of brains, and especially big human brains.

26
00:02:04,160 --> 00:02:08,480
Speaker 1: We're generally willing to grant some intelligence to dolphins and

27
00:02:08,560 --> 00:02:12,280
Speaker 1: chimps and clever birds like ravens, but it's hard to

28
00:02:12,400 --> 00:02:15,680
Speaker 1: know how to think about so many other things happening

29
00:02:15,720 --> 00:02:20,560
Speaker 1: in the world. For example, my skin cells heal a wound.

30
00:02:21,080 --> 00:02:25,080
Speaker 1: Is that intelligence or is that just biochemical cascades? A

31
00:02:25,160 --> 00:02:30,560
Speaker 1: plant grows towards sunlight intelligent? A worm gets its head

32
00:02:30,600 --> 00:02:33,920
Speaker 1: cut off and it regrows it. That's amazing, But we

33
00:02:34,000 --> 00:02:37,239
Speaker 1: don't tend to call that cognition. But what if we've

34
00:02:37,280 --> 00:02:41,480
Speaker 1: been looking at the whole notion of intelligence too narrowly.

35
00:02:41,800 --> 00:02:45,760
Speaker 1: What if intelligence isn't just about neurons and genes, but

36
00:02:45,840 --> 00:02:49,560
Speaker 1: it's about goals, and specifically, it's about the ability of

37
00:02:49,560 --> 00:02:54,920
Speaker 1: a system to navigate towards an objective, to adapt to

38
00:02:54,960 --> 00:02:57,840
Speaker 1: its circumstances, to make decisions.

39
00:02:57,200 --> 00:02:57,960
Speaker 2: Along the way.

40
00:02:58,400 --> 00:03:02,560
Speaker 1: That's a broader definition of intelligence. And if we apply it,

41
00:03:02,600 --> 00:03:07,600
Speaker 1: suddenly intelligence doesn't just belong to creatures with brains. It

42
00:03:07,680 --> 00:03:12,000
Speaker 1: becomes something that shows up in places we didn't expect.

43
00:03:12,120 --> 00:03:16,359
Speaker 1: Think about really simple creatures like a tadpole. It's millions

44
00:03:16,400 --> 00:03:20,919
Speaker 1: of cells collaborate and communicate and organize into an eye

45
00:03:21,000 --> 00:03:25,720
Speaker 1: and a spine and a heart without anybody orchestrating the

46
00:03:25,760 --> 00:03:28,880
Speaker 1: whole thing. There's no central planning, it's just a kind

47
00:03:28,880 --> 00:03:33,519
Speaker 1: of emergent intelligence at work. Or think about a flatworm

48
00:03:33,560 --> 00:03:36,600
Speaker 1: that can be cut into pieces and each piece regenerates

49
00:03:36,640 --> 00:03:40,760
Speaker 1: a complete, properly shaped body. How does each fragment know

50
00:03:41,240 --> 00:03:45,320
Speaker 1: what's missing? Where exactly is that information stored? What is

51
00:03:45,400 --> 00:03:49,600
Speaker 1: guiding the process? And as we ask these questions, that

52
00:03:49,720 --> 00:03:52,640
Speaker 1: leads us to ask how we can learn to talk

53
00:03:52,680 --> 00:03:55,400
Speaker 1: to these systems in the language that they understand, like

54
00:03:55,800 --> 00:03:59,840
Speaker 1: voltage gradients or bioelectric circuits or chemical signals. Can we

55
00:03:59,840 --> 00:04:04,760
Speaker 1: start reprogramming the goals of tissues? Can we tell a

56
00:04:04,880 --> 00:04:08,560
Speaker 1: clump of cells to build something new? And can we

57
00:04:08,720 --> 00:04:13,600
Speaker 1: use this kind of knowledge to regenerate organs, or repair

58
00:04:13,920 --> 00:04:15,200
Speaker 1: birth defects or.

59
00:04:15,560 --> 00:04:17,840
Speaker 2: Create entirely new forms of life.

60
00:04:17,920 --> 00:04:21,200
Speaker 1: These are the kinds of questions that today's guest has

61
00:04:21,240 --> 00:04:24,760
Speaker 1: spent his career exploring and his work leads us to

62
00:04:24,800 --> 00:04:29,680
Speaker 1: the conclusion that we're probably surrounded by minds, almost all

63
00:04:29,680 --> 00:04:33,120
Speaker 1: of which we don't recognize. Minds are pervasive, and they're

64
00:04:33,160 --> 00:04:37,440
Speaker 1: not restricted to brains, but spread across all kinds of

65
00:04:37,560 --> 00:04:45,040
Speaker 1: levels of organization, from single cells to societies. My guest

66
00:04:45,400 --> 00:04:49,320
Speaker 1: is Michael Levin. He's a distinguished professor of developmental and

67
00:04:49,360 --> 00:04:52,760
Speaker 1: Synthetic biology at Tufts University, and I've had him on

68
00:04:52,800 --> 00:04:55,920
Speaker 1: the podcast before because he's really one of my favorite

69
00:04:55,960 --> 00:04:59,279
Speaker 1: thinkers in the field. He's massively creative and always pulling

70
00:04:59,279 --> 00:05:04,039
Speaker 1: off amazing results that the frontier where biology meets information

71
00:05:04,120 --> 00:05:07,400
Speaker 1: theory or computation or philosophy, and as you're going to see,

72
00:05:07,400 --> 00:05:11,960
Speaker 1: his work always challenges our deepest intuitions about agency and

73
00:05:12,040 --> 00:05:16,240
Speaker 1: memory and selfhood. So you've heard of SETI, the Search

74
00:05:16,320 --> 00:05:21,359
Speaker 1: for Extraterrestrial Intelligence. Recently, Levin proposed SUTI, the search for

75
00:05:21,800 --> 00:05:27,000
Speaker 1: unconventional terrestrial intelligence. As we're about to hear, his position

76
00:05:27,160 --> 00:05:30,880
Speaker 1: is that right here on Earth, there are already aliens

77
00:05:31,040 --> 00:05:35,440
Speaker 1: among us that stretch and sometimes break. Are typical ways

78
00:05:35,839 --> 00:05:40,520
Speaker 1: of thinking about minds. So if you've ever wondered where

79
00:05:40,560 --> 00:05:44,120
Speaker 1: intelligence begins, how far it reaches, or whether you might

80
00:05:44,200 --> 00:05:47,719
Speaker 1: share more in common with blobs of cells than you think.

81
00:05:48,000 --> 00:05:56,640
Speaker 1: This episode is for you. Here's my interview with Mike Levin. So, Mike,

82
00:05:56,800 --> 00:06:00,839
Speaker 1: let's start by telling us how you define intelligence.

83
00:06:01,240 --> 00:06:04,520
Speaker 3: Okay, what I use is a definition that helps us

84
00:06:04,640 --> 00:06:06,520
Speaker 3: move forward in the lab. I do not claim that

85
00:06:06,520 --> 00:06:09,000
Speaker 3: it's the only definition or that it captures everything there

86
00:06:09,080 --> 00:06:13,080
Speaker 3: is to capture about intelligence. But I like William James's definition,

87
00:06:13,120 --> 00:06:15,440
Speaker 3: which is some degree of the ability to reach the

88
00:06:15,480 --> 00:06:18,279
Speaker 3: same goal by different means. So it's some level of

89
00:06:18,440 --> 00:06:21,280
Speaker 3: ingenuity to get your goals met when things change. That

90
00:06:21,320 --> 00:06:24,640
Speaker 3: doesn't capture play. It doesn't necessarily capture creativity, things other

91
00:06:24,680 --> 00:06:25,479
Speaker 3: than problem solving.

92
00:06:25,520 --> 00:06:27,560
Speaker 2: But this is what we focus on experimentally.

93
00:06:27,720 --> 00:06:30,360
Speaker 1: So typically when we think about intelligence, we think about

94
00:06:30,440 --> 00:06:33,279
Speaker 1: brains and nervous systems. But you think it doesn't even

95
00:06:33,279 --> 00:06:34,719
Speaker 1: necessarily require that.

96
00:06:35,160 --> 00:06:38,400
Speaker 3: Correct, Because if you're looking at it in this way,

97
00:06:38,400 --> 00:06:41,280
Speaker 3: that it's basically a functional capacity to navigate some kind

98
00:06:41,279 --> 00:06:45,440
Speaker 3: of problem space and meet specific goals under changing circumstances.

99
00:06:45,680 --> 00:06:48,400
Speaker 3: There are apparently a wide range of architectures that can

100
00:06:48,400 --> 00:06:50,680
Speaker 3: do this, and in order to see that what you

101
00:06:50,760 --> 00:06:53,960
Speaker 3: need to do is to relax some really constraining assumptions

102
00:06:54,000 --> 00:06:56,280
Speaker 3: that we often have about the problem space that we're

103
00:06:56,279 --> 00:06:56,680
Speaker 3: working in.

104
00:06:56,839 --> 00:07:00,360
Speaker 1: And so you often describe intelligence as scale free. So

105
00:07:00,440 --> 00:07:01,880
Speaker 1: just give us a sense what you mean by that.

106
00:07:02,040 --> 00:07:04,920
Speaker 3: Yeah, I mean that you know, as humans, we are

107
00:07:05,000 --> 00:07:07,000
Speaker 3: because of our own evolutionary firm where we are so

108
00:07:07,120 --> 00:07:10,040
Speaker 3: obsessed with three dimensional space and moving around in three

109
00:07:10,040 --> 00:07:12,360
Speaker 3: dimensional space, to the point where if people see some

110
00:07:12,400 --> 00:07:14,520
Speaker 3: sort of AI that isn't rolling around in some sort

111
00:07:14,560 --> 00:07:16,920
Speaker 3: of robotic body, they're going to say, this is not embody, right,

112
00:07:16,920 --> 00:07:19,920
Speaker 3: because they're expecting a body. People are expecting a body

113
00:07:19,920 --> 00:07:23,520
Speaker 3: that moves through three dimensional space. But actually the biology,

114
00:07:23,560 --> 00:07:26,400
Speaker 3: for example, has been solving problems and navigating all kinds

115
00:07:26,440 --> 00:07:28,560
Speaker 3: of spaces that are hard for us to visualize. So

116
00:07:28,600 --> 00:07:31,600
Speaker 3: the space of gene expression states, the space of physiological states,

117
00:07:31,600 --> 00:07:34,720
Speaker 3: the space of anatomical possible outcomes, and so on, and

118
00:07:34,760 --> 00:07:38,000
Speaker 3: so in order to understand how we navigate those spaces,

119
00:07:38,440 --> 00:07:40,200
Speaker 3: you have to think in other scales. Some of these

120
00:07:40,200 --> 00:07:42,880
Speaker 3: things happen very slowly, some of these things happen incredibly quickly.

121
00:07:43,400 --> 00:07:45,800
Speaker 3: Some of these things are very small, some of them

122
00:07:45,800 --> 00:07:49,160
Speaker 3: are very large, and we are just you know, focused

123
00:07:49,200 --> 00:07:52,400
Speaker 3: on medium sized objects moving at medium speeds. But you know,

124
00:07:52,560 --> 00:07:55,080
Speaker 3: I'm not claiming it's entirely scale free, but I'm saying

125
00:07:55,080 --> 00:07:59,320
Speaker 3: that there are very deep scale invariant principles that operate

126
00:07:59,360 --> 00:08:01,520
Speaker 3: at many different scales besides the ones we're used to.

127
00:08:01,760 --> 00:08:05,720
Speaker 1: So how have you gone about looking for intelligence at

128
00:08:05,720 --> 00:08:06,440
Speaker 1: other scales?

129
00:08:06,800 --> 00:08:09,640
Speaker 3: So one thing that you that you can do once

130
00:08:09,640 --> 00:08:13,000
Speaker 3: you've bought into the fact that we can't assume how

131
00:08:13,040 --> 00:08:15,240
Speaker 3: intelligent anything is or what kind of intelligence it has,

132
00:08:15,280 --> 00:08:17,680
Speaker 3: but you have to do experiments. Then then what it

133
00:08:17,720 --> 00:08:19,600
Speaker 3: turns out you have to do is you have to

134
00:08:19,920 --> 00:08:22,720
Speaker 3: posit some sort of problem space you have that the

135
00:08:22,760 --> 00:08:25,080
Speaker 3: system is working in. You have to posit some sort

136
00:08:25,080 --> 00:08:27,200
Speaker 3: of goal. That is, these are all hypotheses on some

137
00:08:27,240 --> 00:08:29,600
Speaker 3: sort of goal that it's it's trying to reach. And

138
00:08:29,640 --> 00:08:32,280
Speaker 3: then what you have to do is perturbational experiments to

139
00:08:32,920 --> 00:08:34,840
Speaker 3: prevent it from reaching its goal. And then you see

140
00:08:34,840 --> 00:08:36,880
Speaker 3: what how you know how how smart the system is

141
00:08:36,920 --> 00:08:38,640
Speaker 3: in getting around whatever you did to it, So you.

142
00:08:38,640 --> 00:08:40,480
Speaker 1: Knock it off track and then you see how it

143
00:08:40,520 --> 00:08:42,319
Speaker 1: comes back, or if it comes back.

144
00:08:42,320 --> 00:08:45,240
Speaker 3: Knock it off track is a good one add barriers

145
00:08:45,360 --> 00:08:47,319
Speaker 3: of some sort in whatever space you're working doesn't have

146
00:08:47,360 --> 00:08:49,280
Speaker 3: to be a physical barrier, but whatever space you're working in,

147
00:08:49,320 --> 00:08:51,680
Speaker 3: add a barrier, or in fact manipulate.

148
00:08:51,200 --> 00:08:52,000
Speaker 2: The system itself.

149
00:08:52,120 --> 00:08:54,240
Speaker 3: So change the system, right, It's not all about the environment,

150
00:08:54,360 --> 00:08:56,120
Speaker 3: so it can be about the system itself. And so

151
00:08:56,559 --> 00:08:59,160
Speaker 3: we've done this now in many different context Here are

152
00:08:59,160 --> 00:09:01,280
Speaker 3: a couple of favorites of ours. The biggest one that

153
00:09:01,320 --> 00:09:04,120
Speaker 3: we do most of our work in is morphogenesis. So

154
00:09:04,440 --> 00:09:07,120
Speaker 3: we all make a journey from a single cell to whatever.

