Physicalism as such has been dead since
the time of Pythagoras and probably long
before that. The basement of all this, I
don't think it's math. I think it's a
behavioral science. I think math is a
behavioral science of a certain kind of
pattern. It's a bioelectric code because
literally they preede and actually
instruct the patterns of gene expression
and and cell behavior and morphagenesis.
If you want the the system to make an
eye on a tadpole's tail or or have a
flatworm that has two heads like this or
whatever, you have a chance to do that.
We're dealing with an aential material.
This is not passive matter.
>> Why assume there's this layer where
there's a structure that we can't quite
quite get at? I would say it sounds more
like we just haven't figured [ __ ] out in
this realm. Why assume there's another
one? Uh, so I make a crazy addition to
this that that I'm not sure anybody else
agrees with. My hypothesis is
Michael Lean and Earl Miller are
revolutionaries in the field of bio
electricity, just at radically different
scales of biology. and they've never
recorded a conversation together until
today. Leaven works on single cells earl
with the human brain and they arrive at
the same impossible idea from opposite
directions. Bioelect electrical waves
are the engine of life all the way down.
Memory, intelligence, goals, and
possibly consciousness are no longer
just properties of the brain. Please
remember to subscribe to the channel as
our focus is questions like how far down
does mind really go? And I'm not going
to stop standing on the shoulders of
giants like Earl and Levan until we
figure it out.
>> Go, hey, liver, like why, you know, how
are you doing? Why do I feel like crap?
And what's going on? And our group have
uh have made some some progress in
talking to things that normally don't
talk. You're not talking to genes.
You're not talking to individual cells.
You're talking to the system as a whole.
The system knows things that none of the
parts know. And and and you can talk
about very abstract things like eyes.
No, no individual cell knows what an eye
is, but but the system does. And the
deeper function is of a cognitive glue.
It is what allows the group of cells to
remember that hey we're making a face of
a particular type and to store memories
that no individual cell can access to
store goal states in spaces where none
of the molecular components can access
it.
>> You should have been around 2530 years
ago when you you study the stuff people
thought you were rubbing crystals
together or something like why the hell
would you study electric fields? It's
all magic. It's all [ __ ] When a
paradigm takes hold of a field people
resist it, resist it. Resist it. Resist
it. Kind of kind of what what's going on
now. This is like maybe the craziest
thing I'll say is that
Michael Lean Earl Miller, it is such a
true honor and pleasure to be able to
host you. Welcome back to the Giant
Show, your fan favorites. I'm so excited
for this. Welcome.
>> Thank you.
A really fascinating area of overlap in
your work seems to be bioelect
electricity and its role in driving the
organization and function of biological
systems. But what I'm curious to start
with is how you first got interested in
this question. Michael, maybe maybe
start with you. Where does that story
start for you?
Well, it depends where you want to start
counting. I mean, my earliest experience
is I was really little, probably five or
six, and my dad used to uh take the back
off of our TV set and it was this like,
you know, little screen in the front,
like this giant thing with a vacuum
tubes. And I remember staring at this
thinking and he explained that this was
not just random that all these pieces
are there. Somebody knew how to put all
them put all of them together. So, that
was amazing. And also the fact that the
stuff you see on the front of the TV
isn't in the back of the box, right? And
it's different every time, but the
hardware is always the same. So that was
like that that was a total you know uh
kind of mindbender and so I thought of
that for a really long time and then and
then of course in in biology with
neuroscience trying to understand uh how
how how living systems use those kind of
uh those kind of processes to store and
process information. So I was I always
thought there had to be some connection
and then somewhere around uh well it was
1986 I I happened upon Robert Becker's
multi electric and the best thing about
that book was the bibliography and he
had he had like references going back to
like amazing work right Burr and and and
people in the 40s and and and other
others later on to uh to point out that
yeah this was you know this was critical
for morphagenesis for regeneration for
embryo So, embryionic development. So,
that's that's that's where my interest
dates to.
>> I I'm also old enough to remember when
when TV sets used to have vacuum tubes
and when one would go bad, I'd go with
my dad to the drugstore and there'd be a
tube testing machine. You have to figure
out which tube went went bad and replace
it. But I think my my genesis ma mainly
begins with with with the electrical
environment of the brain begins I was a
physics major for a year when I was an
undergraduate and one of the biggest
mistakes of my career is not going
further in math. Now I got to work with
mathy people instead of doing it myself.
But when I first start got into
neuroscience first I was then I was a
premed major and I I volunteered to work
in a neuroscience laboratory so I can
get into medical school and then I
started doing experiments and fell in
love with the brain. Um, but it always
it I it always struck me that the
dominant view of the brain at the time
was this like telegraph system thing
where where all the neurons are all
wired together into a network and spikes
travel down lines down axons and
influence one another and you know all
the spiking and all this spiking by the
way for those of you who don't know what
spikes are bit of neuroscience jargon is
the is the brief electrical impulses
given off by individual neurons and the
idea back when I uh became a young
neuroscientist This was that and still
is the dominant idea today that the
brain works as like a giant telegraph
system and these little spikes of
neurons are like Morse code going down
the wires and only influencing other
neurons that are connected by the wires.
And I thought but this whole thing is an
electrical system and these spikes exist
in an electrical environment. That's how
that's how these electrical impulses are
generated by neurons having electrical
potential AC difference in electrical
potential across their membrane. I
thought is shouldn't this contribute to
an overall larger electrical
environment? Should shouldn't that play
a role? And when I was a graduate
student, like if you wanted to study
things like um electrical fields or
local field potentials, as we call them
when we're inside the brain, you want to
study electrical fields, people thought,
well, why do you want to do that for?
That's crazy. You could study the
neurons. Why why do you need the uh the
bigger the big stuff is just the humming
of the car engine or the fumes given off
by by the engineer don't really do
anything. And right from the beginning,
I thought that was crazy because this
whole brain exists in this this
electrical environment. How can one
thing be important, then the other thing
not not be important? So, I've been sort
of chipping at this for a long time.
Back when it was very unfashionable, and
now that it's it's becoming more
fashionable, I'm I'm I'm uh if that's
the right word, I'm happy to say.
>> Very cool. Vacuum tubes seem to have
influenced both of you a lot more than
me. I would I wonder why that is. I am
no very cool to hear that background.
Well,
so I was gonna tell you the one of the
biggest influences on me very early on
is I for my dissertation I did one of
the me and a posttock did one of the
first multiple electrode recordings in
visual cortex right and we want we were
using four electrodes wow big deal four
whole electrodes now we use hundreds
electrodes but back then it was four and
we had to sort all the waveforms from
the electrodes into single neuron traces
the state-of-the-art computer we had at
the time was a PDB11 computer in our
laboratory library where at Charlie
Gross's lab where I was a graduate
student and we we want to do a principal
components analysis of the waveforms to
sort them into meaningful signals and
the PD11 computer at the time was wasn't
powerful enough to do it. It would take
him two weeks to to run through the
analysis for a two-hour session, right?
But there was this guy at the University
of Pennsylvania named George Gerson, an
electrical engineer/neuroscientist
who built a waveform principal
components analysis system based on all
analog technology like the kind of uh
technology that Mike has in the receiver
behind him. He did it all with just just
capacitors and resistors and and a
waveform generator and B. So it was an
all analog um principal components
analysis um waveform analysis that was
more powerful than the most powerful
digital computer of the day and that
made a huge impression on me. I think
that's been echoing in my brain ever
since.
>> Extremely cool. Very very cool. Mike,
what's the best way for the listener to
understand the bioelectric code? What's
the right way for the listener to think
about this?
So there's a couple of different levels
that you can appreciate this on. One is
simply the fact that if you look at and
and we'll I'm going to tell a a sort of
particle story and then we could talk
about the field aspect because I think
you know that the field aspect is is
really critical and under
underappreciated but just the part that
we've that we know right now is is
easier to tell is the is the particle
view which is that uh if you if you look
at uh tissues what you see and these are
active tissues during embryionic
development during regeneration during
remodeling repair resistance to aging
resistance to cancer. Any of these
scenarios where large groups of cells
have to be have to cooperate in some way
to build some sort of higher level build
or repair some sort of higher level
structure. When you look at these
things, what you find is that there are
very slowly changing uh patterns,
spatial temporal patterns of resting
potential. So basically, as far as I'm
could as far as I could tell and and and
obviously Earl is more of an expert on
neuroscience than me, but I I believe
that almost anything you get out of a
neuroscience textbook, you can apply to
these examples if you slow down the time
constant and you don't, you know, sort
of or assume that they're neurons, you
just say cells. And in fact, I've had my
students, and now we have a um a a
system online that actually does this
for you. You paste it, you put in a
neuroscience paper, and it changes
neuron. It replaces the word neuron with
the word cell. and then it replaces
millisecond with minutes and you get
yourself to develop biology paper.
There's a few other things you can you
you would change but but the mapping is
like really strong. So the point the
point is that all of these tissues have
this very active biological dimension
that you can measure and uh and more
importantly when you go in and we made
the first molecular tools to read and
write that that information. when you go
in and you change those patterns, what
you find is that uh you then have a
control over the resulting anatomy that
you that you built. So in other words,
if you want the the system to make an
eye on a on a tadpole's tail or or have
a flatworm that has two heads like this
or whatever, you have a chance to do
that. So, so what you learn from this is
two things that there is a it's a
bi-electric code because literally the
patterns that you see encode they preede
and actually instruct the patterns of
gene expression and and cell behavior
and morphagenesis that you're going to
see later on. That's a b you can decode
them. So at least to some extent and we
only scratch the surface but to some
extent you can look at the pattern you
say oh well I know what this thing is
trying to build and then you can
actually change the code and have it
build something completely different.
So, so, so you learn that. You learn
that the biomectric pattern is is
encoding outcomes. You learn that it's
instructive because if you change the
pattern, the system happily builds
something else. You learn that it's a
very top- down control, which I think
has real implications uh for for
neuroscience, which is that much like
when I'm talking to you now, I don't
need to worry about your synaptic
proteins and how you're going to listen
to what I'm saying. You you will take
care of all that. We have this very high
level interface and your own multiscale
architecture will take care of all the
chemistry that that downstream of that.
The same thing is true here. When we
tell the tadpole, a region of the
tadpole to make an eye or make an extra
brain. We don't I I don't know how to
make an eye. I don't know how to control
all the stem cells and all the gene
expression that has to change. We don't
touch any of that. We give a very high
level prompt that says make an eye. And
much like your highle goals that
ultimately change the chemistry of your
muscle cells so that you can get up and
execute on those goals in voluntary
motion, all of this gets filtered down
so that once once I get the tissues buy
in that yes, in fact, we are going to
make this eye versus whatever the
pattern was previously, everything else
falls into place, right? All the all the
uh the the the gene expression, the cell
movements, everything else is taken care
of because we're dealing with an aential
material. This is not passive matter
here. So, so that's that's the third
thing you learn is that you can talk to
um the collective. You're not talking to
genes. You're not talking to individual
cells. You're talking to the system as a
whole. The system knows things that none
of the parts know. And and and you can
talk about very abstract things like
eyes. No, no individual cell knows what
an eye is, but but the system does. And
we can we can we can communicate with it
in that way. So, so that's that's kind
of the first and the most basic thing.
