►
From YouTube: DevoWorm (2021, Meeting 24): GSoC #4, Brain Networks, Differential Growth, Intelligent Materials
Description
Update on Summer of Code activities (Mainak Deb, Improving DevoLearn), Brain Networks: NetNeuro preview, brain network modeling, and parallel networks in development, the generativity of differential growth, vasculature networks, and building (thinking) scaffolds for intelligent materials. Attendees: Susan Crawford-Young, Shruti Raj Vansh Singh, Richard Gordon, Krishna Katyal, Bradly Alicea, and Mainak Deb.
B
B
B
Okay
yeah,
why
don't
we
start
with
a
review?
I
saw
some
things
in
the
slack,
so
you
know
whatever.
However,
you
want
to
present
it.
C
C
C
D
C
Here
actually
told
he
had
actually
said
something
about
this
in
the
club.
We
were.
D
C
D
D
C
B
It
thank
you
good,
very
good.
Thank
you
for
that
presentation.
So,
okay,
so
that's,
let's
see
time
to
show
the
chat.
Okay,.
F
B
C
C
E
E
It's
different,
look
it
up
online,
it's
there's
lots
of
documents,
there's
a
wikipedia
page
on
it,
and
the
code
is
in.
E
D
E
Okay,
it's
an
excellent
technique
for
for
making
a
low
contrast
image
and
making
it
high
contrast
without
edge
detection
or
anything
like
that.
E
B
Yeah
so
I
mean
probably
I
don't
know
if
they
have
those
functions
in
collab
I
mean
I
imagine
they
do
it's
not
that
obscure.
C
Yeah
I
have
received
an
image
that
has
two
samples
like
like:
what's
happened,
has
the
hand
drawn
filled
in
a
diagram
of
the
drosophila.
D
E
E
C
E
E
C
E
D
E
E
Bradley,
can
I
send
a
a
file
good
one.
Can
I
send.
B
The
file
using
the
check,
no
it
I
guess,
email
but
yeah.
I
can't
use
the
chat.
E
Oh
okay,
I'll
I'll,
send
you
I'll,
use
the
email
and
send
you
I
found.
I
found
a
pdf
of
meguru's,
unpublished,
book,
okay,
wow,
good,
okay,
yeah.
So
I'd
like
to
look
at
it
and
see
what
you
think
of
it
and
whether
or
not
we
should
finish
it.
B
So
yeah,
I
think
it
was
shorty
and
krishna
weren't
here
last
week,
but
we
talked
about
steve
mcgrew,
who
used
to
be
a
member
of
the
group
and
he
died,
and
then
he
had
this
unpublished
work
that
he
did
before
he
died.
That
was
the
book
that
dick
has,
and
I
think
it
I
guess
was
on
evolution.
Is
it
or.
E
B
Okay,
that's
great
okay,
so
yeah
back
to
the
thing
about
the
drosophila
eye
image.
We
did
a
paper
on
this
a
couple
years
ago
and
I
I
haven't,
I
should
have
put
it
up,
pulled
it
up
for
today's
meeting,
but
I
didn't,
but
it
was
a
paper
where
we
did
the
it
took
the
eye,
image
and
segmented
it
and
found
you
know
different.
You
know
the
different
the
structure
to
it
and
and
then
did
some
analysis
on
it.
B
So
there's
a
lot
of
data
in
these
images,
they're
they're,
pretty
high
resolution
images,
because
you
have
them
because
you're
drawing
them
in
a
lot
of
ways,
there's
more
information
than
you
might
get
out
of
a
microscopy
image,
especially
if
you
refine
the
image
too
yeah.
E
Here
I
have
to
put
in
the
chat.
E
Camera
obscura,
you
look
through
a
microscope,
but
you
see
two
images:
the
mic,
the
microscope
image
and
a
piece
of
paper.
And
then
you
draw
on
the
paper
which
you
see
in
the
image.
E
Okay
and
you
can
also
move
the
microscope
around
and
move
the
paper
around,
so
they
coincide
okay.
So
it's
a
very
old
technique,
probably
it
it
probably
dates
before
there
were
cameras.
D
B
So
my
knock
put
his
links
in
for
his
his
hero.
I
don't
know
you
pronounce
the
heroku
app
that
he's
using
for
the
testing
for
this.
So
my
knock,
I
had
a
question
about
that.
C
C
C
D
B
Yeah
and
like
well,
we
had
two
years
ago
we
had
vinay
varma,
who
was
a
gsoc
student,
and
he
did
this.
He
did
a
heroku
app
where
he
did.
You
know
he
did
hit,
did
an
algorithm,
he
implemented
an
algorithm
and
we
had
an
interface
for
it,
but
you
know
that
was,
I
mean
you
know
always
ways
to
improve
upon
it
and
the
way
it
you
know,
take
some
data
because
you
have
you
know
it's
also.
I
guess
these
online
ones.
B
You
know
you
can
upload
a
bunch
of
data
and
I
don't
know
if
there's
like
a
serving
limit
for
it
or
not.
But
like
you
know,
how
many
can
you
upload
maximum.
C
B
D
A
C
A
C
C
B
Yeah,
that's
good
well,
thank
you
and
you
have
your
blog
post
here
in
the
chat
as
well
for
week,
three,
your
update
and
then
dick
posted
some
things
about
camera
obscura.
So
we
have
a
wikipedia
article
on
camera,
obscura
and
then
a
wikipedia
article
on
filling
in
algorithms,
which
are
you
know,
I
guess
they
do
a
pretty
good
job
of
describing
the
basic
aspects
of
it,
and
so
then
you
could
take
that
and-
and
I
know
people
like
you
know-
there
are
ways
to
implement
these.
B
So
you
know
the
wikipedia
page
doesn't
necessarily
have
every
option,
but
you
can
find
a
way
to
do
it
so
good.
Well,
thanks
for
your
update
again,
my
knock,
I
wanted
to
give
you
an
update
on
the
network
neuroscience
satellite,
so
I
submitted
the
poster
that
we
had
so
my
knock
and
krishna
were
on
there
as
well
as
mayak
and
jesse
and
myself,
and
so
we
submitted
that
and
they
already
have
up
the
website.
So
let
me
show
you
the
website
that
they
have
now
so.
B
The
first
thing
is
that
they
have
this
well,
okay,
this
one.
So
this
is
the
network
neuroscience
satellite.
B
So
this
is
for
the
network
networks
conference
and
they
have
this
thing
where
you
can
join
the
like
if
you're
joining
the
website,
you
can
register
here
at
with
your
github
account
and
you
can
join
this
network
of
collaborators,
and
so
they
have
this
thing
that
renders,
if
you're
signed,
in
which
I'm
not
I'm
I'm
right
here,
and
this
is,
I
just
happen
to
find
myself,
so
they
list
people's
specialties.
B
B
The
other
thing
I
wanted
to
show
was
this:
this
is
the
poster
wall,
and
so
this
is
where
all
the
posters
are
located.