155
00:09:07,360 --> 00:09:09,760
Speaker 3: You know, we're going to be a human, a giraffe, plant, whatever,

156
00:09:09,840 --> 00:09:12,440
Speaker 3: And that journey, as it turns out, as a matter

157
00:09:12,480 --> 00:09:14,760
Speaker 3: of experimental fact, it turns out that journey is not

158
00:09:15,000 --> 00:09:17,640
Speaker 3: a kind of open loop. What you know, the way

159
00:09:17,640 --> 00:09:20,400
Speaker 3: people model the emergence and complexity, lots of simple things

160
00:09:20,440 --> 00:09:23,199
Speaker 3: happening again and again, and ultimately some sort of complex

161
00:09:23,240 --> 00:09:23,920
Speaker 3: event happens.

162
00:09:24,080 --> 00:09:27,160
Speaker 2: That isn't how it works. It's very contact sensitive.

163
00:09:27,320 --> 00:09:30,800
Speaker 3: If you try to prevent let's say, embryos or regenerating limbs,

164
00:09:31,120 --> 00:09:33,840
Speaker 3: or or you know, in any you know, metamorphosis, any

165
00:09:33,840 --> 00:09:36,240
Speaker 3: of these processes, you try to prevent them from reaching

166
00:09:36,280 --> 00:09:39,240
Speaker 3: their goal. They often have extremely ingenious ways to get

167
00:09:39,240 --> 00:09:42,520
Speaker 3: there anyway, okay, And you can quantify this, and you

168
00:09:42,559 --> 00:09:46,120
Speaker 3: can say, what are kinds of perturbations that it's able

169
00:09:46,160 --> 00:09:48,760
Speaker 3: to deal with? And does it have delayed gratification? Does

170
00:09:48,800 --> 00:09:51,680
Speaker 3: it have planning, does it have a representation of the

171
00:09:52,200 --> 00:09:55,000
Speaker 3: of counterfactual states? Does it have learning and memory? Does

172
00:09:55,040 --> 00:09:57,520
Speaker 3: it store you know, a map of its environment? You

173
00:09:57,559 --> 00:09:59,400
Speaker 3: can you can test all of these things empirically.

174
00:09:59,559 --> 00:10:02,720
Speaker 1: So I mean in terms of let's say an embryo developing,

175
00:10:03,040 --> 00:10:06,559
Speaker 1: what we think traditionally in textbooks is that the genetics

176
00:10:06,559 --> 00:10:09,720
Speaker 1: somehow gives a blueprint and the whole thing just donepacks.

177
00:10:10,040 --> 00:10:13,280
Speaker 1: But you're asking is how is the system intelligent if

178
00:10:13,280 --> 00:10:15,640
Speaker 1: we knock it off track or put barriers in the way,

179
00:10:15,840 --> 00:10:19,000
Speaker 1: how does it figure out how to come together correctly? So,

180
00:10:19,200 --> 00:10:21,920
Speaker 1: what's the specific example of something you've done here? Here

181
00:10:22,080 --> 00:10:24,400
Speaker 1: there are some of my favorites. These first two are

182
00:10:24,440 --> 00:10:26,640
Speaker 1: not my work. These this is like classic, classic work

183
00:10:26,640 --> 00:10:27,120
Speaker 1: in the field.

184
00:10:27,480 --> 00:10:31,440
Speaker 3: So if you take normally, imagine cutting a cross section

185
00:10:31,559 --> 00:10:34,600
Speaker 3: through the kidney tubule of a newt. Normally what you'd

186
00:10:34,640 --> 00:10:36,400
Speaker 3: find is like eight to ten cells and they work

187
00:10:36,440 --> 00:10:40,760
Speaker 3: together to build to build this tube like structure. So

188
00:10:41,000 --> 00:10:42,840
Speaker 3: what you can do is you can make polyploid neutes

189
00:10:42,880 --> 00:10:45,760
Speaker 3: that have multiple copies of their chromosomes, which means their

190
00:10:45,760 --> 00:10:47,920
Speaker 3: cells have to get bigger. Those newts are still the

191
00:10:47,960 --> 00:10:50,840
Speaker 3: same correct size. So that's the first interesting thing. Wow,

192
00:10:50,840 --> 00:10:53,120
Speaker 3: well the cells get bigger, the thing scales down. How

193
00:10:53,120 --> 00:10:56,080
Speaker 3: does it do it by using fewer but bigger cells

194
00:10:56,120 --> 00:10:57,480
Speaker 3: to make the exact same structure.

195
00:10:57,679 --> 00:11:00,000
Speaker 2: So that's an adjustment, right, never mind the environment.

196
00:11:00,080 --> 00:11:01,920
Speaker 3: Your own parts are changing, and this thing is figuring

197
00:11:01,920 --> 00:11:03,480
Speaker 3: out how to get to the exact same goal, the

198
00:11:03,480 --> 00:11:06,840
Speaker 3: same neud same shape, same size, fewer of these bigger cells.

199
00:11:06,960 --> 00:11:08,360
Speaker 2: So let me ask you a quick question.

200
00:11:08,480 --> 00:11:12,480
Speaker 1: Is this analogous to the fact that a mouse's heart

201
00:11:12,520 --> 00:11:15,720
Speaker 1: and an elephant's heart are doing the same thing, but

202
00:11:15,760 --> 00:11:19,160
Speaker 1: they're made of a completely different number of cells. It's

203
00:11:19,200 --> 00:11:21,560
Speaker 1: a massive heart in an elephant, very tiny mouse, but

204
00:11:21,760 --> 00:11:23,760
Speaker 1: it's doing the same function.

205
00:11:24,160 --> 00:11:27,240
Speaker 3: It's similar, but there's only but there's one major difference

206
00:11:27,280 --> 00:11:29,480
Speaker 3: both in a mouse and in an elephant. What people

207
00:11:29,520 --> 00:11:31,960
Speaker 3: will say is, well, both of those have had long,

208
00:11:32,280 --> 00:11:34,320
Speaker 3: long history of being what they are, and so this

209
00:11:34,400 --> 00:11:35,040
Speaker 3: is just mechanical.

210
00:11:35,120 --> 00:11:36,080
Speaker 2: It just does what it does.

211
00:11:36,760 --> 00:11:39,480
Speaker 3: My example is different because you've done something to this

212
00:11:39,600 --> 00:11:40,480
Speaker 3: new that it does.

213
00:11:40,360 --> 00:11:41,040
Speaker 2: Not normally do.

214
00:11:41,240 --> 00:11:45,400
Speaker 3: You've given it a completely novel circumstance, and then it

215
00:11:45,440 --> 00:11:48,040
Speaker 3: adjusts in a new way. And the craziest thing happens

216
00:11:48,200 --> 00:11:51,440
Speaker 3: when you make the cells really gigantic. Okay, these are

217
00:11:51,440 --> 00:11:53,680
Speaker 3: like I think six or eight and the newts so

218
00:11:53,760 --> 00:11:56,920
Speaker 3: massive polyplaty. What happens is the cells are so big

219
00:11:57,040 --> 00:11:58,560
Speaker 3: there's no room for more than one cell.

220
00:11:58,679 --> 00:12:01,440
Speaker 2: One cell will wrap around und itself and give you

221
00:12:01,520 --> 00:12:03,120
Speaker 2: the lumen of the tubule in the middle.

222
00:12:03,280 --> 00:12:06,280
Speaker 3: Now, now this is crazy because because that is that's

223
00:12:06,320 --> 00:12:08,960
Speaker 3: a different molecular mechanism that side of skeletalle bending before

224
00:12:09,000 --> 00:12:12,000
Speaker 3: it was sell to sell communication and tubulo genesis. So

225
00:12:12,160 --> 00:12:13,880
Speaker 3: think about what this means. If you're a nude coming

226
00:12:13,880 --> 00:12:16,360
Speaker 3: into the world. Never mind the environment. You don't really

227
00:12:16,360 --> 00:12:17,840
Speaker 3: know what your environment's going to be. You don't know

228
00:12:17,840 --> 00:12:19,800
Speaker 3: how many copies of your chromosomes you're going to have,

229
00:12:20,160 --> 00:12:22,080
Speaker 3: you don't know how big your cells are going to be.

230
00:12:22,160 --> 00:12:25,200
Speaker 3: You don't know which of your many genetic affordances you

231
00:12:25,240 --> 00:12:27,800
Speaker 3: can use. Right, you have different molecular mechanisms you can use.

232
00:12:28,040 --> 00:12:29,320
Speaker 3: You have to figure out what to do in a

233
00:12:29,320 --> 00:12:31,920
Speaker 3: novel circumstance and still get the job done. I mean

234
00:12:32,000 --> 00:12:34,679
Speaker 3: this sounds like every IQ test you've ever heard of

235
00:12:34,720 --> 00:12:36,640
Speaker 3: when people show you, here's a little box, some tax

236
00:12:36,679 --> 00:12:38,280
Speaker 3: and a candle, and I want you to, you know,

237
00:12:38,360 --> 00:12:41,880
Speaker 3: solve this this particular problem. Right, Yeah, you have genetic affordances,

238
00:12:42,200 --> 00:12:45,840
Speaker 3: and then that morphogenetic process is not just doing the

239
00:12:45,840 --> 00:12:47,000
Speaker 3: same thing every single time.

240
00:12:47,040 --> 00:12:49,000
Speaker 2: You have to solve these problems. That's one of my

241
00:12:49,000 --> 00:12:49,840
Speaker 2: favorite examples.

242
00:12:50,000 --> 00:12:53,440
Speaker 3: Another one that we discovered is tadpoles become frogs, and

243
00:12:53,480 --> 00:12:54,960
Speaker 3: in order to do that, they have to rearrange their

244
00:12:55,000 --> 00:12:57,800
Speaker 3: face because their face looks actually quite different from and

245
00:12:58,120 --> 00:12:59,679
Speaker 3: you know, so the eyes, the jaws of reading has

246
00:12:59,679 --> 00:13:00,440
Speaker 3: to has to move.

247
00:13:00,520 --> 00:13:02,880
Speaker 2: What people thought was that this is a hardwired process.

248
00:13:02,920 --> 00:13:05,559
Speaker 3: Basically, somehow the genetics just tells every organ how far

249
00:13:05,600 --> 00:13:07,800
Speaker 3: to moving, what direction, and then you get from a

250
00:13:07,840 --> 00:13:09,240
Speaker 3: normal tadpole to a normal frog.

251
00:13:09,440 --> 00:13:10,439
Speaker 2: So we decided to test that.

252
00:13:10,520 --> 00:13:12,400
Speaker 3: Because you can't assume these things, you have to test

253
00:13:12,679 --> 00:13:16,120
Speaker 3: and so what we created was something called Picasso tadpoles,

254
00:13:16,400 --> 00:13:20,720
Speaker 3: so basically scrambled all the initial positions, so the mouth

255
00:13:20,800 --> 00:13:22,320
Speaker 3: is off to the side, the eyes on the back

256
00:13:22,360 --> 00:13:25,760
Speaker 3: of the head like everything is completely scramble. And what

257
00:13:25,800 --> 00:13:28,480
Speaker 3: we find is that they make pretty normal frogs because

258
00:13:28,520 --> 00:13:31,120
Speaker 3: all of these things will move in novel paths to

259
00:13:31,400 --> 00:13:33,600
Speaker 3: reach the correct goal, and then they stop. Actually, sometimes

260
00:13:33,640 --> 00:13:35,040
Speaker 3: they go too far and they have to double back

261
00:13:35,040 --> 00:13:36,920
Speaker 3: and stop. This is another example. You start them off

262
00:13:36,920 --> 00:13:38,760
Speaker 3: in the wrong position, they don't just blindly go the

263
00:13:38,800 --> 00:13:41,720
Speaker 3: same distance. They actually go until they meet their goal.

264
00:13:41,760 --> 00:13:42,640
Speaker 3: And you know, it's a goal.

265
00:13:42,840 --> 00:13:44,960
Speaker 2: And when I say goal, I don't mean it's a human.

266
00:13:44,800 --> 00:13:46,520
Speaker 3: Level, like I know what my goal is. That's a

267
00:13:46,559 --> 00:13:48,880
Speaker 3: kind of metacognition that I'm not claiming here. I'm saying

268
00:13:48,960 --> 00:13:51,640
Speaker 3: it's in the cybernetic sets like your thermostat has goals.

269
00:13:51,800 --> 00:13:54,000
Speaker 3: It's a set point, and now how clever are you

270
00:13:54,080 --> 00:13:56,520
Speaker 3: to be able to reach that set point when things change?

271
00:13:58,120 --> 00:14:00,920
Speaker 1: As an interesting analogy, what's going on at the level

272
00:14:00,960 --> 00:14:04,000
Speaker 1: of brain plasticity. We tend to think that, let's say,

273
00:14:04,040 --> 00:14:08,880
Speaker 1: a dog's brain is pre wired to drive a dog's body.

274
00:14:09,240 --> 00:14:11,800
Speaker 1: But one of the examples that I talked about in

275
00:14:11,800 --> 00:14:14,880
Speaker 1: my book Live Wired was this dog who was born

276
00:14:14,960 --> 00:14:18,480
Speaker 1: without front legs, and so she just walks bipedally and

277
00:14:18,520 --> 00:14:20,400
Speaker 1: she moves all around and.

278
00:14:20,400 --> 00:14:21,240
Speaker 2: Gets by that way.

279
00:14:21,280 --> 00:14:25,760
Speaker 1: Why, because she needed to get to her her dog

280
00:14:25,800 --> 00:14:28,360
Speaker 1: bowl and her water and other dogs and so on,

281
00:14:28,600 --> 00:14:30,320
Speaker 1: and so she just figured out. It turns out it's

282
00:14:30,360 --> 00:14:32,840
Speaker 1: not that hard for a dog to walk on back legs.

283
00:14:32,840 --> 00:14:35,160
Speaker 1: And the question is could all dogs walk on their

284
00:14:35,200 --> 00:14:39,000
Speaker 1: back legs? Presumably, but they don't have the proper motivation

285
00:14:39,920 --> 00:14:43,440
Speaker 1: to do so. But the point is that the dog's

286
00:14:43,520 --> 00:14:44,760
Speaker 1: body is very flexible.

287
00:14:44,760 --> 00:14:46,400
Speaker 2: It meets the goals of the world.