And the only thing I'll I'll add to that
is that I think there's a deeper
function to biomectrics. Like it's it's
beyond this. And the deeper function is
of a cognitive glue. And what I mean by
that is that it's a set of mechanisms
that allows the system to be more than
the sum of its parts. It is what allows
the group of cells to remember that hey
we're making a face of a particular type
and to store memories that no individual
cell can access to store goal states in
spaces where none of the molecular
components can access it. So once you
have this and there are other kinds of
biomectric move but but it's it's a
really powerful one that allows you to
align your parts to common purpose in a
space that the parts don't know anything
about and that's why cellular
collectives get to solve problems in
anatomical space and to navigate
anatomical space just like and I'm will
tell you more but but I think this is
very parallel to what happens in in in
the brain where you get to navigate this
this 3D world so adaptedly in ways that
individual neurons don't know about
because this electrphysiology ology
allow allows the scale this you know the
scale up of goal directed so that's
that's what I think we're talking about
>> brief intermission from standing on the
shoulders of giants for today's sponsor
Jenny one of the topics that we cover in
this episode is brain architecture how
the way the brain and biology is
structured and wired really informs how
a system can be made and what it's good
at this is a weirdly good way to think
about Jenny Jenny is built the other way
around to typical LLMs It's an academic
writing tool that grounds everything
directly in your sources, the PDFs you
upload your zotterero or mendlay
library. So when it helps you draft,
it's pulling from research you've
actually vetted. And if you want to
search beyond your library, it also only
pulls from academic work. All of it
traceable back to the exact source. So
no hallucinations, which is always the
biggest fear with Chat GPT or Claude.
One of my favorite things about Jenny is
the reviews feature. For example, you
write your section, run the review, and
claim confidence checks your claims
against the sources that you've already
cited. While peer review simulates an
actual academic review, it marks your
weaknesses, your strengths, and it
scores your draft. Jenny won't think for
you, but it will take all the friction
out of organizing sources and checking
citations so you can spend your
cognitive energy on what actually
matters. Try it for free at jenny.ai.
There is a link in the description with
a special discount code for Giant
Shoulders listeners only. I cannot
recommend them more. Now, back to the
episode.
>> Awesome. It really was grill. I mean,
that's
Oh, it's amazing how parallel that our
thinking is just different systems. I
mean like um what I approach I approach
it I've always been interested in in top
down control, executive control and
that's what our brains are doing which
are special. We we can take control of
our own brains. You could build a a very
complex feed forward network that'll
just respond to inputs in a mind
mindless way. That's not what our brains
are doing. Our brains are taking charge
of our own thoughts. We're controlling
our own processing. We decide what we're
going to pay attention to. We process
it. We respond accordingly. That's our
brains somehow imposing a
self-organization on themselves along
the lines of of our goals and our
knowledge and stuff like that. So, how
does the brain do that? Well, your
cortex contains 20 billion neurons and
10 to the 14 synaptic connections. Under
the the dominant the 20th century model
of all the telegraph system, which I
call connectionism, which is still kind
of, you know, very dominant today. Under
the connectionism model, how do you
produce that kind of top- down control
where where the system can control
itself? Well, under connectionism, the
way of thinking about the brain where is
this all about spikes and how in
synaptic connections somehow the system
has to get control of like 10 to the 14
a gazillion synapses and somehow control
them like a puppet master that just
seems in computationally intractable.
It's just just way too complicated. But
however what you have this when you have
an electric field a fluctuation electric
field and we look at oscilly dynamics
when you have an oscillating electric
field as soon as you have that wave
function you know with a very simple um
equation you know the action of the wave
at a distance and and a time into the
future and that's consistency that's
organization and that's just the kind of
thing evolution is is is going is is
going to take advant advantage of right
it's a way of the system to
self-organize itself so now so you think
about what what um a lot of people are
looking at now in system neuroscience.
Now we have the ability to record from
hundreds or thousands of neurons
simultaneously instead of one at a time
like when I was a young graduate student
and you look across whether you record
from 100, 200 or a thousand neurons what
you do is you do something called
subspace coding. So if I take a thousand
neurons, I can create a thousand
dimensional spa space to explain what
those thousand neurons are doing with
each neuron being a vector and the and
the spike rate, the amount of activity
along the vector. That's just
self-evident, self-descriptive. But then
what we do is we ask ourselves, how many
dimensions do I actually need to explain
what these thousand patterns of spiking
from a thousand neurons are doing? And
the answer is you can explain it in
three dimensions. So what that means, a
spiking cortex in the shared variant
sense isn't like insects buzzing
chaotically in the night sky. They're
like birds flocking together, right?
That's what that's what your cortex is
doing constantly. It's it's highly
highly coordinated. Well, to my mind, I
mean, the best way to produce that kind
of great way to produce that kind of
coordination is throw an oscillating
wave wave, electrical wave across a
network of neurons, and you'll get that
automatic, you'll get that flocking of
of activity. So, I think of it as very
early on, as as Mike pointed out, you
know, the biology is electric. Um, very
early on, nervous systems are full of
recurrent connections both on short
scale, long scale. probably very early
in evolution, simple nervous systems
naturally began to oscillate because
that's what recurrent electrical systems
will do is begin to oscillate. But that
oscillation that's just a simple
oscillation, but it's still it's an it's
it's an internal organization. So before
that, you know, you can imagine a
creature to just swim towards food or
swim away from light. Uh that's like a
thermostat just reacting to the outside
environment. But the moment nervous
systems got complex enough and probably
very early in evolution got cut to begin
oscillating on their own right there
that that's an organization. It's a dumb
organization is a simple simple oscilly
dynamic. But now the system is providing
its own internal organization and that's
something evolution could have and
likely took advantage of find ways to
push around those dynamics and now we
have a system that can control itself in
a computationally tractable way.
You know, one one one thing that's one
thing that's worth noting is that uh
it's so you know, so so Earl and I are
very much on the same page about all of
these things and and the symmetry
between these fields, but it's a really
controversial point. Um, you know, we've
we've we've been used to in in
developmental and cell biology this idea
that we have to have models of that that
that good models in those fields are
models of subunits that don't know
anything. They don't have any goals and
uh they're just mechanical outcomes of
of low-level chemical phenomena. And
that's that's how it's got to be. And
this this idea that um you know when
when I start to talk about goal states
you know and not not high order like
human level goals but just you know
cybernetic kinds of things right like
your thermostat has and so on start
talking about goal states and navigation
of spaces and memory and learning and
these things these other systems
people's people people initially um are
very suspicious about it and they say
well you're trying to smuggle uh some
sort of you know mysterionism into the
process magic the magic right but I'm
like look guys there's a whole
department here that we had a
neuroscience department here where where
this kind of high level control is not
magic there's I don't know how many
people you know and and and and people
like Earl who actually study this stuff
what is what is the problem with testing
out tools that are working so well in
other fields and by the way uh in our
field using exactly the same mechanisms
it's the same stuff it's the ion
channels it's the electrical synapses
it's the neurotransmitters even even the
even the mechanisms are concerned Right.
So, so there is this this weird notion
that on the one hand there isn't any
magic and the neuroscientists will
figure out what's going on in the brain.
On the other hand, the minute you start
talking about um you know learning
memory um goal directedness, high-end go
highend intent and all of that
immediately it's it's you know it starts
to seem like you're you're trying to um
trying to opuscate it. So it's a very
it's a very weird schism I think. Oh,
you should have been around 25, 30 years
ago when you you study this stuff.
People thought you were rubbing crystals
together or something. Like, why the
hell would you study electric fields?
It's all magic. It's all [ __ ] It's
just just it's just a it's just just
kicking the can down the road by
invoking magic. And I mean, basic
electromagnet magnetic principles are
not magic. It's the way the electro
electrical system works. It's not magic
at all. There's nothing magical about
it. In fact, if the if these electrical
field oscillations, waves didn't turn
around and it's a birectional influence.
It has to be. It's the way spikes are
generated. And when spikes generate
change in the electrical fields, they
change the electrical environment around
them and that's going to influence
spiking. In fact, if it didn't work that
way, you'd have to come up with a pretty
good explanation on why it didn't. Well,
does the brain have some sort of Bose
noise cancelling system to cancel all
these electrical influences? Of course,
it it has it has has to work that way.
And because that's just basic theory,
but besides that, I have the philosophy
that first of all, first of all, a lot
of this stuff is just pure Thomas [ __ ]
you know, uh, structure scientific
revolutions. When a paradigm takes hold
of a field, people resist it, resist it,
resist it, resist it, even when it's
obvious it's reaches it expiration date.
And that's kind of kind of what what's
going on now is people are just
resisting for for no because they're
used to doing it one way, right? But
this is not the way. That's not the way
I my philosophy is is two things. One
thing is I'm not going to tell the brain
how it works. I'm going to listen to the
brain. Let me tell how it works, you
know. And secondly, if I can make sense
of these so-called magical electric
signals, electric fields and stuff like
that, local field potentials, things
beyond the spiking, if I can make sense
of them with my crude technology of of
electrophysiology here in the 21st
century, I'm guessing four billion years
of evolution found a way to make use of
it, too.
Yeah, very well said. You know,
interestingly, uh, one, one of my, uh,
favorite, uh, the one one of my
scientific heroes, um, this guy Harold
Harold X Burr, who was working late 30s
all through the 40s, early 50s. Um, he
was he was incredibly precient about
this stuff and and he, you know, he was
an early neuroscientist. He he basically
really didn't have any tools other than
the first good voltmeter, you know, that
and he went around measuring all sorts
of stuff. And one of the things he was
very clear about in his in his book was
the importance of the field aspect of
all of this. So he talked very he talked
a lot about the the symmetry between the
field descriptions and the particle
descriptions and the idea that yeah they
obviously have to work together and and
go back and forth. But the two the two
uh ways of looking at a system have very
different implications in terms of what
how you understand it, what experiments
do next and and so on. And so he was he
was he he foresaw a ton of the stuff
that we've since actually discovered.
But he specifically claimed that until
we deal with the field aspect that bio
electricity whether developmental or
neural that bio electricity was not
going to reach its full um potential no
pun intended uh until we really get hold
of the field aspect not just not just
the particle descriptions but the but
the field aspect.
>> Yeah. and Donald Heb and Walter Freeman,
classic neuroscientists from the 30s and
40s, they talked about um electric
fields and effect coupling all the stuff
we're talking about now, which goes to
show you there's no new ideas, just old
ideas rediscovered.
