HomeVideos

Dr. Michael Levin — Reprogramming Bioelectricity

Now Playing

Dr. Michael Levin — Reprogramming Bioelectricity

Transcript

3001 segments

0:00

Cancer fundamentally involves an

0:02

electrical disregulation among cells.

0:05

It's basically a dissociative identity

0:07

disorder on the part of the cells. It's

0:09

literally a disorder of the cognitive

0:10

glue that binds individual cells towards

0:13

large scale purpose where large scale

0:15

purpose I mean building organs and

0:16

tissues and things like that as opposed

0:18

to being amiebas and doing amieba level

0:20

things. And we've shown in these animal

0:22

models both that we can detect incipient

0:25

tumor information and we can prevent and

0:27

normalize tumors after they form by

0:30

restoring not by fixing the DNA if there

0:32

is any DNA issue not by killing the

0:34

cells with chemotherapy but by

0:36

electrically reconnecting them to the

0:38

group such that they can form again a

0:40

memory of what they're supposed to be

0:41

doing.

0:42

>> Mike, very nice to finally connect.

0:44

>> Yeah, wonderful.

0:45

>> Thanks for making the time.

0:46

>> Of course. Yeah, thanks for having me.

0:48

We have lots of ground to explore and I

0:51

thought we would begin with a book that

0:53

had a spot on my bookshelf when I was a

0:57

kid. It seems like you and I may have

0:59

found it at the same time, but you did a

1:00

lot more with it than I did. The author

1:04

is Robert O. Becker. Is that enough of a

1:07

cue?

1:07

>> Yeah.

1:08

>> To tee it off?

1:09

>> I think it is.

1:10

>> All right. What is the book and why is

1:12

it relevant?

1:13

>> I'm going to guess it's the Body

1:14

Electric.

1:15

>> That's right.

1:16

>> Yeah, it's very relevant. I discovered

1:17

it in an old bookstore that my dad and I

1:20

visited when I was in Vancouver, Canada

1:22

for the World's Fair in ' 86. And I

1:24

found this thing and it's kind of a

1:26

patchwork of a number of different

1:27

things, right? Like he was into applied

1:29

field dangers and things like that. But

1:32

I was just stunned with all the

1:35

references to prior work that revealed

1:38

to me that the kinds of things I'd been

1:40

thinking about were actually real and

1:42

that people had investigated it.

1:44

>> And that book, I guess Dr. Becker was an

1:46

orthopedic surgeon and he was

1:49

effectively penning a scientific memoir,

1:51

right? Describing experiments involving

1:54

salamanders and other animals exploring

1:56

the role of electricity and many many

2:00

different aspects of biology. How would

2:03

you define for folks bio electricity?

2:06

What is a helpful way to define that

2:09

term? and then we'll probably hop to the

2:12

video in a sense that introduced me to

2:14

your work which I will not be alone in

2:16

citing but let's begin with the

2:18

definition bioelectricity what is that

2:20

>> well bio electricity in general is the

2:22

way that living systems exploit physics

2:25

in particular the physics of electricity

2:28

to do the amazing things that living

2:30

systems do and there are roughly

2:32

speaking two kinds of bioelect

2:33

electricity there's the familiar kind

2:35

which is studied by neuroscience and so

2:38

this is the electrical activity of the

2:40

cells in your brain and I think everyone

2:42

has a rough understanding of the fact

2:44

that the reason you know things that

2:46

your individual neurons don't know and

2:47

that you have beliefs and the

2:49

preferences and so on that's that are

2:50

more than just any of the neurons in

2:52

your head is through this amazing

2:54

cognitive glue that electricity provides

2:56

right it binds your neurons into a

2:58

collective intelligence that that

3:00

underlies our mind so that's the that's

3:01

the bio electricity that everybody's

3:03

familiar with and then there's the other

3:05

kind also called developmental bioelect

3:07

electricity which you can get to by

3:10

asking about where did the brain come

3:12

from and where did it learn those

3:14

amazing tricks and very quickly you

3:16

realize that wow some of these things

3:17

have been around for a very long time

3:19

long before we had brains and neurons

3:21

and that the question of what does your

3:23

body think about and before it has a

3:25

brain what how does it use electricity

3:26

is the study of developmental bio

3:28

electricity

3:29

>> the video that I was referencing you

3:31

will not be surprised to hear was an

3:34

older TED talk and then subsequent

3:37

interview on stage and that sent to me

3:38

by Adam Goldstein who's now at Softmax

3:42

and that was probably several years ago

3:45

I would say at this point that it was

3:47

sent to me. Could you perhaps and I know

3:50

a lot has happened since but could you

3:52

describe some of the experiments that

3:55

you covered at TED to give people an

3:58

idea of how this becomes tangible,

4:01

right? This conversation of

4:02

bioelectricity becomes tangible.

4:04

>> When we look at biology, we see lots of

4:05

amazing things. For example, in a

4:07

salamander, if they lose a limb, they

4:09

regenerate the limb and they stop when

4:11

it's complete. Right? And in fact, there

4:13

are many other interesting kinds of

4:15

things that when anybody looks at it,

4:17

the first thing they ask is how does it

4:19

know to do that?

4:20

>> And one of the things I discussed in

4:22

that video was if you scramble the

4:24

cranioacial organs of a tadpole, they

4:26

still make a pretty normal frog. All

4:28

they sort themselves out, they move in

4:30

new paths until they get to a normal

4:31

frog face and then they stop. So anybody

4:33

sees that and immediately the question

4:35

is okay but how do they know what a

4:37

proper frog face look like? And if you

4:38

do know then how do you know how to get

4:40

from here to there right? How do you

4:42

navigate? So the way we're all taught in

4:45

biology is that that's a bad question.

4:47

We are told none of these things know

4:49

anything. They are mechanical machines

4:52

that sort of roll forward according to

4:54

rules of chemistry and in the end some

4:56

cool stuff happens and we'll call it

4:57

emergence and things like that and

4:58

complexity science will will sort of

5:00

catalog them. But but don't worry, none

5:02

of these things actually know anything.

5:03

That's just what they do. And so what I

5:06

was trying to describe in that talk is

5:07

this idea that well actually the idea

5:10

that chemical processes can in fact know

5:13

things. It's not magic. It's not

5:15

mysterionism. We are chemical processes

5:17

that know things. And we've had for for

5:19

many decades mature science of including

5:22

cybernetics and control theory and

5:24

things like that. a mature science of

5:26

figuring out how it is that machines of

5:28

all different kinds can know things and

5:29

they can have goals and so on. So, so

5:31

what I tried to show in that talk are

5:33

some examples by which the living

5:36

tissues for example platforms that are

5:38

cut into pieces and every piece has to

5:39

figure out how many heads should I have

5:41

where do the heads go what should the

5:43

shape of my face be these kinds of

5:44

things that in fact they do know and the

5:47

way they know is because they store

5:48

memories and maybe not shockingly

5:51

although it's certainly shocking to a

5:53

lot of folks the way those memories are

5:55

stored is in an electrical network that

5:57

is very similar to the way that we store

5:59

our goal directed behavioral repertoires

6:02

in our in our brain and that these

6:04

things are sort of widely spread. And so

6:06

regeneration, cancer suppression and

6:08

cancer repair and remodeling, birth

6:11

defects and birth defect repair, all of

6:13

these things are extensively using

6:16

electrical pattern memories. And we now

6:18

have a way to rewrite those pattern

6:20

memories. I've been so excited to have

6:22

you on the show because I am an intrepid

6:26

muggle kind of blindly half blindly

6:30

exploring science to the extent that I

6:32

can. And I every once in a while I'll

6:34

share a resource like I did recently

6:36

this multi-part series called the gene.

6:39

This is a Ken Burns produced documentary

6:42

about genetics, the history of genetics

6:45

starting with Mendle and so on working

6:47

all the way up to modern biotech. The

6:50

underlying framework for that entire

6:53

series is DNA as master copy let's call

6:58

it then RNA then protein and that's kind

7:02

of how it works right you have this

7:05

blueprint that is executed upon and that

7:08

produces what we see in the world on

7:10

some level but as I understand it you by

7:13

manipulating bio electricity have

7:16

produced

7:18

for instance animals that have two

7:20

heads. That trait persists over

7:22

generations. And maybe I'm getting the

7:24

specifics wrong, but that is not by

7:27

virtue of manipulating

7:30

DNA. And I'm just wondering if I'm first

7:32

of all getting that right, but secondly,

7:34

what that says about how we might be

7:39

revising our understanding of biology

7:41

and what the textbooks might look like,

7:44

you know, 5 or 10 years from now or

7:45

further out.

7:46

>> Yeah, you're not wrong. I could list any

7:48

number of scenarios that we and others

7:50

have studied in which the genetics not

7:53

only don't tell the whole story but in

7:54

fact tell a fairly misleading story. And

7:57

the way that I would describe it and and

7:59

there are two pieces to this and I'll do

8:00

the simpler piece first and then we can

8:02

talk about the other piece. The simpler

8:03

piece is we can get there by thinking

8:05

about the distinction between software

8:07

and hardware. And by the way I should

8:08

preface this because some people get

8:10

really upset about this. I am not saying

8:12

that the current way that we think about

8:14

software and hardware is sufficient to

8:16

get everything we need from biology. It

8:17

does not cover all of biology. It covers

8:19

one important piece of biology.

8:21

Reprogrammability is really critical.

8:23

And so if you wanted to make that same

8:25

movie about computers for example, you

8:28

could make a movie that basically goes

8:32

electric fields,

8:35

silicon and germanmanium

8:37

and transistors and the flow the flow of

8:41

energy through circuits. Done. Right?

8:43

That could be your movie. And it's not

8:44

an unimportant part of of the story.

8:46

It's a very important part of the story.

8:48

But the critical part that that doesn't

8:49

get to is that's the hardware. And in

8:52

fact, that's what the genome does. So

8:53

the genome tells every cell what the

8:55

hardware is going to be. So the genome

8:57

gives every cell the little tiny sort of

9:00

protein level hardware that it gets to

9:02

have. But now comes the other

9:04

interesting part which is the

9:05

reprogrammability. And we've known for a

9:07

very long time now that if your hardware

9:09

is good enough and the biological

9:11

hardware is more than good enough then

9:13

that hardware is reprogrammable. So,

9:15

what happens just as an example, what

9:16

happens in these flatworms, these

9:17

two-headed flatworms that you were

9:19

referring to, the flatworm has a

9:21

bioelectric memory in it that says, and

9:24

we can see it. I'm saying these things

9:25

because we can now see these memories

9:27

and we can rewrite them at will. So,

9:28

this is, you know, this is now

9:29

actionable in the lab. It has a

9:31

biomectric memory that says one head.

9:34

That memory is not genetically encoded.

9:38

What is genetically encoded is a bunch

9:39

of hardware that when you first turn on

9:41

the juice, it basically

9:45

acquires that memory as a default. It's

9:47

the way when you buy a calculator from

9:49

the store and you turn on the power,

9:50

they all say zero, right? Reliably 100%

9:52

of the time. They all say zero. Great.

9:54

But that zero is not the only thing that

9:55

that circuit can do, right? As you find

9:57

out very quickly, they can store memory

9:58

and do all these things. the the genetic

10:00

hardware of the worm is very good at

10:03

making sure that every worm starts out

10:04

with a very specific it's a little bit I

10:07

think related to instinct and how you

10:08

know certain birds are are born knowing

10:10

how to make nests and things like that.

10:12

>> Mhm.

10:12

>> The hardware has defaults and by default

10:15

one head but the hardware is

10:17

reprogrammable. So what we were able to

10:19

do is go in and identify the memory that

10:21

actually says how many heads and we can

10:22

change it. And when you change it you

10:24

don't need to change the hardware. You

10:25

don't need to change the genetics any

10:27

more than when we form new memories. You

10:29

don't need to change the genes in your

10:30

brain to form new memories. I always say

10:32

to people on your laptop if you want to

10:34

go from Photoshop to Microsoft Word. You

10:36

don't get out your soldering iron and

10:37

start rewiring. It'd be laughable if if

10:39

you had to, but that's how we used to do

10:40

it in the 40s and 50s. You program a

10:42

computer by pulling and plugging wires.

10:45

Well, you don't do that anymore because

10:46

it's reprogrammable. And that's what the

10:48

biology is. And so that's the first

10:49

thing. And the second thing just very

10:51

quickly and we can get into it if you

10:52

want is that this cellular intelligence

10:55

that exists not only is is

10:58

reprogrammable but it is actually

11:00

creative in the sense that it interprets

11:02

the DNA and we can talk about this. It

11:04

doesn't blindly do what the DNA says and

11:07

this is kind of a deep thing because

11:08

it's the way our cognition works too. It

11:11

interprets memories in a way that is

11:14

improvisational. It does not simply

11:16

follow what they say counter counter to

11:17

what we all learn.

11:18

>> All right. So, I'm going to come back to

11:20

the sort of how the textbooks might be

11:21

revised question in a minute, but before

11:25

we get there, you said we can see

11:27

memories, right? So, this is empirically

11:30

demonstrable in the lab. What does it

11:31

mean to see those memories? What does

11:34

that actually mean and look like? And

11:36

then secondly, with the flatworms with

11:39

the two heads, why does that persist if

11:43

it does into future generations? So what

11:45

we can see directly are the bioelect

11:47

electrical properties of tissues. And

11:49

we've developed tools using voltage

11:52

sensitive fluorescent dyes. And so that

11:54

means you take your embryo or your

11:56

tissues or whatever you've got and you

11:57

soak it in this special chemical that

11:59

glows different degrees or different

12:01

wavelengths depending on what the local

12:03

voltage is. And so back in the olden

12:05

days in electrophysiology, you had an

12:07

electrode then you would have to poke

12:08

like a little needle and you would poke

12:09

every cell and you would get the voltage

12:10

reading. We don't need I mean of course

12:12

we still do that for certain purposes

12:14

but what you can now do is get a full

12:16

map of the whole tissue all at once and

12:18

in fact you can make movies of it and

12:19

watch it change over time and we have

12:21

these amazing videos of embryos changing

12:24

their electrical activities over over

12:25

time. It's basically like what

12:27

neuroscientists do when they do imaging

12:28

in brains but we can do it in the rest

12:30

of the body. So there what you see are

12:32

the electrical patterns. Now from there

12:33

you have to do a lot of experiments to

12:34

prove that what you're looking at are in

12:36

fact memories. And there are many

12:38

different kinds of things we do, but

12:40

functionally what you have to show is

12:41

that you can decode the electrical

12:44

pattern that you're seeing and show that

12:47

what it encodes is the future set points

12:50

towards which the cells will work. In

12:52

other words, I can take a one-headed

12:54

worm. I can change the voltage pattern.