288
00:14:46,720 --> 00:14:51,080
Speaker 1: Another analogy is the world's best archer, as in he's

289
00:14:51,120 --> 00:14:55,240
Speaker 1: got the world record for the longest accurate shot in archery.

290
00:14:55,680 --> 00:14:57,920
Speaker 1: Is a guy named Matt Stutsman who happens to have

291
00:14:58,000 --> 00:15:01,440
Speaker 1: no arms, and he got interest in archery and figured

292
00:15:01,480 --> 00:15:05,160
Speaker 1: out how to pull the bow with his legs, and

293
00:15:05,240 --> 00:15:07,840
Speaker 1: so he shoots with his legs and became a great

294
00:15:07,920 --> 00:15:08,600
Speaker 1: archer that way.

295
00:15:09,000 --> 00:15:12,720
Speaker 2: Amazing, amazing. Yeah, yeah, the plasticity is incredible.

296
00:15:12,800 --> 00:15:15,640
Speaker 3: And you know, the earlier the earliest example that I

297
00:15:15,680 --> 00:15:17,520
Speaker 3: know of the of this hind leg thing is is

298
00:15:17,720 --> 00:15:20,080
Speaker 3: Slipper's Goat, which was this I think it was in

299
00:15:20,120 --> 00:15:23,560
Speaker 3: the forties. This guy Slipper, a published study of a

300
00:15:23,600 --> 00:15:26,120
Speaker 3: goat who, again born without four legs, learned to walk

301
00:15:26,160 --> 00:15:28,720
Speaker 3: on its hind legs. When they dissected the goat, they

302
00:15:28,720 --> 00:15:31,440
Speaker 3: found out that a lot of the adjustments that you

303
00:15:31,480 --> 00:15:33,840
Speaker 3: need for bipedal locomotion, right, So things about the hips,

304
00:15:34,160 --> 00:15:36,480
Speaker 3: you know, the spine, all kind of stuff, we're all

305
00:15:36,520 --> 00:15:38,680
Speaker 3: there right as opposed to what you normally think of

306
00:15:38,720 --> 00:15:40,640
Speaker 3: for the evolution of modern humans as you know how

307
00:15:40,880 --> 00:15:42,840
Speaker 3: I mean, many hundreds of thousands of years you need

308
00:15:42,880 --> 00:15:45,480
Speaker 3: for that. And this is what's really interesting about this

309
00:15:45,520 --> 00:15:47,920
Speaker 3: plasticity is that you can project it into other spaces.

310
00:15:47,960 --> 00:15:50,600
Speaker 3: So so, as you pointed out, you know, can a

311
00:15:50,640 --> 00:15:52,640
Speaker 3: dog brain run an upright body?

312
00:15:52,720 --> 00:15:52,920
Speaker 2: Right?

313
00:15:53,240 --> 00:15:53,480
Speaker 3: Now?

314
00:15:53,800 --> 00:15:55,520
Speaker 2: Look at individual cells?

315
00:15:55,520 --> 00:15:58,600
Speaker 3: Can the same genome run a completely different anatomy and

316
00:15:58,640 --> 00:16:01,200
Speaker 3: set of behaviors? And this is what we've I mean,

317
00:16:01,240 --> 00:16:03,920
Speaker 3: other people have shown other examples of this, but for example,

318
00:16:03,920 --> 00:16:07,680
Speaker 3: in our lab, xenobots anthwrobots, right, these living constructs that

319
00:16:07,760 --> 00:16:10,440
Speaker 3: have a completely different body than what they normally do.

320
00:16:10,440 --> 00:16:13,720
Speaker 3: They have a different behavioral repertoire, no genetic change, same

321
00:16:13,840 --> 00:16:17,000
Speaker 3: gene regulatory networks are running a completely different body.

322
00:16:17,080 --> 00:16:20,280
Speaker 1: For the listenership, can you define anthrobots and xenobots which

323
00:16:20,280 --> 00:16:20,760
Speaker 1: you've built?

324
00:16:20,920 --> 00:16:22,520
Speaker 2: Sure, let's start with the zenobots.

325
00:16:22,520 --> 00:16:25,800
Speaker 3: So in the cases of zennabots, what our team did,

326
00:16:25,800 --> 00:16:28,160
Speaker 3: and this is in collaboration with Josh bond Guard's lab

327
00:16:28,160 --> 00:16:30,320
Speaker 3: at University the e Vermont and this is Doug Blackiston

328
00:16:30,360 --> 00:16:32,560
Speaker 3: in my group and Sam Kregman did a lot of

329
00:16:32,560 --> 00:16:35,040
Speaker 3: the computational work for it. What happens is that when

330
00:16:35,040 --> 00:16:37,640
Speaker 3: you liberate some epithelial cells from an early frog embryo,

331
00:16:37,840 --> 00:16:40,160
Speaker 3: normally what they do is they form this like two

332
00:16:40,240 --> 00:16:43,920
Speaker 3: dimensional outer covering of an embryo and the outer skin layer,

333
00:16:43,960 --> 00:16:46,680
Speaker 3: and they do that because they're induced to do that by.

334
00:16:46,560 --> 00:16:47,280
Speaker 2: The other cells.

335
00:16:47,360 --> 00:16:49,000
Speaker 3: Well, if you get them away from the other cells,

336
00:16:49,000 --> 00:16:50,840
Speaker 3: you sort of liberate, then find out what they really

337
00:16:50,840 --> 00:16:52,600
Speaker 3: want to do on their own and what they do.

338
00:16:52,920 --> 00:16:53,840
Speaker 2: They could do many things.

339
00:16:53,880 --> 00:16:55,560
Speaker 3: They could crawl away from each other, they could die,

340
00:16:55,560 --> 00:16:58,280
Speaker 3: they could make a flat layer like cell culture. What

341
00:16:58,320 --> 00:17:00,320
Speaker 3: to actually do is they form this little ball with

342
00:17:00,720 --> 00:17:02,640
Speaker 3: cilia that are on the outside. He's a little moving

343
00:17:02,680 --> 00:17:05,640
Speaker 3: hairs and they organize them so that the thing can

344
00:17:05,680 --> 00:17:07,160
Speaker 3: swim and it starts swimming around.

345
00:17:07,160 --> 00:17:08,800
Speaker 2: It has all sorts of interesting behaviors.

346
00:17:08,840 --> 00:17:10,159
Speaker 3: A couple of years ago we show that they do

347
00:17:10,200 --> 00:17:12,959
Speaker 3: kinematic cell for replication, which is that if you sprinkle

348
00:17:12,960 --> 00:17:14,760
Speaker 3: a bunch of loose skin cells in their environment, they

349
00:17:14,800 --> 00:17:17,160
Speaker 3: will collect them into little balls and guess what, those

350
00:17:17,200 --> 00:17:19,760
Speaker 3: become the next generation of zenobots. Right, So they can

351
00:17:19,760 --> 00:17:22,440
Speaker 3: do this weird kinematic replication that, as far as we know,

352
00:17:22,520 --> 00:17:25,880
Speaker 3: no other creature does. They express hundreds of genes differently

353
00:17:25,960 --> 00:17:29,320
Speaker 3: than they do within the embryo. No genetic change. By

354
00:17:29,359 --> 00:17:31,320
Speaker 3: the way, this is we're not adding anything. There are

355
00:17:31,320 --> 00:17:34,320
Speaker 3: no scaffolds, no, no synthetic circuits. But they use their

356
00:17:34,960 --> 00:17:38,960
Speaker 3: transcriptional affordances differently. They turn to hundreds of new genes

357
00:17:39,359 --> 00:17:42,119
Speaker 3: and among other things, it turns out they're sensitive to

358
00:17:42,320 --> 00:17:43,440
Speaker 3: acoustic vibrations.

359
00:17:43,560 --> 00:17:45,400
Speaker 2: That's the latest thing that just came out a month ago.

360
00:17:45,440 --> 00:17:47,280
Speaker 3: Is that we get because we found they were turning

361
00:17:47,280 --> 00:17:49,000
Speaker 3: out a bunch of genes related to hearing.

362
00:17:48,760 --> 00:17:50,520
Speaker 2: And we said, is it possible at these things we hear?

363
00:17:50,840 --> 00:17:53,800
Speaker 3: And so Vipofpie my group put a speaker under them

364
00:17:53,880 --> 00:17:55,880
Speaker 3: and showed that, yeah, there's actually sounds you can send

365
00:17:55,880 --> 00:17:57,399
Speaker 3: them that they will respond to.

366
00:17:57,640 --> 00:17:58,600
Speaker 2: So that's zenobots.

367
00:17:58,640 --> 00:18:02,040
Speaker 3: Anthrobots are a similar story because when we first did it,

368
00:18:02,080 --> 00:18:05,280
Speaker 3: some people said, well, you know, they're embryonic cells and

369
00:18:05,400 --> 00:18:07,679
Speaker 3: amphibia are plastic. Maybe that's why this is like a

370
00:18:07,680 --> 00:18:10,280
Speaker 3: frog embryology thing. You know, this is specific to and centebat.

371
00:18:10,440 --> 00:18:12,160
Speaker 3: So I said, okay, what's the furthest you can get

372
00:18:12,160 --> 00:18:13,080
Speaker 3: from an embryonic frog.

373
00:18:13,080 --> 00:18:14,160
Speaker 2: Well, I'll be an adult human.

374
00:18:14,520 --> 00:18:17,879
Speaker 3: And so we went and we took trachael epithelial cells

375
00:18:18,119 --> 00:18:21,040
Speaker 3: from adult human patients and we showed and this is

376
00:18:21,240 --> 00:18:24,040
Speaker 3: Gizem Gumushke's work, PhD student in my group.

377
00:18:23,800 --> 00:18:26,440
Speaker 2: Who developed a protocol whereby.

378
00:18:26,400 --> 00:18:29,600
Speaker 3: Again simply by taking the cells out of their normal context,

379
00:18:29,640 --> 00:18:33,040
Speaker 3: you get to release their the various possible outcomes that

380
00:18:33,080 --> 00:18:35,320
Speaker 3: they can do, and they make anthrobots. It's a little

381
00:18:36,119 --> 00:18:39,360
Speaker 3: round little thing that zips around. It has a couple

382
00:18:39,440 --> 00:18:42,880
Speaker 3: of interesting properties. First of all, it can heal neural wounds.

383
00:18:43,080 --> 00:18:45,000
Speaker 3: So if you played a dish of human neurons and

384
00:18:45,000 --> 00:18:46,840
Speaker 3: you put a big scratch through it with a scalpel,

385
00:18:47,200 --> 00:18:49,600
Speaker 3: they will. When they find the scratch, they settle down,

386
00:18:49,600 --> 00:18:51,359
Speaker 3: a bunch of them. We call it a superbot cluster.

387
00:18:51,440 --> 00:18:54,119
Speaker 3: They settle down and within about four days, if you

388
00:18:54,200 --> 00:18:55,560
Speaker 3: lift them up, you see that what they did. For

389
00:18:55,640 --> 00:18:57,800
Speaker 3: what they did meanwhile is they healed across the They

390
00:18:57,880 --> 00:19:00,200
Speaker 3: healed across the gap. Okay, who would have thought that

391
00:19:00,240 --> 00:19:02,960
Speaker 3: your trachular ethelial cells that sit there quietly dealing with

392
00:19:03,000 --> 00:19:06,199
Speaker 3: you know, mucus and and and the air particles have

393
00:19:06,280 --> 00:19:09,520
Speaker 3: the ability to to to to heal neurons. And these

394
00:19:09,520 --> 00:19:13,840
Speaker 3: guys express about nine thousand genes differently than right, so

395
00:19:13,960 --> 00:19:17,000
Speaker 3: what almost half the genome they express differently than than

396
00:19:17,000 --> 00:19:20,000
Speaker 3: they do in the body. They're, by the way, younger

397
00:19:20,040 --> 00:19:22,240
Speaker 3: than the patient, than the than the cells that they

398
00:19:22,240 --> 00:19:24,679
Speaker 3: came from. So so actually that process of becoming an

399
00:19:24,680 --> 00:19:27,000
Speaker 3: answer what actually rolls back the epigenetic clock.

400
00:19:27,280 --> 00:19:29,000
Speaker 2: So so they're they're they're a bit younger.

401
00:19:29,000 --> 00:19:31,720
Speaker 3: This is fascinating, you know, behaviors and and all of

402
00:19:31,720 --> 00:19:34,200
Speaker 3: this is run by that standard controller. So so that's

403
00:19:34,240 --> 00:19:36,119
Speaker 3: kind of my point is that is that there's amazing

404
00:19:36,160 --> 00:19:39,119
Speaker 3: plasticity in the brain and nervous system. But this goes

405
00:19:39,160 --> 00:19:41,359
Speaker 3: all the way down. This is not just for you know,

406
00:19:41,440 --> 00:19:42,639
Speaker 3: fancy fancy brains.

407
00:19:57,560 --> 00:20:01,840
Speaker 1: So we think about this as problems solved by the system.

408
00:20:02,320 --> 00:20:04,520
Speaker 1: And what's interesting, let's just come back for a second

409
00:20:04,560 --> 00:20:08,119
Speaker 1: to the dog or the goat without fore limbs. We

410
00:20:08,760 --> 00:20:12,880
Speaker 1: generally assume Okay, Look, if you're born with the typical

411
00:20:12,960 --> 00:20:15,680
Speaker 1: structure of the animal, then you just develop in this way.

412
00:20:15,720 --> 00:20:18,600
Speaker 1: But otherwise there's a lot of deep problem solving that

413
00:20:18,640 --> 00:20:21,200
Speaker 1: has to go on. But I know that you think

414
00:20:21,240 --> 00:20:24,679
Speaker 1: about it as, Hey, maybe the system is always problem solving.

415
00:20:24,760 --> 00:20:27,399
Speaker 2: Maybe it's problem solving no matter what if you have

416
00:20:27,520 --> 00:20:30,439
Speaker 2: front legs or not. It's just figuring out what to

417
00:20:30,520 --> 00:20:33,760
Speaker 2: do to get to the goals. Yeah, I think I

418
00:20:33,760 --> 00:20:34,400
Speaker 2: think that's right.