>> So over and advanced on Yeah. Oh, to
totally Yeah. So, but I'm curious is
there is there any fundamental
differences that we can point to towards
bioelectric codes happening at the level
of the plenarian flatworm regrowing a
second head and a human brain choosing
between eating a sandwich or a chocolate
bar for lunch? Is that the same
bioelectric codes operating at different
scales or is there something we can
actually point to that is fundamentally
different at those two levels?
Well, I would say that it's the it's the
electric fields, the biomectric fields
that are providing organization. And
that that doesn't mean by code. I would
say there's there's no difference
between what uh um structural biology
and and what what the brain is doing in
terms of his daily operations. Now, Mike
probably has more of a um insight into
how a plan how a a worm grows a second
head than I I do, but it seems to me
that what we're both talking about is
there's this higher emergent level um of
electricity where where that's providing
an organization that's sorely needed by
both the developing body and the
operating brain.
>> Yeah, I mean there's there are a lot of
things that are very much the same and
then there's a couple of things where we
can start to look for differences. So,
so things that are the same are
the the underlying machinery is
basically the same. So, ion channels,
electrical synapses, neurotransmitters
downstream of that, you know, gene
expression downstream of that. So, so
that that's all that's all there. The
use of these things to navigate a
problem space. So, in in in your
sandwich case, you know, the
threedimensional world, in our case, the
anatomical morph space, right? So the
space of all possible anatomical
configurations and you have to get from
here to there or you have to maintain
your spog where things are trying to you
know push you over and so on. So so
that's all the same the ability to form
recall and uh uh interpret flexibly
interpret memories that's that's all
there. The things that are that are
perhaps a little bit or or that may be
different one is that uh in in in our
system you get a lot of long range um
communication. So, so everything is
connected. Things I do on one end of the
embryo has implications for whether
there's a tumor or a birth defect or
something else like way over on the
other side. So, everything is non-local.
But what you don't see as much, a little
bit, but but not certainly not like
neuroscience is um long range
pointto-point connections. So, very
specific um you know uh I'm going to
connect this point to this point and and
there's something special about this
path. We we don't see as much of that.
It's generally like there's a lot more
um sort of topological connectivity
there. Um although maybe even my picture
of this is not you know Earl can can
correct me on some of that stuff but but
you don't see these extremely long range
connections. And then and then the other
thing is I just want to sort of be very
clear that when when I say that that
aspects of neuroscience apply here this
is not a philosophical or a linguistic
point. I'm not like redefining
intelligence to be complexity or
something like this. I'm making a very
specific experimental point that we can
take these things and use them to do
better experiments and to reach new
competencies in in in morphagenesis. And
so so we've done all we've we've seen
all kinds of stuff, but for example, one
thing we haven't seen and I'm not saying
it's not there. I'm just saying there's
no evidence for it and we haven't seen
it. For example, proper um generative
grammar language. Okay. So, so I see our
system doing a lot of things that brains
do, but we haven't seen them do anything
that I would claim as actual language.
So, we haven't seen that. We've seen
tons of other stuff that you would find
in a neuroscience textbook. If you just
pivot the, you know, pivot the time
scale and the and the and the space from
from 3D space to amorphous space, you
see a lot of it. But there are some
things you don't see. And clearly
there's some reason why, aside from the
time speed up, there's some reason why a
lot of creatures, certainly not all, but
a lot of creatures have neurons and
brains. And so my guess is that they do
that you don't eat milk and so on. There
are going to be some differences but but
massive overlaps.
>> Yeah, more overlapping differences for
sure. And you know the pointto-oint
connecting the way I think about the um
cortex is that all these the traditional
way of thinking of the telegraph system
with the changing synapses and you
change the strength of connections and
therefore you get that that's that which
is the way AI the kind of model that AI
is based on these large language models
all about associations and connections.
I think that's a great way of storing
information. But when it comes to
expressing information, express that
that's where these these fluctuating
electrical fields make make a make a big
difference. They help the brain do
computation. The current the traditional
view of neuroscience is the
connectionism is used for both storing
information and also for computing. And
this is a little bit overly divisive and
and simplistic, but I would say at least
cortex anyway, the this memory storage
is is taking place through the
connections. And it's the it's these
electric field effects where a lot of of
high level computation is going on. So
that's a whole different level of you
know an of computation that that's
largely been ignored by the field. But
I'm happy to say now that the the field
is way more accepting of it than it was
2530 years ago. More people are doing
that kind of work where back in the day
it was just a few of us.
>> So So Earl, um could I follow up on
that? Uh you know you're talking about
storing information. could you give your
view of what you think the basis of of
memory storage is in the you know
classically in in the brain and then and
then I'm going to ask you um for you
know uh how how that might fit with some
some of the sort of weird edge cases
that that that we see.
>> Huh. Well, I think you know in in terms
of um you know I keep saying the 20th
century connectionism models if in a
dismissive way. I don't mean to be
dismissive. A lot of the brain does work
like that and that was certainly
foundational. So you don't get to where
we are now without going through a
building up the system first cataloging
it and fig figuring out the b b b b b b
b b b b b b b b b b b b b b b b b b b b
b b b b b b b b b b b b b b basic units
of it. But I think in terms of memory
storage, that's where that's where the
the connectedness model is is largely
correct. You the brain stores
information via synaptic plasticity
changing changing the weights of
connections to neurons and the
information get baked into the brain
that way. What I'm saying is that then
when when actual thought comes along,
that's not storing information. That's
expressing information. That's
expressing thoughts. And when that
happens, that's where these electric
fields and these these these emerging
dynamics really begin to play a major
role because that that's where the
that's how lot the computation takes
place. And do you think that uh does
that so so the standard view of of the
synaptic plasticity and all that does
that require for these structures to
hang around you know people have
memories that are 80 years old and
things like this. How much how much of
an issue is the is the constant sort of
turnover and of all of this stuff?
>> Well as long as you maintain long as the
turnover maintains the existing patterns
of connections. You could just replicate
the whole brain have a completely
different brain you know next week and
it'll also operate the same. Um, you
know, so that that's a challenge. You
gota either your whole body is, you
know, constantly regenerating itself for
the most part. Um, but you just got to
maintain the pattern of of connectivity.
But you're about to tell me I suspect
some alternative view of memory that
doesn't that that can accommodate this
massive turnover. So go ahead. Well, I
don't I don't I'm not going to try to
float a different view of memory,
although I I have kind of I have some
doubts about the about this the standard
story, but but but there are a few um
there are a few, you know, cases that
that I think are are interesting in in
that respect. Uh first of all, for
example, the scenarios of um memory
transfer so memory transplant. So, so in
a variety of systems and and some of it
was done for simple memories, it was
even done in rats, but but most of it is
in sort of the lower so-called lower
life forms. You get these you get these
experiments where either cells or
oftentimes uh extracts. So, um either
brain extracts or in some cases audio
and brain extracts are moved from a
trained uh from a trained donor to a to
a recipient and there's some degree of
uh some degree of of recall that that
sort of follows along. Right? people
have argued for some sort of a chemical
medium for memory that then has to be
read out by the nervous system. Like
that's one example we've we've done when
and like McConnell found this in the
60s, but we've done this work in
pleneria where you can just you train
them and you drop off the whole head and
they regrow a brand new brain from
scratch, a brand new centralized brain
from scratch and then they have recall
of the original information. So you can
see information then moving we can't say
it's non neuronal but but you can it you
have to move on to the new brain. or you
know those kinds of of things, you know.
>> Yeah, I know. Yeah. And and you're
right. I the thing to remember is that
you know um living organisms had
behaviors and can learn and stuff even
before we had nervous systems. You know
like uh you can do simple classical
Pavlovian conditioning just in single
cells with just chemical gradients. You
don't need a nervous system to do that.
Nervous systems are a way of expanding
on that basic repertoire and getting
more and more more complex behavior. So
yeah, so I I that simple simple memory
simple behaviors can be transferred in a
chemical way. But when I'm talking about
memory, I'm talking about like you know
higher forms of memory like like recall
and and a personal recall and stuff like
that and our knowledge baked in. I think
that's that's got to be somehow involved
in in the patterns of connections
between the um neurons because we know
that they change as a result result of
learning because what I'm saying is
that's that's a great way to to um store
things but not a great way to do to do
uh computation. But yeah, I mean there
memory and learning predates nervous
system. So so some stuff on chemical
transfer from one trained animal to
another does doesn't surprise me that
much although it's beautiful elegant
work. And what do you think about um we
we just we just did a review of clinical
cases where uh you've got humans with
normal in some cases above normal you
know intelligence performance level and
radically reduced brain function. There
are some crazy cases where somebody's
you know somebody's shown a finally you
know an MRI or half their brain's
missing. Yeah. Yeah. Like
>> or much more right there are these very
hydro hydrophille cases and so on like
what what do you think is going on
there? Well, I think brains are very
very plastic and especially if the if
the disruption happens very early in
development, brains can accom can
develop around that and accommodate a
lot. If you see that kind of um damage
or change in an adult brain, you're not
going to get a you're not going to get
normal function. No way. This these the
cases you're talking about tend to
happen when when the when the when the
change happens very early in development
and brains can sort of um develop
develop around it. But you're not going
to see that kind of thing in the adult
brain.
Yeah. Uh yeah, I I mean I'm with you
that the plasticity must be important in
some way. But but what I find curious
about that is that I I my my
understanding is that we are under
considerable um selection not to have
very large heads because of the birth
canal and all these things, right? So
there's some pressure on not having a
giant brain. So, it seems to me that if
if there was a way to get the same level
of performance out of far reduced that
like that's what's weird to me is that
is that some some of these cases you get
you get very normal performance out of a
much reduced uh amount of brain mass.
And it would seem to me that that
evolution would be under and not not to
mention the energetics of the brain,
right? And how much of our energy budget
it's it's chewing up. Like it seems like
if if there was a way to get all of that
with a much smaller brain, it seems like
there's a lot of pressure to do that.
It's just
>> Sure. But but Mike, I think every time
this happens, you don't end up with a
person who grows up to be a normal
adult. A lot of times it happens, the
person ends up with severe learning
disabilities or dead, you know. So
you're hearing about the few successes,
not about not about the a case. Um,
first of all, and second of all, the
thing about the plasticity of turnover,
one thing that I always kind of remind
myself is there's something in the brain
called multiple realization. So for
example, I think it was Eve Martyr at
Brandeise, she studied lobster pyloric
motion. that's um in the lobster gut and
that's just moving food down the
intestine. simple simple simple um
action and that's so just like squeezing
a tube and that that's explained three
three ganglion the neuron in the lobster
gut are responsible for this and I think
you found that you can I maybe get the
exact numbers wrong here but you found
that there's computation there's a
100,000 different ways you could tune up
these three ganglion and get the exact
same pyloric motion so there's lots of
ways in the details to do the same
things in our brains even so even with
this turn over you form a memory and
this these this synapse for example then
it turns over. Well, there's lots of
other synapses and lots of redundancy,
but it but the again the exact details
don't matter so much as the principles
that that arise out of it. So, you know,
you could there you have a 100,000
lobsters all with their ganglion tuned
up 100,000 different ways to get the get
the exact same function. In a lot of
ways, that's where our brains are. Like
you you're born with your quartz at your
bowl with way too many more connections
than you're going to have. What happens
when you're a young infant is the
connections get winnowed away and you're
left behind with connections that
represent the reflect the statistical
regularities of the world like corners
and edges and stuff like that. So you're
born with way you're born with a random
seed of way too many connections and
your own individual experiences will
window those down to fewer number of
connections that have some meaning of
the structure of of the world. Well,
that means all of us are born with
slightly different brains,
different brains in the details and we
all have different experiences. So, all
of our brains are wired up in in a in
the details in a different way, but they
all operate on on the same principles.