12:56

It's still a one-headed worm, but it's

12:59

internal representation of what a

13:01

correct worm should look like now says

13:03

two heads. You don't see it because it's

13:05

a latent memory, but when you cut the

13:07

when you cut the thing into pieces, now

13:09

what the cells do is consult the memory

13:10

and they say, "Oh, two heads." And then

13:12

they build two heads and you get your

13:13

two-headed worm. So, so you don't know

13:15

right away when you're first looking at

13:16

it. You don't know that that's a memory.

13:18

You have to do experiments to prove that

13:19

that's what it actually is.

13:21

>> And then the persistence, the durability

13:23

over generations,

13:24

>> the process of regeneration and repair

13:26

in general is kind of homeostatic

13:28

process. So, it's like a thermostat. You

13:30

have a set point. If the temperature

13:31

gets too low, it tries to go up. If it

13:33

gets too high, it tries to come down. It

13:34

tries to keep a certain that is exactly

13:36

what happens in the body which is

13:38

anatomical homeostasis. So cells come

13:41

and go all the time, right? So the ship

13:43

we're kind of a ship of thesis, right?

13:45

In many ways. So so cells and materials

13:47

come and go. Sometimes drastic kinds of

13:50

injuries for animals that regenerate

13:52

past them. Embryogenesis. I mean look,

13:55

half our population can regenerate an

13:57

entire body from one cell. I mean that's

13:59

amazing, right? That's an amazing like

14:00

development. And embryionic development

14:01

is an incredible example of

14:03

regeneration. The whole body

14:04

regenerating from just one egg cell. And

14:07

in all of those cases, what needs to

14:09

happen is just like a thermostat has to

14:10

remember what's the right set point.

14:12

There has to be a memory mechanism that

14:14

stores it. And so the electric circuits

14:16

in the body that store these patterns,

14:18

they have a memory property as well such

14:20

that when you change it, it stays. Now

14:23

sometimes there are multiple memories.

14:25

And so we've done things like, for

14:27

example, in these flatworms, there are

14:30

different species that have different

14:31

shaped heads, round ones, triangular

14:33

ones, flat ones. We've shown that you

14:34

can take a worm, change the biological

14:37

signaling, and get it to grow ahead of a

14:39

different species. But the fun thing

14:41

about that is it grows the head of a

14:43

different species. You haven't touched

14:44

the genetics, by the way. Again, the

14:45

genome is totally wild type.

14:47

>> So wild,

14:47

>> right? But it'll grow the head of a

14:49

different species and it'll stay there

14:51

for about 30 days and then it goes back

14:54

to its original. It's not permanent

14:56

unlike the the two-headed thing is

14:57

permanent.

14:58

>> That never changes. But the head shape

14:59

after about 30 days they go back. And so

15:02

clearly there are multiple there's more

15:03

than one. There's sort of meta some kind

15:05

of metacognitive thing that says yeah

15:07

you know I know you thought that was

15:09

your memory but actually that's wrong.

15:11

>> It sort of overwrites some kind of error

15:13

correction thing which that one we

15:14

haven't we haven't cracked yet. So so

15:16

there are kind of layers upon layers.

15:17

All right. So, for people who are

15:18

listening and wondering, you know, how

15:20

this translates or might translate to

15:24

humans, sure, I want to get there, but

15:25

I'm going to bridge to that simply by

15:28

saying that this

15:31

topic of bio electricity is is long been

15:34

interesting to me. I mean, it's been

15:35

interesting to humans for a very long

15:37

time, going back to slaves in ancient

15:39

Rome stepping on electric eels and

15:41

finding relief from gout. But in a more

15:43

more modern incarnation, I had Dr. Kevin

15:46

Tracy on the podcast some time ago who

15:49

was

15:49

>> he's incredibly wells cited.

15:51

>> Oh yeah.

15:51

>> Played a part after his experiences with

15:54

patients with septic shock identifying

15:56

TNF alpha and a lot of subtleties around

15:58

that.

15:58

>> Y

15:59

>> and has developed hardware in this case.

16:03

I mean they're programmable but for vag

16:05

nerve stimulation

16:06

>> predominantly for at this point

16:08

autoimmune disorders like rheumatoid

16:10

arthritis and so on. But you can see

16:13

some incredible, incredible clinical

16:15

effects and we're just touching the tip

16:17

of the iceberg. So I'm wondering, it

16:18

took a long time to get here though,

16:19

even with something that is relatively,

16:22

I would say, straightforward to

16:23

identify, which is the Vegas nerve, aka

16:26

Vegas nerves, these sort of

16:29

intercontinental cables running down

16:31

either side of the neck with 100,000

16:33

fibers on either side. So in this case,

16:35

we're talking about flatworms. We could

16:37

certainly talk about other species that

16:39

are known for regeneration, but broadly

16:41

speaking, what might this mean for

16:45

humans? How might this be applied to

16:48

humans? Do humans have this programmable

16:50

layer just as some of these other

16:53

species do? What might therapeutics or

16:56

morphaceuticals or otherwise look like?

16:59

>> Yeah. No, and and that's a great

17:00

connection. Yeah, Kevin's work is

17:01

amazing. I was just talking to him a

17:03

couple weeks ago. It's awesome stuff.

17:05

>> Great guy. Great guy. Yeah, he really

17:06

is. So, a couple of things to explain

17:08

why this is relevant to humans and then

17:10

I'll give you like three three broad

17:11

areas of application.

17:13

>> The reason it's absolutely relevant to

17:14

humans is that we are all basically

17:17

built on fundamentally the same

17:19

principles. People have this idea that

17:21

well frogs are sort of a lower creature,

17:23

but you know, we're we're mammals and

17:25

once once you get past yeast and things

17:27

like that, we are all roughly the same

17:29

as far as this stuff goes. These kind of

17:32

electrical signals were evolution

17:34

discovered them around the time of

17:35

bacterial biofilms like very long ago.

17:37

And so this is all very well conserved

17:39

and for that reason for example there

17:42

are human mutations in ion channels that

17:44

are birth defects. So if you mutate ion

17:46

channels in humans you get birth defects

17:48

just like we see in in frog and and

17:50

chick and zebra fish and things like

17:51

that. So those are all well conserved.

17:53

And with David Kaplan who's a

17:55

collaborator of mine at tufts we've done

17:57

a bunch of work on bielectrics of human

17:59

meenimal stem cells. This stuff works

18:01

for humans as well. It is not some like

18:03

frog of platworm specific thing. This is

18:04

very very broad. I should say this is a

18:06

disclaimer I always have to do. You

18:08

mentioned morphaceuticals. So there are

18:10

a couple of spin-off companies that have

18:11

licensed some of this technology. So I

18:13

need to sort of say that as a as a

18:14

disclosure. So one is specifically

18:16

called morphaceuticals. There's a

18:17

company that is pushing forward our limb

18:19

regeneration work in bielectrics. And

18:21

then there's also this other company

18:22

called Astonishing Labs that is doing

18:24

some of the stuff in aging and so on.

18:26

Having said all that, I firmly believe

18:28

that these things are heading for

18:32

clinical application in humans and

18:33

probably not that far off. I hope you

18:35

know

18:36

>> here are the three applications. So the

18:37

first application is birth defects. So

18:40

we have shown that we can repair a

18:42

number of different birth defects of the

18:45

brain, the face, the heart, what else?

18:48

The gut, these kinds of things by

18:50

restoring correct biological patterns in

18:52

vivo. And so this is now in in animal

18:55

models. We are moving of course to more

18:57

clinical kinds of things and I hope in

18:59

the future this will absolutely be of of

19:01

human application. So birth defects is

19:03

one. Regeneration is another. The name

19:05

of the game here is communicating with

19:06

the cells. This is not about stem cells

19:09

or gene therapy or scaffolds made of

19:12

nanomaterials. Like those are all tools

19:14

that might be useful. But the real trick

19:16

here is to communicate to a group of

19:17

cells what do you want them to build?

19:19

And that's what the bioctric code is all

19:21

about. It's about communicating to the

19:22

collective to the cellular collective.

19:24

And so we've done work on limb

19:26

regeneration. We've done work on

19:28

inducing whole organ formation, eyes and

19:30

things like this. So I think there are

19:32

going to be massive applications

19:33

hopefully clinically in restoring the

19:35

damaged and and missing limbs and other

19:38

other structures like that. And then the

19:40

third thing is going to be cancer. So

19:42

something else and we can get into what

19:44

the kind of more profound aspect is but

19:45

but the bottom line is that cancer

19:48

fundamentally involves an electrical

19:50

disregulation among cells. I'll just say

19:53

it and we can unpack it later, but it's

19:55

basically a dissociative identity

19:57

disorder on the part of the cells. It's

19:59

literally a disorder of the cognitive

20:01

glue that binds individual cells towards

20:03

large scale purpose where large scale

20:05

purpose I mean building organs and

20:07

tissues and things like that as opposed

20:08

to being amiebas and doing amoeba level

20:11

things. So cancer is another thing and

20:13

we've shown again in in these animal

20:15

models both that we can detect incipient

20:18

tumor formation and we can prevent and

20:20

normalize tumors after they form by

20:23

restoring not by fixing the the DNA if

20:25

there is any DNA issue which they're not

20:27

doesn't have to be not by killing the

20:29

cells with chemotherapy but by

20:30

electrically reconnecting them to the

20:33

group such that they can form again a

20:35

memory of what they're supposed to be

20:36

doing. So those three things,

20:38

regeneration, birth defects, and cancer,

20:40

I think are going to be of great value

20:42

in humans. Now, there's also issues of

20:44

aging. So we also have an aging program

20:46

in our lab and looking at why it is that

20:48

over time cells forget how to upkeep a

20:50

proper organism and we have some some

20:52

interesting thoughts about that as well.

20:54

>> Let's dive in. I would love to hear more

20:55

about the interesting thoughts on aging

20:57

and then we're we're definitely going to

20:58

get to cognition, which is

21:01

I mean that can go in a lot of

21:02

directions, but let's start with the

21:04

aging piece. what are some of the

21:06

implications or experiments

21:09

or just maybe conceptual frameworks that

21:12

are are sort of due as an as a revision

21:16

of what we've thought to date.

21:18

>> First of all, one of the things that

21:20

we've seen is that over time and by the

21:23

way this is fairly recent work. So this

21:26

is in no way is this the final story.

21:27

This is just kind of what we know now.

21:29

I'm sure this will this will be updated

21:31

over time. the electrical prepatterns

21:34

that tell the cells and tissues what

21:37

large scale structure we're supposed to

21:38

look like, they get fuzzy. They degrade

21:41

over time. And so much like what we do

21:43

with birth defects is we try to

21:45

reinforce the correct patterns. And this

21:47

is this is one of the ways we're we're

21:49

addressing aging as well is by

21:50

reinforcing these patterns. Now, one

21:52

question you might ask is why over time

21:54

are these things getting fuzzy, right?

21:56

What's going on? And there are a couple

21:58

of schools of thought. One is that this

22:00

is the consequence of accumulated noise

22:03

and damage. So molecular damage entropy

22:05

basically right over time you just

22:06

accumulate damage and every everything

22:08

kind of gets degraded over time. And

22:10

then there's also these kind of what

22:12

they call programmatic theories where

22:14

basically the idea is that you're

22:15

programmed to age for whatever reason

22:16

evolution has favored a decline and

22:19

death. So we have an interesting third

22:21

alternative to offer which is the

22:23

following. We did a simulation

22:24

experiment where we had a kind of a

22:28

virtual a virtual body where the cells

22:31

cooperate together to build an embryo.

22:33

Okay? And so they work really hard to

22:35

work together. They build to a

22:36

particular pattern memory. So you know

22:38

this thing I've been telling you about.

22:39

And then I said, "Let it run. Just just

22:40

leave it alone and let it run." And so

22:42

what you see is something very

22:43

interesting. They work really hard

22:44

together and they make they make the

22:47

correct body. Then it sort of stays that

22:50

way as they defend it and then it falls

22:53

apart and it begins to degrade. Now

22:55

what's interesting is that in our

22:56

simulation there was no evolution for a

22:59

limited lifespan. There was no noise.

23:01

There was no damage. It was sort of

23:03

perfect. You know everything was perfect

23:04

and still it degraded. Why would it do

23:06

that? I had this interesting thought and

23:08

I'll back into it this way. Just imagine

23:10

this standard Judeo-Christian version of

23:12

heaven, right? So you get to heaven and

23:14

you get there. Let's say let's say you

23:16

your pet snake and your dog get get to

23:17

heaven. So okay, everything is great.

23:20

There's no more damage. There's no

23:21

decay. Nothing is damaged. Everything is

23:23

great. Everything's fantastic for the

23:25

next trillion years. What happens? So,

23:29

the snake may be fine doing snake things

23:32

for every day is the same as every other

23:34

day. Maybe fine. The dog, not sure.

23:37

Probably okay chasing rabbits on the

23:39

farm, you know, like maybe fine for

23:40

forever. Basically, the human though,

23:43

what do you think? I' I'd be interested

23:44

in your thoughts like what what are the

23:46

odds that a human cognitive system can

23:48

be sane for an infinite like okay I'll

23:51

keep myself busy for the first 10,000

23:53

years maybe 100 thousand years but like

23:55

a billion years in are we still sane

23:57

like what happens

23:59

>> what do you think like what do you think

24:00

would happen

24:01

>> that's interesting so well if I'm

24:03

hearing you correctly I I don't really

24:05

have a passing through the pearly gates

24:08

timeline prediction for like the the

24:11

halflife of sanity but I if I'm hearing

24:14

you correctly that the biological

24:16

programmed I mean death I suppose is

24:19

basically to

24:21

intended to ensure biological

24:25

death before insanity. Am I misharing

24:28

that?

24:28

>> So may maybe that's not the claim I was

24:30

going to make but but it's not

24:31

impossible.

24:32

>> Not a claim but I guess I'm trying to

24:34

squint and look through the exercise.

24:36

>> What I took away from that work that we

24:38

did was the following. You have a goal

24:41

seeking system

24:42

>> that has met its goal.

24:44

>> Yeah,

24:44

>> it's achieved the goal. It made the body

24:46

it was supposed to make. The error falls

24:47

to zero. Everything is great. Hangs out

24:49

there for a while. But what does a goal

24:51

seeking system do when there are no new

24:53

goals?

24:54

>> Because we're looking at a system that

24:55

may or may not be able to give itself

24:57

new goals. I mean, cognitively I think

24:58

we can, but it's not clear yet that this

25:02

system can do that. And so what we were

25:04

able to do is we were able to give it

25:06

new goals by having interventions and

25:09

going back in and saying okay now this

25:10

is your new pattern and it will do that.