419
00:20:34,480 --> 00:20:36,240
Speaker 3: And and you know, in the last couple of years

420
00:20:36,480 --> 00:20:40,240
Speaker 3: we've really emphasized this and started to develop this idea

421
00:20:40,280 --> 00:20:42,400
Speaker 3: that you know, you can think about it as beginner's

422
00:20:42,400 --> 00:20:44,640
Speaker 3: mind basically, the way that the reason that all these

423
00:20:44,640 --> 00:20:48,280
Speaker 3: incredible plastic the plasticities exist. You know, when when we

424
00:20:48,359 --> 00:20:51,520
Speaker 3: make a Doug Blackiston years ago in our lab, many

425
00:20:51,560 --> 00:20:54,359
Speaker 3: tadpoles with eyes on their tails, and these these guys

426
00:20:54,359 --> 00:20:55,480
Speaker 3: could see they were.

427
00:20:55,359 --> 00:20:56,240
Speaker 2: Not connected to the brain.

428
00:20:56,280 --> 00:20:58,159
Speaker 3: They make an optic nerve that connects sometimes to the

429
00:20:58,160 --> 00:21:00,880
Speaker 3: spinal corse, sometimes to the gut, sometimes no where. They

430
00:21:00,920 --> 00:21:03,440
Speaker 3: can see it, and they can learn visual tasks. Why

431
00:21:03,440 --> 00:21:05,159
Speaker 3: does that work out of the box. Why don't you

432
00:21:05,240 --> 00:21:08,159
Speaker 3: need you know, new rounds of selection mutation? You know,

433
00:21:08,200 --> 00:21:09,160
Speaker 3: basically adaptation.

434
00:21:09,640 --> 00:21:10,440
Speaker 2: All of these things.

435
00:21:10,520 --> 00:21:14,960
Speaker 3: Plasticities work, I think because it never expected everything to

436
00:21:14,960 --> 00:21:16,600
Speaker 3: be in the right place to begin with. It has

437
00:21:16,640 --> 00:21:19,479
Speaker 3: to solve the problem from scratch every single time. And

438
00:21:19,520 --> 00:21:22,360
Speaker 3: that goes back to the idea that biology is fundamentally

439
00:21:22,400 --> 00:21:25,080
Speaker 3: dealing with an unreliable medium. Think about the way we

440
00:21:25,119 --> 00:21:28,639
Speaker 3: build computers today. So we have error correcting codes, we

441
00:21:28,640 --> 00:21:29,800
Speaker 3: have abstraction layers.

442
00:21:29,880 --> 00:21:30,040
Speaker 2: Right.

443
00:21:30,320 --> 00:21:32,840
Speaker 3: The reason that we do, you know, our microchips can't

444
00:21:32,880 --> 00:21:35,159
Speaker 3: can't scale down easily, is because you don't want the

445
00:21:35,680 --> 00:21:37,439
Speaker 3: data interfering with each other. Right when you get to

446
00:21:37,480 --> 00:21:41,080
Speaker 3: that atomic limit that you know, the memory, the bits

447
00:21:41,080 --> 00:21:42,800
Speaker 3: that are in there are stars starting to you know,

448
00:21:43,080 --> 00:21:44,600
Speaker 3: interact with each other, and you don't want that.

449
00:21:44,960 --> 00:21:46,800
Speaker 2: All of all of our current.

450
00:21:46,520 --> 00:21:49,560
Speaker 3: Computer technology is built around the fidelity of the data.

451
00:21:49,920 --> 00:21:52,239
Speaker 3: And that's because the interpreter of that data is us,

452
00:21:52,280 --> 00:21:54,359
Speaker 3: the user. We don't you know, the computer has no issues,

453
00:21:54,720 --> 00:21:56,960
Speaker 3: It doesn't need to interpret the data. We interpret the data,

454
00:21:57,040 --> 00:21:59,000
Speaker 3: so all the computer has to do is keep the

455
00:21:59,080 --> 00:22:02,320
Speaker 3: data still. Biology is exactly the opposite. First of all,

456
00:22:02,359 --> 00:22:04,440
Speaker 3: you have no hope of keeping anything still in biology.

457
00:22:04,480 --> 00:22:06,720
Speaker 3: You have no idea, never mind your environment but you're

458
00:22:06,760 --> 00:22:08,720
Speaker 3: going to mutate as a lineage, you're going to mutate.

459
00:22:08,760 --> 00:22:09,840
Speaker 2: You can't count on your parts.

460
00:22:09,840 --> 00:22:12,320
Speaker 3: You can't count on, you know, knowing how many copies

461
00:22:12,359 --> 00:22:14,760
Speaker 3: of any protein you're going to have. Things degrade, the environment,

462
00:22:15,000 --> 00:22:18,760
Speaker 3: you know, internal millire plus or minus, you know whatever, homeostasis.

463
00:22:18,760 --> 00:22:22,240
Speaker 3: But things are always changing. So I think what biology

464
00:22:22,480 --> 00:22:25,480
Speaker 3: really cranks on. And we've done computational simulations showing how

465
00:22:25,480 --> 00:22:27,639
Speaker 3: this happens. That the minute you have this kind of

466
00:22:28,240 --> 00:22:31,000
Speaker 3: problem solving material, I call it an agential material because

467
00:22:31,000 --> 00:22:34,000
Speaker 3: it's not just the computational material, it's actually an agential material.

468
00:22:34,040 --> 00:22:36,760
Speaker 3: And the minute you have that material, evolution it starts

469
00:22:36,760 --> 00:22:39,359
Speaker 3: to hide information from selection because you're not looking at

470
00:22:39,400 --> 00:22:42,400
Speaker 3: the genome. You're looking at what's going on after you've

471
00:22:42,600 --> 00:22:44,280
Speaker 3: you've solved the problem using whatever.

472
00:22:44,080 --> 00:22:45,320
Speaker 2: Tools the genome has given you.

473
00:22:45,640 --> 00:22:47,920
Speaker 3: And that means that evolution starts to spend a lot

474
00:22:47,960 --> 00:22:50,800
Speaker 3: of its time cranking on that problem solving capacity.

475
00:22:50,960 --> 00:22:54,720
Speaker 2: It spends you know, less of its time on the.

476
00:22:54,640 --> 00:22:56,800
Speaker 3: On the hardwired mechanisms, and more of its time on

477
00:22:56,840 --> 00:22:59,280
Speaker 3: that creative, confabulatory problem solving.

478
00:22:59,560 --> 00:23:01,120
Speaker 2: So I see all of these.

479
00:23:00,920 --> 00:23:04,720
Speaker 3: Things, you know, behavioral memories, genetic memories, meaning you know

480
00:23:04,720 --> 00:23:07,040
Speaker 3: your genome, of your lineage. All of these things are

481
00:23:07,080 --> 00:23:10,720
Speaker 3: basically messages. They're messages from your past self. They're prompts,

482
00:23:11,240 --> 00:23:12,879
Speaker 3: but at any given moment it's up to you how

483
00:23:12,920 --> 00:23:16,440
Speaker 3: you're going to interpret them. And the biological material has

484
00:23:16,480 --> 00:23:20,400
Speaker 3: eons of pressure to learn to tell good stories with

485
00:23:20,440 --> 00:23:22,480
Speaker 3: whatever it's given, whatever information it's given.

486
00:23:22,560 --> 00:23:24,600
Speaker 2: And that's morphogenesism, behavior and so on.

487
00:23:26,520 --> 00:23:29,840
Speaker 1: Yeah, you know, as an analogy, this is exactly the

488
00:23:29,960 --> 00:23:32,199
Speaker 1: argument that I made in Live Wired, is that the

489
00:23:32,359 --> 00:23:37,560
Speaker 1: genes do not specify the blueprints for making the brain. Instead,

490
00:23:37,600 --> 00:23:41,280
Speaker 1: it's just specifying how to build this problem solving organism.

491
00:23:41,280 --> 00:23:43,719
Speaker 1: And as you know, you know, children can get a

492
00:23:43,800 --> 00:23:46,800
Speaker 1: hemisphere ectomy, which means half of their brain is removed.

493
00:23:47,000 --> 00:23:47,960
Speaker 2: For example, if they.

494
00:23:47,880 --> 00:23:51,000
Speaker 1: Have an epilepsy that affects an entire half of the brain,

495
00:23:51,200 --> 00:23:53,960
Speaker 1: So the surgeon removes half the brain and the kids

496
00:23:53,960 --> 00:23:56,520
Speaker 1: grow up to be just fine because the other half

497
00:23:56,840 --> 00:23:59,679
Speaker 1: that remains takes over the missing functions.

498
00:24:00,040 --> 00:24:01,959
Speaker 3: So it's actually I wanted to ask you about that.

499
00:24:02,240 --> 00:24:04,119
Speaker 3: I want to see what your take on it is.

500
00:24:04,119 --> 00:24:07,240
Speaker 3: So we reviewed recently. I've got this Karina Coffin that

501
00:24:07,280 --> 00:24:11,320
Speaker 3: I reviewed these cases where people have massive amounts of

502
00:24:11,320 --> 00:24:14,879
Speaker 3: brain missing, like the part that's left is incredibly small.

503
00:24:15,000 --> 00:24:16,879
Speaker 3: So most of them, of course have very reduced function.

504
00:24:16,960 --> 00:24:19,920
Speaker 3: But the interesting cases, and there are some amazing cases

505
00:24:20,040 --> 00:24:22,600
Speaker 3: where it's a massive reduction on both sides, right, so

506
00:24:22,600 --> 00:24:24,480
Speaker 3: it's not a hemisphere, I mean, and yet they have

507
00:24:24,600 --> 00:24:27,600
Speaker 3: normal or in some cases above normal intelligence. What do

508
00:24:27,640 --> 00:24:29,439
Speaker 3: you think is going on in these in these you know,

509
00:24:29,480 --> 00:24:32,040
Speaker 3: fairly unique but still have to be explained cases, What's

510
00:24:32,040 --> 00:24:32,600
Speaker 3: going on there?

511
00:24:32,720 --> 00:24:33,000
Speaker 2: Yeah?

512
00:24:33,080 --> 00:24:35,399
Speaker 1: So this is the uh, this is the magic of

513
00:24:35,520 --> 00:24:37,640
Speaker 1: live weiring. I think one of the cases that used

514
00:24:37,640 --> 00:24:41,439
Speaker 1: in that paper was people with hydrocephalis, which means you

515
00:24:41,440 --> 00:24:44,440
Speaker 1: get this build up of this pressure in the ventricles,

516
00:24:44,480 --> 00:24:46,600
Speaker 1: these fluid filled spaces in the brain, and the whole

517
00:24:46,720 --> 00:24:49,320
Speaker 1: ring gets squished up against the sides of the skull.

518
00:24:49,400 --> 00:24:52,800
Speaker 1: So when you look at it on MRI, it looks

519
00:24:52,880 --> 00:24:56,000
Speaker 1: like it's essentially empty space and the little bit of

520
00:24:56,000 --> 00:24:59,159
Speaker 1: brain is squished up against it. What that demonstrates is

521
00:24:59,160 --> 00:25:02,240
Speaker 1: exactly what you I both love, which is how flexible

522
00:25:02,280 --> 00:25:05,760
Speaker 1: this material is. Because you know, you can't run over

523
00:25:06,200 --> 00:25:10,479
Speaker 1: half your laptop and expected to still function, but you

524
00:25:10,520 --> 00:25:13,480
Speaker 1: can squish this stuff anyway you want, and it just

525
00:25:13,560 --> 00:25:16,920
Speaker 1: figures out how to accomplish the goals, in this case,

526
00:25:17,000 --> 00:25:20,560
Speaker 1: the cognitive and movement goals that it needs to do.

527
00:25:20,720 --> 00:25:23,399
Speaker 1: There's one of the cases in the medical literature this

528
00:25:23,840 --> 00:25:25,600
Speaker 1: guy who was forty years old and he went to

529
00:25:25,640 --> 00:25:27,119
Speaker 1: the doctor because he was having a little bit of

530
00:25:27,240 --> 00:25:30,600
Speaker 1: leg pain. And the doctor couldn't figure out why the

531
00:25:30,600 --> 00:25:32,600
Speaker 1: guy's leg was hurting, so he said, hey, why don't

532
00:25:32,640 --> 00:25:35,080
Speaker 1: we just take a brain scan, And that's when they

533
00:25:35,119 --> 00:25:37,800
Speaker 1: discovered that most of the brain scan just looks like

534
00:25:38,320 --> 00:25:40,240
Speaker 1: empty or fluid filled space.

535
00:25:40,640 --> 00:25:42,800
Speaker 2: But you know, he was married, he had a job,

536
00:25:43,000 --> 00:25:43,640
Speaker 2: normal IQ.

537
00:25:44,720 --> 00:25:50,000
Speaker 1: It's quite remarkable how different this livewear is from the

538
00:25:50,040 --> 00:25:52,520
Speaker 1: way that we think about building things in Silicon Valley.

539
00:25:52,800 --> 00:25:54,960
Speaker 1: And of course you're what you do which is so

540
00:25:55,000 --> 00:25:59,320
Speaker 1: remarkable is is look at all the cells, the whole system,

541
00:25:59,760 --> 00:26:03,280
Speaker 1: and this massive flexibility and the collective intelligence of all

542
00:26:03,320 --> 00:26:05,000
Speaker 1: the pieces and parts all the way up. Tell me

543
00:26:05,080 --> 00:26:08,080
Speaker 1: you think this is a good analogy about collective intelligence.

544
00:26:08,160 --> 00:26:09,879
Speaker 1: I was thinking about I was just trying to think

545
00:26:09,920 --> 00:26:13,560
Speaker 1: of an analogy, and I was thinking about with Wikipedia,

546
00:26:13,760 --> 00:26:17,240
Speaker 1: everybody's doing their little thing depending on their own expertise,

547
00:26:17,359 --> 00:26:20,960
Speaker 1: they put in some things, and nobody who's doing this

548
00:26:21,200 --> 00:26:24,640
Speaker 1: knows the giant shape of the full Wikipedia.

549
00:26:24,680 --> 00:26:27,320
Speaker 2: It's much too big for any given human.

550
00:26:28,800 --> 00:26:31,640
Speaker 1: But nonetheless everyone's doing their things, and what you get

551
00:26:31,720 --> 00:26:35,280
Speaker 1: is this collective intelligence out of it. And I was

552
00:26:35,280 --> 00:26:38,560
Speaker 1: thinking about whether I could stretch its analogy. You know,

553
00:26:38,680 --> 00:26:42,000
Speaker 1: if some part of Wikipedia got cut off, like an

554
00:26:42,040 --> 00:26:45,480
Speaker 1: Axi Lotel's limb, it would grow back and it would

555
00:26:45,520 --> 00:26:49,280
Speaker 1: take the right shape again, because that knowledge is somehow

556
00:26:49,400 --> 00:26:52,119
Speaker 1: stored in all the individuals who are writing the stuff.