And if you think about in terms of less
about exact anatomy than the principles
that arise out of the way the anatomy
interacts with these with these uh
emergent properties, then the the the
details um um and the turnover and
things like adaptability seem like less
of a problem.
If you're anything like me, then you
thought that skin irritation was just a
part of shaving. Two out of three men
believe this. And it's because it's what
these multi- head cartridges do. They're
built for flexibility, but that flexing
creates uncontrollable movement on the
skin, which causes deep damage and
irritation. Well, Hensen shaving did the
science. And you know, on this channel,
we love a good experiment. 59 volunteers
shaved one side of their face with a
Hensen and the other shaved using a
multi-blade cartridge. They then used
multisspectral near infrared
spectroscopy. This is the same imaging
tech they use in neuroscience research.
Quite shockingly, they found that the
multi-blade cartridges that are
advertised to us constantly did 50% more
skin damage than the Hensen blade. Why
is this, you ask? Well, the team behind
the Hensen are literally aerospace
engineers. They build parts for the
International Space Station and the Exom
Mars rover. This razor is machined to
tolerances of 5 10,000th of an inch. The
blade sticks out less than a width of
hair. 30° fixed angle, no wobble, no
flex. When the blade can't vibrate, it
can't flex and cause that inflammatory
cascade in the skin. It's metal. It has
a really premium feel and honestly when
I use it I feel like James Bond in that
scene where Money Penny is using the
cutthroat razor just has a really nice
feel to it as opposed to these garbage
plastic ones that I've been using my
whole life that are actually causing
more damage. Hensson are giving Giant
Shoulder listeners 100 free blades with
your purchase of a razor using the link
in the description with the code the
Giant Shoulder. Once again, 100 free
blades. That's over 2 years worth. use
the link in the description. Now, back
to the episode.
Earl, do you think there's evidence that
humans can store memories outside of the
brain? I was speaking to Nikolai
Kokushkin from NYU. He seems fairly
convinced that, you know, we can argue
that maybe cells on by themselves have
memories related to their own specific
goals. But I'm talking about actual
human memories here. The memories that
we have as a human unified system. Is
there any evidence that those memories
can exist outside of the brain?
Memory is a broad term. Do you mean the
um what I did yesterday? What what I
watched on TV last night? You're talking
about sort of more broad kind of um form
of memory because certainly we're
learning a lot more about interactions
between the brain and body. You know,
the brain is not isolated from the body.
It's constantly getting metabolic and
other information from from the body and
then feeding back and changing things.
So, it's it's an interacting system and
little things like metabolism make make
a huge difference in in the brain
function. So, it doesn't surprise me
there's going to be memory in the
system. The brain is the system together
with the body. There's going to be
memory in that system. But I'm but I'm
trying to understand how specific or
high level of a memory are we talking.
What about an actual autobiographical
memory? Like an actual memory of of your
life.
>> I don't know of any evidence and Mike or
you can correct me. You know where where
there where actual autobiograph episodic
memories we call them where they're
taking place somehow outside the brain.
Is there a study that shows that? I
don't I don't know.
>> I mean, so so I'll tell you the the kind
of study that that could I I don't I'm
not sure the evidence is is very good
yet. I I think the the cases are are few
and there needs to and I'm not aware of
an animal model unfortunately for it. So
um we have to kind of wait till these
clinical cases show up. But the kind of
case that I've seen I've seen case
reports of um memory trans or I don't
know if you call it memory but but
transfer in heart lung transplant
patients. So one of the things that you
you occasionally see in cases of uh
heart lung transfer is not specific
episodic memories but um personality
traits. So there have been claims of
specific personality traits moving over
from from a donor. Now this is obviously
like a like a wild claim, right? So I'm
not saying the evidence is is is strong
at this point. But if but if that were
to be true, I think that would be the
kind of thing you're talking about. And
then and then and and I'll tell you I'll
tell you an experiment that we're doing
now that's that's kind of fun along
those
>> just made up just one second but we keep
talking about case studies remember case
studies are they're not study study what
there are other descriptions of of
oneoffs. So case studies are are
examples here's something interesting we
should we should investigate they're not
not strong evidence in and of
themselves.
>> Yeah. Yeah. 100 100%. Yeah. I just, you
know, I think I think if if thea, if the
case studies pile up, then then I think
it suggests that we need a model system
to really study this. So, so the closest
the closest thing that that I can think
of, and so we're going to we're going to
do this uh shortly. I don't know if if
you've heard of these um amphibots. So,
we make these we take um adult human
patients donate tracheal epithelial
biopsies when they go in for, you know,
for for for um sampling. And we take the
cells and we allow them in the culture
conditions. What they do is they re
assemble into a little little
protoorganism that basically moves
around on its own because it has celia.
So they round up, they move, they run
around on their own. They have all kinds
of cool properties. They express 9,000
genes differently than than they
otherwise would. They can heal neural
scratches and all kinds of stuff. But
one of the things we're interested in
doing is, you know, when when you buy
these things, you get a little you get a
little sheet on the patient that um that
donated them. And what I want to do and
and some of them were smokers. So what
I'd like to do is to uh get some get so
you see where this is going. I'd like to
I'd like to make from smokers. I'd like
to see if the anthrobots show an
affinity to nicotine. They don't have
any nervous they don't have a nervous
system. They're tired
work out absolutely may not regardless
it's amazing. But that's the kind of
extra I like to do, right? That if it
does work out then then then you know
you you know we're we're overlooking
something. So
>> sure. But a chem a chemical sensitivity
isn't the same as an ep episodic memory
though.
>> No doubt. No doubt. That would be a very
that would be a very simple low-level
you know low-level kind of thing. But
but but but still what what I think but
if it does work what I think would be
cool about that that idea is that you
typically so so what you started with is
a patient in which uh some kind of cells
presumably in the brain are addicted to
nicotine then they drive muscle-based
behaviors to go seek out nicotine right
in a rat that's the presumably that's
the story you would tell that there are
neurons that are that are addicted to
nicotine and then you have this like
muscle driven behavior to go get the
nicotine in the anthrobiot there are no
muscles there are no nerves So you're
remapping whatever is going on there.
You're remapping it onto a new
architecture where you now have
epithelial cells and you have little
cyia which is how which is how you row
which I think is cool and biology tells
us that works because of the caterpillar
butterfly memory transfer. So there are
experiments right where again
caterpillar butter caterpillar me
trained caterpillars pass on their
memories to the butterfly which never
mind how the information survives the
the tear down to the brain. The to me
the more interesting thing is that
you've now remapped it on a completely
different uh sensor and array. So you no
longer move like a soft body creature
like a butterfly. You move now at three
dimensions. You fly with a hard body
kind of architecture. You don't care
about the the leaves that you chewed as
a caterpillar. Now you want the nectar
and like all these things are different
but you get you but but the memories get
remapped. And so I would love it if I I
think it would be fascinating if we
could show in a mamalian system that
that those kinds of the the the uh
addiction to the nicotine would then
remap onto a different body plan which
is what the anthropods do. And then what
I'm going to do if that does work, what
I'm going to do is take the anthropods,
I'm going to stick them stick stick
those into a rat and I'm going to see if
we can then pull it out pull the the
information back out and have the rat
now back to the to the musclebased
motion. Right. So this is our
>> if the rattle will pop down to 7-Eleven
for a pack of smokes.
>> Get a pack of them or a red. That's
that's absolutely insane. Would you try
it for other addictive things as well,
other drugs? Would you get, you know,
pull tracheal cells out of opiate
addicts and try and see if they have a,
you know, an affinity for cappa opioid
agonism or like would you extend that
then beyond nicotine into other
receptors into other addictions and see
how how broad spectrum that could be? If
this works and again just to underscore
I'm not making the claim that this is a
thing where this is just something that
we're going to try right this is
>> you try gotta try
>> it's a it's an experiment that's born by
having you know weird ideas about how
memories can be transferred across
tissue and so on but if it works then
obviously um not only not only would I
try different kinds of addictions forget
addiction I would then I would then go
fullon and see what other kinds of
behavioral propensities you can move
right what else what else is movable are
are you know our memory, right? Our our
our fear-based memory skills, uh,
experiences, like what else what else
can you move? So then then then you have
an assay, right? Then then then you have
an assay for these things.
>> Yeah. I mean that that that would be
fantastic. I really really hope it works
out. I really want to see that. Um, and
it would be it's a great idea for
experiment. It's a fantastic experiment.
them and and it wouldn't surprise me if
it did work out because like I said we
had
multisell single cell creatures had
learning and memory and they had it all
by chemicals chemical waves of of
chemical flow um long before we
developed um nervous systems. So I think
it's quite possible a lot of these
chemical sensitivities can be kept can
be kept or chemical affinities can be
kept across creatures and then you then
have a downstream downstream effect on
behavior. Why not? I love it.
>> When are you gonna do these experiments?
Are you doing them now?
>> We are. We are beginning them now. So So
we're going to start we're going to
start this summer. Um yeah. Yeah. We we
have we we had a lot of other we just
finished this this uh this well the
pre-print is up. The real paper
hopefully will be out soon. Uh looking
at memory and zenobots. So zenobots are
a similar kind of idea. This is Vikov
Kai's work in my group. Zenobots similar
kind of idea with frog epithelial cells.
and we've been giving them experiences
and then reading out their memories in
terms of their behavior, their calcium
signaling and their gene expression. So
we can read out which ones have had an
experience a particular experience and
they can distinguish between two
different experiences and so on.
>> Right? So, but we do need to keep this
kind of in perspective in the sense you
know a lot we we keep saying memory is a
memory is one thing and actually memory
is a it's a different thing so it's a
continuum and it's a wide range of
things and you know um we had um u
memories and learning before we really
developed um complex brains or anything
that we would like I'm not going to
define just try to define what
consciousness is but anything like human
human consciousness and you know so
things like like people often attribute
a lot of a lot of highle stuff to
behavior or that may not be um um
warranted. Like you know, people common
thing is that you see the bear and you
run away because you're scared of the
bear. When in actuality, you know, the
conscious mind is about half a second
behind in uh in time and in what's going
on in the real world. What actually
happens is you you run away and then
later on you get you later you get
scared and you use that as episodic
memories for in the future I'm going to
learn to avoid bears. But your behavior
at the time was a simple automatic
reaction that even multiple cell
creatures on nervous system can can do.