25:13

But I think you know part of the you

25:14

could call it the boredom theory of

25:16

aging basically not cognitively

25:17

sematically like if your body cells over

25:19

a long period of time they they've

25:21

completed their job they've created a

25:23

body during adulthood but at some point

25:26

they start to degrade the cells don't

25:28

degrade the collective does the the

25:29

cohesion the alignment between them

25:31

because there's no longer a common goal

25:34

I mean this is what makes for an embryo

25:35

or a body as opposed to just a billion

25:37

independent cells is they're all aligned

25:39

towards the same set point towards the

25:41

same goal

25:42

>> and so When that isn't there,

25:44

regeneration, repair, maybe remodeling

25:47

becomes something else. I don't know

25:48

how, you know, maybe you need to sort of

25:50

change up the body every once in a

25:51

while. That's also a possibility.

25:53

Pleneria, do a pleneria are immortal.

25:55

>> And the pleneria or the flatworms we

25:57

were talking about earlier.

25:58

>> The flatworms. Yeah. Yeah. They're

25:59

immortal. Every two weeks they rip

26:00

themselves in half and regenerate. So

26:02

they give themselves a challenge every

26:03

two weeks. And so they've been, you

26:05

know, they've been that way for half a

26:06

billion years or so. We can see evidence

26:08

of this. For example, if you look at

26:10

there's a way to look at the age of

26:13

certain genes, the evolutionary age of

26:14

certain g of genes to see like when did

26:16

they show up. The gene expression of a

26:19

young person, all the cells are in all

26:24

the different tissues have the same idea

26:26

of what evolutionary stage they are,

26:28

meaning a human. When you look at old

26:31

tissue, and this is something we just we

26:32

just published recently, when you look

26:34

at we call it adivistic dissociation.

26:36

When you look at the tissues of old age,

26:38

the genes that they express start to

26:40

float backwards in evolution and they're

26:43

discordant. They're out of sync. So your

26:45

your liver versus your neurons, they may

26:47

all have different start to get

26:48

different ideas in terms of the genes

26:50

they express of of where on the

26:51

evolutionary tree they are,

26:53

>> right? It's like starts to float off and

26:55

in the absence of a compelling set point

26:56

or goal state, all the subunits start to

26:59

sort of float off and do their own

27:00

thing. And this is this is I think an

27:03

important component of aging. So if you

27:06

were put in charge of,

27:08

for lack of a better term, the Manhattan

27:11

project style initiative related to

27:15

aging, right? That was your sole

27:16

directive was to really do a deep dive

27:20

with the intention of developing some

27:22

type of therapeutic for humans.

27:26

What might that look like? I mean, for

27:28

all extensive purposes, infinite

27:29

funding, but you have the resources, you

27:32

can get the talent.

27:35

Where would you take it if you had a

27:37

similarly pressing deadline? And I'm not

27:40

asking for the impossible, but if you

27:41

had a reasonably tight deadline by which

27:44

you needed to try to come up with

27:46

something, where would you take it?

27:49

>> How would you think about it?

27:50

>> Tight deadlines for aging are tough

27:51

because you're not going to know for

27:53

decades whether your thing works. But I

27:55

get the idea.

27:56

>> This is what I would say. Fundamentally,

27:58

I think that aging,

28:01

cancer, birth defects, lack of

28:04

regenerative repair throughout our

28:05

lifespan,

28:07

all of these kinds of things are

28:08

downstream of one fundamental

28:12

pressure point that if you solve that,

28:13

all of these things get solved by by you

28:15

know sort of by side effect and that is

28:17

regeneration. More specifically, that

28:19

that in turn is everything there hangs

28:23

on the cognition of groups of cells. In

28:26

other words, how do groups of cells know

28:28

what to build, when to stop, how do we

28:31

communicate with them, and what kind of

28:33

intelligence do they have? And I'm being

28:35

very specific about this. When I say

28:36

they have intelligence, I don't mean

28:38

complexity. I don't mean some sort of

28:40

linguistic project where I'm going to

28:42

take things that are beautiful and

28:43

fascinating and I say, well, that's the

28:45

intelligence of life. That's not what I

28:46

mean. I'm using a very specific

28:48

definition of intelligence which is what

28:50

behavioral scientists use which is

28:51

problem solving, memory, different

28:54

degrees of a cognitive light cone of

28:56

goal directed the size of your goals,

28:58

things like that. So specifically

29:00

figuring out what are the competencies

29:03

of the living material that we're made

29:04

of and how do you communicate new goals

29:06

to them. There are lots of amazing

29:08

people in the aging field doing

29:09

interesting things and that's cool. If I

29:11

had a lot of money specifically for

29:13

aging, I would put everybody on on that

29:15

question. I would say you're not

29:17

studying aging. What you're studying is

29:19

the goal directedness of of

29:20

multisellular systems. Figure out how

29:23

they know what to do and how we

29:24

communicate goals with them. If you

29:25

solve that, all of these other things

29:27

get get taken care of as a as a side

29:29

effect.

29:30

>> What might an example or sample new

29:34

directive be to give human cells or

29:38

groups of cells a a new goal? What might

29:42

that new goal look like? I'll give you

29:44

an example and then we can talk about

29:45

what the human case might look like.

29:46

What we can do is we can take a frog

29:49

embryo and

29:52

induce a particular electrical pattern

29:56

somewhere in the body that we already

29:58

know that pattern codes for make an eye.

30:01

That's how the other cells interpret

30:03

that pattern. It means make an eye. Very

30:05

interesting in the sense that we don't

30:07

have to say which cells do what. We

30:09

don't have to say which genes you need

30:11

to turn on. These are all microlevel

30:12

details. We don't need to worry about

30:14

them because the material is competent.

30:17

Just like when I'm talking to you, I

30:18

don't need to worry about how your

30:20

synaptic proteins are going to, you

30:21

know, you're going to take care of all

30:22

of that, right? All I need to do is give

30:24

you the prompt and vice versa. And we're

30:26

having this amazing conversation, but

30:27

our hardware takes care of all all the

30:29

all the molecular detail.

30:31

>> And the same thing here. So, we provide

30:32

a biological pattern that says make an

30:35

eye here and the cells make an eye. Now,

30:39

the first thing that happens, it's

30:40

interesting. The first thing that

30:41

happens is there's a battle of world

30:43

views that takes place. We inject a few

30:44

cells. They tell their neighbors, "Let's

30:46

make an eye." The neighbors actually

30:48

say, "No, we're supposed to be skin or

30:50

gut. Don't do it." And some case,

30:51

sometimes they win and sometimes we win.

30:53

And so the goal of regenerative medicine

30:55

is to be as convincing as possible so

30:56

that you win 100% of the time. But in

30:59

the cases where we are convincing, and

31:01

we have amazing videos of cells like

31:03

convincing each other to have different

31:04

voltages and whatnot, they make an eye.

31:06

And so what you've done is you've taken

31:08

a bunch of cells that were going to be

31:09

for example gut and you've now pushed

31:12

them to be an eye at a very high level.

31:14

We don't tell I don't know how to build

31:16

an eye. I don't know all the genes that

31:17

have to be turned on. You do that. I'm

31:19

telling you something at the level of

31:21

organs. This is going to be an eye. The

31:23

eye is of the right size. It has all the

31:26

right layers to it. It is functional,

31:28

right? So you can see out of these

31:30

ectopic eyes. It's really really

31:31

amazing. And so that is an example of

31:33

giving these cells a new goal. How do I

31:35

know it's a goal? because I did not

31:36

micromanage you to do it. I was not

31:38

there saying turn on this gene, turn on

31:40

that gene. I gave you a faroff set

31:42

point, by the way, in a wild space that

31:44

no individual cell knows anything about.

31:46

The anatomical space of organ

31:47

structures, no individual cell knows

31:49

what an eye is, but the collective does.

31:51

And they stop when it's done. I don't

31:53

need to be there to tell them to stop.

31:54

They stop when it's done.

31:56

>> And so this is autonomous goal- directed

31:58

activity. It's a navigation of

31:59

anatomical space.

32:01

>> And so and so we can do this. And we

32:02

can't make everything. We can make

32:03

portions of the brain. We can make eyes.

32:05

We can make in some cases limbs. We can

32:07

make some other some other structures.

32:09

So in the human you could imagine two

32:11

ways to go and I don't know which is

32:12

going to be correct and we need to do a

32:14

lot of experiments in mammals to nail

32:16

this down. One possibility is that it

32:19

might be enough to simply reinforce the

32:23

existing human pattern. Every so often

32:26

you would get like a tuneup that reminds

32:28

the all the cellular collectives what

32:30

we're supposed to look like.

32:31

>> That's one possibility. There's another

32:33

possibility and I don't know which is

32:35

correct. I hope the first one is right

32:36

but I think it wouldn't be the end of

32:37

the world if it's the latter. Maybe it

32:40

really does get too boring with the same

32:42

pattern meaning that okay you can go a

32:45

few hundred years with the reminding of

32:46

the standard human pattern but

32:48

eventually you have to do something

32:49

unique. Now the plenary are telling us

32:51

that actually it's hundreds of millions

32:53

of years that you can make the same

32:54

thing. So I'm kind of optimistic that

32:56

you can do that but let's say that's not

32:57

the case. If that's not the case in

32:59

humans, maybe you have some number of

33:01

the hundreds of years or whatever of the

33:03

standard human body plan. But then if

33:05

you want to keep going, you got to make

33:07

some changes. What does that mean? Maybe

33:09

you wanted some wings. Maybe you want

33:10

some tentacles. Maybe you want a third

33:12

hemisphere, you know, to crank your IQ.

33:14

Maybe you want, I don't know,

33:16

>> a third eye. Who knows?

33:17

>> Sure. Sure. Sure. Infrared vision out

33:19

the back of your head. I don't know. You

33:20

know, people email me all the time

33:21

asking for all kinds of weird

33:23

peripherals. So maybe at some point it

33:26

means that you really got to change

33:27

things up a little bit. You know,

33:28

caterpillar butterfly style.

33:30

>> Mhm.

33:31

>> Maybe.

33:31

>> Wow. And just to come back to a piece

33:36

that we covered through the thought

33:38

exercise of the the pet snake, the pet

33:41

dog. Do you think we have evolved to die

33:45

or to age? I mean, if so, why? What

33:49

might be a strongman argument for that?

33:51

I'm just curious. There certainly are

33:53

theories, reasonable theories of why

33:56

evolution wants you dead

33:58

>> and there have been a number of them.

34:00

Overall, I think there may well be

34:02

tradeoffs of the kind that for example,

34:05

we're not going to put a lot of

34:07

evolution, you know, would not put a lot

34:10

of effort into maintaining something if

34:13

something else is going to go off and

34:15

you're going to die anyway, right? So,

34:16

there are these ecological trade-offs.

34:17

I'll give you an example of something

34:19

like that. People ask, "Hey, why can't

34:21

humans regenerate their limbs the way

34:22

that, you know, axelis can and things

34:24

like that?" Nobody knows. But here's a

34:26

plausible theory, right? Imagine you're

34:27

an early mammal. You're running around

34:29

the forest. Somebody bites your leg off.

34:31

Now, you have you have a high blood

34:34

pressure. You're going to bleed out. If

34:36

you don't bleed out, you're going to

34:37

walk around and grind that thing into

34:38

the forest floor. It's going to get

34:40

infected. You're never going to have

34:41

time to regenerate. What you might do is

34:43

scar, seal the wound, inflammation, so

34:46

that you might live, you know, to fight

34:48

another day. But you're definitely not

34:49

going to have time to regenerate the way

34:50

that an axelottle might, you know, sort

34:52

of floating around in water for three

34:53

weeks or whatever. Basically, what you

34:55

might say is that evolution just kind of

34:57

decided that it's not worth it. It's

34:59

never going to work. It's not worth it.

35:00

Right? And by the way, deer antlers,

35:02

deer antlers are the one amazing

35:04

mamalian example of regeneration. Plus,

35:06

the liver, I mean, liver regenerates,

35:07

stupid, but deer antlers, right? Adult

35:09

large adult mammal that regenerates this

35:11

like huge structure of of vascule.

35:13

>> The rate of regrowth is just incredible.

35:16

centimeter and a half per day of new

35:18

bone.

35:19

>> So nuts

35:20

>> nuts bone vascule intervation and you

35:23

don't put weight on it. It's not

35:25

loadbearing. It's the one, you know,

35:27

appendage that's not loadbearing. So

35:29

anyway, why I'm saying that is because

35:31

you can imagine evolutionary trade-offs

35:32

like that where evolution just didn't

35:34

bother optimizing for long age. You

35:36

could you can imagine that. But

35:37

fundamentally, I do not believe that we

35:40

are inevitably mortal. I think that at

35:43

some point if we knew what we were

35:44

doing, if we had appropriate

35:46

regenerative medicine, I don't see any

35:49

particular reason why we have to age and

35:51

die. And then you face interesting

35:53

questions about for example mental

35:54

plasticity. We all know with advanced

35:57

age people get a little a little less

35:58

plastic mentally, right? That that kind

36:00

of stuff. Is that a hardware problem or

36:02

a software problem? We don't know. You

36:04

know, if you had somebody with a

36:06

physically young brain at, you know, at

36:08

100, would they be like an 18-year-old

36:11

in terms of their ability to take on new

36:12

ideas and and focus and pay attention,

36:15

whatever? Would that still stay or is

36:16

there some kind of a cognitive tiredness

36:19

that happens that is not a hardware

36:21

issue? Like I don't I don't think we

36:22

know, but but we need to find out.

36:24

>> So, I was going to ask you about

36:26

computer science and AI and concepts

36:29

that you would like biologists to learn.

36:31

Well, let's start there and then I'm

36:32

going to ask a question that might

36:33

destroy any shred of respect that you

36:36

have for me, but I'll save that for

36:38

after this one. Do any concepts come to

36:40

mind because you certainly have spent a

36:43

lot of time in computer science that you

36:46

wish you could require biologists to

36:50

become familiar with or to study. I'm

36:52

I'm wondering about cross pollination

36:55

between disciplines within which you've

36:57

spent a lot of time. It could go the

37:00

other way as well and this could be, you

37:02

know, concepts from developmental

37:04

biology or biology writ large that you

37:07

think computer scientists should pay

37:09

more attention to. But, uh, does

37:10

anything come to mind for either of

37:12

those?