557
00:26:52,440 --> 00:26:56,080
Speaker 1: But again, nobody knows what they're doing. Everyone's just contributing

558
00:26:56,080 --> 00:26:58,000
Speaker 1: where they see a gap. Does that seem like an

559
00:26:58,040 --> 00:26:58,840
Speaker 1: interesting analogies?

560
00:26:59,119 --> 00:27:01,320
Speaker 3: It is and what it suggests to me, And I'll

561
00:27:01,359 --> 00:27:03,879
Speaker 3: actually I'll actually talk to Eric Hole about this and

562
00:27:03,920 --> 00:27:06,000
Speaker 3: see I see if this is a good analysis to do.

563
00:27:06,240 --> 00:27:11,119
Speaker 3: Because there are now computational tools from from information theories

564
00:27:11,200 --> 00:27:12,680
Speaker 3: and then Eric Hole was one of the one of

565
00:27:12,720 --> 00:27:15,439
Speaker 3: the key developers of some of this where you can

566
00:27:15,520 --> 00:27:18,240
Speaker 3: actually in a specific given circumstance, you can actually ask

567
00:27:18,280 --> 00:27:21,840
Speaker 3: whether the higher level has caught more causal power than

568
00:27:21,880 --> 00:27:24,240
Speaker 3: the lower level. And so so it's actually an amazing

569
00:27:24,240 --> 00:27:26,560
Speaker 3: advance because it means that questions that before used to

570
00:27:26,560 --> 00:27:28,440
Speaker 3: be philosophy, and people argued about this for you know,

571
00:27:28,600 --> 00:27:32,040
Speaker 3: probly thousands of years, whether the reductionism or you know,

572
00:27:32,200 --> 00:27:34,199
Speaker 3: was was all you need or whether sometimes you have

573
00:27:34,200 --> 00:27:36,399
Speaker 3: these higher level things that are causally powerful. Now, now

574
00:27:36,480 --> 00:27:38,639
Speaker 3: now there's actual maths to answer that question. It's it's

575
00:27:38,680 --> 00:27:40,280
Speaker 3: it's quite amazing. And so so you know, there are

576
00:27:40,320 --> 00:27:43,359
Speaker 3: Python toolkits now to estimate in your given system, is

577
00:27:43,400 --> 00:27:46,440
Speaker 3: everything explainable by the lower levels or is there a

578
00:27:47,080 --> 00:27:49,800
Speaker 3: higher level that does something that the lower levels don't do.

579
00:27:50,200 --> 00:27:54,720
Speaker 3: And actually, people like Julio Tononi and Laris Albontakis in

580
00:27:54,760 --> 00:27:57,320
Speaker 3: his group, they apply this to all kinds of human patients. So,

581
00:27:57,680 --> 00:28:01,360
Speaker 3: so coma you know, locked in a sleep, anesthetize the wake, right,

582
00:28:01,600 --> 00:28:03,199
Speaker 3: are you dealing with a pile of neurons or is

583
00:28:03,200 --> 00:28:03,920
Speaker 3: there a human.

584
00:28:03,720 --> 00:28:05,520
Speaker 2: Being in there, you know, some kind of collector?

585
00:28:05,600 --> 00:28:05,800
Speaker 1: Right.

586
00:28:05,920 --> 00:28:08,520
Speaker 3: So, so now what you're making me think, is this whole,

587
00:28:08,720 --> 00:28:12,159
Speaker 3: this whole process of Wikipedia, we could we could in

588
00:28:12,240 --> 00:28:15,240
Speaker 3: theory apply those tools and and and really empirically asked

589
00:28:15,240 --> 00:28:17,920
Speaker 3: the question is there a collective there that's bigger than

590
00:28:18,720 --> 00:28:21,360
Speaker 3: just the individual processes? That go on when people get

591
00:28:21,400 --> 00:28:24,320
Speaker 3: on and and edit those you know, at those those

592
00:28:24,320 --> 00:28:25,160
Speaker 3: Wikipedia entries.

593
00:28:25,240 --> 00:28:26,840
Speaker 2: Let's do this experiment. I love it.

594
00:28:26,920 --> 00:28:28,440
Speaker 1: I want to make sure that we have enough time

595
00:28:28,440 --> 00:28:34,280
Speaker 1: to talk about diverse intelligences. So one of your interests

596
00:28:34,400 --> 00:28:38,320
Speaker 1: is in not just looking at human brains and thinking about, okay,

597
00:28:38,320 --> 00:28:41,000
Speaker 1: how is this intelligence? So on, but saying, what are

598
00:28:41,080 --> 00:28:44,160
Speaker 1: other systems that are intelligence? So let's let's dive into that.

599
00:28:44,280 --> 00:28:47,000
Speaker 1: Tell us about how you think about diverse cognition.

600
00:28:47,440 --> 00:28:50,520
Speaker 3: Yeah, so so because of some of the things that

601
00:28:50,520 --> 00:28:53,520
Speaker 3: we've already discussed, meaning that problem solving is something that

602
00:28:53,560 --> 00:28:55,880
Speaker 3: biology has to grapple with at the very beginning, you know,

603
00:28:55,920 --> 00:28:58,200
Speaker 3: at the very origin of life, and in fact, probably

604
00:28:58,360 --> 00:29:01,160
Speaker 3: long before that. You know, you can't afford to be

605
00:29:01,360 --> 00:29:03,680
Speaker 3: like a Laplacian demon that's going to track micro states.

606
00:29:03,720 --> 00:29:05,840
Speaker 3: You have to coarse grain your environment. You have to

607
00:29:05,920 --> 00:29:10,080
Speaker 3: start telling yourself kind of agential stories about what's going on.

608
00:29:10,120 --> 00:29:11,600
Speaker 3: In other words, you have to make models of the

609
00:29:11,680 --> 00:29:14,360
Speaker 3: environment where you course grain a whole bunch of different

610
00:29:14,360 --> 00:29:15,959
Speaker 3: things that are happening, and you say, okay, I'm going

611
00:29:16,000 --> 00:29:18,200
Speaker 3: to treat all those as one thing, and this is

612
00:29:18,280 --> 00:29:20,680
Speaker 3: danger or this is food, or this is a conspecific

613
00:29:20,760 --> 00:29:22,520
Speaker 3: or this is you know, low pH you know for

614
00:29:22,560 --> 00:29:25,520
Speaker 3: an ambarrow or whatever it is. So that kind of thing,

615
00:29:25,560 --> 00:29:28,000
Speaker 3: having to having to tell these kind of agential stories

616
00:29:28,080 --> 00:29:30,320
Speaker 3: that you can then eventually turn on yourself and say wait.

617
00:29:30,400 --> 00:29:32,560
Speaker 3: And I also am an agent that does things and

618
00:29:33,000 --> 00:29:37,360
Speaker 3: the need to improvise, continuously improvise meaning for the information

619
00:29:37,440 --> 00:29:39,240
Speaker 3: that you get, because you're not told nobody's going to

620
00:29:39,240 --> 00:29:41,600
Speaker 3: interpret anything for you. You have to interpret your own genome,

621
00:29:41,720 --> 00:29:44,400
Speaker 3: your own physiological states, your own memories. And so I'm

622
00:29:44,400 --> 00:29:46,840
Speaker 3: really interested in the different ways that this gets amplified

623
00:29:46,840 --> 00:29:49,479
Speaker 3: in evolution. And of course, you know brains, you know,

624
00:29:49,560 --> 00:29:51,720
Speaker 3: the familiar brains are one way that that happens, but

625
00:29:51,720 --> 00:29:53,520
Speaker 3: there are many other ways that that happens. And I

626
00:29:53,600 --> 00:29:55,360
Speaker 3: want to want to just briefly give you two quick

627
00:29:55,400 --> 00:29:58,600
Speaker 3: analogies that I think illustrate some of the aspects of

628
00:29:58,640 --> 00:30:00,680
Speaker 3: what the field of diverse and tell legence is about,

629
00:30:00,720 --> 00:30:02,720
Speaker 3: at least the way I see it. First of all,

630
00:30:03,040 --> 00:30:05,360
Speaker 3: think about the electromagnetic spectrum. So back in the day

631
00:30:05,360 --> 00:30:08,440
Speaker 3: when we didn't have a proper theory of electromagnetism, we

632
00:30:08,520 --> 00:30:13,000
Speaker 3: had lightning and static electricity, and light and magnets and.

633
00:30:13,240 --> 00:30:15,120
Speaker 2: Various things like that, and we thought those were all

634
00:30:15,120 --> 00:30:15,640
Speaker 2: different things.

635
00:30:15,720 --> 00:30:19,200
Speaker 3: We thought they were all categories, like sharp crisp categories.

636
00:30:19,240 --> 00:30:21,840
Speaker 3: Nobody thought that light and magnets were the same, and

637
00:30:21,920 --> 00:30:24,800
Speaker 3: those are you know, we have categories for all those things.

638
00:30:25,160 --> 00:30:27,240
Speaker 2: And also, so that's the first thing. We thought these

639
00:30:27,240 --> 00:30:29,840
Speaker 2: were all distinct and because of our own.

640
00:30:29,680 --> 00:30:32,720
Speaker 3: Evolutionary history, we were only sensitive to a tiny part

641
00:30:32,720 --> 00:30:35,440
Speaker 3: of that spectrum. There were huge examples of this phenomena

642
00:30:35,480 --> 00:30:38,960
Speaker 3: that we were completely blind to. And then we eventually

643
00:30:39,000 --> 00:30:40,920
Speaker 3: we ended up with a good theory of electromagnetism.

644
00:30:40,920 --> 00:30:41,560
Speaker 2: We did two things.

645
00:30:41,560 --> 00:30:43,800
Speaker 3: First of all, unified it says, no, these are all

646
00:30:43,840 --> 00:30:47,480
Speaker 3: actually in a very meaningful way. They are all examples

647
00:30:47,480 --> 00:30:50,320
Speaker 3: of the same underlying phenomena. Okay, so a deep unification,

648
00:30:50,440 --> 00:30:51,000
Speaker 3: so that's great.

649
00:30:51,360 --> 00:30:54,480
Speaker 2: And two, they allowed us to make technology useful, technology

650
00:30:54,560 --> 00:30:55,240
Speaker 2: that allows.

651
00:30:55,040 --> 00:30:57,040
Speaker 3: Us to operate across the spectrum, to be able to

652
00:30:57,080 --> 00:31:00,360
Speaker 3: detect and modulate things that before were completely in visible

653
00:31:00,360 --> 00:31:02,120
Speaker 3: to us, and meaning we didn't think they existed, but

654
00:31:02,240 --> 00:31:04,600
Speaker 3: now we know better. So something like this is what

655
00:31:04,640 --> 00:31:07,280
Speaker 3: I think is going to happen for cognition. I think

656
00:31:07,320 --> 00:31:10,520
Speaker 3: we are sensitive to an extremely narrow spectrum among the

657
00:31:10,600 --> 00:31:14,520
Speaker 3: gigantic space of possible minds. I think they are all

658
00:31:14,520 --> 00:31:17,680
Speaker 3: around us, but we are totally mind blind to most

659
00:31:17,680 --> 00:31:19,760
Speaker 3: of them, you know. I think that's a good term

660
00:31:19,760 --> 00:31:22,160
Speaker 3: mind blindness, is that we just don't recognize these things

661
00:31:22,280 --> 00:31:25,160
Speaker 3: because we don't have a good theory that explains why

662
00:31:25,600 --> 00:31:28,800
Speaker 3: the problem solving of an amoeba, of a thermostat, of

663
00:31:29,520 --> 00:31:31,719
Speaker 3: you know, of an organ of a human, of a

664
00:31:31,800 --> 00:31:34,560
Speaker 3: collection of humans doing Wikipedia whatever, why these are all

665
00:31:34,560 --> 00:31:37,240
Speaker 3: actually on the same spectrum. We don't have a good

666
00:31:37,240 --> 00:31:40,280
Speaker 3: theory yet. And and the second thing is we don't

667
00:31:40,280 --> 00:31:42,480
Speaker 3: have the technology. And this is something else that I

668
00:31:42,480 --> 00:31:44,000
Speaker 3: think we have a lot to talk about in terms

669
00:31:44,040 --> 00:31:47,960
Speaker 3: of prosthetics, Okay, cognitive and physical, bodily prosthetics that would

670
00:31:47,960 --> 00:31:49,800
Speaker 3: allow us to interact with these other beings that are

671
00:31:49,840 --> 00:31:50,440
Speaker 3: all around us.

672
00:31:52,160 --> 00:31:55,360
Speaker 1: So let's dive into some examples of diverse intelligence. Sure,

673
00:31:55,400 --> 00:31:57,200
Speaker 1: so let's just start from from the beginning and work

674
00:31:57,200 --> 00:31:59,960
Speaker 1: our way up. So, so brains, okay, we all know brain.

675
00:32:00,000 --> 00:32:02,080
Speaker 1: Any kinds of animals exist.

676
00:32:02,160 --> 00:32:06,280
Speaker 3: Then, because of what we understand about navigating other biological spaces,

677
00:32:06,600 --> 00:32:09,720
Speaker 3: we can think about plants, and we can think about cells,

678
00:32:09,880 --> 00:32:13,000
Speaker 3: and we can think about tissues and organs, which also

679
00:32:13,480 --> 00:32:14,280
Speaker 3: solve problems.

680
00:32:14,320 --> 00:32:17,280
Speaker 2: They store memories, they can learn, they can be communicated with.

681
00:32:17,400 --> 00:32:19,720
Speaker 3: This is what all of the biomedical efforts in my

682
00:32:19,800 --> 00:32:23,560
Speaker 3: lab are pointed at, which is learning through in particular

683
00:32:23,640 --> 00:32:28,080
Speaker 3: bioelectrical interface. They're all oriented towards communicating our goals to

684
00:32:28,120 --> 00:32:31,400
Speaker 3: cells and tissues. So for full on regenerative medicine, it

685
00:32:31,480 --> 00:32:33,680
Speaker 3: is not going to be sufficient to try to micromanage

686
00:32:33,920 --> 00:32:36,960
Speaker 3: the receptors or genetic states. We are going to have

687
00:32:37,000 --> 00:32:39,400
Speaker 3: to get the buy in of the cells, respecify their

688
00:32:39,400 --> 00:32:41,680
Speaker 3: goals at a high level, and get them to do

689
00:32:41,720 --> 00:32:44,200
Speaker 3: these complicated things that we can't possibly micromanage.