So
this
is,
let
me
grab
the
link
and
I'll
put
it
in
the
chat.
B
So
you
know
you
have
like
brain
networks
derived
from
different
imaging
data
sets
like
this
one
here.
B
This
is
human
brains,
where
they're
looking
at
networks
within
the
brain,
using
neurons
or
using
what
they
call
voxels
or
you
know
they
image
the
brain
and
they
find
like
three-dimensional,
cubes
called
voxels
and
they
use
those
as
the
criterion
for
the
the
a
single
data
point
and
then
they
find
the
adjacency
matrix.
So
you
do
like
these
pairwise
comparisons,
and
then
you
find
the
ones
that
are
strong.
You
know
with
you
know
above
a
threshold
of
strength,
and
then
those
are
your
connections,
and
then
you
can
generate
maps
like
this.
B
Where
you
can
show
the
connections
you
can
show
the
you
know
different
data
coming
out
of
those
networks,
things
called
rich
club
nodes,
which
we
have
in
c
elegans,
which
are
nodes
that
are
highly
connected.
B
In
other
words,
you
have
nodes
that
have
a
lot
of
connections
to
other
nodes
relative
to
the
rest
of
the
network,
so
they
they
actually
do.
There
are
a
lot
of
attributes
of
rich
club
nodes,
virtual
neurons
that
are
unique,
and
you
know,
there's
a
whole
literature
on
this.
B
People
are
doing
things
like
so
there's
another
network
visualization
here
where
they
have
a
circular
network.
Where,
again
you
make
these
pairwise
comparisons
and
you
find
the
connect
connectivity
between
all
of
them.
The.
B
Matrix
basically
plotted
out
like
this
and
visualized,
let's
see
where's
our
poster,
I'm
going
to
show
part
of
our
poster
is
sorry
with
all
the
posters
here.
It's
hard
to
find
ones
that
are.
B
They
have
see
if
I
can
find
her
as
though
I
was
really
trying
to
find
this
one
here.
So
this
is
our
poster
here,
neuromorphogenetic
patterns
in
the
theory
of
deep
learning,
and
we
have
our
organizational
logos.
We
have
some
a
lot
of
it
was
just
putting
in
images
to
make
it
like.
You
know,
have
some
visual
elements
in
the
poster
so
feature
discovery
network
depth,
network,
heterogeneity?
B
Those
are
all
factors
in
this
thing
that
we're
arguing
we're,
arguing
that
you
know
they're
parallels
between
the
brain
and
or
in
between
deep
learning
net.
You
know
architectures
and
the
brain
and
development
of
the
brain,
and
so
we're
kind
of
like
putting
these
things
out
here.
B
As
I
would
have
liked,
but
you
know
this
is
a
poster-
you
can't
really
get
into
a
lot
of
detail,
but
this
is
something
maybe
we
can
figu
and
I
did
take
some
things
for,
and
I
did
cite
the
ann
and
bnn's
paper
that
we
put
out
on
the
archive
this
one
here
with
krishna
and
jesse
and
myself.
B
So
I
put
a
couple
citations
of
that
in
this.
Certainly
we
can,
you
know
that's
kind
of
where
the
paper
is
for
this,
so
it's
not
really
a
companion
paper,
but
it's
very
similar.
B
So
I
don't
know
what
we'll
do
with
that
from
well
we'll
see
if
I
get
any
feedback
when
I'm
in
the
session
on
this
poster,
but
you
know
just
to
remember
recall
that
we
had
that
paper
and
then
we
have
this
poster
and
now
we
want
to.
If
we
want
to
move
forward
on
this
type
of
research,
we
have
those
two
streams
we
can
figure
out.
The
next
step
is
there.
We
might,
you
know,
make
a
different
version
of
the
paper
that
exists.
Now
we
might
write
a
separate
paper.
D
B
How
you
get
interesting
things
you
just
have
to
play
around
with
your
your
research
elements
and
see
what
you
get
anything
else
interesting
here
I
mean
this
is
just
it's
it's.
Basically,
you
have
theoretical
work
and
you
have
you
know
data
work.
A
lot
of
it
is
in
the
human
brain.
B
I
don't
really
see
anything
in,
say
c
elegans.
So
actually
I
mean
I
should
have
put
more
about
c
elegans
in
the
poster,
but
that's
okay,
that's
okay,
yeah!
So
that's
that's
the
poster
wall
and
I
like
that
that
they
have
the
posters
kind
of
out
and
you
can
like-
and
I
guess
some
people
recorded
videos
which
I
didn't
know
was
a
thing
you
could
do
but
well
I
think
it's
just
kind
of
like
yeah.
B
I
think
it's
just
like
if
you
want
to
zoom
in
on
it,
but
so
I
like
the
way
they
presented
this
on
this
website
and
then
the
program
for
this
is
actually
interesting.
There's
some
interesting
talks
here.
It
may
be
recorded
and
put
on
like
made
public,
so
that
might
be
good
for
people
if,
if
they
get
a
chance,
if
they
get
a
chance
to
do
that
I'll,
let
you
know
so.
You
know
there
are
a
number
of
different
talks.
There's
a
talk
on
network
and
brain
disorders
to
kick
things
off.
B
Then
they
have
some
contributed
talks
so
they're
on
a
number
of
different
topics,
mostly
the
human
brain.
There's
this
one,
that's
interesting
here
by
alexander
gulas
and
claus
silk
attack,
bioinstantiated
recurrent
neural
networks.
B
That
should
be
kind
of
interesting,
I'm
not
really
sure
what
they're
getting
at,
but
they've
done
a
lot
of
work,
interesting
work
with
like
networks
that
grow
in
size
and
how
those
are
wired
together.
So
I
mean
this.
This
is
probably
going
to
be
a
pretty
interesting
talk
and
I
don't
know
what
I
don't
know
what
that's
about,
but
you
know
we'll
see
and
so
again
there's
a
lot
of
stuff,
there's
a
spatial
developmental,
generative
network
model
of
human
brain
structural
connectivity.
B
So
there's
this:
there
people
are
doing
generative
network
modeling
people
are
doing
different
types
of
graph
analysis,
so
they're
doing
different
types
of
analysis.
There's.
B
Clinical,
some
that
are
more
what
they
call
conictomic,
which
is
where
you
have.
Of
course,
we
have
a
connectome
of
c
elegans,
which
is
302
neurons
in
human
brains.
You
know
you
have
potentially
millions
of
nodes,
and
so
you
have
these
huge
connectives,
which
are
very
different
from
the
c
elegans
connectome.
B
You
have
regional
connectomes,
you
have
all
sorts
of
structure
and
it's
operates
at
different
scales,
so
you
have
neurons,
you
have
what
they
call
voxels,
which
capture
like
a
maybe
a
it
could
be
a
square
millimeter
of
tissue
which
has
a
lot
of
things
in
it
or
you
have
regions
which
are
even
bigger
and
more
diverse,
and
so
all
those
different
layer
levels
of
analysis
exist
simultaneously
in
a
brain.