37:12

>> My original background is in computer

37:14

science. Computer scientists are amazing

37:17

generally at

37:20

compartmentalizing coarse graining you

37:23

know sort of modularizing like hiding

37:26

details and asking okay but what's it

37:28

actually important here you know and and

37:30

like blackboxing things biologists

37:32

generally think everything is important

37:34

and if you ask a biologist you get a

37:36

list of you know 30 genes like these are

37:38

you know hard one details right they're

37:39

all important but a computer scientist

37:41

is like okay but but what is it actually

37:43

doing you know and that's really

37:44

important. The most basic thing is this

37:46

issue of reprogrammability is that

37:48

understanding that certain kinds of

37:49

hardware is reprogrammable and why that

37:52

that I think is really key. The other

37:54

thing that I wish and there's not really

37:55

time unfortunately for almost any

37:57

biologist to do this but one thing I

38:00

really love for my students to do if

38:02

they can is to take a course in

38:03

programming languages. And here's why

38:05

not so they could code that that doesn't

38:07

matter. It's not the coding aspect. What

38:08

happens in a typical course of

38:10

programming language is that so let's

38:12

say in a single semester you'll spend

38:13

three weeks doing different languages

38:15

and the thing about those languages and

38:16

maybe this is true of some human

38:18

languages as well but it's definitely

38:19

true of computer languages is that each

38:22

language is a different way of looking

38:24

at the world you start off with

38:26

something that makes sense and you're

38:27

like ah step by step you know you sort

38:29

of tell it what to do okay and then all

38:30

of a sudden bam now there's this other

38:32

thing where every piece of data there's

38:34

this language called lisp where every

38:35

piece of data is also instructions and

38:37

you can execute any piece of data. Like

38:39

what? And then you get into this other

38:40

thing and it's functional programming.

38:42

Now there are no variables. You don't

38:43

get to have any variables. You have to

38:44

like everything is just a function call.

38:46

And every time you do this, it sort of

38:48

rips the foundation of your of your

38:49

world out from under you. And it says

38:51

this universe works in a very different

38:53

way than you thought before. Forget

38:54

everything you knew before. Now you got

38:56

to do this. And how are you going to

38:57

solve this problem? Now there's

38:58

recursion or now you know there's no

39:00

global variables or whatever. And what

39:01

it's really good for is that mental

39:03

plasticity that reminds you that the way

39:06

you think things are and the tools you

39:08

think you have are not the only things

39:10

in town. And so when you do that in a

39:12

lightning and things go fast and then

39:14

the final exam comes and that's this

39:15

other thing you've never seen before,

39:16

like being able to do that quickly, I

39:18

think is super valuable. And I would

39:21

love that to be more known in biology.

39:23

But the final thing I'll say is and this

39:25

is I think this is true but but just to

39:27

be clear this is very controversial and

39:29

almost nobody else thinks this is true.

39:31

So you know who knows but the

39:33

interesting thing that a lot of people

39:35

not just biologists but a lot of people

39:36

think is is something like this. Okay,

39:39

there's something going on with humans,

39:41

maybe other animals where

39:44

biochemistry does not tell the whole

39:45

story, right? You you read the

39:47

biochemistry textbook and you say,

39:48

"Okay, that's that's cool, but there's

39:50

something about my mind and my my

39:53

ability to solve problems in abstract

39:55

spaces and my inner perspective and all

39:57

this stuff is just not captured in these

39:59

low-level details." So, that's a little

40:01

disturbing. It's like what is that then?

40:02

If it's not captured in the chemistry,

40:03

like where is that coming from? But

40:05

don't worry, we have this other thing

40:07

over here which are machines. Dumb

40:08

machines, dead matter, dumb machines,

40:10

algorithms, computers, and those things

40:13

do only exactly what the algorithm tells

40:16

them to do. They are perfectly captured

40:17

by our formal model. So, we have a

40:19

formal model of of chemistry and the

40:21

rules of chemistry. And that we think

40:23

does not capture all what it is to be an

40:25

entire, you know, full-on human. But we

40:27

have these other formal models of

40:29

touring machines and programming and

40:31

code and and you know mechanics and

40:33

those things capture exactly what the

40:35

machines do. Those get the whole thing.

40:36

Okay. I think and this is the part

40:38

that's very sort of controversial and

40:39

not a widely shared opinion. I think

40:40

that's false. I think our formal models

40:43

never capture all of what's going on.

40:46

And some of the some of the craziest

40:47

stuff coming out out of our lab recently

40:49

is showing how much even in very simple

40:52

sorts of machines, how much interesting

40:55

novelty, not just complexity, not just

40:57

unpredictability, but things that any

40:58

behavioral scientist would recognize as

41:00

some kind of a protocognitive capacity

41:03

shows up in even minimal systems where

41:06

you don't expect it. And so what I'd

41:09

like the biologist to sort of

41:10

eventually, you know, once we can show

41:12

this this widely, the biologist to

41:15

understand is that biological systems

41:17

are amazing and awesome, but it's a kind

41:19

of a larger degree, not kind of what's

41:22

already going on in inanimate systems.

41:24

And for this reason, I this is this is

41:26

also kind of a crazy claim is that I

41:28

think the circle, if you make a circle

41:30

of cognitive things and living things, I

41:33

think cognition is wider than life. I

41:35

think cognition predates life and I

41:36

think it's bigger than life. And

41:37

normally people do that the other way

41:38

around. They say here's the inanimate

41:40

universe, some chunk of that is living

41:41

and some tiny piece of that is

41:43

intelligent. I think that's exactly

41:45

backwards and that's something we need

41:46

to understand both on the biology end

41:48

and on the computer science end is like

41:51

is there a distinction between what

41:54

people commonly think of as living

41:55

things and machines? Like are there any

41:58

actual machines in the sense that we

42:00

like to think that there are? You know,

42:02

that's a deep set of questions for both

42:04

fields in the future.

42:05

>> All right, that's a super tempting

42:08

opening to take and I might come back to

42:10

it, but I wanted to take the

42:13

opportunity, as promised, to destroy any

42:15

credibility I might have with you and my

42:17

audience.

42:17

>> Great.

42:19

All right. So, I'm going to try to give

42:21

myself some some air cover

42:25

by going back, sorry to drag you into

42:27

it, Kevin, but to go back to Kevin Tracy

42:30

and also actually years before my

42:33

interview with Kevin won with Martin

42:35

Rothblat and in both cases, Martin is

42:38

just an incredible polymath on a lot of

42:40

levels. People should look into Martine.

42:42

We were chatting Martine and I about

42:47

transuricular

42:48

stimulation of the vagus nerve and

42:51

there's quite a bit of mechanistic

42:54

debate around this. How many fibers are

42:57

you hitting? Is it actually possible to

42:58

do through the skin etc. But suffice to

43:00

say, the clinical outcomes of certain

43:05

types of placement of certain types of

43:08

currents with on the ear seem to produce

43:13

pretty dramatic anti-inflammatory

43:16

>> effects. And so then that raised the

43:18

question for me of wait a second

43:21

>> do those maps I've seen in Chinese

43:26

medical offices have anything to them

43:29

right now chatting with Kevin he's like

43:30

well funny thing about that is that it

43:32

was a Frenchman who actually put that

43:33

together after taking like a ballpoint

43:35

pen and pressing on patients ears and

43:37

then it made its way back to China. I

43:39

don't know the full history but as we're

43:42

talking about bio electricity I have to

43:45

ask and again this might be a dead end

43:47

but if you look at traditional Chinese

43:49

medicine I went to two universities in

43:50

China and the took a pretty close look

43:53

at this at the time in 1996 but is there

43:56

anything to meridian'sqi did they get

43:59

anything right or was it just

44:01

coincidence

44:02

is there really nothing defensible to it

44:06

I'm just wondering if there's any

44:07

overlap

44:08

>> I I was wondering how wild you were

44:09

going to get that with that question,

44:11

like where that was going to go, but

44:13

that's not too bad. Okay. I don't know

44:15

the epidemiological data on acupuncture

44:19

and how it works in clinical trials or

44:21

any of that stuff. I I don't know. What

44:23

I do know is that I personally know an

44:25

amazing there's a guy in Boston called

44:28

Tom Tam and I've known him since the

44:30

80s, you know, my whole life since I was

44:32

a kid and he's treated me, he's treated

44:34

my family. I've seen people advance

44:36

cancer patients in his clinic. Don't

44:38

know anything about the wider

44:39

epidemiological aspect of it. To me, as

44:42

someone who's interested in practical

44:43

results, I would say I can't say

44:46

anything other than 100% that I I think

44:47

there's something very powerful here.

44:49

Very significant. So the next question

44:52

is what are those meridians and do they

44:55

have any functional overlap with the bio

44:57

electricity that we're talking about? I

44:59

don't know. We actually had back in 2006

45:01

I think we had a little bit of a

45:03

collaboration with the New England

45:05

School of Acupuncture to try to figure

45:06

that out. I wanted an animal model. I

45:08

wanted to see if we can do a frog model

45:09

of acupuncture or something and it

45:11

didn't work for a number of reasons. If

45:13

I had to guess and I don't know, right?

45:14

The real answer is I don't know. But if

45:16

I had to guess, what I would say is that

45:19

whatever it is that acupuncturists are

45:21

managing with their treatments, it has

45:23

the same relationship to the

45:24

bioelectricity that the bioelectricity

45:26

has to the chemical signaling. In other

45:27

words, you know, chemical, physical

45:30

protein signaling pathways,

45:33

bioelect electrical state, there's some

45:36

otherformational state. Maybe it has to

45:38

do with a biomechanics of tissues. And

45:40

again, like disclaimer, I I still get

45:41

acupuncture, right? So, Vanessa Grimes

45:43

here in Beverly, like, you know, every

45:45

month I get a tuneup like I I I think it

45:46

really works. So, you know, take it all

45:48

with a grain of salt, but I don't think

45:50

they're managing bioelectricity

45:51

directly. I think they're managing

45:52

something else which is no doubt

45:55

relevant to the bielectric layer because

45:57

it then has to transduce through that to

45:58

the rest of the body, but I suspect it's

46:01

not bioelectricity per se. I suspect

46:03

it's it's something additional. That's a

46:05

that's a guess on my part.

46:06

>> Yeah. Cool. I'm glad I asked. Thanks for

46:08

answering, too. On the acupuncture side,

46:11

I don't get a whole lot of acupuncture.

46:13

And you you can look at sham studies and

46:16

so on where yes in the case of for

46:18

instance my one of my pts in Texas you

46:22

can use something called dry needling

46:24

instead for muscle spasms and that's

46:27

very very effective but then you can

46:28

also conversely look at data in say

46:33

canines or pain control in animals where

46:36

the as far as we know placebo is going

46:39

to be pretty tough to defend. And well

46:43

>> well maybe I guess you tell me maybe not

46:45

or surgery with I mean this is probably

46:48

not the right term but sort of

46:49

anesthesia via acupuncture also pretty

46:52

interesting

46:53

>> right so I don't know where to take that

46:55

I don't have any domain expertise but it

46:58

>> continues to be interesting I suppose

46:59

and also pregnancy data

47:01

>> acupuncture for conception

47:04

>> which may intersect with vagus nerve

47:07

stimulation who knows

47:08

>> yeah the deal with placebo like I don't

47:10

see placebo as a confound Mhm.

47:12

>> I mean, it can be if you're trying to

47:14

calculate certain things, but I think

47:15

it's kind of the main show in a lot of

47:18

ways. Yeah.

47:18

>> And some of the placebo research like

47:21

Fabriio Benedeti is, you know, one of my

47:23

favorites and he has a talk where he

47:24

says words and drugs have the same

47:26

mechanism of action. And it's amazing

47:29

because he actually does the experiments

47:30

of giving patients drugs that he tells

47:33

them what they are and then he looks at

47:35

molecular markers in their blood and in

47:36

their you know in their cells and yeah

47:38

they turn on the downstream like Yeah.

47:40

Except that they didn't get any of the

47:41

drug.

47:42

>> Yeah.

47:42

>> Right. So there's something very

47:44

interesting going on here and we already

47:46

know if I were to come here and tell you

47:48

that hey did you know that with the

47:49

power of my mind alone I can

47:51

electrically depolarize like up to 30%

47:54

of of the body right? You'd say what is

47:56

that yoga mind matter? like like mind

47:58

body inter like what kind of thing is

47:59

that they say no it's voluntary motion

48:01

we do it every day so so it's an it's an

48:04

amazing thing that nobody talks about

48:05

like think about this you wake up in the

48:06

morning you have these very abstract

48:09

highle goals you have social goals

48:11

financial goals research like whatever

48:13

and in order for you to do any of that

48:14

you have to get up out of bed so what

48:16

has to happen is these these incredibly

48:18

high level abstract intent has to change

48:21

the way that calcium and potassium ions

48:24

go across your muscle cell membranes

48:26

right these abstract mental things have

48:28

to change the chemistry of your body

48:30

cells. We know that's true. That's every

48:31

time you, you know, you lift your arm up

48:33

or you take a step voluntarily, that is

48:35

what's happening. So, we know that

48:36

works. So, if that works, why is it so

48:39

bizarre to think that our other mental

48:41

states might not affect either through

48:43

the electrical transduction of the

48:44

nervous system or through other

48:46

non-neural bioelectricity or through

48:48

other pathways yet could affect ways

48:50

that other cells act. It just it it

48:53

doesn't seem weird to me at all. It

48:55

seems like it would have to be that way.

48:57

But what we need to figure out is how it

48:58

works and how to communicate. I think

49:00

that's an incredibly powerful. If

49:02

acupuncture is some kind of entry point

49:04

into figuring that out, great. You know,

49:06

it's not a compound. It's a it's a

49:08

feature.

49:08

>> Yeah, I totally agree with the placebo

49:11

not necessarily being a compound, as you

49:13

mentioned, depending on kind of what

49:15

you're optimizing for measuring and so

49:16

on.

49:17

>> Yeah. I mean as someone who's funded a

49:19

lot of basic science

49:20

>> and clinical research involving

49:22

psychedelic compounds which are just

49:25

notoriously difficult to blind. It's

49:27

like yeah give someone a meggaos and

49:29

nasin plus x y and z or rolin or

49:31

something like that but

49:33

>> generally the control group knows that

49:35

they are the control group

49:37

>> but that doesn't invalidate the research

49:38

right it just points out maybe some

49:40

methodological

49:42

revision or or tweaking that might be

49:44

helpful. Well, just to add it, there's

49:46

something else here that's re that's

49:47

really interesting and I haven't seen

49:49

anybody in the field, maybe you know

49:50

folks that have looked at it. A lot of

49:52

times, at least what I understand in

49:54

some of Fabriio's data, like what both

49:57

for the efficacy and for the side

49:58

effects because there's the no SIBO

50:00

effect, right?

50:01

>> People they start, oh yeah, definitely

50:02

like headache or whatever. But what's

50:04

interesting to me anyway is that unless

50:06

you're like if you're a scientist and I

50:07

tell you that okay I just gave you an

50:09

SSRI you may know what the what the

50:11

downstream steps are going to if you're

50:13

a regular person off the street right

50:15

participating in this study. Now how how

50:17

do you know what the actual consequences

50:19

should be?

50:20

>> That's the wild part. How do you

50:21

actually implement the instructions?

50:24

>> That's right. That's right. I think

50:25

animal studies should actually be very

50:27

this is how we got here is talking about

50:28

animal placebo because there are studies

50:30

in experimental effects in animals.

