►
From YouTube: DevoWorm (2021, Meeting 33): Hacktoberfest, Roles and Opportunities, Criticality and Morphogenesis
Description
Planning for Hacktoberfest and Neuromatch 4.0. Review of group meetings board and new DevoLearn contributors/pull requests/issues. Discussion of Bacillaria non-neuronal cognition paper and Collective Pattern Generators (CoPG). Papers on the mathematical modeling of the virtual worm, universal routes to explosive (phase-transitive) phenomena, and models of developmental change, phase transitions, and transformation. Attendees: Bradly Alicea, Akshay Nair, Ujjwal Singh, and Shruti Raj Vansh Singh.
A
Hello
welcome
to
the
meeting
we're
on
a
different
day
today.
So
maybe
no
one
will
show
up.
It's
not
clear.
I
didn't
get
any
messages
from
people,
but
we
do
have
a
lot
of
things
to
talk
about.
We
missed
last
week
because
of
the
labor
day
holiday
in
north
america,
so
we
wanted.
I
wanted
to
go
this
week
and
do
something
talk
about
some
issues
and.
A
A
So
some
of
you
have
been
involved
in
the
neuromatch
academy
this
last
summer
and
the
summer
before,
and-
and
some
of
you
have
also
been
involved
in
the
narrow
match
conference,
and
especially
the
last
one,
which
was
neuromatch
three-
and
this
is
something
that
arose
in
response
to
covet,
but
is
also
a
very
good
idea.
From
the
standpoint
of
having
a
nice,
all-purpose,
neuroscience
oriented,
work
conference
and
so
they're
going
to
have
another
conference.
A
A
Match
is
because
it's
they,
they
originally
had
a
matching
system
where
people
would
register
they'd
put
in
some
abstracts
from
their
papers
and
then
there
would
be
an
algorithm
that
would
match
them
with
collaborators,
but
that
that
part
has
kind
of
fallen
by
the
wayside
in
in
favor
of
some
of
the
more
traditional
work
conference
or
workshop
setting
interactions
that
you
can
make.
But
still
it's
it's
a
it's
a
very
good
experience.
A
So
this
is
the
fourth
version
of
neuromatch
they'll
feature
traditional
talks,
lecture
style,
keynotes,
tutorials,
flash
talks
and
posters
and
they
encourage
local
meetups
during
the
conference,
which
is
something
that
actually
we
might.
It
would
be
nice
if
we
could
exploit
this
it'd
be
nice.
If
we
could
have
like
a
developmental
session
that
people
could
attend
and
just
kind
of
like
a
drop
in
or
maybe
we
could
schedule
some
content
during
that
time,
it's
coming
up
fairly
soon,
so
that
might
not
be
something
we
can
pull
off.
A
But
who
knows
if
people
are
interested?
Let
me
know
so
that
might
be
good
if
we
had
a
local,
maybe
like
not
a
local.
I
think
what
they
mean
is
like
locally
in
your
community,
but
we
could
also
have
a
what
they
call
satellite
meeting,
which
is
where
you
have
a
group
of
people
who
are
interested
in
a
topic
get
together
and
discuss
it,
give
like
informal
talks,
or
things
like
that,
so
we
could
have
a
diva
worm
satellite
for
neuromatch.
That
might
be
good.
A
You
know,
I
don't
know
again.
I
want
to
gauge
interest
on
this.
I
don't
want
to
force
people
into
it,
but
I
also
want
to
be
able
to
do
it
if
we
can
so
the
registration
date
for
the
main
conference
is
pretty
flexible.
It's
open
in
october
register
before
november
30th.
I
think
critically,
though
the
submission
date
is
october
15th.
A
A
They're
interested
in
computational,
neural
science,
so
they're
interested
in
traditional
neuroscience,
but
also
in
computational
neuroscience
and
then
a
lot
of
deep
learning,
machine
learning
and
so
forth,
and
but
this
will,
you
know
this
kind
of
work
would
have
to
be
a
little
bit
more
connected
to
either
theory
or
the
brain.
So
I
don't
want
to
just
have
you
think
that
you
know
any?
A
A
The
main
conference,
then,
is
from
december
4th
to
6th,
so
it's
towards
the
end
of
the
year.
So
all
you
need
for
this
submission
is
like
an
abstract.
It's
like
250
words.
It's
very
simple!
You
just
you
know,
describe
your
work.
Usually
it's
like
the
intro,
the
main
body,
the
the
main
idea
of
the
talk
and
then
at
conclusion,
which
is
usually
like
future
directions
and
then
that's
it
and
there's
no
paper.