B
So
this
is
going
to
be
an
interesting
set
of
sessions.
Okay,
so
I
guess
I
wanted
to
move
on
to
the
submissions
document.
So
did
we
have
any
questions
on
any
of
that.
B
Nothing
from
my
side,
okay,
I
just
wanted
to
know
if
you
were
had
it
yeah,
because
I
had
mentioned
the
the
a
n's
bnn's
work.
So
if
you
had
any
thoughts
about
that
more
recently,.
F
Not
really
didn't
get
a
chance
to
work
on
that
thing,
but
yeah
I
submitted
a
book
chapter
I'll.
Send
you
the.
D
F
B
No
problem
yeah.
I
look
forward
to
looking
that
over
yeah.
So
in
the
chat
we
had
dick
had
a
link
to
his
hierarchical
genome
and
differentiation
waves
book
and
then
he
says,
number
of
cell
types
in
the
brain
may
be
huge,
let
alone
connections
yeah.
So
there
are
a
lot
of
different
types
of
cells
that
are,
you
know,
doing
different
things
and
in
the
human
brain
you
have
probably
a
lot
more
types
of
cells
than
the
c
elegans
brain
and
you
know
they
they.
B
You
know
you
can
get
into
a
conversation
about
cell
type.
You
know
you
can
use
a
an
anatomical
definition,
which
is
just
kind
of
like
finding
markers
in
the
phenotype
or
people
also
use
genotypic
profiles
where
they
will
take
awake
they'll,
do
some
next-gen
sequencing
or
something
and
they'll
find
like
what's
up
regulated
in
certain
cells,
and
then
they
can.
You
know,
define
the
different
cell
types
from
that.
It's
it's
a
little
tricky,
because
there
are
a
lot
of
things
that
are
transiently
expressed.
B
What
they
call
transient
expression,
which
is
where
they're
just
kind
of
turned
on
for
you
know,
may
not
have
a
functional
significance,
but
it's
you
know,
there's
something:
that's
upregulated,
so
you
know,
but
but
finding
the
right
sort
of
mix
of
things
that
define
cells
and
that's
that's
the
key,
but
it
so
yeah
there's
we.
I
know
that
dick
and
I
were
trying
to
come
up
with
an
estimate
of
cell
types
and
it's
very
hard
and
there
haven't
been
very
good
estimates
made
historically.
B
You
know
there's
a
vast
range
of
values
that
people
have
estimated
from,
like
you
know
like
in
the
human
brain
or
in
the
human
body.
It's
like
you
know
from
a
hundred
to
a
thousand
on
that
scale
of
difference,
so
yeah,
so
you
have
so
dixie's
differentiation
tree
may
organize
levels
in
the
brain,
but
mostly
so
most
cells
look
the
same,
but
they
have
different
functions
potentially
and
we
can
use
a
developmental
approach
using
the
differentiation
tree
to
organize
levels
in
the
brain.
So
you
can
organize
the
cells
in
that
way.
B
D
B
Like
the
different
parts
of
the
brain
emerge
phylogenetically
over
evolutionary
time,
and
so
that
kind
of
almost
links
in
some
ways
to
developmental
emergence-
and
you
know
making
that
connection
but
yeah-
it's
that's
something
I
have
to
spend
a
while
a
lot
more
to
discuss
it
fully.
But
if
you're
interested
we
can
talk
about
that.
B
So
dick
says
that
the
estimated
mouse
is
256
to
7
000
types
of
cells,
so
you
can
see
that
there's
really
no
consensus
on
the
number
of
cells.
It's
you
know
a
pretty
wide
range,
but
that's
because
people
are
using
different
techniques
and
a
lot
of
the
techniques
rest
on
like
that
looks
this
way
or
it
looks
that
way
or
you
know,
there's
some
marker,
that's
expressed.
So
it's
yeah
there's
a
lot
of
work
there.
That
could
be
done.
I
think
you
know
with
deep
learning
and
machine
learning
might
help
with
that.
B
But
you
know
that's
something
that
I
think
people
are
doing
it.
I
just
don't
know
how
successful
they've
been
so
susan
says,
I'm
taking
a
course
on
coursera
on
deep
learning,
since
I
don't
know
much
about
it
at
the
beginning
of
the
course,
there
was
a
lecture
by
jeffrey
hinton
and
he
talked
about
brain
organization,
yeah,
there's
a
parallel,
of
course,
between
brain
organization
and
and
neural
networks,
and
you
know
we
that
was
the
a
n's
bnn's
paper
in
part
and
they're.
B
The
holy
grail
there,
of
course,
is
that
you're
trying
to
mimic
brain
organization
but
you're
not
trying
to
make
the
network
too
complex,
and
so
that's
that's
kind
of
been
the
way
people
approach
this,
and
so
they
do
talk
about
like.
If
you
take
a
course
on
neural
networks,
we'll
talk
about
the
brain,
but
it's
like
okay
and
then
we're
going
to
do
it.
This
way,
you
know
the
the
the
lecture
isn't
like
how
diverse
the
brain
is,
but
it's
like
this
is
the
like.
B
B
It's
like
the
phylogeny
of
mammalian
brains,
but
the
person
who
wrote
it
is
trying
to
get
a
handle
on
sort
of
like
the
origin
of
different
brain
regions.
So
they're
doing
this
in
a
way,
that's
like
very
reminiscent
of
a
differentiation
tree
and
I
was
actually
take.
I
took
that
paper
or
one
of
the
figures,
and
I
was
trying
to
draw
a
differentiation
tree
from
it,
and
I
don't
have
it
up
here,
but
I
I
that's
something
that
I
think
could
be
done.
B
I'm
not
sure
how
accurate
it
would
be
or
how
detailed,
but
it's
definitely
something
that
is,
I
think
it
fits
into
sort
of
the
way.
You
know
you
get
these
structures
that
emerge
in
different
parts
of
the
phylogeny,
and
you
have
to
imagine
that.
There's
like
this
process
of
differentiation,
where
you
get
these
parts
of
the
brain
that
emerge,
they
would
have
to
emerge
in
development
in
a
certain
way,
like
you
know,
there's
certain
parts
of
the
brain
that
all
mammals
share,
for
example
and
you'd.
B
Imagine
those
would
be
developmentally
sort
of
early
and
then
they're
other
parts
of
the
brain.
Like
you
know,
there's
a
in
birds
there's
something
called
a
pallium,
that's
enlarged
and
of
course,
in
mammals
you
have
a
a
cortex,
that's
enlarged,
and
then
you
have
different
size
differences
across
species.
So
you
imagine
in
development
those
things
are
there's
those
tissues
are
differentiating
and
then
you
have,
you
know,
growth,
size,
regulation
of
those
areas.
So
there's
there's
a
way
to
draw
that
out.
That
would
be
quite
interesting,
so
krishna
says
susan.