50:32

There are whole books on this where in

50:34

behavioral science they do these

50:35

experiments on rats and whatever the

50:37

experimentter believes is what the rats

50:39

end up doing. They don't need to

50:40

understand the place. They're going to

50:42

they're going to do it anyway if the

50:43

experimentter believes, right? So trying

50:45

to understand some of these subtle cues

50:47

and influences and how does your body

50:49

know things I think is like super super

50:52

interesting.

50:53

>> Okay, I can't let that one go. So what

50:56

do you think is actually happening there

50:58

between the experimentter and the rat? I

51:01

mean, is it just the subtle body

51:03

language, etc. that's being transmitted

51:06

to an animal who's perceiving that? That

51:08

seems like a stretch even as I say it,

51:10

but I don't know what the alternative

51:11

explanation would be.

51:12

>> Yeah.

51:13

>> What might be a theory or two for what

51:15

is actually happening?

51:16

>> Yeah, good question. I don't have a

51:17

theory, but I will mention some things

51:19

to think about. One of the remarkable

51:22

things that living systems are good at

51:25

is in credit assignment, in selective

51:28

attention. So for example there's this

51:30

old work on bio feedback from I think

51:32

the 70s where they can show that a rat

51:35

can generate a temperature difference of

51:37

a few degrees Celsius between its ears

51:39

if you reward for that right and so now

51:42

just think and it doesn't take years of

51:44

practice it's pretty quick and just

51:47

think you're a rat you just got some

51:48

reward so let me see while my tail was

51:50

pointing north and my whiskers were kind

51:52

of vibrating and my gut was doing this

51:54

and my toes were like what the hell did

51:56

I just get rewarded for right you would

51:58

think in This in computer science, this

52:00

is called the frame problem because

52:01

trying to get robots and AIs to focus on

52:03

the important thing. There's an old I

52:05

forget who did this example, but imagine

52:08

there's a robot and it's in a room with

52:10

a bomb and the robot says, "Oh, there's

52:13

a bomb. I got to get out of here." And

52:14

it leaves. Except the bomb was on a cart

52:16

that was connected to the robot, right?

52:18

So it goes with him and of course he

52:19

blows up. So what does the next robot

52:20

do? This maybe Dan Dan, I don't

52:22

remember. So the next robot is like,

52:24

"Okay, okay, we have to have them

52:25

consider all the options, right?" So now

52:27

this robot, he goes in, there's the So

52:29

the robot's like, "Well, let me see. The

52:30

walls are pretty vertical and the paint

52:32

is dried." Yeah. And it's a 90° angle.

52:34

Cool. And so by the time it's considered

52:36

like all these things, of course, it

52:37

blows up again. So that's no good. And

52:39

so biologicals are like amazing at

52:42

knowing what to pay attention to. What

52:44

was I just rewarded for? What was the

52:46

thing I did which I'm never going to do

52:47

again, which, you know, turned out

52:48

poorly. Like we don't know how that

52:51

works. And that I think is going to be a

52:54

major part of that puzzle that you're

52:55

asking about. And I I'll just give you

52:57

an example from our work. Flatworms

52:59

again, plenarium. We put pleneria in a

53:02

solution of barium. Barryium is a

53:04

non-specific potassium channel blocker.

53:06

It like blocks all the potassium

53:08

channels. So that makes it very hard for

53:10

cells to do their physiology, especially

53:12

the neurons freak out. Their heads

53:13

explode. Literally overnight their heads

53:16

explode. But as it turns out, so it's

53:19

called deep progression is a way to put

53:20

it. But basically the cells just like

53:22

stage.

53:22

>> Very polite way.

53:24

>> Yeah. Yeah. cuz it sort of deep

53:25

progresses.

53:25

>> It's like negative treatment in

53:28

>> like special ops assassination. Oh yeah.

53:30

It's just a negative treatment. Yeah.

53:31

>> Yeah. Yeah. Basically, it's a deep

53:33

progression. Yeah. But here's the

53:34

amazing part. So you take the part

53:36

that's left, right, the tail and the mid

53:38

the midbody. You leave it in the barium

53:41

and within about 14 days, they grow a

53:43

new head and the new head doesn't care

53:45

at all about the barium. No problem

53:46

whatsoever. Right? So the new head is

53:48

fine. They said, "How is this possible?"

53:49

So what we did was a very simple-minded

53:51

experiment. We took all the genes that a

53:53

normal head expresses, all the genes

53:55

that and and it and but sure this

53:57

doesn't have to be in the genes. This is

53:58

just a simple thing we did to start with

54:00

and what genes does the barium adapted

54:02

head express? And we found less than a

54:05

dozen genes that make the difference.

54:07

Now think about this. Pleneria don't

54:09

normally see barium in the wild. You

54:11

don't have an evolutionary response to

54:13

what happens when I get hit with barium.

54:14

You're sitting there. I view that you

54:16

you have something like 20,000 genes.

54:19

you're hit with this new stressor that

54:21

you know you don't you've never seen

54:23

before. How do you know which of those

54:24

20,000 genes are going to help? I always

54:26

visualize this as you're sitting in one

54:28

of those nuclear reactor control rooms,

54:29

right? There's buttons everywhere. The

54:31

thing's melting down. You don't have

54:33

time to start flipping switches sort of

54:34

randomly. Like you'll be dead long

54:36

before that. How did they zero in on the

54:39

correct 12 things out of a space of

54:41

20,000 dimensions that they could have

54:43

like it's a very highdimensional search

54:45

problem. We we don't know. No, nobody

54:47

knows. And that aspect of it, biology,

54:51

finding solutions to problems they

54:52

haven't seen before, knowing what's

54:54

salient, figuring out what to pay

54:55

attention to. There are aspects here

54:57

that we haven't even come close to

55:00

replicating in our engineering

55:02

technologies. I think it's going to be

55:03

part of all that.

55:04

>> This is a pretty close hop to and I

55:08

don't this is a term that has very

55:10

specific meaning for you, so it may not

55:12

be the right term for me to use, but

55:15

cognition. Let's talk about human

55:17

cognition in the way that most people

55:21

would think about it, right? We have

55:23

this big ball of fat inside our skulls.

55:26

Bunch of magic seems to happen and we've

55:29

got these amazing tools. We've got these

55:31

MRIs, PET scans, etc. that we can EEGs

55:34

and so on that we can use to try to

55:36

study the brain and what's actually

55:38

happening. And my question is, and not

55:41

to belabor this type of question, but

55:43

it's just a forcing function for

55:45

conversation sort of 10 years out, 10

55:47

years from now, how the textbooks, and

55:50

textbooks may or may not even exist at

55:52

that point, but how the teaching of

55:54

neuroscience might have fundamentally

55:55

changed as it relates to cognition.

55:57

Because I I look at, for instance,

56:00

funding a lot of neuroscience over the

56:02

last 10 years. It's like okay sometimes

56:04

the scientists are attracted to whatever

56:06

the fanciest tools might be. There's

56:09

some prestige in that. They produce a

56:11

lot of beautiful images. You can slice

56:13

and dice the data from a single study 15

56:16

different ways and get a lot of

56:17

publications.

56:19

But and this is not something I could

56:22

kind of technically defend. I'm left

56:24

feeling as a lot of people do that

56:26

there's something missing. it's not

56:28

quite capturing the full picture, pun

56:31

intended, not just with the MRIs, but

56:33

with a lot of these these tools that

56:35

we're using. And I'm bringing this up

56:37

because of the comment you made about

56:39

the gap between the biologics and

56:41

current engineering. And this certainly

56:44

relates to AI and so on, but I don't

56:46

have the the technical chops to

56:49

understand quantum effects. But if I

56:51

think about some of the cursory reading

56:53

I've done about sort of quantum effects

56:56

and all faction, let's just say, right?

56:57

Smell. I'm just left wondering what we

57:01

might be missing fundamentally about how

57:03

cognition works and also ties into not

57:07

turn this into my own TED talk. I'll try

57:09

to wrap this up in a second, but having

57:10

conversations with my friend Kevin

57:11

Kelly, who is the founding editor of

57:14

Wired magazine, who's an avid beekeeper

57:16

and about just the collective memory of

57:19

hives and properties that you would

57:21

never be able to predict and that I'm

57:25

not entirely sure you you can at least

57:28

at this point engineer from the ground

57:29

up. But how do you think our view of

57:31

cognition, thinking, mind

57:35

change in the next 5 10 years? I want to

57:38

talk about two things. One of which I'm

57:41

pretty sure is going to be very

57:43

different in in that time frame and

57:45

another thing which is more fundamental

57:46

that may take longer or may not.

57:48

>> The one thing that I think for sure is

57:50

going to change is that there's a

57:51

thriving emerging field out there now

57:53

called diverse intelligence. This is the

57:55

idea that biology and as I've been

57:58

pushing it also nonbiology has been

58:01

doing intelligence of different kinds

58:04

long before brains and neurons appeared.

58:07

It's been solving problems, navigating

58:08

spaces, having memories, anticipating

58:10

the future long before neurons appeared.

58:13

The biggest barriers to this are these

58:15

ancient categories that we got saddled

58:18

with from pre-scientific times. This

58:20

idea that everything is binary. People

58:22

ask, is it intelligent? Is it conscious?

58:24

Is it this or like that binary framing

58:27

has been holding everything back for a

58:29

really long time.

58:30

>> Is it holding it back because it's

58:32

bifurcated between inanimate and animate

58:34

or is it something else? It's the idea

58:36

that it hides it obscures the fact that

58:40

we don't have a good story of scaling.

58:44

Just two quick examples. When you go to

58:46

court, there's this notion of an adult.

58:48

Okay, we all know if you really think

58:50

about it, nothing happens on the night

58:51

of your 18th birthday. Like literally

58:53

nothing. And that's a and b, we don't

58:56

actually have a good story of a

58:58

scientifically grounded story of what

59:01

does it mean to have personal

59:03

responsibility? How does that change

59:05

over time? How is it impacted by

59:07

neurotransmitters, brain tumors,

59:09

Twinkies, uh, society, like whatever? We

59:11

don't actually have those questions

59:13

answered, but you've got to get traffic

59:16

court done or whatever. And so, we've

59:18

just decided we're going to have this

59:20

thing called adult. We're going to clock

59:21

it on the 18. The car rental industry

59:23

actually does better because they look

59:25

at statistics and they'll say, "No,

59:26

actually, it's 25 is when you're like

59:28

more fully cooked is when you can rent a

59:29

car." And so, they do a little better.

59:30

But regardless, the idea is that we and

59:33

we all say it's an adult. And so what

59:35

those kind of binary terms do is they

59:39

obscure the fact that yeah, but

59:41

underneath we actually still don't have

59:43

a proper understanding of what's going

59:45

on. And so by saying that something is

59:48

or isn't intelligent, what you're

59:51

basically assuming is that somewhere

59:53

some developmental biologist can tell

59:55

you what happened from the time that you

59:57

were an oasite, a little blob of

59:59

chemicals that presumably was well

60:02

handled by biochemistry and physics. And

60:04

then eventually, well, now you're the

60:06

subject of physiology. And then

60:07

eventually you're the subject of

60:08

developmental biology. And then, oh

60:10

look, now you're the subject of behavior

60:12

science. Oh wait, psychoanalysis. You

60:13

know, so like each of us made that

60:15

journey. It's a smooth continuous

60:17

journey. Developmental biology offers no

60:20

support for this idea that somewhere

60:21

there's a bright, you know, flash of

60:23

light and that, okay, now you used to be

60:25

just chemistry, but now you've got a

60:26

real mind, like that never happens.

60:28

>> Because here's the other thing they do.

60:30

If I were to say that it's a continuum,

60:32

right? If cognition is a continuum from

60:34

the most primitive passive matter to,

60:36

you know, humans and above,

60:38

>> what I could say is I'm going to take

60:39

some tools from behavioral neuroscience

60:42

and I'm going to apply them to all kinds

60:43

of weird things and see how that works

60:45

out for me. And that's how we're going

60:46

to know what's cognitive and what's not.

60:48

And this in fact is what my lab is

60:49

doing. That that project is very

60:51

disruptive and there are a lot of people

60:52

who really think that's crazy because

60:55

what they will say is look, it's a

60:57

category error. brains and humans think.

61:00

Cells and tissues can't think. How do

61:03

you know? Well, because the way the word

61:04

is defined, right? So, so what they've

61:06

done is they've taken something that's

61:08

actually should be an empirical

61:10

experimental science. Take the tools and

61:12

see where they give you benefits and

61:14

where they don't. But instead, they've

61:15

made it into a philosophical or

61:17

linguistic project where these ancient

61:19

categories that we got saddled with, you

61:20

know, oh, don't make a category error,

61:22

you know, that kind of thing. So, I

61:23

think it's very disruptive. So I think

61:25

what's going to happen in the future is

61:27

that all of the applications now that

61:30

are coming out from active matter

61:31

research from basil cognition from work

61:33

in slime molds and single cells and

61:35

materials with learning capacity and all

61:37

this stuff we're going to realize I

61:39

think this is you know again one of

61:41

these claims I think that neuroscience

61:43

is we're going to realize neuroscience

61:45

is not about neurons at all okay and

61:47

what neuroscience is really about is

61:48

cognitive glue neuroscience is the

61:50

question of what kind of architectures

61:52

add up to larger scale minds from

61:55

aligned simpler components. Now

61:57

neuroscience has a lot to teach us about

61:58

that because that that's basically what

62:00

they've been studying but I think the

62:03

majority of them not everybody because

62:05

we have all kinds of collaborators in

62:06

this field who are doing something else

62:08

but the vast majority of traditional

62:09

neuroscience think they're studying

62:10

neurons that this is something unique to

62:13

these you know cellular systems that

62:14

they're studying and I think this field

62:16

of diverse intelligence combines

62:18

artificial intelligence and engineering

62:19

and cybernetics and evolutionary biology

62:23

and AI and exobiology right in the

62:26

search for alien life. Like all of these

62:28

things are together asking what are

62:30

actually the common threads of being an

62:34

agent. No matter what your origin story

62:36

whether you were designed or evolved or

62:38

you know engineered or evolved or

62:39

whether you were made of squishy

62:42

proteins or whether you were you know

62:44

made of silicon or something else right

62:46

I don't know I think science fiction

62:47

prepares you for that nicely for that

62:49

kind of stuff to really have a broader

62:51

conception of it. And so I think really

62:53

understanding what neuroscience is

62:55

actually about I think is going to be a

62:57

massive change. And the final thing I'll

62:58

say is in this I don't know how long

62:59

it's going to take. Hopefully not that

63:01

long. But you might remember the story

63:02

that at one point I think in the late

63:04

1800s

63:06

I think was Lord Kelvin who said that

63:08

yeah physics is you know kind of done.