690
00:32:44,280 --> 00:32:46,160
Speaker 2: So give us some specific examples.

691
00:32:46,320 --> 00:32:49,360
Speaker 3: So one of the things that we have learned to

692
00:32:49,360 --> 00:32:52,960
Speaker 3: do is much like neuroscientists read electrical patterns in the

693
00:32:52,960 --> 00:32:54,360
Speaker 3: brain and they try to decode them.

694
00:32:54,440 --> 00:32:55,640
Speaker 2: So this is neural.

695
00:32:55,360 --> 00:32:59,160
Speaker 3: Decoding, where people want to read the electrophysiology of your

696
00:32:59,160 --> 00:33:01,200
Speaker 3: brain and say here's your memories or goals or preferences

697
00:33:01,360 --> 00:33:03,320
Speaker 3: and be able to read that out. We've learned to

698
00:33:03,320 --> 00:33:05,400
Speaker 3: do that, and we developed the first tools to do

699
00:33:05,440 --> 00:33:07,480
Speaker 3: it in the early two thousands for the rest of

700
00:33:07,480 --> 00:33:10,160
Speaker 3: the body. So when I say that the early embryo

701
00:33:10,360 --> 00:33:14,800
Speaker 3: navigates anatomical MorphOS space to the shape of whatever it's

702
00:33:14,840 --> 00:33:16,960
Speaker 3: going to be, and that it is an active agent

703
00:33:17,000 --> 00:33:18,760
Speaker 3: that has a memory of where it's going, it has

704
00:33:18,760 --> 00:33:21,840
Speaker 3: a representation of where it's going, that's a very big claim.

705
00:33:21,880 --> 00:33:23,880
Speaker 3: You then have to say, well, what's the mechanism for

706
00:33:23,920 --> 00:33:25,920
Speaker 3: storing the representation where is it?

707
00:33:25,960 --> 00:33:28,080
Speaker 2: Can you decode it? And can you rewrite it? And

708
00:33:28,120 --> 00:33:30,080
Speaker 2: so this is what we've done. We've developed tools to

709
00:33:30,480 --> 00:33:33,800
Speaker 2: read the electrical memories of collections of cells. This goes

710
00:33:33,840 --> 00:33:34,760
Speaker 2: right back to what you said.

711
00:33:34,800 --> 00:33:36,800
Speaker 3: No individual cell knows what a face is, or what

712
00:33:36,840 --> 00:33:38,720
Speaker 3: an eye is, or how many fingers you're supposed to have,

713
00:33:38,840 --> 00:33:41,280
Speaker 3: but the collective absolutely knows. And we can read this

714
00:33:41,320 --> 00:33:44,360
Speaker 3: out now. In a few cases, we can literally see

715
00:33:44,160 --> 00:33:48,280
Speaker 3: the in images and videos, the memory, the electrical pattern

716
00:33:48,320 --> 00:33:50,560
Speaker 3: that is of the future shape that is guiding the

717
00:33:50,600 --> 00:33:53,520
Speaker 3: sell activity. Moreover, it serves as a kind of cognitive

718
00:33:53,520 --> 00:33:56,440
Speaker 3: glue that binds all the cells towards one story, one

719
00:33:56,520 --> 00:33:59,040
Speaker 3: story of what a correct embryo is supposed to look like.

720
00:33:59,160 --> 00:34:00,880
Speaker 3: This is why you say it's an embryo and not

721
00:34:00,960 --> 00:34:02,720
Speaker 3: a pile of cells because they've all committed to the

722
00:34:02,760 --> 00:34:06,040
Speaker 3: same journey in that space. This actually, this idea is

723
00:34:06,160 --> 00:34:08,800
Speaker 3: at least as old as Harold Burr in the thirties.

724
00:34:08,840 --> 00:34:11,480
Speaker 3: He without anything other than a good vaultmeter, he was

725
00:34:11,520 --> 00:34:13,319
Speaker 3: able to kind of already figure this out.

726
00:34:13,440 --> 00:34:13,840
Speaker 2: Amazing.

727
00:34:14,000 --> 00:34:16,960
Speaker 3: And so now we can read those memories, we can

728
00:34:17,040 --> 00:34:19,920
Speaker 3: decode those memories, and we can rewrite those memories.

729
00:34:20,080 --> 00:34:20,839
Speaker 2: Because if I take.

730
00:34:20,760 --> 00:34:23,080
Speaker 3: A plenarian flatworm and I say, oh, look, this is

731
00:34:23,080 --> 00:34:25,239
Speaker 3: where it says that you should have two heads if

732
00:34:25,239 --> 00:34:28,400
Speaker 3: you're injured, we can rewrite that. And this is Falon

733
00:34:28,440 --> 00:34:30,560
Speaker 3: Durant's work when she was a PhD student in my group.

734
00:34:30,800 --> 00:34:36,200
Speaker 3: We can rewrite that electrical pattern, no genetic modification, just

735
00:34:36,719 --> 00:34:39,360
Speaker 3: brief application only takes about three hours, a brief application

736
00:34:39,400 --> 00:34:42,200
Speaker 3: of ion channel drugs that we've chosen specifically in tune

737
00:34:42,239 --> 00:34:44,480
Speaker 3: with a computational model of how you would.

738
00:34:44,239 --> 00:34:46,640
Speaker 2: Do that, and we change that pattern. Instead of saying

739
00:34:46,640 --> 00:34:47,799
Speaker 2: one head and now it says two.

740
00:34:48,480 --> 00:34:51,440
Speaker 3: Now that becomes a false memory because the worm currently

741
00:34:51,480 --> 00:34:53,160
Speaker 3: doesn't have to It has one and it'll sit there

742
00:34:53,160 --> 00:34:55,560
Speaker 3: perfectly happy. The anatomy does not match the memory. It's

743
00:34:55,560 --> 00:34:58,319
Speaker 3: a latent memory until you injure the thing, and when

744
00:34:58,320 --> 00:35:00,560
Speaker 3: you cut it, bang, that's when the cells consult the

745
00:35:00,560 --> 00:35:02,719
Speaker 3: memory and memories says, build two heads. Well, that's their

746
00:35:02,719 --> 00:35:04,759
Speaker 3: ground truth. I don't know any different, and so they

747
00:35:04,760 --> 00:35:06,720
Speaker 3: will go ahead and they will build this new vision

748
00:35:06,760 --> 00:35:07,520
Speaker 3: of what a worm is.

749
00:35:07,800 --> 00:35:09,880
Speaker 2: And it's a memory because it is permanent.

750
00:35:09,920 --> 00:35:11,880
Speaker 3: If you take two headed animals and keep cutting them,

751
00:35:11,920 --> 00:35:15,680
Speaker 3: they will continue regenerating as two headed, even though their genome.

752
00:35:15,600 --> 00:35:16,880
Speaker 2: Is a perfectly standard genome.

753
00:35:16,880 --> 00:35:18,279
Speaker 3: If you were to sequence that, you would have been

754
00:35:18,280 --> 00:35:20,399
Speaker 3: none the wiser that this thing has two heads. So

755
00:35:20,440 --> 00:35:23,320
Speaker 3: this kind of thing, the ability to put new goals

756
00:35:23,360 --> 00:35:26,319
Speaker 3: into the mind of the collective is the kind of

757
00:35:26,320 --> 00:35:29,279
Speaker 3: an earliest example of communicating with it because we can,

758
00:35:29,600 --> 00:35:31,880
Speaker 3: we can in some cases, we can train it. Another

759
00:35:31,880 --> 00:35:34,480
Speaker 3: thing we're really working on is to actually ask it questions.

760
00:35:34,560 --> 00:35:36,439
Speaker 3: That would be really cool because sells have all kinds

761
00:35:36,480 --> 00:35:37,680
Speaker 3: of problem solving capacities.

762
00:35:37,680 --> 00:35:39,120
Speaker 2: I would love to be able to actually ask them

763
00:35:39,160 --> 00:35:41,080
Speaker 2: questions in a way. And AI is a.

764
00:35:41,120 --> 00:35:43,000
Speaker 3: Very powerful tool that we're now using to start to

765
00:35:43,000 --> 00:35:46,520
Speaker 3: communicate with these things. So that's kind of the first

766
00:35:46,560 --> 00:35:49,279
Speaker 3: weird kind of mind, meaning in our body we have

767
00:35:49,880 --> 00:35:51,359
Speaker 3: I can't you know, I don't think you can count them.

768
00:35:51,360 --> 00:35:53,640
Speaker 3: I think that you know, it's not probably not really infinite,

769
00:35:53,680 --> 00:35:57,520
Speaker 3: but but a very large number of different cognitive units

770
00:35:57,560 --> 00:35:59,759
Speaker 3: inside your body, solving their own problems in their own

771
00:35:59,760 --> 00:36:00,640
Speaker 3: time scales and so on.

772
00:36:00,800 --> 00:36:02,640
Speaker 2: But you can get weirder than that. Which is which

773
00:36:02,680 --> 00:36:04,319
Speaker 2: is this? You know?

774
00:36:05,080 --> 00:36:07,040
Speaker 3: I'll start with a very quick story, and this goes

775
00:36:07,080 --> 00:36:09,319
Speaker 3: back to us, to US sci fi story from from

776
00:36:09,680 --> 00:36:12,319
Speaker 3: that I read years ago. Imagine, these creatures come from

777
00:36:12,320 --> 00:36:13,719
Speaker 3: the core of the earth. They live, they live in

778
00:36:13,719 --> 00:36:15,719
Speaker 3: the center of the earth. They're super dense. They come

779
00:36:15,800 --> 00:36:18,200
Speaker 3: up to the surface. What do they see, Well, they

780
00:36:18,200 --> 00:36:20,520
Speaker 3: don't see physical objects as far as they're concerned.

781
00:36:20,600 --> 00:36:22,880
Speaker 2: Everything here is like a thin gas. It's like a plasma.

782
00:36:22,880 --> 00:36:23,520
Speaker 2: They are so dense.

783
00:36:23,520 --> 00:36:25,160
Speaker 3: They walk right through us the way that we walk

784
00:36:25,239 --> 00:36:27,319
Speaker 3: through you know, patterns of pollen in the garden, and

785
00:36:27,360 --> 00:36:29,040
Speaker 3: we don't even we don't even notice it. And so

786
00:36:29,120 --> 00:36:30,719
Speaker 3: one of them is a scientist and he's looking and

787
00:36:30,719 --> 00:36:32,960
Speaker 3: he says, you know, this gas that we're that we're

788
00:36:32,960 --> 00:36:35,400
Speaker 3: walking through. I kind of if you actually look at

789
00:36:35,440 --> 00:36:38,040
Speaker 3: patterns within the gas, it almost looks like they're doing something.

790
00:36:38,080 --> 00:36:40,640
Speaker 3: It almost looks like they're agential. They like these patterns,

791
00:36:40,680 --> 00:36:42,800
Speaker 3: you know, they walk around, they have behaviors, they're doing stuff,

792
00:36:43,040 --> 00:36:44,920
Speaker 3: and and the others say, well, that's crazy.

793
00:36:44,920 --> 00:36:47,320
Speaker 2: We're real physical beings. Patterns can't be agents.

794
00:36:47,360 --> 00:36:49,759
Speaker 3: Patterns, you know, patterns and an excitable medium can't have,

795
00:36:50,120 --> 00:36:52,239
Speaker 3: you know, their own their own memories and their own goals.

796
00:36:52,280 --> 00:36:54,160
Speaker 3: And by the way, how long these patterns last? He says, well,

797
00:36:54,200 --> 00:36:56,200
Speaker 3: they dissipate after about one hundred years. He's like, yeah, no,

798
00:36:56,200 --> 00:36:58,920
Speaker 3: it's not anything, right, So okay, So so what that

799
00:36:58,960 --> 00:37:02,000
Speaker 3: reminds us of is that the distinction between you know,

800
00:37:02,200 --> 00:37:04,680
Speaker 3: we too our patterns, right, we're metabolic patterns that hold

801
00:37:04,680 --> 00:37:06,960
Speaker 3: ourselves together for some amount of time and then we dissipate.

802
00:37:07,400 --> 00:37:11,080
Speaker 3: And this distinction between patterns and objects is in the

803
00:37:11,120 --> 00:37:13,479
Speaker 3: eye of the beholder. And so that leads you to ask,

804
00:37:13,600 --> 00:37:16,080
Speaker 3: what are the things that we think of as mere

805
00:37:16,200 --> 00:37:19,120
Speaker 3: patterns and an excitable medium that might be agents themselves.

806
00:37:19,320 --> 00:37:21,239
Speaker 2: And so that's the second that's the next kind of

807
00:37:21,320 --> 00:37:22,800
Speaker 2: level is can we communicate?

808
00:37:22,840 --> 00:37:26,840
Speaker 3: Can we can we recognize and communicate with patterns, patterns

809
00:37:26,840 --> 00:37:29,719
Speaker 3: of gene expression, patterns of bioelectric state. You know, this

810
00:37:30,000 --> 00:37:32,560
Speaker 3: whole thoughts are thinker's idea from William James.

811
00:37:50,120 --> 00:37:52,839
Speaker 1: So you look around, you see these patterns everywhere, and

812
00:37:52,880 --> 00:37:56,160
Speaker 1: you think, which of these are agential, which have what

813
00:37:56,200 --> 00:38:00,640
Speaker 1: we might call intelligence unpack the thoughts or things idea

814
00:38:00,680 --> 00:38:01,080
Speaker 1: for us?