B
If
you
need
any
help,
feel
free
to
contact
me.
Susan
said
thank.
You
is
a
book
on
genetics
and
machine
learning
if
someone
needs
an
email.
So
if
you're
interested
in
that
topic,
yeah,
okay,
put
differentiation
tree
of
brain
via
fly,
lodging
and
paper
on
wishlist
paper,
wishlist
I'll,
write
it
down.
Well,
that's
what
we'll
get
to
next
here
actually
and
I'll.
B
Put
that
on
the
list-
and
then
susan
says
I
don't
know
if
they
have
your
email
and
then
yeah
so
krishna,
I
guess
you
could
put
your
email
and
or
they
could
put
their
emails
in
the
chat
and
krishna
can
send
it
to
you
all
right.
I'm
going
to
talk
now
about
where
we'll
go
to
the
submissions
document
so
again,
there's
a
submissions
document
as
we've
seen
many
times
before.
B
We
have
a
lot
of
things.
These
things
are
completed,
but
there
are
a
couple
things
we
need
to
kind
of
think
about.
So
let's
see
this,
there
was
a
poster
that
was
presented
at
the
international
c
elegans
conference,
this
one
number
eight.
B
So
I
got
him
porter
gleason
who
who
actually
presented
it,
he's
a
senior
contributor
in
open
worm
and
he
told
us
that
it
went
well
that
there
are
a
lot
of
different
people
coming
up
to
the
booth,
the
virtual
booth
and
asking
about
open
worm
and
the
state
of
the
project.
And
there
are
a
couple
people
interested
in
diva
worm
because
there
was
a
diva
worm
section
on
the
poster.
B
So
we
saw
a
couple
weeks
ago,
and
so
that's
gonna,
be
you
know
that
that
was
a
pretty
successful
thing
to
put
out,
and
so
I
don't
know
if
porg
is
going
to
put
it
up
in
in
the
public
domain,
but
yeah
it's,
I
think,
it'll
be
recycled.
Definitely
for
another
conference.
B
I
know
that
there
have
been
we've.
We've
had
a
diva,
warm
poster
in
the
past,
where
it's
just
been
kind
of
the
diva
worm
group
like
different
things
about
the
actually.
I
presented
a
poster
on
that
in
2015
to
the
the
c
elegans
conference,
so
that
still
exists.
It
needs
to
be
updated
if
we
want
to
have
a
newer
version
of
it,
but
it's
still
there
so
that
was
successful.
B
Then
there's
this
number
20
mathematics
of
diva
worm.
So
that's
this
putative
addition
to
the
worm
book,
and
so
this
is
a
this-
is
a
right
now.
It
exists
kind
of
as
like
a
poster
or
a
large
document
that
has
a
number
of
different
sort
of
diagrams
with
equations.
Next
to
it.
So
it's
like
you
have,
you
know
like
the
fundamental
equations
of
of
diva
worm
or
of
computational
worm
development.
B
We
could
call
it
that,
and
so
we
have
those
and
so
we're
still
working
on
this,
maybe
soliciting
different
fundamental
data
structures
or
equations
that
we
can
put
in,
and
then
we
want
to
write
up
like
a
little
a
short
chapter
on
this
or
something
it
could
also
be
useful
as
a
as
a
poster.
I
don't
see
why
not,
but
you
know
we'll
work
on
this
in
the
near
future.
B
We
have
these
papers
on
the
test
of
volumes
and
symbiosis,
which
is
this
project
that
involves
bioinformatics
if
you're
interested
in
genomics.
That
might
be
something
you'd
want
to
be
involved
with.
We
have
these
molecular
level
simulations
diatom
motion
jerkiness.
This
is
this
rafa
project.
We
also
have
the
diatom
movement.
Smoother
jerkys
is
predicting
motion
from
images.
B
We
have
the
quantitative
comparison
of
our
key
and
shape
droplets.
This.
Of
course,
this
toponet's
session
is
coming
up
this
week
as
well.
So
I'll
get
some
feedback
from
that,
then
we
can
also
have
maybe
next
week
have
a
conversation
about
this,
how
this
might
be
done,
and
then
we
talked
about
this
book
by
steve
mcgrew.
F
D
B
B
B
We'll
figure
out
what
to
do
with
that
I
mean
we
could
publish
it.
I
guess
you
could
publish
it
on
something.
Like
you
know,
amazon
has
a
way
to
publish
things,
but
you
know
there
are
also
ways
to
publish
it
as
a
series
of
chapters.
Perhaps
I
don't
know
you
know,
that's
that's
something
that
is
well.
There
are
different
ways
to
do
that.
We
can
talk
about
it,
I'll,
take
a
look
at
it
and
see
what
I
think
about
what
we
should
do.
B
And
then
there's
this
article
that
I
don't
have
the
citation
for
right
now,
but
this
is
the
article
this
guy
paul
cyzek.
He
has
this
article
where
he's
worked
out
this
phylogeny
of
the
brain
in
in
this
way
that
would
blend
itself,
I
think,
to
a
differentiation
tree.
So
that's
good!
So
again,
if
we
want
to
add
anything
else,
let's
add
it
in.
B
Let
me
know,
or
you
can
just
access
the
document
here.
Let
me
actually
put
the
link
in
the
chat
and
we
have
a
couple
more
notes
here.
Dick
says
I
have
a
brazilian
geneticist,
possibly
interested
in
the
williamson
problem.
I
think
I
saw
some
emails
on
that.
You
said
you
forwarded
them
to
me
so
yeah.
That
would
be
great.
Maybe
we
can
get
them
in
touch
or
or
we
can
talk
about
this
more
so
this
would
be
the
williamson
symbiosis.
B
Let
me
see
if
I
can
add
in
a
note
here
all
right.
So
that's
good!
So
good!
That's
that's!
It's
all
coming
along
nicely
any
questions
before
we
move
on
any.
B
So
I
have
yeah,
I
have
some
papers.
I
have
this
thing
that
I'm
gonna
talk
about
here,
and
this
is
kind
of
weird.
I
don't
know
if
it's
weird,
but
it's
it's
a
little
bit
different
way
of
thinking
about
these
things.
So
let
me
start
from
the
top
here.
B
So
no,
maybe
not
okay,
yeah,
let's
start
from
here
so
there's
this
topic
that
I
got
interested
in
it
was
this
weekend
it's
called
differential
growth
generativity,
and
so
this
is
going
to
be
like
I'm
going
to
show
you
some
pictures
or
some
points
about
like
development
and
how
that
works,
and
then
I'm
going
to
show
you
some
graphic.
I
guess
you
could
call
it
graphical
art
or
something
that
starts
to
replicate
this
stuff.
B
So
the
first
thing
is:
is
that
you
have
this
is
in
in
response
to
an
article
and
they
talk
about.
This
is
a
journal.
It's
it's!
You
know
a
journal
account
where
they
talk
about
this.