You
don't
need
to
submit
a
paper
for
this,
so
this
is
pretty
flexible.
A
They
don't
have
the
agenda.
Yet
registration
fees,
you
can
pay
fifteen
dollars
or
pay
as
you
wish,
after
the
registration.
So
it's
like
you,
don't
you
know
it's
they're,
not
gonna
hunt
you
down
for
money.
If
you
can't
afford
it
and
then
the
local
meetups,
of
course
they're
calling
for
that.
So
that's
you
know,
that's
that's
something
to
keep
on
the
calendar.
A
Another
thing
I'd
like
to
bring
up
here-
and
this
is
my
other
group
and
I'd
like
to
talk
about
this
a
little
bit.
This
is
the
orthogonal
research
and
education
laboratory
group
and
I
work.
This
is
something
that
I
host
as
the
head
scientist
and
I
work
a
lot
with
jesse
parent
and
I
think
a
number
of
other
people
from
this
group
have
also
gone
over
to
that
group
and
done
things
so
we're
a
pretty
flexible,
open
science
group.
A
We
do
a
lot
of
different
types
of
research
and
I
I
bring
this
up
here
in
this
meeting,
because
I
want
people
to
kind
of
get
a
flavor.
For
you
know,
research
and
in
the
url
group
you
can
do
it
in
a
lot
more
broad
fashion
than
in
this
group.
So
we're
oh,
we
have
a
bunch
of
opportunities
open
now
we
don't
really
have
any
money
for
this,
but
the
idea
here
is
to
gain
experience
of
the
research
and
to
engage
in
research
and
do
other
things.
A
It's
it's
a
very
good
opportunity
for
young
researchers,
especially
people
who
don't
have
phds
non-traditional
scholars.
People
like
that
to
really
kind
of
get
a
taste
of
research
and
do
it
at
their
own
speed.
We
have
a
lot
of
people
who
come
and
go
in
the
group.
They
they
come
in.
They
do
some
work
and
then
they
leave
and
then
they
may
come
back
and
do
more
work.
A
So
it's
it's
pretty
flexible
from
that
standpoint,
but
right
now
we're
looking
for
people
who
are
a
little
bit
more
sort
of
devoted
to
their
work,
and
they
want
to
do
this
on
a
weekly
or
monthly
basis.
So
we
have
a
couple
of
opportunities
here.
We
have
an
advanced
opportunity
here
for
a
research
assistant.
A
B
A
So
this
is,
though,
this
this
position
is
a
little
bit
higher
up
on
the
on
the
ladder.
This
isn't
just
like
you
know
a
research
position
like
taking
ownership
of
a
project.
This
is
actually
doing
some
management
activities,
and
so
you
would
lead
or
generate
your
own
project
with
support
from
existing
lab
managers,
or
you
know
in
that,
but
then
also
on
top
of
that
doing
administrative
things.
So
this
could
be
things
like
maintaining
our
medium
posts.
We
have
a
lab
medium,
developing
educational
materials,
literature
reviews.
A
You
know,
coordinating
efforts
to
engage
people
things
like
that.
So
we
have
a
lot
of
different
things
that
you
would
do
here.
You
wear
a
lot
of
hats
and
it's
critic.
The
one
critical
aspect
of
this
is:
you
have
to
have
a
real
commitment
to
interdisciplinary
research.
So
it's
a
very
good
opportunity
to
get
that
up.
You
know
to
learn
how
to
do
interdisciplinary
research,
especially
like
crossing
over
from
one
discipline
to
another,
and
you
know
seeing
how
those
things
combine
in
a
way.
A
That's
you
know,
sustainable
and
that
you
can
actually
build
projects
off
of
so
and
then
there
are
a
whole
set
of
opportunities
related
to
project
management
that
you
can
gain
from
this
kind
of
position
as
well,
there's
model
builder,
which
is
a
different
position.
This
is,
I
think,
a
lot
of
people
in
this
group
might
be
interested
in
this.
A
This
is
where
we
have
model
builders
who
you
know
you
attend
a
meeting
and
you
can
attend
this
meeting
and
do
this
where
you
attend
the
meeting
you
listen
to
what's
going
on,
and
then
you
take
some
idea
and
you
model
it
and
what
I
mean
by
model.
Is
you
want
to
turn
it
into
something
that
you
can
understand
with
a
computational
model.
B
A
You
know
we're
interested
in
people
who
have
some
background
in
reinforcement,
learning,
genetic
algorithms,
deep
learning,
networks,
complex
networks,
agent-based
models
and
other
types
of
models
as
well.