815
00:38:01,400 --> 00:38:03,920
Speaker 3: Yeah, yeah, so so this is uh and and I

816
00:38:03,960 --> 00:38:06,960
Speaker 3: admit I haven't I haven't looked for the actual reference,

817
00:38:06,960 --> 00:38:08,600
Speaker 3: but but I'm pretty sure I saw this in in

818
00:38:08,920 --> 00:38:12,399
Speaker 3: James's book, where what he's pointing out is that, look,

819
00:38:12,440 --> 00:38:15,000
Speaker 3: you have fleeting thoughts. They come and they go, right,

820
00:38:15,040 --> 00:38:17,440
Speaker 3: they sort of run through your your your memory medium,

821
00:38:17,440 --> 00:38:19,000
Speaker 3: and then they and then they go. Then you have

822
00:38:19,120 --> 00:38:21,520
Speaker 3: persistent thoughts, and these are a little harder to get

823
00:38:21,560 --> 00:38:23,440
Speaker 3: rid of. They do a little niche construction, as you know,

824
00:38:23,480 --> 00:38:25,279
Speaker 3: they they kind of change some of your brain to

825
00:38:25,400 --> 00:38:27,200
Speaker 3: enable it to be to make it easier for them

826
00:38:27,239 --> 00:38:30,120
Speaker 3: to to persist. Right, these these intrusive, persistent, you know

827
00:38:30,239 --> 00:38:32,959
Speaker 3: kinds of thoughts. And then you have you go further

828
00:38:33,000 --> 00:38:35,680
Speaker 3: on the spectrum and you have personality fragments like from

829
00:38:35,719 --> 00:38:38,800
Speaker 3: a you know, like from a dissociated identity kind of situation.

830
00:38:39,200 --> 00:38:40,759
Speaker 3: And then you keep going and then you have a

831
00:38:40,760 --> 00:38:43,000
Speaker 3: full coherent human personality and you say Okay, well that's

832
00:38:43,040 --> 00:38:45,560
Speaker 3: the thing we're we're kind of used to. But but

833
00:38:45,719 --> 00:38:47,640
Speaker 3: it's on a spectrum. And then and then who knows, right,

834
00:38:47,680 --> 00:38:50,040
Speaker 3: some people claim there's like a superhuman you know, sort

835
00:38:50,040 --> 00:38:52,960
Speaker 3: of a larger, larger superman mind and so on.

836
00:38:53,000 --> 00:38:55,600
Speaker 2: I don't know. So that's the idea. And so there

837
00:38:55,640 --> 00:38:57,760
Speaker 2: are two there are two ways to to think.

838
00:38:57,640 --> 00:38:59,839
Speaker 3: About any of these situations that were sort of given

839
00:38:59,840 --> 00:39:02,360
Speaker 3: to us by by the by the touring paradigm. You

840
00:39:02,400 --> 00:39:07,600
Speaker 3: can say that the cells those that's that's your touring machine.

841
00:39:07,600 --> 00:39:10,239
Speaker 3: That's your that's your machine. That's the real agent. And

842
00:39:10,360 --> 00:39:13,360
Speaker 3: the patterns that move through it, the information, the energy

843
00:39:13,360 --> 00:39:14,719
Speaker 3: slash information patterns that.

844
00:39:14,640 --> 00:39:16,560
Speaker 2: Move through it. They're they're just they're just patterns.

845
00:39:16,600 --> 00:39:18,879
Speaker 3: They're passive data and and and it's the agent that

846
00:39:19,080 --> 00:39:21,800
Speaker 3: processes the data. Right we you know, our brain moves

847
00:39:21,800 --> 00:39:24,800
Speaker 3: around the information that moves the energy, and our body

848
00:39:24,800 --> 00:39:25,319
Speaker 3: does the same thing.

849
00:39:25,360 --> 00:39:25,640
Speaker 2: Okay.

850
00:39:26,120 --> 00:39:27,920
Speaker 3: Or you can flip the whole thing, which is what

851
00:39:27,960 --> 00:39:30,040
Speaker 3: we're working on now, which is to say, what if

852
00:39:30,560 --> 00:39:32,880
Speaker 3: it's the patterns that are the agents and everything that

853
00:39:32,920 --> 00:39:36,080
Speaker 3: happens to the machine, meaning all the outcomes of gene expression,

854
00:39:36,080 --> 00:39:39,160
Speaker 3: of protein movement, of cell behavior of morphogenesis. What if

855
00:39:39,160 --> 00:39:41,759
Speaker 3: that's those things are just a scratch pad. It's the

856
00:39:41,880 --> 00:39:43,520
Speaker 3: it's kind of a stigma gee the way that any

857
00:39:43,520 --> 00:39:46,279
Speaker 3: ant colony will eventually you know, particles and pheromones and

858
00:39:46,320 --> 00:39:48,560
Speaker 3: things will move around because the ant colony mind is

859
00:39:48,640 --> 00:39:50,120
Speaker 3: kind of doing its thing as the ants, you know,

860
00:39:50,160 --> 00:39:51,239
Speaker 3: send messages to each other.

861
00:39:51,480 --> 00:39:52,080
Speaker 2: What if the the.

862
00:39:52,360 --> 00:39:56,000
Speaker 3: The anatomy and physiology that we see and and and

863
00:39:56,080 --> 00:39:57,880
Speaker 3: the and the body of the touring machine is the

864
00:39:57,880 --> 00:40:01,080
Speaker 3: scratch pad of of the actual age, which are the patterns,

865
00:40:01,440 --> 00:40:04,399
Speaker 3: you know, working out their dynamics in the physical world.

866
00:40:05,000 --> 00:40:09,360
Speaker 3: And it actually has some real implications just very quickly,

867
00:40:09,360 --> 00:40:11,480
Speaker 3: for example, in our program on aging, right, so we're

868
00:40:11,480 --> 00:40:15,280
Speaker 3: trying to understand an address agent. So imagine the classic

869
00:40:15,360 --> 00:40:17,799
Speaker 3: way of thinking about aging from a bioelectric standpoint is

870
00:40:18,160 --> 00:40:20,920
Speaker 3: we know that during embryogenesis there's a bielectric pattern that

871
00:40:20,920 --> 00:40:25,279
Speaker 3: guides morphogenesis. And so probably what happens is that those

872
00:40:25,320 --> 00:40:28,160
Speaker 3: memories become fuzzy in adulthood, and as the age, they

873
00:40:28,200 --> 00:40:29,320
Speaker 3: just get fuzzy and fuzzier.

874
00:40:29,320 --> 00:40:30,719
Speaker 2: Their cells have no idea what to do.

875
00:40:30,960 --> 00:40:34,319
Speaker 3: The memory degrades, and the agent, the physical body doesn't

876
00:40:34,320 --> 00:40:35,080
Speaker 3: know what to do anymore.

877
00:40:35,120 --> 00:40:37,120
Speaker 2: That's the standard approach, and that's what you know. That's

878
00:40:37,160 --> 00:40:38,000
Speaker 2: one thing we're doing.

879
00:40:38,239 --> 00:40:40,000
Speaker 3: But you can flip it and you can say, what

880
00:40:40,040 --> 00:40:42,680
Speaker 3: if the agent is actually the pattern that it's trying

881
00:40:42,719 --> 00:40:45,839
Speaker 3: to the vocabulary kind of fails us year, but it's

882
00:40:45,840 --> 00:40:49,240
Speaker 3: trying to ingress into the physical world through our medium.

883
00:40:49,520 --> 00:40:52,120
Speaker 3: And maybe what happens as we age is that the

884
00:40:52,160 --> 00:40:54,680
Speaker 3: cells become less and less able to implement it, They

885
00:40:54,680 --> 00:40:59,200
Speaker 3: become unresponsive, the machine slows down. Maybe the mind of

886
00:40:59,239 --> 00:41:02,319
Speaker 3: the agent, of the morphogenetic intelligence is perfectly fine, but

887
00:41:02,400 --> 00:41:06,000
Speaker 3: the machine doesn't respond. And so those are experimentally distinguishable,

888
00:41:06,040 --> 00:41:08,160
Speaker 3: and we're doing those experiments. We actually have some data

889
00:41:08,160 --> 00:41:09,920
Speaker 3: for this now and so those are those are just

890
00:41:10,040 --> 00:41:12,600
Speaker 3: very different. And the way you then would address aging

891
00:41:12,640 --> 00:41:15,920
Speaker 3: from two different from those two different viewpoints is quite different.

892
00:41:16,040 --> 00:41:17,400
Speaker 3: So that's what we would love to do, is to

893
00:41:17,640 --> 00:41:21,920
Speaker 3: is to learn to recognize and communicate with other kinds

894
00:41:21,960 --> 00:41:24,160
Speaker 3: of agents that are not even physical objects as such.

895
00:41:24,160 --> 00:41:25,000
Speaker 2: They are they are.

896
00:41:25,080 --> 00:41:29,400
Speaker 3: Persistent patterns that may have all kinds of energe, you know,

897
00:41:29,440 --> 00:41:30,280
Speaker 3: their own agendas.

898
00:41:31,920 --> 00:41:34,480
Speaker 1: So let me ask you a couple of rapid fire questions.

899
00:41:34,560 --> 00:41:38,799
Speaker 1: If they're diverse intelligences everywhere. If we can start understanding

900
00:41:38,800 --> 00:41:42,400
Speaker 1: these patterns around us as being cognitions of their own,

901
00:41:42,760 --> 00:41:44,240
Speaker 1: what does this mean for ethics?

902
00:41:44,560 --> 00:41:47,360
Speaker 3: Yeah, this is a huge problem. This is an absolutely

903
00:41:47,480 --> 00:41:50,560
Speaker 3: huge problem. I think that it is foundational to the

904
00:41:50,600 --> 00:41:54,480
Speaker 3: development of ethics as a mature species to learn to

905
00:41:54,600 --> 00:41:58,440
Speaker 3: recognize and ethically relate to minds that are nothing like ours,

906
00:41:58,680 --> 00:42:01,759
Speaker 3: that are basically not on the you know, at least

907
00:42:01,800 --> 00:42:03,840
Speaker 3: in some cases, because you can actually believe or now

908
00:42:03,840 --> 00:42:06,359
Speaker 3: you could get much even weirder than this pattern thing

909
00:42:06,400 --> 00:42:09,400
Speaker 3: that I'm talking about, And so it's certainly above my

910
00:42:09,560 --> 00:42:12,080
Speaker 3: remit to try and formulate the ethics. But what is

911
00:42:12,280 --> 00:42:15,400
Speaker 3: very clear is that we need to learn to recognize them,

912
00:42:15,440 --> 00:42:17,360
Speaker 3: we need to learn to communicate with them, and we

913
00:42:17,400 --> 00:42:19,560
Speaker 3: need to start thinking about what do we owe other

914
00:42:19,640 --> 00:42:22,480
Speaker 3: beings that live with us, that live in spaces and

915
00:42:22,520 --> 00:42:24,960
Speaker 3: have goals that are really hard for us to visualize.

916
00:42:25,200 --> 00:42:25,680
Speaker 2: What are the.

917
00:42:25,640 --> 00:42:29,080
Speaker 1: Implications for AI at this moment that we're in.

918
00:42:29,239 --> 00:42:32,600
Speaker 3: People tend to have a very kind of a binary

919
00:42:32,760 --> 00:42:35,440
Speaker 3: view on this. They will either say, oh, yeah, it

920
00:42:35,480 --> 00:42:37,439
Speaker 3: talks like us and therefore it's like a human brain,

921
00:42:37,800 --> 00:42:39,520
Speaker 3: or they'll say, oh no, this thing is a machine,

922
00:42:39,560 --> 00:42:40,880
Speaker 3: and therefore it's nothing like us.

923
00:42:40,960 --> 00:42:43,200
Speaker 2: So I think both of those are terrible.

924
00:42:43,560 --> 00:42:46,239
Speaker 3: And first of all, because in order to be intelligent

925
00:42:46,480 --> 00:42:49,879
Speaker 3: and have meaningful cognition and maybe moral worth, you don't

926
00:42:49,920 --> 00:42:51,560
Speaker 3: need to be like a human mind.

927
00:42:51,880 --> 00:42:54,000
Speaker 2: There are many minds that are nothing like a human mind.

928
00:42:54,000 --> 00:42:55,120
Speaker 2: You don't have to be like humans.

929
00:42:55,120 --> 00:42:56,759
Speaker 3: And I don't believe at this point, as far as

930
00:42:56,840 --> 00:42:58,440
Speaker 3: I know, we don't have any ais that are like

931
00:42:58,480 --> 00:43:00,640
Speaker 3: a human mind, but that doesn't mean they're not minds.

932
00:43:01,040 --> 00:43:03,040
Speaker 3: And the other problem is that there is no such

933
00:43:03,120 --> 00:43:06,680
Speaker 3: thing as a machie. And if you believe that algorithms

934
00:43:06,800 --> 00:43:10,960
Speaker 3: and the facts of physics around the silicon and copper,

935
00:43:11,000 --> 00:43:13,040
Speaker 3: and the kinds of things we make computers out of,

936
00:43:13,320 --> 00:43:15,680
Speaker 3: if you think that those things tell the entire story

937
00:43:15,719 --> 00:43:19,040
Speaker 3: of artificial intelligence, then you should think that the story

938
00:43:19,040 --> 00:43:22,200
Speaker 3: of biochemistry is everything you need to know about the

939
00:43:22,239 --> 00:43:24,520
Speaker 3: human mind. And that's you know, I think that's blatantly false.

940
00:43:24,760 --> 00:43:27,200
Speaker 3: And so for both in both cases, I think we

941
00:43:27,280 --> 00:43:29,880
Speaker 3: have to be extremely open to the idea that we

942
00:43:29,960 --> 00:43:33,480
Speaker 3: do not understand how different kinds of minds ingress into

943
00:43:33,520 --> 00:43:36,279
Speaker 3: the world through different interfaces. And I realized this is

944
00:43:36,440 --> 00:43:38,640
Speaker 3: a weird way of putting it. This is not the standard,

945
00:43:38,680 --> 00:43:41,759
Speaker 3: the kind of neuroscience way where intelligence is created by

946
00:43:41,840 --> 00:43:42,360
Speaker 3: the hardware.

947
00:43:42,400 --> 00:43:43,959
Speaker 2: I don't actually believe that's true.

948
00:43:44,000 --> 00:43:47,239
Speaker 3: I think I think consciousness is separate and what we

949
00:43:47,280 --> 00:43:51,960
Speaker 3: see what we provide when we make you know, ais, robots, embryos,

950
00:43:52,239 --> 00:43:54,800
Speaker 3: the biobots, all of this stuff. We make interfaces different

951
00:43:54,800 --> 00:43:58,799
Speaker 3: interfaces for it, and we are currently very bad at

952
00:43:58,960 --> 00:44:02,640
Speaker 3: guessing ahead of time what is going to appear when

953
00:44:02,680 --> 00:44:05,480
Speaker 3: we make certain kinds of interfaces. And you know, I

954
00:44:05,600 --> 00:44:09,040
Speaker 3: think I think one of the most relevant pieces of

955
00:44:09,080 --> 00:44:12,200
Speaker 3: our work for this is the stuff that we detaining.