Are
these
articles
that
they're
reviewing,
so
it
turns
out
that
mechanical
instabilities
may
play
an
important
role
in
creating
surface
convulsions,
so
they're
talking
right
now
about
pattern
formation
in
the
brain.
So
there
this
is,
there's
this
paper
where
they're
looking
at
like
the
human
cortex,
which
is
wrinkled,
and
it
has
these
things
called
convolutions.
B
And
so
the
question
is:
how
do
you
get
these
convolutions?
And
so
there's
this?
What
they're
doing
here
is
they're
showing
in
development
how
you
get
these
two
layers,
one
on
top
of
the
other
and
then
there's
no
growth
on
the
bottom
layer
relative
to
the
top
layer
which
is
growing,
and
so
as
the
top
layer
grows
compared
to
no
growth
in
the
bottom
layer.
It
kind
of
scrunches
up
the
bottom
layer
with
the
top
layer
and
it
forms
these
waves,
which
then
ultimately
become
these
folds
and
what
they
call
surface
convulsions.
B
So
the
convulsion
is
like
a
deep
fold.
That's
you
know
not
like
a
wave
like
this,
but
it
folds
inward.
Like
this,
you
know
there
are
a
lot
of
you
know
what
they
call
deep
folds,
where
it's
like.
You
know
you
pull
something.
B
You
know
you
fold
it
and
the
fold
is
like
more
than
a
couple
inches
deep.
In
this
case,
it
would
be
like
millimeters,
but
if
you
took
like
a
something
like
a
piece
of
paper
or
some
other
thing
in
your
hands-
and
you
did
this-
you
could
replicate
this.
So
you
get
these
deep
folds,
which
you
know
sort
of
deform
these
waves
and
they
become
these,
what
they
call
convulsions.
B
So
that's,
basically,
what
they're
proposing
is
the
mechanism.
In
particular
differential
growth
of
the
cortex,
is
thought
to
drive
a
wrinkling
instability,
so
this
is
a
physical
process.
That's
like
an
instability
like
we
think
of
entering
instabilities
and
things
like
that
in
chemical
morphogenesis,
as
demonstrated
both
in
silicon,
what
they
call
ingello,
which
I
guess
is
where
they
build
a
model
of
the
physical
model
of
the
brain.
B
So
then,
what
about
the
finer
structure
of
our
body,
the
orientation
orientational
ordering
of
feathers,
is
crucial
for
for
flight
and
birds.
So
this
is
the
bird
wing
where
they're
the
feathers
are
oriented
in
a
certain
way,
an
orientation
of
hair
cells
in
your
ear.
So
this
is
a
microscopy
image
of
that
is
important
for
hearing
this
requires
breaking
rotational
symmetry
uniformly
across
large
tissues.
B
D
B
Of
different
states-
and
they
have-
and
you
know,
they're
aligned
in
different
ways,
so
there's
an
alignment
problem
in
a
network,
and
so,
in
this
case,
they're
showing
this
on
a
lattice.
This
phenomenon
is
called
planar
cell
polarity,
where
cells
align
themselves
over
large
distances,
and
this
is
something
you
could
simulate
a
cellular
automata
where
you're
trying
to
get
them.
You
know
you,
you
have
different
states
of
these
cells
and
you're,
trying
to
align
them
into
a
certain
set
of
states
that
are
aligned.
So
you
know
you
can
think
of
this.
B
Arrow
is
shifting
its
state
in
different
orientations,
and
you
want
to
make
sure
that
at
some
point
all
these
cells
are
oriented
to
the
right
and
this
you
know
this
involves
a
lot
of
interactions
and
alignment
on
the
part
of
individual
cells
and
synchronization.
So
you
have
this
model
and
then
lastly,
tissues
can
break
translational
symmetry
by
spontaneous
interactions
between
chemical
morphogens.
B
So
this
is
the
classic
turing
chemical
morphogenesis,
so
turing
models
dating
back
to
1952,
offer
potential
explanations
for
stripe
spots
and
labyrinthine
textures
and
skin.
This
is
a
puffer
fish
skin.
Where
you
see
this
sort
of
labyrinthine
texture
that
results
from
this
chemical
morphogenesis,
these
are
different.
These
are
molecules
that
are
interacting
and
they're
forming
these
stripes.
B
B
So
people
have
done
this
with
like
pens
and
plotter
ink.
So
you
know
they
can
print
these
things
out
on
paper.
They
simulate
them
in
a
computer
and
then
print
them
out
on
paper.
So
this
is
one
example
of
something
that
was
generated
artificially.
This
is
a
generative
model
of
differential
growth,
so
this
is
a
it's,
not
a
fractal,
but
it's
something
that
has
been
like.
They've
been
they've
applied
this
algorithm
and
then
they've
plotted
this
out,
and
this
is
an
animation
of
the
plot.
B
B
So
this
is
something
that
if
you
run
this
code-
and
this
is
javascript
code-
it's
very
short
but
it'll
generate
this.
This
generative
differential
growth
pattern,
and
so
it's
again
it
just
starts
from
the
middle
and
it
radiates
out-
and
the
middle
of
course
is
undifferentiated.
It's
this
black
sort
of
blob
and
then
over
time
you
get
these
sort
of
branching
patterns
or
you
know,
non-uniform
branching
patterns.
I
guess
you
call
them
coming
out
from
the
center.
B
This
is
a
little
bit
different.
This
is
someone
who's
done.
This
he's
doing
this
plotting
so
he's
like
drawing
these
things,
and
you
know
kind
of
drawing
them
into
actually
with
a
pen
on
paper,
and
so
this
is
differential
growth
done
after
three
hours
and
51
minutes,
109
paths
of
257
850
on
points
1.29
kilometers
of
pen
travel
on
this
piece
of
paper,
so
these
are
the
different
patterns
that
were
yielded
from
this.
I
guess
it's
a
human
algorithm.
B
You
follow
a
certain
instruction,
you
do
this
over
a
wide
area
and
you
can
get
these
kind
of
patterns.
Let
me
see
if
I
have
a
close-up
of
it
now.
This
is
this
is
a
different
one.
This
is
back
to
differential
growth,
so
this
is
how
you
know
you're
using
this
on
a
this
is
a
hand-drawn
version
of
this,
where
you're
drawing
differential
growth
using
this
sort
of
procedure,
so
you're
generating
this
using
procedure
in
this
case
an
artistic
procedure.
This
is
the
same
one,
and
this
might
be
the
same
one.
B
I
think
I'd
uploaded
this
a
couple
times
and
then
finally,
we
have
this.
This
is
the
one
that
I
just
showed
you
before
of
differential
growth,
where
he
was
drawing
it
by
hand.
So
this
is
what
happens
when
you
put
it
on
a
as
you
can
see
that
you
put
it
into
an
animation.
You
can
see
it's
radiating
out
from
the
center.
This
actually
looks
a
lot
like
it's
reminiscent
of
a
bacterial
colony,
that's
kind
of
growing
outward
and
how
that
works.