So, but we tend to see it and we look at
through our lens of a conscious mind now
and say, "Oh, that person ran away
because they were scared. They had a
conscious experience of the bear and
they they they ran away." But it's
really not the way most memories work.
Most memories are kind of unconscious
and and and uh a lot of times your your
uh conscious mind is just kind of kind
of along for the ride to build these
higher level episodic autobiographical
memories that you can use for things
like deliberation later on. So we
shouldn't over interpret um behavior as
being something approaching the kind of
like conscious kind of memories that we
have. That's all.
>> Yeah. I I mean I I agree with all of
that. I'm I'm not planning to make any
uh any claims about consciousness from
any of these data because because that's
a whole other kettle of fish that
>> didn't she see the rats would go down to
the 7-Eleven for a pack of smokes. Oh,
yeah. I said that.
>> That was you. Yeah, that was me. Sorry.
So, you know, so yeah. Yeah, I'm not
planning to make any any claims on
consciousness based on those particular
experiments for the exact the reasons
that you said. However, um I would say
that I I don't think that for for for me
anyway, human level consciousness is not
the bar here. I'm interested sort of
much more broadly and and something else
is that I'm I'm very suspicious of
>> the of um things like unconscious
learning and and unconscious any kind of
unconscious process in general because
the way we the way we distinguish them
typically is by asking the human were
you conscious of this and the human says
well I wasn't well great you weren't
conscious of my my experience either so
you know it's Yeah, that's the problem.
Humans are the only creatures that we
can confirm consciousness in because we
can ask them. Other creatures we have to
sort of infer it, you know.
>> Well, well, even but but even in a
standard human, you have you have one
voice that makes a really good case for
himself or herself linguistically,
right? So, you got probably a left
hemisphere that puts up a great sort of
uh story about how look, I'm the I'm the
I'm the real boss in here and like I
have, you know, I have consciousness and
whatever. you got a bunch of other stuff
that you don't really know how to
interrogate and and and you can ask uh
you know they kind of like I say the
history books are written by the winner
you know I feel like neuroscience books
100% you are 100% correct consciousness
is often the story your brain makes up
to explain what is just at least part of
your brain makes up to explain what just
happened that is that actually what's
what's what's going on in many ways
consciousness is overrated it's really
only good for planning the future
behavior and stopping yourself from
doing something stupid you just decide
to do. Other than that, consciousness
just kind of along for the ride. But
that doesn't mean it's irrelevant.
Doesn't mean it's there passively. I
It's there for a reason. Planning and
deliberating are important. But but it's
amazing how much of our life we do
without planning and deliberating.
>> Yeah. I just I you know uh what for me
I'm I'm I'm cautious in uh being uh sort
of um constrained as to how we
understand a first-person experience of
structures other than ones that have
language. You know there's all kinds of
structures in our bodies and other
things that we make that may well have
certain all kinds of experiences that
they're just not able to tell us tell us
a story about. And we're part of the
efforts in our lab is actually to make
um communication interfaces to all these
all these kinds of things that we can we
can talk to them. So I'd like to be able
to you know on your phone instead of hey
Siri I want to be able liver like why
you know how are you doing? Why do I
feel like crap and what's going on and
and and be able yeah have have a
language have a language interface.
We've got we've got some so um Yambo
Jang and other folks uh in our in our
group have uh have made some some
progress in talking to things that
normally don't talk and and I think it
you know we I think it's important. No,
>> I love it. I mean I I completely agree
with you. I mean consciousness is kind
of overrated. I mean I I'm studying
consciousness now but the it's because
I'm interested in executive functions,
how the brain controls itself. In a more
broad sense, consciousness is not really
all that important all the time, but
it's important if you want to do things
like figure out how executive functions
work because they're they're kind of the
pinnacle of that, the apex of that
executive top down control process, but
they're not the beall endall of
everything that's going on in the brain.
Not by a long shot. So, what do you
think? Um, uh, Earl, uh, you so, so you
were very kind to give a talk to my son
recently. I watched it was super cool. I
was I was interested in um uh your uh
your your your stuff on anesthesia and
readings from the brain. You know, you
talk
>> um
>> could you could you talk a little bit
about that? And what I'm interested in
is
applying some of that some of that same
kind of analysis to the things that we
study that don't have a brain. So, for
example, we've been study we've been
doing calcium re and calcium recordings
from from zenobots and anthrobots. Um,
no neurons, but we're hitting them with
we're hitting them with anesthetics and
uh and and and hallucinogens and all all
kinds of stuff. And you can record Yeah.
If you want to, you know, see Zenobots
on hallucinogens that that's going to be
a thing.
>> I do.
>> Yeah.
>> Hallelujah. It's coming.
>> You do. I asked Hi about that and she
said she hadn't thought about it. I
said, "When are you giving the neurobots
DMT?" And she laughed in my face.
>> Yeah. Well, okay. But we're never on it.
I I don't have I don't have a permit for
DMT yet, but but we're we're on it
though. Um we're on top of it. But uh
but but but really like like being able
to do those things and then get the same
kind of readings uh as you're getting
from brains. What do you think how does
one interpret those in in that case when
uh you're dealing with a system that's
not the same but your tools are
perfectly applicable to this to the data
set that you're getting?
>> Yeah. How applicable is the different
systems is an empirical question and I
would love to know the answer to that.
But what we're doing with anesthesia is
that um well may be disturbing to know
because anesthesia has been around what
what was the ether dome 1849 the first
demonstration of uh over here at MGH. So
anesthesia has been around for like
since then but up doctors still don't
know why it works. It just all they know
is it's like a lot of things in
medicine. It works you know we don't
know why it works but it works. So let's
we'll keep using it. Thank god we have
it because I don't want to go through
surgery without a uh anesthetic. But the
tacid assumption for many years is just
just simply shuts off your cortex. just
turns it off like turning off a light
bulb. And and what I've been talking
about with electric fields, we study
electric fields um emerging electric
field dynamics in the brain. And what
they take the form of are these oscilly
dynamics, these brain waves, these
squiggly lines you see when you record
EEG from the scalp. And probably
everybody's familiar with them. When
they're squiggly, you're you're awake
and you're okay. If they're ever flat,
you're in a lot of trouble. You're
either dead or in a in a coma. It turns
out that what anesthesia does, it does
shut off your cortex. It changes the
brain wave dynamics. It shifts them from
higher frequencies we associate with
cognition to lower frequencies. And the
other thing it does is it misalign the
waves. Before when waves were in sync
together and you want a system to be in
sync so it can influence itself. But
then what happens anesthesia shifts
everything low frequency then misalign
the waves. Now the waves are out of sync
one another. And out of sync means this
set of neurons in an excitable state and
while this one's quiet and the other way
and if they're going to be doing that
they can't talk to one another. or
anything that's what causes disconnects
cortex and causes unconsciousness. But
the more general point is that you know
we've done we've seen this same effect
with three different three different
anesthetic drugs that have three
different effects um propol anesthesis
of ketamine and dmetondine and not vex
isn't really an anesthetic it's more of
sedative but the point is they render
you into this state and we're seeing
these effects in these three different
drugs that act on three different
receptors in the brain three different
parts of the brain if you talk to our
reductionistic colleagues or into
connectionism and in detail details of
snapses and ask how do all three of
these drugs do the same thing make you
unconscious. they would have to be
forced to say they operate three
different pathways, right? Well, we're
seeing all three of these drugs despite
their different pathways on the redistic
level do the exact same thing or very
similar thing on on this higher higher
emergent electrical field field level.
And right there, that's an explanation
on this electric field level that
something the correction effects of all
these different of these three different
anesthetics that can't be explained on a
reduistic level. That doesn't tell you
there's that's a functional functionally
important level. I I don't I don't know.
And I forget why we talk started talking
about anesthesia, but I will mention
that this is not just an academic thing
about making people what causes
conscious or unconsciousness. We're now
we developed a closed loop um anesthesia
delivery system. You record EG from the
scalp and it feeds into the system and
now you get tighter control over
anesthesia.
And we're we're developing this now and
trying to get introduced for clinical
use. And the advantage of that is when
you get to be old about my age,
anesthesia is something you should ab
general anesthesia should absolutely
avoid when when you get older because
there's these um they're associated with
both short-term dementia and then uh
lower relative long-term dementia. You
hear these stories all the time. People
say, "Oh, grandpa was doing great till
he had that surgery um for the heart and
then uh you never bounce back from that
cognitively." This is anesthesia is
really bad for you when you're old. And
part of that may be because what
anesthesiologists are doing is they're
monitoring your heart rate and your and
your respiration and your blood pressure
to see how unconscious you are instead
of measuring the actual thing that
that's going unconscious. So we think if
we could actually have a better system
for understanding the electrical
signatures of unconscious that you could
read in the alarm with simple electrodes
on the scalp then you could reduce
exposure to anesthesia and solve a major
p public health problem. And I forget
why we started this conversation about
anesthesia. Well, I brought it up for I
mean it that sounds that sounds awesome
because uh I'd love to use your system
with our with our Zenobots because
because I think I think it would be
totally usable. And then the question is
what exactly are we doing to them when
we put them in that state? Like what
what what is the sort of uh either
either a qual qualitative or functional?
>> Do your little creatures have
oscillating electric field dynamics?
>> They they sure do.
>> Yeah.
>> All right. All right. Let's talk.
>> Yeah. So we should write we should we
should we should do some of the
analyses. I mean the other thing the
other thing is you know what you just
said about um the dangers of general
anesthesia for for older folks. I
frankly I
>> young folks too by the way.
>> Well well this is the thing. If somebody
said to me hey I I I invented this
thing. We're going to we're going to
we're going to uh uh break up a lot of
the electrical connectivity of your
brain and it'll go into a different
state. But don't worry eventually when
when we sort of wash out the gas
everything will come back and you'll be
the same person you were. I would have
been like are you kidding me? like that
that does not seem like that should
work. I'm amazed that anybody comes out
of it the same way they went in. Does
that
>> does that surprise you at all or am I am
I off off base here?
>> Well, you know, brains are very
resilient. The biological systems are
are resilient. You you need I need to
tell you that. And then what's what's
going you're probably changing the EI
balance in in the system. Then it's it's
it's making it go into this domain where
the waves are are are uh and but once
you get rid of the agent that's changing
EI balance the natural system of the
brain with this recurrent connections EI
balance bounces back. I know I'm I'm
kind of trivializing it but I agree it's
amazing that it actually works and you
know thank goodness it does because uh
uh thank goodness for anesthesia is all
I'm going to say. We don't want to be in
a world without it.
>> No no for sure. It's just wild to me and
and I mean I don't know if all of them
do but certainly some of them um disrupt
gap junctional connections and right and
and when you do that at least in our
simulations u of non- neural cells you
don't come back to the same state
necessarily and I mean sometimes you do
but but often times you don't so it just
it's it it it's it boggles my mind that
it that this ever works for anybody.