956
00:44:12,280 --> 00:44:14,880
Speaker 3: Zhang and Adam Goldstein and I wrote this paper on

957
00:44:15,520 --> 00:44:20,040
Speaker 3: unexpected competencies in sorting algorithms like bubblesort. These are things

958
00:44:20,080 --> 00:44:22,160
Speaker 3: that that computer science students have been studying, you know,

959
00:44:22,200 --> 00:44:24,399
Speaker 3: in first year CS for I don't know, sixty years,

960
00:44:24,440 --> 00:44:27,440
Speaker 3: I guess, and nobody had actually looked at it the

961
00:44:27,440 --> 00:44:29,000
Speaker 3: way that we had looked at it, and we found

962
00:44:29,080 --> 00:44:32,160
Speaker 3: this thing has delayed gratification and has these weird little

963
00:44:32,200 --> 00:44:34,160
Speaker 3: side quests that it goes on that are not in

964
00:44:34,200 --> 00:44:36,239
Speaker 3: the algorithm at all. In other words, if you just

965
00:44:36,280 --> 00:44:38,120
Speaker 3: stare at the algorithm. You know, it's six lines of code.

966
00:44:38,160 --> 00:44:40,480
Speaker 3: It's a deterministic algorithm. There's no magic that, there's no

967
00:44:40,520 --> 00:44:42,680
Speaker 3: new biology to be found. You know exactly what it's doing,

968
00:44:42,920 --> 00:44:45,440
Speaker 3: and yet it does things that we do not expect

969
00:44:45,440 --> 00:44:47,319
Speaker 3: it to do in the algorithm, not just randomness, not

970
00:44:47,360 --> 00:44:50,680
Speaker 3: just complexity, not just unpredictability, but things you would recognize

971
00:44:50,680 --> 00:44:52,120
Speaker 3: as cognitive competencies.

972
00:44:52,560 --> 00:44:55,080
Speaker 2: And that means that if we don't, if we can't,

973
00:44:55,120 --> 00:44:55,399
Speaker 2: you know.

974
00:44:55,400 --> 00:44:59,040
Speaker 3: Sometimes people say me, well, I build I build language models.

975
00:44:59,040 --> 00:45:00,799
Speaker 3: It's just linear alogib I know what they're doing. There's

976
00:45:00,840 --> 00:45:02,880
Speaker 3: nothing as if look, we don't even know what bubble

977
00:45:02,880 --> 00:45:04,520
Speaker 3: story is doing. If you can't, if you don't know

978
00:45:04,520 --> 00:45:06,160
Speaker 3: what bubble story is doing, you sure as hell don't

979
00:45:06,200 --> 00:45:08,279
Speaker 3: know what these language models are doing. And so we

980
00:45:08,320 --> 00:45:11,280
Speaker 3: need to treat all of these things as empirical questions,

981
00:45:11,320 --> 00:45:14,319
Speaker 3: not philosophical decisions that we can make, and we have

982
00:45:14,400 --> 00:45:17,520
Speaker 3: to get much better at understanding how new minds ingress

983
00:45:17,600 --> 00:45:21,759
Speaker 3: even in tiny interfaces like low complexity, very simple kinds

984
00:45:21,800 --> 00:45:22,480
Speaker 3: of interfaces.

985
00:45:26,880 --> 00:45:29,920
Speaker 1: That was my interview with Mike Levin, biologist at Tufts.

986
00:45:30,160 --> 00:45:32,879
Speaker 1: Every time I talk with Mike, it's hard to look

987
00:45:32,880 --> 00:45:35,480
Speaker 1: at the world the same way. So we started by

988
00:45:35,520 --> 00:45:40,320
Speaker 1: asking what is intelligence? But instead of finding a crisp,

989
00:45:40,680 --> 00:45:44,680
Speaker 1: singular answer, we were handed something far more powerful, which

990
00:45:44,719 --> 00:45:48,799
Speaker 1: is a new lens, a new way of thinking about intelligence,

991
00:45:48,880 --> 00:45:52,880
Speaker 1: not as a static property that some lucky creatures have

992
00:45:52,960 --> 00:45:56,799
Speaker 1: and others lack, but as a multi dimensional space of

993
00:45:57,080 --> 00:46:02,440
Speaker 1: gold directed behavior, shaped by evolution and context and purpose.

994
00:46:02,640 --> 00:46:06,000
Speaker 1: And this is of course a reframing of life itself

995
00:46:06,080 --> 00:46:11,719
Speaker 1: because through this lens, intelligence isn't only confined to a cranium.

996
00:46:11,880 --> 00:46:15,440
Speaker 1: It's not restricted just to animals with brains. Instead, it's

997
00:46:15,440 --> 00:46:19,480
Speaker 1: something that shows up wherever you have systems that are

998
00:46:19,520 --> 00:46:23,480
Speaker 1: solving problems and adapting to things they didn't expect and

999
00:46:23,960 --> 00:46:28,000
Speaker 1: correcting errors. Wherever there are goals, there may be something

1000
00:46:28,040 --> 00:46:31,600
Speaker 1: in play that is like a mind. And when we

1001
00:46:31,640 --> 00:46:35,560
Speaker 1: look through this lens, the universe becomes alive in strange

1002
00:46:35,560 --> 00:46:40,480
Speaker 1: and beautiful ways. Cells are more than dumb building blocks.

1003
00:46:40,520 --> 00:46:44,759
Speaker 1: We can see them as decision makers. Organs are more

1004
00:46:44,800 --> 00:46:47,600
Speaker 1: than machine units that are chugging along. We can see

1005
00:46:47,640 --> 00:46:52,239
Speaker 1: them as negotiating parties in their own societies. A regenerating

1006
00:46:52,280 --> 00:46:56,120
Speaker 1: flatworm is more than a textbook collection of cells. It's

1007
00:46:56,160 --> 00:46:59,880
Speaker 1: an entity that knows what it's missing and takes action

1008
00:47:00,080 --> 00:47:02,839
Speaker 1: to restore itself. In a sense, it remembers what it

1009
00:47:03,000 --> 00:47:05,960
Speaker 1: used to be, and it holds that shape in its

1010
00:47:06,000 --> 00:47:09,480
Speaker 1: future and moves towards it. So, Mike Levin's work suggests

1011
00:47:09,520 --> 00:47:14,240
Speaker 1: that the basic machinery of cognition, like memory and problem

1012
00:47:14,360 --> 00:47:19,080
Speaker 1: solving and preferences, this all might emerge way earlier in

1013
00:47:19,120 --> 00:47:24,120
Speaker 1: evolution then we've assumed. Because cognition might not require neurons,

1014
00:47:24,160 --> 00:47:27,720
Speaker 1: it might not even require consciousness in any familiar sense.

1015
00:47:27,760 --> 00:47:31,640
Speaker 1: What it does require is something more basic that you

1016
00:47:31,680 --> 00:47:36,239
Speaker 1: can achieve with lots of architectures, a capacity to act

1017
00:47:36,280 --> 00:47:38,839
Speaker 1: in service of a goal. And this raises all kinds

1018
00:47:38,880 --> 00:47:42,920
Speaker 1: of great questions. If we accept that intelligence exists in

1019
00:47:42,960 --> 00:47:48,080
Speaker 1: a multi dimensional space, what else around us counts as intelligent?

1020
00:47:48,640 --> 00:47:52,320
Speaker 1: How about a tree sending resources through its root network.

1021
00:47:52,360 --> 00:47:55,960
Speaker 1: How about a colony of ants adjusting its forging behavior.

1022
00:47:56,320 --> 00:47:59,760
Speaker 1: How about your immune system adapting to a virus.

1023
00:48:00,200 --> 00:48:01,399
Speaker 2: How about a cluster of.

1024
00:48:01,360 --> 00:48:05,479
Speaker 1: Engineered cells navigating a maze. And one thing I think

1025
00:48:05,560 --> 00:48:07,840
Speaker 1: is really important here is thinking about what all this

1026
00:48:08,040 --> 00:48:11,640
Speaker 1: means for the future of AI. At the moment, we're

1027
00:48:11,640 --> 00:48:15,359
Speaker 1: only building machines inspired by brains. But I think when

1028
00:48:15,400 --> 00:48:18,560
Speaker 1: we look back in twenty years, that will seem quaint,

1029
00:48:18,680 --> 00:48:21,560
Speaker 1: and we will be seeing a lot more emulation of

1030
00:48:21,600 --> 00:48:27,400
Speaker 1: the distributed, adaptive self regulating qualities of other more spread

1031
00:48:27,400 --> 00:48:32,280
Speaker 1: out and sometimes more creative biological systems. Could we design

1032
00:48:32,440 --> 00:48:36,680
Speaker 1: machines that do physical things, not just like minds, but

1033
00:48:36,880 --> 00:48:40,600
Speaker 1: like cell assemblies and bodies. And finally, with everything that

1034
00:48:40,640 --> 00:48:43,759
Speaker 1: we talked about today, what does this all say about.

1035
00:48:43,680 --> 00:48:45,160
Speaker 2: Who you really are?

1036
00:48:45,400 --> 00:48:49,319
Speaker 1: Because when we're being honest, we are not individuals in

1037
00:48:49,360 --> 00:48:50,480
Speaker 1: the traditional sense.

1038
00:48:50,560 --> 00:48:52,200
Speaker 2: We are collectives.

1039
00:48:52,280 --> 00:48:57,360
Speaker 1: We are billions of cells and trillions of microbes, all

1040
00:48:57,400 --> 00:49:04,040
Speaker 1: operating with partial autonomy some goals, and this vast ballgame,

1041
00:49:04,080 --> 00:49:07,520
Speaker 1: which is much larger than we can conceive, is somehow

1042
00:49:07,920 --> 00:49:13,960
Speaker 1: coordinated into the illusion of a unified self. The story

1043
00:49:14,000 --> 00:49:18,520
Speaker 1: of you is a kind of consensus reality emerging from

1044
00:49:19,040 --> 00:49:24,000
Speaker 1: many smaller parts, most of which have no idea you exist.

1045
00:49:24,239 --> 00:49:27,120
Speaker 2: To my mind, this is a call for awe.

1046
00:49:27,320 --> 00:49:29,320
Speaker 1: I don't know why we'd only talk about this stuff

1047
00:49:29,360 --> 00:49:32,960
Speaker 1: occasionally on a podcast. Why aren't airplanes flying around with

1048
00:49:33,040 --> 00:49:36,319
Speaker 1: banners celebrating this kind of stuff? Why aren't we talking

1049
00:49:36,400 --> 00:49:40,360
Speaker 1: about this on CNN instead of local political cycles, because

1050
00:49:40,400 --> 00:49:43,680
Speaker 1: the lesson from today's episode is that intelligence is probably

1051
00:49:43,719 --> 00:49:47,200
Speaker 1: not rare but common. We always look at it as

1052
00:49:47,239 --> 00:49:49,800
Speaker 1: a strange exception to the rules of nature, but maybe

1053
00:49:49,800 --> 00:49:52,759
Speaker 1: it is the rule. And if this is the right

1054
00:49:52,880 --> 00:49:55,080
Speaker 1: lens to look through, what it means for us is

1055
00:49:55,080 --> 00:49:59,960
Speaker 1: that the world is full of minds, strange and ancient,

1056
00:50:00,239 --> 00:50:04,520
Speaker 1: and in many ways alien minds, some fast, some slow,

1057
00:50:04,880 --> 00:50:09,040
Speaker 1: some huge, some microscopic, some we've built ourselves, and most

1058
00:50:09,280 --> 00:50:09,640
Speaker 1: that have.

1059
00:50:09,640 --> 00:50:12,720
Speaker 2: Been here all along waiting.

1060
00:50:12,400 --> 00:50:15,640
Speaker 1: For us to notice. We're just starting to map this territory.

1061
00:50:15,640 --> 00:50:18,280
Speaker 1: And I think one of the lessons from Levin's lab

1062
00:50:18,480 --> 00:50:21,520
Speaker 1: is that the boundary between mind and matter is more

1063
00:50:21,600 --> 00:50:23,160
Speaker 1: porous than we generally assume.

1064
00:50:23,520 --> 00:50:25,080
Speaker 2: And the more we study this, the.

1065
00:50:24,960 --> 00:50:28,400
Speaker 1: More we're going to need to update our science textbooks.

1066
00:50:28,440 --> 00:50:30,879
Speaker 2: But more importantly, we're going to need to update our.

1067
00:50:30,800 --> 00:50:34,480
Speaker 1: Intuitions about what it means to be alive and to

1068
00:50:34,520 --> 00:50:35,759
Speaker 1: be intelligent.

1069
00:50:36,000 --> 00:50:37,520
Speaker 2: We'll need to tune into the.

1070
00:50:37,480 --> 00:50:40,400
Speaker 1: Fact that the whole world around us might be more alive,

1071
00:50:40,600 --> 00:50:44,879
Speaker 1: more curious, more goal seeking than we thought to imagine.

1072
00:50:45,280 --> 00:50:48,720
Speaker 1: In that light, the story of intelligence isn't a peak

1073
00:50:48,800 --> 00:50:52,719
Speaker 1: that we have reached, but a vast landscape where agency

1074
00:50:52,800 --> 00:50:56,840
Speaker 1: is common, and every living system, no matter how small

1075
00:50:56,960 --> 00:51:01,840
Speaker 1: or strange, might be solving problems that we have yet

1076
00:51:01,920 --> 00:51:09,880
Speaker 1: to understand. Go to eagleman dot com slash podcast for

1077
00:51:09,920 --> 00:51:13,360
Speaker 1: more information and to find further reading. Join the weekly

1078
00:51:13,400 --> 00:51:16,719
Speaker 1: discussions on my substack, and check out and subscribe to

1079
00:51:16,840 --> 00:51:20,560
Speaker 1: Inner Cosmos on YouTube for videos of each episode and

1080
00:51:20,600 --> 00:51:24,719
Speaker 1: to leave comments until next time. I'm David Eagleman, and

1081
00:51:24,760 --> 00:51:28,200
Speaker 1: this is Inner Cosmos