B
So
you
know
there's
a
lot
going
on
in
the
biology,
but
you
can
draw
these
out
by
hand
and
they're
actually
quite
interesting
shapes,
and
so
I
just
wanted
to
show
that
off,
because
I
thought
that
was
interesting.
The
parallels
there
between
the
two
systems
or
the
types
of
things
you
can
do.
You
can
actually
look
at
nature
or
you
can
look
at
like
you
can
draw
them
out
by
hand
and
actually
looks
it's
interesting.
B
What
kinds
of
things
you
can
do
if
you
follow
some
procedures
and
you
build
these
animations,
so
there's
this
paper
that
came
out
recently,
it's
on
what
happens
in
the
brain
in
the
neonatal
mouse
brain,
not
with
respect
to
neurons,
but
with
respect
to
capillary
networks,
and
so
this
relates
somewhat
to
differential
growth.
B
But
this
is
this
is
about
capillary
networks
growing,
so
the
brain
and
the
in
the
bloodstream
don't
necessarily
interact,
there's
something
called
a
brain
blood
barrier,
but
nevertheless
that
tissue
needs
blood
to
perfuse
through
the
structure.
B
In
order
for
the
brain
to
operate
properly,
because
the
brain
generates
a
lot
of
heat,
it
consumes
a
lot
of
glucose,
and
so
you
need
to
have
a
blood
system
or
a
circulatory
system
that
deals
with
it,
at
least
in
you
know,
have
some
indirect
interactions
with
the
brain
itself.
So
the
abstract
reads:
capillary
networks
are
essential
for
distribution
of
blood
flow
through
the
brain
and
numerous
other
homeostatic
functions.
B
So
this
is
regulation
of
the
operation
of
the
brain,
including
neurovascular,
signal
conduction
and
blood-brain
barrier
integrity
according
accordingly,
the
impairment
of
capillary
architecture
and
function
lies
at
the
root
of
many
brain
diseases.
So
if
you
don't
have
proper
capillary
architecture
that
could
be
due
to
some
mutation,
it
could
be
due
to
like
some
sort
of.
If
you
have
a
stroke,
for
example,
you
can
have
damage
to
some
of
these
capillaries.
B
B
So
these
vascular
networks
are
also
important
in
brain
repair
as
well,
and
enabling
that
here
we
use
longitudinal
two-photon
imaging
through
non-invasive
thin
skull
windows,
and
these
are
in
mice
to
study
a
burst
of
angiogenic
activity
during
cerebrovascular
development
in
mouse
neonates.
So
these
are
mouse.
I
guess
they're
newborn
mice,
that
they're
they're
doing
this
study
on
they're,
looking
at
angiogenic
activity,
they're
using
an
imaging
tech
set
of
imaging
techniques
and
they're
able
to
look
at
the
development
of
this.
B
The
first
are
early
long-range
sprouts,
that
directly
connected
venules
to
upstream
arterial,
arteriolar
input,
establishing
the
backbone
of
the
capillary
bed
and
two
short-range
sprouts
that
contributed
to
expansion
of
antistomatic
activity
within
the
capillary
bed.
All
nascent
sprouts
were
prefabricated
with
an
intact
endothelial,
lumen
and
perisic
coverage,
ensuring
their
immediate
profusion
and
stability
upon
connection
to
their
target
vessels.
B
The
bulk
of
the
this
capillary
expansion
spanned
only
two
to
three
days
of
development
and
contributed
to
an
increase
of
blood
flow
during
a
critical
period
in
cortical
development.
So
this
development,
you
have
two
different
types
of
activities:
you
have
these
early
long-range
sprouts
and
then
the
short-range
spread.
B
So
you
have
two
different
types
of
sort
of
modes
of
activity,
and
then
you
have
this
activity
being
restricted
in
time
to
only
a
couple
days,
and
this
is
contributing
to
an
increase
of
blood
flow
during
a
very
short
window
of
cortical
development,
which
they
call
a
critical
period.
And
so
this
is
a
time
perhaps
when
the
brain
is
acquiring
information,
but
this
is
so.
This
is
the
idea.
I
think
it's
interesting,
that
you
have
these
capillary
networks
and
they're
sort
of
analogous
to
brain
networks,
as
we
talked
about
before
so.
D
B
Things
also
form
networks
and
the
networks
have
critical
components
to
them.
They
have
central
components,
but
they
also
operate
at
certain
points
in
time.
So
here's
some
images,
here's
the
imaging
platform
that
they're
using
they're
imaging
there's
the
mouse.
I
don't
know
if
you
can
see
it
in
here.
B
Let
me
see
if
I
can
zoom
in
on
it.
Okay,
so
here's
the
mouse
here,
here's
the
mouse
brain
and
here's
the
imaging
technique,
so
they
put
the
mouse
down
on
this
heat
pad
and
the
image
at
the
top
of
the
brain
and
they're
able
to
fix
this
rig
to
the
mouse
and
they're
able
to
look
at
these.
This
vasculature
here,
cerebral
vascular
development
in
the
mouse.
You
have
these
different
phases.
B
You
have
the
formation
which
is
embryonic
and
then
the
mouse
is
born
and
that's
the
neonatal
period
where
there's
expansion
of
the
cerebrovascular
system
and
this
actually
in
mammalian
brains
is,
is
so
very
similar
to
the
sort
of
where
you
get
a
lot
of
connections
forming
in
the
embryo
and
then
later
embryonic
development
and
before
birth.
B
And
then,
once
you
know
once
there's
birth,
there's
this
neonatal
period
where
there's
actually
pruning
of
connections
in
the
brain,
so
maybe
those
two
things
are
related
in
some
way
and
of
course
this
is
the
neonatal
period
where
they
did
the
imaging
it
was
p7
through
p12.
So
this
is
right
up
to
the
point
where
the
pups
open
their
eyes.
So
this
is
a
very
early
point
in
development
and
of
course,
before
they
open
their
eyes,
they're
not
using
their
eyes.
B
So
that's
another
interesting
sort
of
connection
there
then
there's
this
refinement
stabilization
period,
which
is
adolescents
and
adult,
and
this
is
where
this
network
is
sort
of
expands,
but
then
it
sort
of
matures
and
then
there's
after
it's
matured.
There's
this
refinement
and
stabilization,
and
this
happens
in
both
adolescence
and
adulthood.
B
B
You
have
these
capillary
junctions
and
then
you
have,
you
know
connected
versus
non-connected,
so
you
can
see
that
the
flowing
actually
helps
the
connectivity
in
the
network
when
there's
not
flowing
the
connectivity
dies
off
and
you
get
fewer
connections
across
the
network.
So
this
is
an
example
of
how
sort
of
these
capillary
networks
work.
It's
very
much
like
if
there's
no
flow,
then
that
that
capillary
dies
off.
E
B
If
there
is
flow,
then
there's
a
capillary
that
forms-
and
that's
you
know
that's
it's.