>> I agree
trying to figure out the details.
I'm really curious on uh anesthesia
experiments with xenobots compared to
neurobots. Maybe first give a little
breakdown of what neurobots are and then
I'm really curious to know would you
have any difference in predictions on
how anesthesia might interact with both
of those systems.
>> Yeah. Um so so first the definition so
so xenobots are self assembled from frog
embryoepithelial cells. neurobots which
were um first created by Hale Fawat in
my lab who's now at Harvard. Uh she
introduced neurons into them. So they're
basically zenobots with a core of neural
cells inside. So they have a nervous
system. So you know the reason the
reason I wanted to do that work was I'm
curious like like if if you look at a
standard animal with a standard nervous
system and you say why does it have that
kind of nervous system usually you say
well because evolutionarily here's the
environments it had to deal with. here
are the ancestors that it had. And
therefore, the nervous system has been
shaped by selection pressure to have
specific properties and whatever. Well,
there have never been any zenobots.
There have never been any neurobots.
There's never been selection pressure to
be a great neurobot. So, my question
was, what the hell does a nervous system
look like in a creature that's never
existed before? I mean, obviously frogs
have existed, but but but zenobots are
not frogs. They don't look like frogs,
have the same body plant. So, so what
would that look like? So, that's what I
was interested. And and and we've done
some behavioral analysis. um we need to
do tons more. I mean, so there are
obviously some differences, but the
thing the thing about knowing what to
predict is I don't necessarily know what
to predict from anesthesia even in
Zenobots in the first place. Like one of
the things that we're doing is um and
this is this is some work with the Neil
Seth's group and and some other folks uh
to look at metrics that people are using
to judge you know conscious awareness
wakefulness sleep coma nobody home you
know all the different options that use
different kinds of metrics you know
there's causal emergence and whatever
that all these tools so so just to
understand first can we even say what is
it that that these anesthetics are going
to do in a in a zenobi is it going to
stop moving Is it going to not be able
to uh solve certain problems that it
could solve before which we're still you
know characterizing all the things that
it can do. Um it's it's it's not super
clear to me. Uh what the one the one
prediction that that I will will will
give is that I think whether with
neurons or without neurons I think we're
going to find important differences. And
I think that yes, no doubt the neurons
are going to add some some cool like
bells and whistles on it, but the
fundamental phenomenon I think we're
already going to see with epithelial
cells. I I don't think it's going to be
unique to neurons. That would be great,
but I mean like things like propol,
which is a common anesthetic, acts on
GABA receptors in the neurons. So what's
the equivalent of that in a in your
xenobot?
>> I don't remember specifically if
xenobots have GABA, but but GABAS like
like those receptors are
>> robbots do.
>> Yeah, they're present. Let's do that
experiment. Yeah,
>> it's goblet cells, right? Gobblet cells
have I believe I believe goblet cells
have GABA receptors.
>> I think I'm embarrassed. I don't
remember exactly. Uh because I know we
did well like we've we've did the
transcript almost. We know exactly
what's there. Uh lots of let's put it
this way. Uh in general lots of
machinery that you find that that are
that's neural machinery like all the
serotonin stuff um you know glycine
receptors gavo like that stuff is
present in all kinds of cells. So it may
it may well be there. Yeah, let's
anesthetize. Let's start with neurobots.
Let's anesthetize them. I mean the uh so
this by the way this week anesthesia I'm
doing with Emry Brown who's not only a
professor at MIT like me. He's he's a
practicing anesthesiologist. He knows.
>> I think I you know what? I think I
emailed him for for for a um uh for a
dumb uh dumb question like the and maybe
you guys have an answer to this so I'll
follow up but a lot of the anesthetics
that we try to use are not water soluble
and right and and and I I don't know if
that's if that's a standard thing or not
but but because these because all of
these biobots they they function
underwater basically.
>> Oh that would be that's a problem.
>> Right. So so either we're using it or we
need to we need to get it into the water
somehow. So maybe maybe I'll pick your
brain afterwards about how to how to do
it.
>> But your neurobots, so they have a they
they're showing these um emerging like
electric field dynamics, oscilly
dynamics. What's what's the what's the
frequency profile, right? Is it like the
human brain or animal brain where it's
like anywhere from one hertz to 100
hertz or more?
>> I'll send you I'll I'll I'll get you a
full profile and send it to you. I don't
remember sitting here today. I don't
remember what the but but there are lots
of we've done we've done uh some some
calcium signaling imaging and I bet you
know I bet we we have the kinds of
things you're looking for but let let me
see
>> if if they're like you know the primate
brain uh a human brain then um we should
see these profiles get basically shift
down to a lower frequency. Let's start
there. Let's let's see see if uh that
happens. And also it isn't just simple
you know lower fre there's this whole
cascade of events that happens in the
primary brain. first this mid-level
frequency of alpha goes up in your
frontal cortex and then that's at the
start of induction then it cascades down
to so there's a whole like cascade of
ostory events and if we could see you
know first of all whether the state
change is the same as in the neurobots
then whether the cascade is is the same
or which elements are there which ones
aren't that would be wicked cool. Yeah,
we should. Yeah, we should look and and
and you know, it's also possible that
the whole thing might because because in
general um non-neural cellular bio
electricity functions several orders of
magnitude slower like a lot of the same
but it's just much slower. So what you
might see is that the whole thing is
slow but then it gets even slower,
right?
>> Yeah,
>> we might you know it yeah it might be a
relative.
>> Well, you know these frequencies are not
magic. or not a certain band is good for
something different just like saying a
spike is always does one thing or
another but even the exact frequencies
don't really matter like like um um
human brains like animal brains they
tend to be higher frequencies because
the brains are smaller so the recurrent
connections are shorter scale what's
really important is relationship between
the waves it's more like the different
brains can play the same song in
different keys basically it's the
harmonic relationship between the waves
are going to be important you see this
kind of nested frequency coupling or
octave effects and stuff like that
because that's wave music is sound waves
and electric waves are the follow the
same principle. So what's important is
the relationship between the waves not
exactly what their frequency is that
what the exact frequency is.
How do you explain or do you have any
explanation for the visual expression
genes in this neurobot? So we you keep
we keep seeing this strange uh gene
expression in these novel um little
creatures as Earl said and now I will I
will use that phrase forever now to
describe any of these things as as
little creatures. And we saw hearing
gene expression in xenobots, right? And
now we're seeing visual gene expression
in neurobots. And to me, this is so
funny because it's almost like you have
this novel
accumulation of cells being dropped into
an environment and it's like it's
reaching for a sensory apparatus. It's
like, right, what the hell do we do,
guys? Completely new environment. How
can we reach for something to make us
help us navigate this environment a
little better? That's I'm probably
extrapolating a lot there, but that's
kind of seems like what's happening to
me. Do you have any explanation for
this?
>> So, so there's a couple of interesting
things. So, so just to say what the
basic observation is, we were interested
to know uh what is the transcriptto of
beings that have never been here before,
right? So, so we know the you know
standard transcripttos have an
evolutionary backstory that tells you
why certain genes are up or down if we
we we we think they have that back we
assume they have that backstory. And so
in these things we we just ask the
question, okay, so so what what new
genes do xenobots and anthrobots express
that their parent tissues did not
express? Okay, you can't do the reverse
because they're obviously missing a ton
of genes because they don't have, you
know, endoderm, they don't have
misoderm. They're missing a bunch of so
you can't do that. But you but you can
ask what extra like what additional
things do they express. So several
hundred new genes, anthrobots about
9,000, so like half the genome. And and
the thing is it's it's you know, you
said they're in a new environment. In a
certain sense they are, but but just to
be clear because because a lot of people
say, "Well, if you put things in a weird
enough environment, you know, of course
they'll turn on, you know, different
genes." The environment for the zenobots
is basically the exact same environment
the embryo was in. They're not in some
crazy new um, you know, set of
parameters. They're in weak saltwater
just like the the frog embryo was. The
one thing that's missing, however, are
the other cells. So normally what
happens is that these epithelial cells
are basically bullied by all the other
cells that that give them a very boring
life as this like outer covering of a
frog embryo that's just going to keep
out the bacteria. And so those are gone.
So you so it's like engineering by
subtraction. You remove those influences
and now the tissue gets to do whatever
it natively wants to do. Well, this is
what it wants to do. So um so so why are
they expressing these hearing gen? So so
they express a whole cluster of genes
related to um sound perception. And we
saw that we said, "Well, let's stick a
speaker under the dish and see if they
can really, you know, sense sense
vibration." And sure enough, unlike
embryos, xenobots, when you when you
give them, and this is like this is just
sine, like a simple sine wave, we
haven't done, you know, bak and
whatever. You give them a, you know, you
give them a simple sine wave and and and
they absolutely react differently when
when the sound is on. Um, why are they
doing it? I don't know. But I kind of go
with what what you just said, Evan,
which is that I think all all
morphagenesis, even normal standard
embryionic morph, like all morphagenesis
is trying to answer the question of what
the hell am I? What and and what is the
most um uh uh uh you know, sort of
efficacious uh coherent story that I can
tell in anatomical space and and
typically that ends up, you know, dogs
have puppies and cats have kittens
because usually they answer things the
same way. But but but what evolution
made in that in most cases are
problem-solving systems that when they
are in a different configuration will
solve the problem in a different way and
if they have access and I can give you
other examples of uh complex systems
using different molecular affordances to
solve problems when situations get
really weird. Um I think they're trying
to uh yeah they're trying to have a more
coherent life in their environment and
and uh this is now what we're studying
you know very very vigorously is to find
out um what what exactly that that
plasticity is
>> that what life is local reverses of
entropy using as little energy as
possible. I mean life kind of wants to
be organized right
>> I I think that's true. I you know I
don't know if I'm I'm a little
skeptical. I mean I know it has to be
tried and and and there's been good work
on it. I'm a little skeptical that
that's kind of that that those kind of
um entropic explanations going to be the
whole story or like even the best
perspective. But but obviously it's it's
a useful
>> Oh, it's not really an explanation. It's
kind of just more of a philosophy, but
uh you know
>> Yeah. Yeah. Yeah. It's a you know, it's
a question of it's a
>> how it happens is the whole story.
>> Well, it's a question of drive, right?
What is the fundamental what what what
the you can ask people like what do you
see as the fundamental drive of biology?
And some people will say uh it's it's
entropy, you know, reduction. And some
people will say it's um uh uh
propagation of information, right?
Reproduction. Yeah. Yeah. I I'm
skeptical about either of those as as
like I I'm not sure that's going to be
enough, but but but those are the kinds
of questions we have to ask.
>> Pretty cool.
>> Yeah. I have nothing to add.
Mike, when I asked my community you were
coming back, all of the questions they
wanted to ask you were related to the
platonic space of minds.