63:09

there's just like these two black clouds

63:11

or something but but mostly it's just

63:13

about like uh more decimal you know more

63:15

digits past the decimal point right but

63:16

there's these two clouds you know and

63:18

the two clouds basically you know became

63:20

quantum mechanics right and relativity

63:22

and all of that and so I think

63:23

neuroscience has a couple of black

63:25

clouds I'll just describe one of them

63:27

Karina Copman and I she's an amazing she

63:29

started as a high school student working

63:31

with me remotely we just did a review of

63:33

this clinical cases in humans of normal

63:36

or above normal IQ while having very

63:40

minimal brain volume. I'm sure you've

63:41

heard some of these cases, but there are

63:43

many to look at right now.

63:45

>> It's not that you can't add a bunch of

63:47

epicycles to standard neuroscience and

63:50

somehow try to squeeze these things into

63:53

the mainstream paradigm.

63:54

>> Maybe you can, but to me the most

63:56

important thing is that it doesn't

63:58

predict that that should be possible.

63:59

There's nothing we learn, at least that

64:01

I've ever seen in neuroscience courses

64:03

that tells you that, oh, and by the way,

64:04

yeah, you should be able to do all this

64:06

with like less than a third of the brain

64:07

volume of a chimpanzeee. So, there's

64:10

something going on here which I think is

64:12

really fundamental. It's one of these

64:13

like, you know, observations that you

64:16

can try to sweep under the rug, but I

64:18

think it's actually telling you that we

64:19

have some very, very seriously wrong

64:22

assumptions somewhere in the theory.

64:24

I've looked at some of that research or

64:26

in some cases brain adaptations

64:28

around severe injury and they just

64:32

raised a lot more questions than we can

64:34

currently answer. This could be a

64:36

quagmire I'm about to create, but I'm

64:40

going to take a stab at it anyway.

64:43

A lot of people talk about consciousness

64:46

maybe in the same way that people argue

64:47

about God without defining it very well.

64:49

But then even the best intentions to

64:51

define it can end up slipping on banana

64:54

peels. But I am curious. You've spent

64:56

time with Daniel Dennett who I think you

64:58

mentioned a little bit earlier. We're

65:00

talking about and I think you can keep

65:02

most people probably on the same page

65:04

when you're talking about intelligence

65:07

as very carefully defined in a specific

65:11

way. Right? And I'm paraphrasing here

65:13

from memory, so I apologize if I get it

65:15

wrong, but you know, goal seeeking

65:16

systems that maybe can satisfy those

65:18

goals in multiple ways. Maybe this is

65:20

kind of along William James lines. Feel

65:24

free to fact check that, but I'm

65:26

wondering where you go from there or how

65:29

you think about consciousness, if you do

65:31

at all. Maybe that's just one of those

65:32

terms. It's like, well, it's like

65:34

success or happiness. It's like so

65:36

poorly defined. I don't spend a lot of

65:37

time thinking about it because it's a

65:39

dead end. But if that's not the case,

65:41

how do you think about consciousness?

65:43

Because as you're talking, and some

65:44

people may have been thinking of this,

65:46

they're like, "Well, wait a second. Is,

65:48

you know, is Mike a pans psychist?" Is

65:50

like, "Where where are we going here?"

65:52

>> Yeah. Oh, I'm I'm a I don't know, some

65:54

some sort of super pansychist or

65:56

something. I don't think it's

65:58

unimportant. I think it's a very

66:00

important question. Big picture. Like, I

66:02

think it's really important. I'm not a

66:04

consciousness researcher and in my lab

66:06

we haven't done pretty much any

66:08

experiments on consciousness. So I want

66:10

to preface everything I'm about to say

66:12

by saying that first of all this is not

66:15

something I typically work on and the

66:18

reason I don't work on it right now and

66:19

and I do have some stuff cooking but

66:21

that's sort of not ready yet for public

66:23

consumption. The reason I don't focus on

66:25

it now is that there's so much that can

66:27

be done without delving into that with a

66:30

third person perspective on observable

66:33

problem solving you know cognition and

66:35

even that has been such a slog you know

66:37

I've been at this for now what 20 years

66:38

and it's been so difficult to get people

66:40

to to shift in that way like I don't

66:42

need to get into consciousness to do the

66:44

things that I need to do now

66:45

nevertheless and so for that practical

66:48

strategic reason I haven't been talking

66:49

about it except for when people ask and

66:51

so if you ask I would say that for the

66:54

purposes of defining what we're talking

66:56

about now, I would say simply something

66:58

like firsterson perspective of the kind

67:01

that makes my toothache really quite

67:04

different in import than anybody else's

67:06

toothache. There's something there's

67:08

something about my toothache that's

67:09

quite different than when other people

67:10

like it's terrible when other people

67:11

have a toothache, but but there's

67:12

something different when I have it. And

67:14

so that's I think the kind of thing that

67:16

we're talking about here. So here's what

67:19

I would say about it. First of all,

67:21

again, I really can't understand how

67:24

anybody can maintain a binary view about

67:27

this both on an evolutionary scale and

67:30

on a developmental scale. If you think

67:32

you are conscious, and I realize that

67:34

some people don't even think that, but

67:36

let's assume that we think that we are

67:38

conscious. You have to tell me when that

67:40

showed up in development. Development is

67:41

slow and gradual. And either the oite

67:44

had something that got scaled up in some

67:46

way and then what we really owe is a

67:48

story of scaling which is what I think

67:50

or something some sort of people will

67:52

say phase transition and that's a fine

67:56

hypothesis. You have to show me what the

67:57

phase transition is and why I can't zoom

68:00

into it because the nice thing about

68:01

those graphs that goes like this is that

68:03

if you just stretch the the horizontal

68:05

axis they all become smooth and flat

68:08

eventually. So like what exactly

68:10

happened that you weren't conscious and

68:12

then you and then you began like I think

68:13

that's a total non-starter. So I think

68:15

the question about consciousness is what

68:17

kind and how much? So let's just start

68:19

there and then I would say that there

68:21

are roughly four reasons why people give

68:25

each other the benefit of the doubt

68:27

about consciousness. Right? So the

68:28

problem of other minds. How do I know

68:29

that you're that you're conscious? Yeah.

68:31

There's there's usually about four types

68:32

of reasons that people give. What I can

68:34

say is that if you like any of those

68:36

reasons, for any of those four reasons,

68:39

you should take very seriously, for

68:40

example, the idea that other organs in

68:42

your body have their own consciousness

68:44

for those exact same reasons. For the

68:46

same reason, we can dive into it if you

68:47

want, but for the same reasons that you

68:49

and I think each other is conscious, you

68:51

should take very seriously the idea that

68:52

there are other parts of your body that

68:54

are. Now, at this point, people usually

68:55

say, "Well, that's weird. I don't feel

68:56

my liver being conscious." Right? your

68:59

left hemisphere that's verbal puts up a

69:00

very nice story about how it's the only

69:03

one that's conscious and of course you

69:04

don't feel your liver being conscious.

69:06

You also don't feel me being conscious.

69:07

That's because you are not that

69:09

consciousness. But that doesn't mean

69:11

that there aren't any number of other

69:13

consciousnesses inhabiting your body and

69:15

you would not have primary access to

69:17

them. Some people disagree, but that's

69:18

that's what I think. So I think that we

69:23

should take very seriously the idea that

69:26

certainly all kinds of other minimal

69:28

biologicals have some degree of I'm not

69:31

saying I'm not saying you know every

69:32

cell is sitting there having you know

69:33

hopes and dreams like we are but little

69:35

ones right little tiny ones that I think

69:37

I can say reasonably strongly the thing

69:40

that is a total conjecture is the

69:43

following something that I've said more

69:45

recently just this year I've started

69:46

talking about this notion of this

69:48

platonic space and if you want talk

69:50

about that. We can get into it. But I

69:52

think that in many ways all the things

69:55

that we are looking at, so bodies,

69:57

computers, robots, embryos, the biobots,

70:02

all of those things are in an important

70:04

sense thin client. They're front-end

70:07

interfaces for patterns, patterns of

70:10

behavior, patterns of information

70:13

processing, patterns of form, and so on.

70:15

For patterns that come from a different

70:17

space, they don't come from this

70:18

physical space. and we can dig into

70:20

that. If that's the case, then what you

70:22

could say is, and again, this is just

70:26

conjecturing here. I'm not saying this

70:27

is like useful in the lab yet or

70:29

anything like that. I like to keep those

70:30

things separate. But if you had to say

70:32

something about consciousness, what you

70:34

might say is that consciousness is the

70:37

point of view of the pattern projecting

70:39

into the physical space. In other words,

70:41

third person observable behavior problem

70:44

solving like normal science is what we

70:47

see with each other doing within the

70:48

space. But consciousness is the

70:50

viewpoint of the pattern that is

70:53

fundamentally like you and I on that

70:55

view and many other things are

70:57

fundamentally patterns that live in this

70:59

other space and we sometimes project

71:01

through various interfaces like physical

71:03

bodies, robots, androids you know what

71:05

whatever machines you know embryos we

71:08

sometimes project through these physical

71:09

interfaces and consciousness is what it

71:12

is like the experience that it is like

71:14

to be one of those patterns projecting

71:16

into space. That's that's one way you

71:18

might think about it.

71:19

>> Could you explain that again as if I'm a

71:22

like smart sixth grader very interested

71:24

in technical stuff?

71:27

And I suppose what I'm trying to

71:28

triangulate on is are you getting

71:32

into kind of Donald Hoffman territory of

71:36

sort of reality as user interface? I'd

71:38

love to hear you explain the other space

71:41

or like not coming from physical space

71:43

just maybe to to put it a different way.

71:45

>> Sure. Okay. Let's run through it. I

71:47

think Don's work is is very interesting.

71:49

For the purposes of what I'm about to

71:50

say, we don't need to worry about it.

71:52

Let's assume a perfectly conventional

71:53

physics. I think Don's on to something I

71:56

think for sure. But let's assume that we

71:57

don't need to worry about that. A

71:58

perfectly conventional physics. One

72:01

thing that scientists nowadays like is a

72:04

view called physicalism. Physicalism

72:06

says that look, there's only one realm

72:07

that we need to worry about. It's this

72:09

physical realm. Physics tells us

72:10

everything you need to know about this

72:12

realm and there it is. A lot of people

72:14

like that, but I actually think that

72:15

view is a non-starter for the following

72:17

reason. There are all kinds of important

72:20

facts that are simply not facts about

72:24

physics. They are not discovered by

72:25

physicists. They will never be

72:26

discovered by physicists. They are not

72:28

changed by anything we do in physics.

72:30

And those are certain facts of

72:31

mathematics. So for example, the exact

72:33

value of e the natural logarithm. The

72:36

fact that complex numbers behave

72:37

differently than quitterians that behave

72:40

differently than octonians under certain

72:42

rights like you know the truths of

72:44

number theory certain facts of topology

72:46

and the distribution of prime numbers.

72:47

You can't just dissolve the math

72:49

department and hope that don't worry the

72:51

physicists will figure out why this is

72:52

this is not what what they will ever do.

72:54

The math department does things that are

72:57

different and additive to what physics

72:59

does. And both in physics and biology

73:01

and I think in cognitive science too,

73:03

there's an interesting phenomenon which

73:04

is that if you're like a 5-year-old and

73:06

you do that thing where you keep asking

73:07

but why, right? So this is the but but

73:10

why if you keep asking but why long

73:12

enough eventually you always end up in

73:14

the math department. It's the damnest

73:16

thing. Like imagine cicas, right? They

73:17

come out every uh whatever 13 and 17

73:20

years or something they come out, right?

73:21

And you say, "Hey, hey, why is that?"

73:22

The biologist says, "Hey, why is that?"

73:24

Ah, because that way they don't time

73:25

their predators because if it was every

73:27

12 years then every two year, three

73:28

year, four year, six year predator would

73:30

would get you, right? So 13 and 17 like

73:32

that's cool. Why are those numbers so

73:34

special? They're prime numbers. But why

73:36

13 and 17? Why isn't there one in

73:38

between? Now you got to go to the math

73:39

department, right? Because they're the

73:41

only ones that understand why that is.

73:42

So it's like this with everything with

73:44

physics. You, you know, you keep

73:45

digging, but why do the firmians do this

73:46

or that? Oh, because this like amplitude

73:48

has like this symmetry group or

73:50

whatever. So, so there's something

73:52

interesting going on where even from the

73:53

basic most basic math that you learn in

73:55

high school up through these very

73:57

complicated things. There are a bunch of

73:58

facts that are simply not facts of

73:59

physics.

74:00

>> Yeah,

74:00

>> this I think is just how it is. Now,

74:02

from here you have a choice to make. You

74:05

could say, well, these are just random

74:09

regularities that are true in our world.

74:12

It's just a random grab bag of of

74:13

interesting things. Mathematicians don't

74:15

treat it that way, right? They think

74:17

it's an ordered structure space that

74:18

they are exploring. Especially plaintist

74:21

mathematicians think they are

74:22

discovering. They're not inventing that.

74:23

You don't have a choice. You start with

74:25

set theory. Eventually you find out the

74:26

value of PE. You didn't have a choice

74:28

about that. Like that's what you found

74:29

out. You discovered that. So I think

74:31

more optimistically that this is not a

74:33

random grabag of stuff. This is some

74:35

kind of structured space of patterns,

74:37

mathematical patterns. Now you can take

74:39

one other step and you say interesting.

74:41

How do we know that these patterns are

74:43

only of relevance to math? Is it

74:45

possible? Well, we know they're of

74:46

relevance to physics because they

74:48

constrain how physics go. What about

74:50

biology? Well, biology is interesting.

74:52

Imagine that there's some planet and on

74:54

this planet the highest fitness belongs

74:56

to a triangle of a very specific shape.

74:58

Okay, so here comes evolution and it

75:00

cranks a bunch of generations and it

75:01

finds the first angle. Cool. And it

75:03

cranks a bunch more generations. It

75:04

finds a second angle. Does it need to do

75:07

it again to find the third angle? Why

75:08

no? Because once you know two angles of

75:10

the triangle, you know the third one.

75:12

Why did evolution just get to save

75:14

one-third of the time that it would take

75:15

to figure this out? Why? You get a free

75:17

gift for mathematics. And so I think

75:20

that physics is what we call things that

75:22

are constrained by these patterns.