It's
actually
much
more
plastic
than
the
brain
networks,
in
fact,
but
this
is
necessary
to
sort
of
you
know,
figure
out.
You
know
where,
where
blood
is
needed
and
it
makes
it
adaptable
so
that
blood
can
flow,
maybe
to
where
they're
I
mean
you
know,
there's
a
need
for
repair
or
there's
a
need
for
you
know,
maybe
there's
local
activity
that
needs
to
be.
B
You
know
we
need
to
provide
metabolic
support
to
that
region.
So
that's
pretty
much
all
there
is
about
that
paper.
It's
interesting
it
if
you're
interested
I'll,
send
you
the
link
to
this.
Actually,
let
me
put
the
link
in
the.
B
Chat.
Okay,
so
let
me
actually
go
to
the
comments
here
in
the
chat
susan
said
about.
I
think
about
the
first
thing.
Our
whole
body
is
formed
by
folding.
That's
true!
We
do
have
a
lot
of
there's
a
lot
of
folding
and
morphogenesis.
That
goes
on,
find
some
youtube
on
active
matter.
Oh
well,
that's
a
good
point
because
that's
the
next
paper,
so
we're
going
to
talk
about
active
matter
next,
thanks
for
bringing
that
up,
I
can
do
a
presentation
on
active
matter.
Well,
that
would
be
good.
B
A
B
But
that
yeah,
that
sounds
interesting,
surety
says
the
deep
learning
specialization
is
really
a
nice
way
to
learn
about
deep
learning.
Yeah.
D
B
And
then
there's
this
drive
full.
This
is
the
drive
that
this
is
the
papers.
Okay,
dick
says:
contrary
work,
steve
smith,
you've
got
to
have
heart,
or
do
you
and
then
susan
says
blood
vessel
endothelial
cells
are
very
sensitive
to
blood
flow
or
shear
stress
yeah.
So
I
think.
B
So
that
the
capillaries
are
formed
in
part
by
sort
of
blood
flow
and
shear,
and
things
like
that
and
there's
a
whole
yeah,
there's
an
ear,
endothelial
cells
involved
and
all
that
so
anyways,
I'm
going
to
go
back
to
the
next
paper
here,
and
that
is,
oh,
I'm
still
sharing
my
screen.
Okay,
let
me
share
my
screen
again,
all
right,
so
the
next
paper
I'm
going
to
talk
about
is
this
rise
of
intelligent
matter.
B
So
this
is
about
soft
materials
and
soft
matter,
as
they
mentioned,
or
susan
mentioned
in
her
comment-
and
this
is
an
interesting
take
on
it.
This
is
about
intelligent
matter.
So
what
is
intelligent
matter
so
artificial
intelligence
is
accelerating
the
development
of
unconventional
computing
paradigms.
B
So
there's
this
whole
area
called
unconventional
computing
and
involves
things
like
slime,
molds
and
other
types
of
maybe
natural
systems.
There
are
all
sorts
of
ways
that
people
have
approached
unconventional
computing,
but
this,
I
think,
is
one
of
them,
and
so
some
of
these
paradigms
have
been
expired.
Inspired
by
the
abilities
and
energy
efficiency
of
the
brain,
the
human
brain
excels,
especially
in
computationally
intensive
cognitive
tasks
such
as
pattern,
recognition
and
classification.
B
A
long-term
goal
is
decentralized,
neuromorphic
computing,
relying
on
a
network
of
distributed
cores
to
mimic
the
massive
parallelism
of
the
brain
thus
rigorously,
following
a
nature-inspired
approach
to
information
processing,
with
the
gradual
transformation
of
interconnected
computing
blocks
and
the
continuous
computing
tissue.
So
these
are
like
these
nodes.
These
decentralized
nodes
are
being
transformed
into
continuous
computing
tissue,
which
I
assume
is
like
going
from
like
neurons
to
like
a
tissue
in
the
brain.
It
would
be
an
analogy.
B
B
Such
intelligent
matter
would
interact
with
the
environment
by
receiving
and
responding
to
external
stimuli,
while
internally
adapting
structure
to
enable
the
distribution
and
storage
as
memory.
So
there's
this
you
know
ability
to
store
information
and
retrieve
it
of
information.
B
We
review
progress
towards
implementations
of
intelligent
matter
using
molecular
systems,
soft
materials
of
solid
state
materials
with
respect
to
applications
in
soft
robotics,
the
development
of
adaptive,
artificial
skins
and
distributed
neuromorphic
computing.
So
this
is
quite
the
paper.
This
is,
I
think,
a
nature
review
or
something
yeah,
it's
in
nature.
Actually,
so
so
they
get
into
this
paradigm
of
intelligent
manner
matter
so
for
the
design
of
intelligent
matter,
inspiration
from
nature
is
beneficial.
B
A
bottom-up
assembly
is
nature's
way
of
achieving
material
properties
that
outperform
properties
of
their
individual
constituent
units,
and
so
you
can
use
this
bottom-up
approach,
which
is
some
people
call
it
emergent.
Some
people,
you
know
refer
to
it
as
opposed
to
top-down,
which
is
like
having
a
blueprint
and
designing
things.
B
You
can
use
bottom-up
assembly,
which
takes
advantage
of
self-organization,
and
you
know
doing
it
that
way,
letting
letting
nature
sort
of
organize
it
for
you.
Basically,.
B
Know
you've
read
anything
in
the
artificial
life
literature.
You
know
that's
one
of
the
advantages
of
like
say,
an
artificial
life
approach
versus
an
artificial
intelligence
approach
that
you
know
the
artificial
life
approach
you
can
just
you
know
organize
behaviors
based
on
the
you
know,
interactions
of
agents,
rather
than
saying
I'm
going
to
design
an
algorithm
that.
B
Impose
intelligent,
behavior,
you
know
you
have
maybe
an
algorithm,
but
it's
it's
not
as
it's
not
as
central
to
the
design
of
the
of
the
way
of
manipulating
it.
So
in
artificial
matter.
A
combination
of
bottom-up
and
top-down
methods
enable
architectures
of
the
variety
of
novel
characteristics
and
functionalities.
B
And
then
you
have
this
adaptive
matter
and
then
this
intelligent
matter.
So
it's
a
matter
of
you
start
off
with
structural.
You
move
to
the
responsive,
you
go
to
an
adaptive
framework
and
then
you
go
to
this
intelligent
matter.
So
in
structural
you
just
have
the
structure
you're
in
responsive
you're
at
you
adding
an
actuator
and
a
sensor.
So
it's
taking
in
basic
information
with
an
adaptive
network
you
have,
or
with
an
adaptive
matter.
You
have
a
network
of
these
structures
with
sensors
and
actuators
and
they
behave
in
a
network
and
then
intelligent.
B
The
intelligent
version
is
the
adaptive
version,
but
you're
adding
memory
to
it.
So
you
have
the
sensor,
the
actuator,
you
have
the
structures,
they're
networked
and
then
there's
some
memory
component
within
it.