So
if
you
have
this
background,
say
you
have
a
background
in
deep
learning.
You
know
my
this
position
might
allow
you
to
broaden
your
skill
set,
and
you
know
pick
pick
problems
from.
A
You
know
the
wide
range
of
problems
we
talk
about
and
you
know,
work
out
some
sort
of
implementation,
but
then
also
you
have
the
opportunity
to
learn
other
methodologies
and
then
work
on
those
as
well.
So
this
is
a
very
good
opportunity
for
this,
and
then
there
are
these
other
project-based
internships
which
are
just
based
on
our
projects.
A
Some
of
the
projects
are
well
far
along
and
some
aren't
the
ai
ethics.
Of
course,
a
couple
of
people
in
this
group
have
done
this.
We
have
something
on
something
called
cognition
futures,
so
this
is,
if
you
want
to
find
out
more
about
this
contact
myself
or
jesseparent,
and
we
can
put
you
in
proper
put.
B
A
A
A
B
A
A
A
The
idea
is
that
people
contribute
during
hacktoberfest
and
if
they
make
five
or
more
contributions,
they
get
some
swag
or
something.
So
we
don't.
I
don't
know
if
we
ever
really
have
any
swag
to
give
I'll
have
to
figure
this
out,
but
definitely
it's
something
that
will
lead
to
better
bigger
and
better
things
for
people.
There's
definitely
going
to
be
some
social
media
recognition
for
people
who
participate
I'll,
prepare
a
blog
post,
so
people
will
be
recognized.
A
So
this
was,
this
is
a
big
big
event
for
open
source
enthusiasts
held
every
october
this
year,
it's
from
the
first
31st
of
october,
it's
organized
by
digital
ocean,
along
with
other
sponsoring
companies.
So
every
organization
has
their
own
sort
of
specific
hectoberfest
organization
and
their
rules.
So
you
know
we
want
to
do
this.
We
have
to
start
planning
this
out
now,
which
is
why
I
bring
it
up
today,
and
I
know
that
I
think
it
was
krishna
and
I
think
who
talked
about
this
last
time.
A
We
had
a
meeting,
and
so
I
don't
know
if
there's
but
their
interest
level
is
in
this.
We
definitely
are
accepting
we're
always
accepting
contributions
to
diva
learn
we're
also
accepting
contributions
to
some
of
our
other
repos,
but
diva
learn
might
make
the
most
sense
in
terms
of
focus
this
year,
so
this
is
just.
We
just
have
to
come
up
with
some
of
these
different
things.
Here
we
need
to
come
up
with.
A
I
need
to
come
up
with
a
social
media
post
which
would
lay
out
some
of
these
rules,
so
I
guess
the
rules
are
just
kind
of
like
what
do
you
need
to
do
to
be
recognized
and
what
I've
what
I
did
last
year?
Actually
we
didn't
well,
I
guess
we
did
formally
participate
in
it
last
year
my
rule
was
like.
If
you
make
one
pull
request
and
it's
merged,
then
you
get
recognition
and
then
we,
you
know
mayak.
A
Dev
has
been
working
with
people
to
make
sure
that
their
submissions
are
successfully
pulled.
So
thank
you,
maya
for
that.
They,
so
people
will
get
at
least
one,
probably,
unless
it's
just
egregious
but
still,
and
if
you
get
one
pull
request
accepted,
then
you
usually
get
some
recognition.
It
would
be
nice,
though,
if
I
could
get
some
sort
of
like
swag
for
people,
but
that's
not
something
that
you
know
it's
kind
of
one
of
those
things
that
I
don't
know
how
motivating
that
is,
but
anyways.
A
You
know
if
someone
contributes
like
15
really
good,
you
know
pull
requests
that
are
merged,
maybe
we'll
you
know
give
them
some
sort
of
swag.
I
don't
know
what,
but
you
know,
that's
that's!
That's
the
so
we'll
make
the
rules
we'll
figure
out.
You
know
what
I'll
kind
of
work
through
the
rules,
and
maybe
next
week
I'll
show
them
to
the
group,
and
then
this
is
just
more
about
like
how
this
organization
is
running
their
hacktoberfest.
A
I
think
when
you
have
new,
you
know
when
we
have
gsoc
and
then
we
also
have
a
hacktoberfest,
although
oktoberfest
isn't
really
self-sustainingly
g-shock,
it's
still
an
opportunity
for
people
to
come
in
and
contribute.
So
that's
good.
Now,
speaking
of
divo
learn,
I
want
to
go
over
some
of
the
different
things
that
are
going
on
with
that.