75:24

Biology are the things that are enabled

75:26

or facilitated by these patterns. I

75:27

think biology uses the hell out of these

75:29

things and we'll talk about what they

75:30

are momentarily. But now you say, okay,

75:32

so they're relevant in physics. They're

75:33

relevant in biology. What kinds of

75:35

patterns are there? Well, there are

75:36

passive things like the value of E and

75:38

some fractals and things like that. But

75:40

could it be that there are other

75:41

patterns in the space that look a lot

75:44

like things that are not studied by

75:45

mathematicians? Maybe they look a lot

75:47

like things that are studied by

75:48

behavioral scientists. Could they be

75:50

patterns that have some capacity for

75:53

memory or patterns that have capacity

75:55

for problem solving? Could they be

75:56

recognizable as kinds of minds? And so

75:59

maybe and so this is the kind of crazy

76:02

claim that I'm making. Maybe the

76:04

relationship between the mind and the

76:05

body is exactly the same relationship as

76:09

between the truths of mathematics and

76:10

physics. So this is an old idea. Decart

76:13

for example in the west is associated

76:14

with this that okay the mind is this

76:16

non-material thing somewhere and then of

76:18

course immediately you know the princes

76:20

of Bohemia and other people immediately

76:22

nailed him on this idea. Yeah but how

76:23

does the interaction happen? How do you

76:25

have a non-physical pattern making the

76:28

brain sort of dance like a puppet? Like

76:30

how, you know, energy conservation laws

76:32

like how how could that possibly work?

76:34

And I don't think he said this, and I

76:36

don't know why he didn't say this cuz he

76:38

was a mathematician. He could have said,

76:40

I think you already have this problem

76:42

since the time of Pythagoras. You have

76:44

this problem that you have these

76:45

immaterial truths of mathematics are

76:47

constraining the physics of our

76:48

universe. We already have this

76:50

interaction. This is not new. This isn't

76:51

right. This has been around for forever.

76:53

This is a kind of interaction where some

76:55

of these truths that come from a

76:56

different space of of facts absolutely

76:59

constrain and enable things that happen

77:00

in the physical world. So one thing you

77:03

might think about is whether some of

77:06

these patterns and we have right now if

77:08

anybody's interested I give you a link

77:09

to it. We're having this thing I

77:11

organized called the symposium on the

77:12

platonic space and we've got about 26

77:15

people. I I initially thought it was

77:16

going to be three people me and these

77:18

two other groups. They turned out

77:19

there's like 26 people who gave awesome

77:21

talks about this stuff talking about

77:23

this notion. I I think it's going to be

77:25

huge and I think it has all kinds of

77:28

very practical implications because what

77:31

do you get? Well, maybe you get static

77:33

patterns but maybe you get dynamic

77:36

patterns that are more like behavioral

77:37

policies or or even you know

77:40

competencies but maybe you also get

77:42

compute and if you get compute and we

77:44

can talk about this because we've

77:45

actually done some experiments on this.

77:47

If you actually get compute this way,

77:49

maybe the way we've been totally sort of

77:52

adding up the cost of computation isn't

77:54

right. Right? Because we've been looking

77:55

at the front end. And I actually think

77:57

this is what's happening here is that

77:59

the theories of computation that we have

78:01

are mostly about the front end interface

78:03

and they're kind of been neglecting some

78:05

stuff that happens on the back end. And

78:07

we've just begun. We published a couple

78:09

of things on it. There's lots more

78:10

coming. So I think that's an exciting

78:12

new area that that may have all kinds of

78:13

implications for cognition and and

78:15

behavioral science more generally.

78:17

>> All right. So people will definitely be

78:19

interested in the symposium on the

78:21

platonic space. So we'll include links

78:23

to that for sure. Separately, lots of

78:25

things I want to ask you offline but

78:28

that relate to this. But I will say just

78:31

a confession briefly which is one of my

78:33

biggest regrets is that in 10th grade I

78:36

and my brother had very different

78:38

experiences with math. I was very good

78:40

at math up to that point. My brother

78:41

also, he had a great math teacher in

78:44

10th grade. I had a really, let's call

78:46

her abusive teacher in 10th grade. I at

78:50

that point retired from mathematics. My

78:52

brother went on to get a PhD in

78:53

statistics and has done computer science

78:56

and data science. And it's to this day

78:58

one of my biggest regrets that I

79:00

stopped. But it's wild how these

79:02

inflection points. Same school, two

79:04

different teachers.

79:05

>> Amazing.

79:05

>> Yeah. So, never too late, I guess. go

79:08

pick up a textbook. I wanted to ask you

79:12

to expand on on the compute piece that

79:14

you alluded to at the end. Could you say

79:16

more about that?

79:17

>> There are two pieces to this that people

79:19

should know about. One is this idea

79:21

called polycomputing and this is

79:23

something that Josh Bongard and I and

79:25

his student who's now a posttock in my

79:27

group Atusa parsa has taken on and it's

79:29

this idea that when there's a physical

79:31

event something is physically happening

79:33

it might be a current going through a

79:35

logic gate in your computer or it might

79:37

be you know something else like that the

79:39

question of what is it actually

79:41

computing is in the eye of the beholder.

79:44

So multiple observers could be looking

79:46

at the same exact physical thing going

79:48

on and seeing different things being

79:50

computed. Okay? And yeah, I can go into

79:52

details, but I'll give you a very simple

79:54

example of this. And this is a paper

79:56

that my group put out about a year and a

79:58

half ago. There are these things called

80:00

sorting algorithms. And these are very

80:03

simple sets of rules. They're usually

80:05

about six lines of code, something like

80:07

that, that are designed to they're

80:09

they're recipes that you follow. It's an

80:10

algorithm. So you follow the steps. And

80:12

the idea is you're handed a list of

80:14

numbers and these numbers are all

80:15

jumbled up. They're out of order

80:17

randomly and the algorithm is designed

80:20

to sort them so that everything is

80:21

sorted. You might think of the way you

80:23

know if somebody gives you a bunch of

80:25

names and you need to do a phone book,

80:26

you want to put them alphabetical like

80:27

that or numbers, right? That kind of

80:29

thing. These sorting algorithms, they

80:30

have a couple of features. One feature

80:32

is that they're short. They're fully

80:34

deterministic, meaning that there's no

80:36

randomness. There's no question about

80:38

what to do. You just follow step by

80:39

step. That's it. Right? and people have

80:42

been studying them for about 80 years.

80:43

Every computer science 101 student has

80:46

had to deal with these sorting

80:47

algorithms. Okay. So, what we showed,

80:51

long story short, is that if you

80:54

actually watch what they're doing,

80:56

yeah, they're sorting numbers, but if

80:59

you watch carefully, and apparently

81:00

nobody has actually looked, and I think

81:02

this goes back to the thing I said

81:04

earlier, if you're completely convinced

81:06

that these things are dumb machines that

81:08

only do what you ask them to do, why

81:10

would you look at what else they're

81:11

doing while they're sorting? And that's

81:13

exactly this kind of thing where the

81:15

paradigm that you're using or the

81:16

formalism that you're using constrains

81:18

what experiments you do or what you can

81:20

see right like this matters. So if

81:22

you're not so sure as I wasn't that

81:24

these things are only doing what you ask

81:26

them to do. What you find is two general

81:28

classes of things. One is that the way

81:31

they do them has extra behavioral

81:34

competencies. Things like delay

81:36

gratification you know things that a

81:37

behavioral scientist would recognize

81:40

that you never coded in the algorithm.

81:42

you know because it's not some big hairy

81:44

like three billion parameter you know

81:46

neural net or whatever it's six lines of

81:48

code you can see all the code you know

81:50

what is there unlike biology there's

81:52

nothing there's no new mechanisms to be

81:54

discovered like there it is it's all

81:56

there that's that's why I picked it for

81:57

the shock value of exactly that that no

81:59

no one could say that well there's

82:00

probably some mechanism that you just

82:01

haven't found yet so that's the first

82:03

thing and the second thing is that while

82:05

they are sorting the numbers which of

82:07

course they do they are also doing some

82:10

other stuff that again you never ask

82:11

them to

82:12

And these other things, I've called them

82:14

like side quests. They're like these

82:16

little side quests. You can also call

82:18

them intrinsic motivations because like

82:20

with any system, like with a kid in

82:22

school, as you were saying, there's

82:23

things you force them to do. And then

82:25

within that, within the space in between

82:28

that, the time they have or whatever,

82:29

you get to find out what they really

82:31

want to do, right? If you don't overdo

82:32

it, if you if you give them a little bit

82:34

of room, you find out that but what is

82:36

the intrinsic? What is their, you know,

82:37

sort of inner nature or their, you know,

82:40

that kind of thing. So basically what we

82:42

found is that there is a simple minimal

82:44

version of that even in the most dumbest

82:48

fully deterministic.

82:50

This is nothing about determinism or

82:52

randomness or indeterminism. This is the

82:54

idea that our view of what an algorithm

82:57

is and how much of what the thing is

82:59

doing it captures is incomplete. It

83:01

captures very well the thing you asked

83:03

it to do but it does not provide a good

83:06

view of yeah but what else does it want

83:08

to do? And apparently in a very minimal

83:11

way even extremely simple systems have

83:14

this. So here's what it means. And there

83:16

was a cool Andrea Morris wrote a really

83:17

good story before Forbes about all this

83:19

that's like I think very generally

83:20

understandable. On my blog I have a

83:22

couple of pieces you know trying to

83:24

explain this in a very simple way. The

83:26

bottom line is this. One observer likes

83:28

the sorting and you pay for the steps of

83:30

the algorithm. Of course every step you

83:32

do you pay for it. So you pay for the

83:33

sorting but all the other stuff it's

83:35

doing that's all free because there are

83:37

no extra steps. you didn't have to do

83:39

the other steps. It does it while it's

83:41

doing the other thing. So if you had a

83:43

different observer that's interested in

83:44

the other thing, they got it for free.

83:46

And so now the question is how much of

83:48

that these I call them well this is a

83:50

word that exists ingressions, you know,

83:52

into the physical world of some of these

83:54

patterns like how many of them actually

83:56

are there and how much extra oomph do

84:00

you get when you don't know that that

84:03

you got it? And in some cases that might

84:05

be great because that might be

84:07

facilitating things you want to do. In

84:08

other cases you might have a machine

84:11

that has this going on where you don't

84:12

want that happening. You'd rather that

84:14

not be happening. And we have a very

84:16

active research program right now trying

84:17

to figure out basically better ways to

84:20

detect it, better ways to facilitate it

84:22

and ways to suppress it because there

84:23

will be situations where you don't want

84:24

this thing doing other stuff. So that's

84:26

the question like what are we getting?

84:28

Are we getting free compute here? Are we

84:30

getting something else? I don't we I'm

84:32

not even sure we have the vocabulary for

84:34

it yet because that's just not been the

84:36

way people have thought about these

84:37

things.

84:38

>> To dig a bit deeper on that as you

84:43

develop the vocabulary, the better

84:46

understanding of how to

84:49

measure, understand, inhibit or

84:51

facilitate this type of offging isn't

84:54

the right term, but

84:56

>> that's cool.

84:57

>> Sort of like secondary activities. Well,

85:00

I'm thinking of this technology. I think

85:01

it's called lamprey, which is this

85:03

device. It's a hardware device they

85:04

throw on long haul trailers and so on to

85:07

basically take the exhaust and convert

85:09

it into something useful, right? It's

85:11

not the best metaphor for what you're

85:13

mentioning, but as we flash forward 5

85:16

years or or however long it is, I mean,

85:18

compute is a very pressing problem,

85:20

right? So, there are tremendous

85:21

incentives.

85:22

>> Yep. If there were a pot of gold at the

85:23

end of the rainbow, so to speak, with

85:25

this, if it were even 5% possible that

85:30

the metas of the world and so on, would

85:33

need fewer fision, let alone like fusion

85:36

reactors to produce the power they need,

85:38

then this is of great commercial

85:41

interest, right?

85:41

>> Correct.

85:42

>> Intellectual certainly. What might, and

85:45

I know I'm asking for some real

85:47

speculative leaps here probably at this

85:49

point, but what might that look like in

85:52

the future for compute within just for

85:56

the time being, compute within the

85:58

context of like hyperscalers who are

85:59

like, okay, we need 20x the capacity of

86:02

the current power grid or whatever to do

86:04

what we want to do.

86:05

>> A couple things. So, first, this is very

86:07

late breaking stuff, so take everything

86:09

I say here with a grain of salt.

86:11

>> We'll see how it shakes out. But I think

86:13

you're right. I think this is going to

86:14

have massive implications. Oh, and first

86:16

of all, the offging actually thing is

86:18

important because one thing about that

86:19

metaphor, the lamprey metaphor is that

86:22

there is a main thing that it's doing

86:24

and then there are these side effects.

86:26

But what's interesting about

86:27

polycomputing is that you actually don't

86:28

know which is the main thing. So I look

86:31

at this and I say it's a sorting

86:32

algorithm and oh my god, it does this

86:34

other thing we call clustering. Aliens

86:37

come down, they look at and they go,

86:38

"Oh, that's a cool clustering

86:39

algorithm." Wait, it sorts too. Holy

86:41

crap. Right? So like it's important that

86:43

it's not obvious at all which is the

86:45

main thing. Right? Okay. So let's just

86:47

say we have a set of things that it

86:48

does. There's two possibilities how it

86:50

could come out. I think one possibility

86:51

is that multiple of these are useful as

86:55

they are and people can sort of siphon

86:58

off actionable information you know

87:01

valuable utility out of them how they

87:03

are. We're certainly investigating how

87:05

to do that. That's one possibility.

87:06

Another possibility is that there is the

87:10

thing you forced it to do, but there's

87:12

also a bunch of other stuff which is

87:14

much more whatever it quote unquote

87:17

wants to do. And that stuff may not

87:20

actually be what you ever wanted or

87:22

needed. In other words, there is no

87:24

guarantee, right? So, you know, you have

87:26

a student and you make them study, you

87:27

know, math or whatever, something

87:28

useful, you know, accounting, like you

87:30

got to get a job or whatever. And then,

87:32

well, in my spare time, you know, I I

87:33

make, I don't know, figurines or

87:35

something. And there is no guarantee

87:37

that this other thing is ever going to

87:38

be commercially valuable. It might be

87:40

really important in understanding the

87:42

true nature of what you have. But

87:44

there's no saying that whatever it

87:46

actually wants, we would find

87:48

commercially valuable, right? I don't

87:49

think you can guarantee that. I think

87:50

it's going to be a combination of both

87:52

of these things. But this latter thing

87:53

has an implication for AI. And the

87:55

implication is this that when we are

87:58

looking at a language model for example

88:00

and people are debating is it this is it

88:03

that I asked it how it was feeling and

88:04

it told me that it had an inner you know

88:06

world and all of this okay but what we

88:08

don't know is whether the talking right

88:10

the language use is at all related to

88:13

what the actual intelligence is in this

88:16

thing

88:16

>> maybe but I'll just say that in our

88:18

sorting algorithm the additional thing

88:20

it's doing is not sorting it's something

88:23

else So, it's entirely possible that in

88:26

these AIs, the thing we have forced them

88:30

to do, which is to talk, and the thing

88:31

that we're all obsessed about are the

88:33

things it says, could be a complete red

88:35

herring as far as what kind of

88:37

intelligence is actually there, what

88:38

does it want, how do we communicate with

88:40

it, like the verbal interface that we're

88:43

all sort of so glued in on might not be

88:45

the interesting part of that equation.