So
you
can
have
you
know
something
that
will
recall
information
from
the
past
and
so
that
that
allows
you
to
have
this
this
sort
of
thing.
So
there
are
examples
here
they
give.
So
this
is
structural.
B
B
So
that's
that's
responsive,
but
that's
not
adaptive
because
it
just
basically
is
a
you
know,
pushing
a
button.
There's
no
real!
You
know
it
can't
change
its
behavior
based
on
its
prior
interactions
or
on
recent
interactions.
So
that's
where
the
adaptive
part
comes
in
and
I'm
not
sure.
Oh,
this
is
a
micro
swarm.
So
this
is
a
swarm
of
something
I
don't
know
what
it
is.
But
this
is
where
it's
exhibiting.
B
Behavior
with
respect
to
some
stimulus
and
the
swarm
is
adapting
in
certain
ways
and
then
there's
this
intelligent
behavior
that
takes
that
capacity,
but
adds
memory,
and
you
get
things
like
a
an
octopus
arm.
So
this
is
where
you
have
something
that
can
retract
and
fold
around
objects
and
manipulate
them.
B
So
that's
so
this
is
the
arm
of
an
octopus
with
its
embedded.
Sensors,
actuators
and
nervous
system
represents
intelligent
matter,
and
so
you
need
all
these
four
components
to
do
this,
so
they
have
yeah.
This
one
was
an
actually
nanoparticle
assemblies
here,
this
micro
swarm,
so
these
are
nanoparticles,
so
this
can
operate
on.
Many
scales
can
operate
on.
You
know
mechanical
skill.
It's
at
the
scale
of
someone's
hand.
You
know
something
sitting
in
their
hand.
You
have
the
skill
of
an
organism,
an
octopus.
B
D
B
B
So
you
know
you
can
get
molecular
materials
to
behave
in
these
ways
and
the
idea
is
you
want
to
build
like
artificial
systems
like
machines
or
robots,
or
even
things
that,
like
at
the
nano
scale
where
you
can
like
organize
materials
in
certain
ways.
So
this
is
an
example
of
where
they
have
a
target,
shape
they're,
trying
to
teach
these
swarms
of
nano
materials,
how
to
build
structures
for
a
target
shape
so
they're
trying,
I
guess,
they're
training
it
with
different
stimuli
or
I'm
not
really
sure
how
they're
doing
this
but
they're.
B
Basically,
the
shape
is
emerging
from
this
non-uniform
group
of
particles
and
then
it's
gradually
forming
the
target
shape
and
you
can
see
they're
doing
the
same
thing
with
star
here
where
these
particles
are
forming
aggregating
into
the
star,
and
so
you
can
do
this
with.
You
know
things
that
don't
have
a
brain,
they
don't
yeah.
So
there
are
paramagnetic
nanoparticles,
removing
a
microswarm
in
an
oscillating
magnetic
field.
An
external
programmer
can
change
the
field
such
as
the
adaptive
storm
can
split
and
circumvent
obstacles.
B
The
insects
show
an
overview
of
the
path
and
current
location
of
the
swarm.
So
this
is
this
green
line
here
in
in
this
figure
and
then
so.
This
one
here
is
a
so
air
autonomous,
individual
or
responsive
robots
merely
follow
their
programmed
algorithm,
communicate
with
their
closest
neighbors
in
a
storm
of
a
thousand
robots.
They
self-assemble
into
complex
2d
patterns,
since
an
external
programmer
predefines
the
target
shape
the
swarm
is
adaptive
and
not
intelligent.
B
Where
there's
a
target
shape
and
then
it's
like
it
eventually
adapts
to
this
target
shape.
Now,
if
it
were
intelligent
it
by
contrast,
you
could
form
a
k
or
a
star
without
really
having
the
target
there.
So
this
is
like
reminiscent
of
some
of
the
work
that
they
did
in
artificial
life
on
artificial
embryonics
back
in
the
90s,
where
they
would
do
this
with
a
cellular,
automata
and
they'd
have
the
cellular,
automata
and
one
of
the
ways
they
you
know
they
tried
to
build
like
these.
B
They
tried
to
take
like
something
that
was
undifferentiated
and
form
like
predefined
shapes
and
so
one
of
the
ways
they
do,
that
is
use
target
shapes
where
they
define
a
target
shape
and
then
they'd
allow
the
lattice
to
sort
of
model
the
target
shape.
But
that's
not
really
like
you
know
something
that
you
could
replicate.
If
you
didn't
have
the
exact
target
shape
in
place,
you'd
have
to
know
what
the
target
shape
was,
and
so
you
know
that's
that's
one
way
of
doing
it,
but
wasn't
necessarily
the
best
way
of
doing
it.
B
So,
as
you
can
see
this
in
in
these
intelligent
materials
that
this
is
what
they're
trying
to
do
just
more
examples
of
this
so
yeah,
I
think
this
is
a
really
interesting
paper.
B
I
don't
know
I
don't
want
to
go
through
much
more
of
it,
because
it's
just
there's
a
lot
here,
but
if
you're
interested
in
actually,
if
susan's
interested
in
presenting
on
it,
would
be
great
we're
presenting
on
this
this
topic
there's
a
lot
here
to
talk
about
so
dick
had
a
comment
here:
no
heart
equals
no
blood
flow,
but
heartless
axolotls
survive
past
hatching,
that's
interesting
so
that
they
actually
are
able
to
like
regulate
their
blood
flow
in
some
other
way
or
they
just
kind
of
they
just
survive.
B
I
guess
oh
it's
so
that
that's
interesting
stuff,
so
yeah
again,
this
is
so
I
have
the
papers
here.
If
you
want
me
to
send
you
the
paper
separately,
let
me
know
so
I
think
that's
good
for
this
week.
B
Did
anyone
have
anything
else
they
wanted
to
say
before
we
go?
No
blood
flow,
that's
what
he
said.
So
there's
no
blood
flow
in
the
axle
bottles,
but
they
can
survive
past.
Hatching
it's
interesting
so
next
week
we're
going
to
talk.
I
guess
we're
going
to
have
might
not
present
his
weekly
update
as
usual,
and
if
people
want
to
present
something,
let
me
know
we
have.
B
I
also
wanted
to
talk
about
some
things
lean
from
the
network
networks
conference,
so
nick
this
coming
week,
there's
the
toponets
workshop
or
the
toponet
satellite
and
the
neuro
net
neuro
satellite.
So
there
are
going
to
be
two
satellites
where
we
have
material
that
are
going
to
be
presented
and
then
I'll
report
back
on
what
the
feedback
was
and
then
we
can
talk
more
about
you
know
other
things
as
well.
So,
let's
see
we
have
some
more
things
in
the
chat
here:
okay,
oxygen
through
skin;
okay.
So
that's
that's!
B
B
Okay,
well
thanks
for
attending
and
talk
to
you
next
week
or
see
you
on
slack
if
you're
on
slack.
If
you
have
questions.