Could you lay out in a way that I really
want as many of the listeners to
possibly understand because this is so
cool and so weird and so sort of hard to
wrap your head around, but can you give
us like a good model for thinking about
what the platonic space of minds is? How
should we think about this?
All right. Uh let's we can we can we can
get into it this way. Um
the the the things things that we want
to explain in all sorts of arenas in
behavioral arena, in physiological
arena, in anatom and an anatomy, gene
expression, whatever, there are certain
there are certain patterns that that we
see. And one question we like to ask in
science is where did those patterns come
from? Okay, with the put putting aside
for a moment the whether whether that's
that's you know whether whether come
from is is is well is well but but but
what is the origin and and very
specifically the spec the why you got a
certain pattern versus some other kind
of pattern right that's where we're
interested in knowing and so typically
when you when you're looking at that in
a standard um plant or animal uh this
and you say okay why why this versus
that the the answer is because there's a
long history of selection against
specific environments that that got you
right. So, uh, now what we can do is we
can make novel beings where you can't
really tell that story that that same
story falls apart. Yes, the cells were
part of embryos that that do have an
evolutionary history, but they weren't
selected for the kinds of crazy things
we see them now doing now. And so, you
might ask, so where do those come from?
So, then we take a step further back and
we say, well, where do things come from
in general? So, so where where where you
know where do what are what are good
explanations? Well, let's see. We have
um there are facts of genetics and
heredity. So, so history and then there
are facts of physics. Is that it? Like
are those typically the only things and
typically by like cell biologists would
say yeah those are the two things you've
got you've got a history you've got
you've got basic physics and you've got
and and together those things um
sometimes give you emergent features.
And what are emergent features? Well,
there are things you didn't see coming,
things that surprised you, and and if
you did have a good explanation for
them, they wouldn't be emergent anymore,
but they would, you know, they would
have some underlying some underlying
explanation, but but emergent are are
features that that were surprising. So,
so you ask that question. Okay. Is that
does that really uh exhaust the the list
of explanations? And I would say that
that what the what what at least some
healthy sector of mathematicians would
tell you is that no in fact that that is
not the only place where in important
information comes from. There are also
important facts that are not facts of
phys. You say what what what what are
those? Well, those are uh c certain
certain facts of mathematics and the
truths of number theory and and the
exact value of the natural logarithm and
things like that and and you know fen
bounds constant and various rules about
uh you know how spheres will will will
pack together in certain dimensional
spaces and so on. Uh these are things
that these are truths that a you you you
don't discover them in physics. So, so
you can't just fire the math department
and hope the physicists will tell you
why the quitterians don't obey the same
laws as the complex numbers. Like that's
the physicists don't come up with that
even though they really care about the
outcome of those things. Uh and if you
were to change the constants at the
beginning of the big bang, you couldn't
make them any different. In other words,
there is there is no fundamental
constant of of physics that you could
change that would that would alter um uh
you know the you know touring the truths
about the halting problem and things
like that. These things just don't don't
affect that. So that suggests that there
is a space uh of of uh very specific
facts you know it's the four color
theorem it's not the nine color theorem
it's you know and it's and e is has a
specific value and so on there is there
is a set of facts that are not facts and
so now so so now you can say this uh we
we already know that those things are
important for things that happen in
physics and biology if you if you start
asking why like a 5-year-old keep saying
yeah but why why why eventually you
always end up in the math department. So
if you keep asking the physicist is why
it's white out why do the particles do
this or that eventually it's okay it's
because this this mathematical object
has a certain symmetry about it and that
that's why you know that's why the
firmia if you ask why did the cicas come
out at 13 13 and 17 years it's because
it's because they're trying to avoid
predators uh to time them and you're
cool so why those numbers not because
they're prime why are those prime go to
the math I think so
>> so there's a s there's a very specific
fact that are not facts to physics and
and so those facts uh are important for
things that happen in physics and
biology. So I make only one additional
claim on top of that that this is like I
know not everybody agrees with it but
but this is so far there there's large
sectors of the of the scientific
community that that agree with that. So
I make I make a crazy addition to this
that that I'm not sure anybody else
agrees with but but here's a hypothesis.
My hypothesis is this that once you've
uh taken the basic scientific um uh
optimistic assumption that the that the
latent space of these facts is not a
random grabag of of emergent surprises
that we just get to catalog when they
show up, but it's a structured ordered
space that we can systematically
investigate. I think that's that's just
the basic scientific exception. Then we
can say that I don't want to argue about
whether it's a realm or not. People hate
additional realms. I don't care if you
call them realm. I don't you know people
ask what the onlogical status is. I
don't care about that either. I I'm an
engineer. It's real. If I have to worry
about it or if I can exploit it in some
way. So that's that. Uh but what I what
I do hypothesize is that whatever that
latent space is that contains these
truths, it doesn't just have things that
mathematicians care about. Those are
perhaps the lower level of of of low
sort of low agency patterns that just
kind of sit there and and and maybe
don't change. But I I I hypothesize that
it also has other patterns in there that
are highly dynamic complex patterns that
we would recognize as behavioral
propensities aka kinds of minds. So what
I'm so all I'm saying is we take basic
mathematical platism and the relevance
it has for biology which I think is huge
and then we just say there really isn't
any uh reason to have this founding
axiom that okay but but all of those
things are only suitable for math.
There's no other there's no other kinds
of patterns and I suspect that other
patterns in that same space are readily
recognizable to believer scientists and
we've done some very crazy experiments
along those lines that I think are
important.
>> Can I play devil's advocate for a
moment?
Isn't the more personalist explanation
we haven't just haven't figured [ __ ]
out?
I mean, in the end, why why assume
there's this um up layer where there's a
structure that we can't quite quite get
at when when actually we just haven't we
just we're we're scientists haven't
figured everything everything out yet is
where we have I know you don't want to
use a realm, but I would say it sounds
more like we just haven't figured [ __ ]
[ __ ] out in this realm. Why assume
there's another one?
>> Uh so so I'm I'm 100% in agreement in
the sense that what I
>> just to be devil's advocate by the way.
I'm not saying we're wrong. I'm just
saying that
>> I I don't think you've said anything yet
that I have that I disagree with in the
sense that what I am what I'm not
claiming this very very important
because because I people people do um
complain all the time that I'm you know
sort of trying to make things
mysterious. That's not at all what I'm
saying. I am not saying that this realm
is where you push things that you're
never going to understand and and that's
why there's a bunch of mystery. That is
not what I'm saying. What I'm saying is
that we seem to uh in in in in the
sciences that we care about in physics
and computation in in in biology, we
seem to be dealing with at least at
least two different uh domains of things
that are that are um amendable to
different tools. You can
>> I hear I hear but that that's the whole
point, isn't it? Is it more like just
what there's science and math is just
running ahead of everything else because
everything else is slowed down by
technological limitations that math is
less uh constrained by. So math is just
ahead of everything else doesn't mean
doesn't mean they're talking about some
third thing we don't know about.
Well I think I mean look uh it's
entirely possible that I'm not sure if
that's more parsimonious. It's a kind of
a promisory note to say eventually all
of these things will be encompassed by
physics as well. I can't disprove that.
That may well be true, but I suspect
that the kinds of uh
>> That's my other issue. It kind of kind
of sounds defeist to say we we um you
know, we're having trouble figuring it
out here, so it must be something else.
>> No, I I think it's exactly I think it's
exactly the opposite because I think
this is a very exciting um the
beginnings of a very exciting research
program because because when I argue
with people, I say, look, there is a
structured latent space of patterns that
we can we can study this and we can
study that by making novel embodiment as
we've done. We've done some crazy stuff
recently where where we can make we can
make novel systems where where these
patterns come through and you can study
them and you can try to understand the
metric of the space. You can study you
can try to understand the mapping
between the structures we make and the
patterns that we get like all of this is
it's a perfectly good research program.
But the
>> you have a research program to study
this thing and I I retract my statements
then
>> we we absolutely have a research
program. This would be I would be I
would be nowhere near this this idea if
uh if this was some way of pushing
mysteries off into another into another
realm. I'm
>> I good.
>> Yeah. What what I would what I would uh
actually I I turn it around and and and
typically um uh the way people resist
this kind of thing is to say there's no
there's no structured space. There's
just regularities that hold in our
world. They say well well they say they
say it's emergent and they say well what
what does that mean? And they say well
these are regularities that hold in our
world. That's all. They say, "But but
but but if if you don't think there's a
structured space of these regularities,
what you're basically saying is it's
random or a causal and when they show
up, we'll write it down in our big book
of emergence and and and and that'll be
the end of it." Well, you know that that
I think that's much more defeist. I
would much rather think that that there
is a structured space as the
mathematicians think there is and as
they study you know going to you find
one thing and you use that as the
jumping point for the next thing and
then we can we can actually use them to
uh to improve our our experiments. I
like I think this is this is you got to
assume it's it's tractable.
>> Just you give an example catch up
>> this crazy stuff. Could you give an
example of this in platonic this
platonism in biology? Can you give an
example that the listener can really
latch on to here? is this, you know, sea
shells and, you know, the Fibonacci
sequences. Is is this sort of what we're
talking about?
>> No. No. Uh, I mean, that stuff has been
done, uh, but lots of people, lots of
people have done things like that, but
that that already exists. I don't I
don't need to add to that. Um, our So,
so let's let's put it that So, so here's
here's what I'm interested. Here's
here's part of part of the research
program. Um
to the extent that we think that the
standard stories of of physics, of
computer science, whatever uh uh tell
tell the story of what we need, what
what what we get in certain
circumstances. Uh we should be able to
use the existing frameworks to make
predictions about how much effort you
put in to creating something and what do
you get out of it. Okay? And there
should be there should be a match. And
we have we have all kinds of all kinds
of current paradigms and like what the
cost of certain computations for
example. And so and and what I'm
interested in is is for looking for
mismatches. I want to I want to find
situations where the standard paradigm
uh that the accounting that the standard
paradigm says between what you've what
you've put in when you created something
versus what you got out of it when
there's a delta. And and I want to
understand what that delta is. First of
all, can we show that there is a delta
which basically just tells you I mean
that's compatible with what Earl said
which all that tells you is that the
current methods are incomplete. I mean
clear enough everything is incomplete
eventually you sort of get to the end of
it right so so so find out find out
where where these where these frameworks
are incomplete and b quantify by how
much they're incomplete and what is the
if there is a kind of free lunch here
and I'll define that in a minute what
kind of thing is it and how much can can
we can we quantify and the thing with
biology is that it it often it gives you
very impressive examples but biology is
so complex you can't really prove
anything it's always you know Well,
maybe there's some quantum microtubial
in there or something or there's always
there's always something, you know, some
kind of a mechanism that you don't know
about or you haven't found yet that that
it's it's very hard to like to prove
anything. Um although although it's good
for starting to ask those questions like
we know when we paid the computational
cost to design a frog or a human, you
pay that in the eons of of bashing
against certain environments. When did
you pay the computational cost to have
uh you know a zenobot that does
kinematic self-replication? Right? So,
so you we already know that there's that
there's something not quite complete in
the stories that we tell about these
things, but you can't really quantify
any anything there. Uh, so what we've
done is create very minimal
computational systems where you know
exactly what you've put in. You can see
the entire algorithm. You know, there's
no magic under the hood because that
there's no it's not a quantum event.