88:47

Yeah. And so that's my only thing is

88:49

that some of this may very well be

88:50

commercially viable but some of it may

88:52

have implications that are very

88:54

different that are not about the utility

88:56

of the compute but about teaching you

88:58

about what do you really have when you

88:59

have a system like that and I think

89:01

that's where a lot of surprises are

89:02

coming. Yeah, folks can go back and

89:04

watch Xmachina,

89:06

but uh I do want to ask you about sci-fi

89:09

in a moment and your most recommended

89:11

sort of sci-fi books or films,

89:13

favorites. But before we get there, you

89:15

know, this is me just ruminating and I'm

89:18

going to apologize in advance for for

89:20

anthropomorphizing, but thinking about

89:22

the the school child example, right?

89:25

Studying math or accounting and making

89:27

the figurines. I wonder if the quote

89:30

unquote unproductive

89:33

side activities in some cases might

89:35

prove to be really critical to the

89:39

forced function in the sense that that

89:41

student who's studying math

89:43

>> needs to let off some steam and do

89:45

something different in order to have the

89:48

endurance and periods of focus to

89:51

actually do the mathematics. So if you

89:53

split the baby and get rid of the

89:55

figurine, do you accidentally handicap

89:57

the main function at the same time? I

89:59

don't I don't know.

90:00

>> That's a great question and that is

90:01

exactly what we are studying like right

90:03

now. I have people working on the this

90:05

exact question and specifically what is

90:07

the relationship among the different

90:09

things that are happening here. Are they

90:10

living in completely parallel universes

90:12

such that they don't really touch each

90:14

other or are they entangled in a way

90:17

that when you mess with one you're going

90:18

to have implications somewhere else? We

90:20

we don't know. That's a that's a great

90:21

question. I I don't know the answer to

90:23

that. I guess

90:24

>> I'm tempted to chew on that word

90:25

entangled with you. This probably

90:27

another two-hour conversation. Sci-fi,

90:29

as I believe you do. I I just think it's

90:32

so powerful in so many ways. Do you have

90:34

any

90:36

books, movies, anything at all, essays

90:39

that are just favorites of yours or that

90:42

you recommend to students or friends?

90:44

>> Let's see. Well, I grew up on all like

90:47

classic sci-fi from the ' 50s, 60s, 70s,

90:50

that kind of stuff. So all the all the

90:52

favorites. One particular author that I

90:54

love is Lem. Stannislav Lem. L E M.

90:57

>> I've never read Lemm.

90:58

>> Oh, he's amazing. Solaris was his, but

91:01

also he has a ton of very humorous short

91:03

stories, like really funny stuff. I like

91:05

him a lot. He's a master of the absurd

91:07

and releasing the assumptions that we

91:09

all have in ways that kind of illustrate

91:11

how narrow thinking and things like that

91:13

is just beautiful. I'll give you two

91:14

short stories that I like. One is

91:16

They're Made of Meat by Terry Bison,

91:19

right? You know that one? Yeah, it's a

91:20

great one.

91:21

>> Yeah, that's a great one.

91:22

>> Very fast read for people.

91:23

>> Yeah. Yeah. Very fast read. It's like a

91:25

page and it just like Yeah. Reminds us

91:27

reminds us all how silly some of our

91:29

preconceptions are. There's another one

91:31

I like which I'm going to butcher it cuz

91:33

I use this example, but I'm sure I've

91:35

added on things that weren't really

91:37

there. I think it's The Fires Within by

91:39

Clark. The version that I have in my

91:40

head, which probably isn't really close,

91:42

is the following, but I think it's

91:43

valuable. Imagine there's some creatures

91:45

that live in the core of the Earth and

91:47

they come out to the surface. So they're

91:49

incredibly dense. They're hot. They're

91:50

incredibly dense. They use gamma rays,

91:52

you know, for vision, whatever. They

91:54

come up to the surface. What do they

91:55

see? Well, everything that we see here

91:57

that's solid is like a thin gas to them.

92:00

Like this isn't solid to them. They're

92:01

walking through. It's like walking

92:02

through a garden of, you know, smells

92:04

that you like you walk right through,

92:05

disturb everything. You don't even know

92:06

what's there. And one of them is the

92:08

scientist and he says, you know, there's

92:09

like this thin plasma around the surface

92:11

of our planet. And they go, "Yeah." He

92:13

says, "Yeah." And it's got little little

92:15

patterns in it. And I've been watching

92:17

these patterns with my instruments. And

92:19

these patterns, they almost look

92:20

agential. They almost look like they're

92:21

doing things. They almost look like they

92:23

have little lives. You know, they move

92:24

around and you know, well, how long do

92:26

these patterns stick together? Well,

92:27

about a hundred years. Ah, that's

92:29

stupid. And then nothing interesting can

92:30

happen like that. And I have a story on

92:32

my blog based around this. He says, uh,

92:34

you know, we are real beings. We are

92:37

real agents. We're physical agents.

92:39

Patterns in the gas can't be anything.

92:41

So, you get the idea. The point is that

92:43

even the distinction between an agent

92:45

and the patterns within their cognitive

92:47

system, right? Thoughts versus thinkers,

92:49

right, as as William James said, and

92:52

what's data and what's the machine. Like

92:54

all of this to me is a continuum, a very

92:55

observer dependent continuum. And you

92:57

can get there with a science fiction

92:59

story.

92:59

>> What fun. You mentioned the blog a few

93:01

times. You've got some great stuff on

93:02

the blog. I I've shared some of your

93:04

writing in my newsletter before,

93:06

specifically your advice to students.

93:08

>> Oh, thanks. which has some fantastic

93:10

advice in it. And for folks who are

93:13

listening, even if you are not in the

93:14

world of science and academia, there's a

93:17

lot in that piece that they can

93:19

recommend it. But where would you

93:21

suggest people start if they've enjoyed

93:23

this conversation within the landscape

93:26

of your blog? Are there one to three

93:28

articles you might suggest they start

93:29

with?

93:30

>> Yeah, I have like a starter pack article

93:32

and things like that. I can provide some

93:34

links for sure.

93:35

>> Great. Okay, we'll put those in the show

93:36

notes, folks, as per usual. You know,

93:39

we're going to lay on the plan cuz I

93:40

know you've got another engagement

93:42

coming up, but I'll tell you what, I'm

93:44

going to make it dealer's choice, but in

93:45

this case, you're the dealer. So, you

93:47

can pick which question you want to

93:48

tackle and then we'll wind up. But super

93:52

curious

93:54

what you picked up from the late Daniel

93:57

Dennett. I have a bunch of his books.

94:00

Really fascinating guy. Option number

94:02

two is this is a quote from the New

94:05

Yorker beast in 2021, but this is a

94:08

congratulatory toast from Clifford

94:10

Tabin, if I'm pronouncing that

94:11

correctly. Quote, "You're the most

94:12

likely to crash and burn and never be

94:14

heard from again. You're also the most

94:16

likely to do something really

94:17

fundamentally important that no one else

94:19

on Earth would have done that will

94:21

really change the field." So, I'm

94:22

curious about that first part,

94:24

especially most likely to crash and

94:26

burn, never be heard from again, and why

94:27

that hasn't happened. And I suppose last

94:30

and you can answer more than one of

94:31

these two, but if you could put a giant

94:33

billboard out in front of and this is

94:35

metaphorically speaking, right? Just to

94:36

get a message in front of a lot of

94:37

people in front of

94:40

departments of biology or just even more

94:44

broadly for lots of people to see and

94:46

understand

94:47

what that might be. So,

94:50

>> I'll leave it to you to pick how you

94:51

want to.

94:52

>> Yeah, that last one, you know, it's hard

94:53

cuz if there's just one billboard, I

94:55

don't know. There's a lot to choose

94:57

from. You can have more than one if you

94:58

want.

94:58

>> Well, yeah. I mean, that's basically the

95:01

blog and the website and everything, but

95:03

I'll say just a couple things about the

95:04

first two. I guess Dan was an amazing

95:06

person. We agreed on a lot. We disagreed

95:08

on a lot of stuff. He was always an

95:10

incredibly generous thinker. One of the

95:13

great things that he always insisted on

95:15

was Steel Manning.

95:16

>> Mhm.

95:16

>> This is the idea that if you're going to

95:18

shoot down somebody's viewpoint or

95:20

disagree with it, you first need to

95:22

articulate the absolute strongest

95:24

version of it that you can. Right. And

95:26

for people who don't have context, I

95:28

suppose we should just establish who Dan

95:30

Dennett was. How would you describe him

95:32

in brief? Philosopher, cognitive

95:34

scientist.

95:35

>> Yeah,

95:36

>> it's understatement.

95:38

>> He passed away I think in the last year

95:39

and before that I think he was widely

95:42

written about as maybe one of the most

95:43

important living philosophers today. I

95:45

think I've seen that.

95:46

>> And so yeah, he was a professor at Tufts

95:48

where I am and he was an incredible um

95:50

thinker and he wrote many many

95:51

interesting and popular books and and so

95:53

on. Yeah. So it's the opposite of straw

95:55

man. this idea that there's no point

95:57

critiquing a bad argument. You should be

95:59

critiquing the best possible version of

96:01

an argument that you can. And so I think

96:02

that's extremely valuable is to take the

96:04

view and understand it so thoroughly

96:06

that you can give it a really strong

96:08

defense and then if you want go back and

96:10

shoot it down after that. But but first

96:12

you got to do the first part.

96:13

>> Mhm.

96:13

>> I thought that was really really

96:15

important and you know and I guess I

96:16

guess the second part so Cliff Tabin is

96:18

a great scientist. He's a geneticist. He

96:20

was my PhD mentor. I did my PhD with him

96:23

at Harvard and uh yeah I mean I don't

96:26

know I'm you know getting old now

96:28

getting into retirement like at some

96:29

point we ought to call it which way it's

96:31

going to be. I don't remember how long

96:32

ago it was that he said it but you know

96:34

it could still happen. It could still

96:35

crash and burn I suppose. Why not?

96:38

>> Did he say that just because of an

96:39

intrinsic intensity that you have? What

96:41

what would lead him to say something

96:43

like that? I don't want to put words in

96:44

his mouth, but what I hear him saying is

96:46

that I mean I'm very strategic in what I

96:48

say when. But I don't really have a

96:50

filter on what I think.

96:51

>> No halfway measures.

96:53

>> Yeah. I'm just not very constrained as

96:55

far as what I'm willing to think and

96:56

eventually say if I think there's good

96:58

reason to say it. And I think that's

97:00

what he was talking about. That's a very

97:01

dangerous thing, right? Because let's

97:03

face it, in science, most of what we say

97:05

is wrong. And I'm clear on that with

97:08

people all the time. Like I'll say what

97:09

I think now and I'll say it as strongly

97:11

as I possibly can, but I'm under no

97:13

illusions that we have the right answer

97:15

to any of these extremely difficult

97:16

questions. So most of it is probably

97:18

wrong in some important way. And I think

97:20

he was just commenting on the fact that

97:22

I say a lot of things that are counter

97:25

paradigm and not in agreement with what

97:27

the mainstream thinks. Occasionally that

97:28

goes well. Usually that goes very

97:30

poorly, which is what I think he was he

97:31

was pointing at.

97:33

>> Mike, thank you so much for the time. I

97:35

really have had so much fun in this

97:37

conversation. And I want to make sure we

97:39

point people to the right places. I've

97:41

got a few websites in front of me here.

97:43

Thought forums.life, that's one. We've

97:46

got drdmicholven.org

97:49

as well. Are there any other websites or

97:52

profiles you'd like to point to? Are you

97:54

active on X or any other platforms?

97:56

>> Yeah, at DRMich 11 on X. The thought

97:59

forms.life is the blog. That's my

98:01

personal blog. So I say things there

98:02

that I wouldn't put on the website,

98:03

which is my official lab website. and

98:06

you can sign up for updates on the book

98:08

and all that kind of stuff. The

98:09

drmike1.org is the official lab website.

98:12

So that has all of our papers, all of

98:14

the software, you can download the data

98:16

sets, like all the stuff to back up all

98:18

these crazy things that I'm saying. All

98:19

of that is on drmichol.org. There are

98:22

also lists of books that I recommend to

98:24

my students and and things like that.

98:26

There is a YouTube channel which also

98:28

has some conversations. I've been for

98:30

the last five or six years I've been

98:31

hitting record on some meetings I've had

98:33

with some amazing people, you know, some

98:35

really interesting collaborators and and

98:37

all of that is there for you for you to

98:38

sort of be apply on the wall with. So

98:40

that's that's fun, too.

98:41

>> And the YouTube channel is linked to

98:42

from thought forms.life

98:44

>> probably. I'll send you the link. I

98:46

don't even know if I remember what

98:46

exactly the URL is. So no problem.

98:48

>> I'll send you the link.

98:49

>> Mike, thank you so much. I hope this is

98:52

>> not our last conversation. Absolutely.

98:54

For people listening or watching, we

98:57

will link to lots of things. Everything

99:00

that we can possibly link to from this

99:03

conversation and more at

99:05

tim.blog/mpodcast

99:06

as per usual. Just search Michael Leven

99:09

or probably Lean. I think you might be

99:10

the only Levan. Lev and it will pop

99:13

right up. So you'll have plenty of

99:15

resources to do more digging and more

99:18

thinking, more assumption testing,

99:20

assumption bending in a lot of ways. And

99:24

until next time, as always, be a bit

99:26

kinder than is necessary to others, but

99:28

also to yourself. Thanks for tuning in.

Interactive Summary

The video discusses the concept of bioelectricity and its role in biological processes, moving beyond the traditional view of DNA as the sole determinant of biological function. Dr. Michael Levin explains that bioelectricity, particularly developmental bioelectricity, governs how cells organize and form tissues and organs. He highlights that cancer, regeneration, and birth defects are all influenced by electrical signaling patterns, which act as a form of memory for cells. Levin contrasts this with the central dogma of molecular biology, suggesting that while DNA provides the 'hardware,' bioelectric patterns represent the 'software' that directs development and function. The discussion touches upon experiments with flatworms and other organisms to illustrate how manipulating bioelectric fields can lead to significant developmental changes, even across generations, without altering the underlying genetics. The conversation also explores the potential applications of this research in human health, including treating birth defects, promoting regeneration, and combating cancer. Furthermore, it delves into the implications for understanding aging, cognition, and even consciousness, proposing that intelligence and cognitive processes may be more fundamental and widespread than previously thought, existing beyond just complex organisms.

Suggested questions

7 ready-made prompts