It's a purely classical deterministic
system. There's no, you know, this
there's no funny business. And yet what
you find out uh is that uh you get out
more than this you you in terms of in
terms of behavioral competencies you get
out more than you put in. And so the
question is by how much and and what do
you get? Do you get static patterns? We
already know that's true from the from
the shells and then you know and
sunflowers and all that. We we already
know that. But do you also get problem
solving capacities? Do you get compute?
Do you get um uh uh you know do you get
there's a million other things it could
be anyway. So so that so that's the
idea. So so what we're doing is we're
we're we're creating very simple
systems. We're characterizing them using
the conventional mech the conventional
means of saying okay here's what here's
the algorithm here's the mechanism
here's what this thing should do and
then we put them in novel circumstances
actually they can do a bunch of other
stuff that we never put in. See the
thing the thing is uh in the standard
paradigm there are typically three uh uh
ways that you put in effort. Either
somebody to to make a controller that
does something useful. You either write
an algorithm that means there was some
engineer that knew how to solve it and
he wrote the algorithm or you evolve it.
So you look for high large numbers of
variants. You don't know how to solve it
but you look for large through large
number of variants and eventually you
find one that does a good job. Or three
you learn. So you've had contact with
the problem before. you took your lumps
when you got it wrong and now now you
now you can do it th those are typically
three three places where where you can
put an effort. So we are specifically
making systems biological but also very
simple computational systems where none
of those three things happened and yet
the system has some sort of competency
that you would not expect and that ask
that that enables you to quantify what
did you get uh and and why did you get
it in this case and not other cases? Why
did you get this competency versus some
other competency? How much effort did
you save? And that's something that's
that's really cool is that we can
actually quantify like how much extra
did we just get and where the hell did
it come from? And um and those those are
the kinds of things. So uh I'll give you
specific like you'll be able to see
we've got we we have one thing that's
published. We have a couple more that's
going to come out this summer. Um you'll
see examples of it. But that's but
that's the point. The research program
is to make systems where you think you
know what you should get out of it and
actually you get much more out of it.
And that's the delta that I would like
to understand. And that delta is not
just to finish I'll shut up delta is not
complexity. It's not ability. It's not
just perverse instantiation. It is
problem solving competencies that would
be recognizable to any behavior
scientist and they show up in substrates
that you would not expect under the
standard paradigm.
>> So you want to measure andor figure out
the structure of our scientific
prediction error is what you're saying.
>> Uh that's that's one way to put it.
Sure. Yeah. That's one way to put.
>> All right.
>> And are you are you claiming that that
delta is caused by non-material forces?
Now, I know we can get into a debate on
what you really material or physical
means there, but if I'm interpreting it
right, is that delta or that anomalous
what recording that you're seeing, is
that fundamentally non-physical or
non-material? Is that part of the claim?
>> See, that's the thing I struggle with is
this this difference because if if
there's nothing non-physical, then it
must mean our error or or as it or it
seems like you're saying there's
something else going on. That's what I'm
struggling with. What I'm saying is what
I'm saying among other things is that
our notion of physical causation I mean
typ typically when people say
non-physical we we already I I think I
think physicalism as such has been dead
since the time of Pythagoras and
probably long before that because we
already have situations where the
explanation for why this physical thing
does what it does the the best
explanation for it is oh it's because
the amplifed has this symmetry group
like that is a non-physical explanation
and you could say ah that's you know
some some people uh say well that's a
just scripture or eventually physics
will get there or something that's that
that's possible but I think we already
right right now and for thousands of
years I think we've already had examples
where the best explanation for something
that's happening in the physical world
living or non-living bad explanation is
not a physical event it is so so
causation so we have yes
>> as far as we know
>> as as far well everything right
everything is as far as we know so so
like every other scientific theory. I I
am completely open that that you know
next year somebody could figure all this
out and be like hey you know what the
value of E really does come from physics
not math we can close the math
department and the physicist will get
everything and it'll be fine like that
might happen that might well happen I
bet any money on that
>> maybe not tomorrow but yeah
>> yeah yeah I I mean right we all have to
place our bets I personally wouldn't bet
any any money on that I think I think I
think math is is as real if not more
real um a science than than physics. And
I suspect actually that the basement of
all of this and this is like maybe the
craziest thing I'll say is that the
basement of all this I don't think is
math. I think it's behavioral science. I
think math is a behavioral science of a
certain kind of pattern from that space.
Um math is just what we call the
behavior of certain objects in that
space and and other disciplines or other
you know they study the impact of other
levels of of that latent space. But I
think I think I mean really that
billiard ball of uh definition of
causation where everything has to be
physical that thing's been beaten up by
by by physics itself like quantum theory
right
>> we have a um a Laur Lauren Ross who's a
who's a philosopher of causation she and
I are writing a paper like taking all
this apart very carefully but uh I think
be I think being able to say that the
cause of important things is
non-physical I am hardly the first
person to say that and I I think we're
already in that land for for a long
time. The the real question is what's
the what what how far do the
implications of that go? Because a
standard assumption is that yeah well
it's a little bit important in in in you
know in particle physics but after that
you're down to like you know
conventional stuff and I don't think
that's true at all.
>> Fascinating. I find all this so cool. I
know we're running out of time, but I I
really want to understand the
relationship between bio electricity and
the platonic space of minds and how
though how you envision those two things
relating to each other causally
impacting each other because there is
this question of what establishes the
bioelectric network and sort of what I
read from your work and maybe this is
totally wrong and you can tell me if
it's wrong please do is that somewhat
thought of the of the setting of the
biomectric network, the dynamic
landscape is maybe somewhat established
from that platonic space. Is that right?
Or is that misreading thing?
>> Evan, it sounds like you're trying to
search still searching for something
physical.
>> Yeah. I mean, so so I'll tell so so I'll
give you the conventional story and then
and then how how I see it. The
conventional story would be would be
this. This is this is how you would you
would tell this from the conventional
point of view. uh look if you if you
track let's say let's say frog
development we can track from the one
cell stage from the fertilized state we
have a simulator that uh simulates ion
channels and voltages and and and
electric fields and stuff like that and
you can see that from a homogeneous
distribution of ion channels uh you get
you get uh electrical dynamics that have
symmetry breaking they have
amplification positive feedback loops
eventually you get some other kind of
pattern and it becomes more complex and
then and and so That's that's a
conventional story. So on that
conventional story you say okay um I see
I see why why it becomes more complex
and why you get symmetry breaking. And
then over here it's this thing that has
uh bilateral symmetry or it has
four-fold symmetry or it has you know an
axis with one end higher than or
something. And you say well why exactly
did that happen? And if you look the the
real answer of course is well the
mathematics of how electricity works in
this kind of medium. And so in the end
there is again I hate to say it but
again you're you find yourself in the
math department because the real answer
to why it is the way it is is because
the solutions to certain equations have
specific forms and this you know e is
over here and it's like this you know
it's 2.7 whatever instead of being nine
and and and that is the that is the
answer. So, so and then you say, "So, so
what does that mean?" And the
conventional answer then stops and says,
"Don't mean anything. Uh, it's
emergent." That that's it. These things
are the way they are. There's really
nothing we want to do about that. Uh
they're they're just emergent and you
just have to live with it and and you
can use it and you can do cool things
with it. But but that but that's it. Um
that you you know there's no there's no
why question to be asked beyond that. So
that's that's that I think is is what
the conventional story is. um uh uh on
that view the biomectric network is no
different than any other kind of
chemical you know the BZ reaction or or
or touring patterns in chemical media
like again well why do they have these
waves because that's how the math works
out that that's it that the distance
between the you know between the crest
like that's how the equations work so so
that's fine uh the only change I'll make
to that is that I I think that uh
bioelectric networks whether in the
brain or in body are particularly good
at hosting particular kinds of patterns
selected from that space. And that's why
we see them doing amazing things because
because they're tuned partially by
evolution. They're tuned to be really
really good at at hosting specific kinds
of patterns. And some of those patterns
are static things like hey this end be a
better bolarized while this end is
hyperpolarized. Kind of boring and
static but important because that's an
axis. or they can host really
interesting dynamics like hey that's
active inference or that's basian um you
know inf you know basian computations or
that's memory or that's whatever uh that
you know and that that's it otherwise
it's the it's the same story
>> I I mean that all sounds great I mean I
would say though when people say it's
emergent they're not offering as an
explanation or as a a mystery it's just
a fact and then the fact that we um you
have to live with it I don't think
anybody says that I mean it's more like
our models can't explain it yet. But
emergence isn't a explanation. It's a
simple property. I mean there are it's
something that can't be behavior that
can't be explained from the behavior of
its uh components unless unless they're
all working working together, but it's
not an explanation. I don't think any
would offer it as an explanation for
anything.
>> Well, I I agree with you 100% but but I
think people absolutely offer it as an
explanation when they didn't.
>> Well, yeah, I I think so. But but but
but let's put it this way. uh you know
uh they offer it as a which this is the
part that that I think goes back to your
point um about about being a defeist is
they often they often uh use it as an
alternative to uh to to to asking the
next question. Yeah, but why this
pattern versus that pattern? Like there
there is there is no why. There is no no
space of possible patterns that could
have come from there. There's no there's
no latent space. It's just these are
regularities that hold in our world.
Like that's it. And and and I I I get
that answer all the time from people.
>> I just figure we'll get there one day.
Our models our models will evolve.
>> Jens, you're you are incredible. Amazing
communicators, amazing amazing
scientists. Just so fun to talk to. I
could go for hours and hours. I'm kind
of glad that we disagreed a little bit
at the end because we're agreeing on far
too much and it was very interesting at
least a little bit at the end to to to
to find some parts of uh you know
division which I think is fascinating
but such the focus is on this station.
>> I'll just thank devil's advocate. I wish
Mike all the luck in this research part.
I hope I'm I'm dying to see the results.
>> Well, I appreciate that. I mean, look, I
I you know, I I say I say these things
because because I think they're
actionable now, but but I'm I'm
completely open to finding out that
there's better ways to do things. So,
yeah, Devil's is critical at at all
points. So, it's all good.
>> Yeah, it's been such a pleasure to host.
Thank you so much, guys. Really, really
such an honor.
>> Thank you. Thank you so much, Evan.
Earl, great to see you. I appreciate it.
Thank you. Great to see you, Mike. Great
to see. The giant shoulder mission is to
explore radical ideas in biology,
neuroscience, and consciousness and
elevate those stories to the highest
possible level while keeping them
accessible to everyone. If this
interests you and you want to support
independent science, then please
consider subscribing to the clips
channel. Check out our 26 neuroscience
book. Can download it for free.