A
So,
as
I
said,
thank
you
to
my
oak
and
my
knock
deb
for
being
maintainers
of
divalearn.
Now
that
gsoc
is
over
we're
going
to
be
working
on
a
paper
on
this,
but
we're
also
going
to
be
continue
to
accept
pull
requests,
and
so
we
already.
We
have
one
pull
request
open
now
and
we
had
one
a
while
back,
and
I
see
that
we're
getting
some
new
contributors
here.
A
A
That
means
that
this
project,
you
know,
is
still
going
pretty
strong.
We
have
a
number
of
different
issues
here,
six
issues
that
you
can
address,
so
these
will
be
maybe
we'll
add
a
couple
more
for
in
time
for
hacktoberfest
in
the
next
in
the
next
weekly
email
I'll
send
out
I'll
make
a
call,
for
you
know,
sort
of
coming
up
with
different
things
that
we
can
put.
B
A
A
To
follow,
I
think
so.
That's
that's
the
way
we'll
do
this
and
we
can
do
this
in
any
of
the
other
repos.
It's
fine
yeah,
so
that's
diva,
learn
then
I'll
go
to
major
tasks.
We
haven't
talked
about
this
in
a
while.
We've
got.
I
think
we
went
over
a
couple
weeks
ago,
but
it's
getting
kind
of
unwieldy
here
I
was
trying
to
figure
out
where
we
are
exactly
on
this.
I've
been
trying
to
update
it
accordingly,
but
it's
I
come
up
with
ideas
for
things.
A
A
So
I
think
actually,
this
christian
paper
review
is
finished
all
right.
So
we
have
this
ct
computed
tomography
via
cloud
computing.
We
have
this
update,
divorm
ml
lectures.
This
is
something
that's
kind
of
fallen
off.
My
radar,
because
I
was
doing
that
earlier
this
year,
but
I
just
haven't
had
time
recently
issue
30
follow
up
on
neural
organoids.
A
So
we
talk
a
lot
about
organoids
in
the
group,
but
we
haven't
really
done
anything
scientific
with
them,
and
I've
been
talking
for
like
two
years
with
people
about
how
to
make
like
how
to
do
sort
of
organoid
data
science,
and
that
is
like
can
I
can
we
take
like
secondary
data
sets
and
analyze
them
and
bring
out
like
insight
from
these
data
sets
a
lot
of
the
experimental
data
sets
with
organoids.
Are
you
know
people
grow
organoids
in
a
culture
they
put
in
some
fluorescent
markers,
they
look
at
the
organoid
itself.
A
Sometimes
they
do
some
cell
tracking,
but
most
of
the
time
it's
like
looking
at
different
physiological
functions
and
identifying
gene
expressions.
So
there
really
is
an
open
area
here
for
other
methods
or
for
taking
those
data
and
finding
you
know
constructing
measures
and
doing
some
other
types
answering
other
types
of
questions.
A
B
A
A
A
Living
networks
paper,
so
this
is
based
on
our
our
embryo
networks
work.
This
is
just
something
that
is
a
next
step
in
that
direction.
There's
been
some
progress
on
this.
I
I
don't
know
where.
To
put
this,
it
could
be
an
action
item,
because
we've
done
some
presentations
this
year
for
these
networks
and
how
to
model
networks,
not
as
sort
of
static
entities
but
entities
where
you
get
new
cells
coming
into
the
network.
A
So
you
get
new
nodes
and
you're
growing
the
network
and
you're
gaining
connections,
but
the
rules
of
connectivity
aren't
like
most
other
networks
because
they
don't
behave
like
most
other
networks,
like
I'm
talking
about
social
networks
or
telecommunication
networks
or
power
grid
networks,
things
like
that
they
have
a
different
behavioral
mode
and
we
don't
really
know
what
that
is.
So
this
is
a
this
paper
would
sort
of
lay
out
what
exactly
those
rules
would
be.
What
kind
of
net
it's
a
special
you
know.
A
A
Those
are
the
kinds
of
things
that
people
try
to
use
to
characterize.
You
know
these
different
types
of
complex
networks,
but
biological
networks
might
be
very
different
from
that,
especially
like
embryo
networks
or
developmental
networks,
where
you're,
adding
in
nodes
and
there's
a
spatial
aspect
to
it.
That's
very
different
from
other
types
of
networks.
A
A
A
We
have
this
axolotl
embryo
animations
and
segmentation,
we're
working
with
susan
to
get
data
we're
working
with
akshay-
and
you
know
maybe
some
other
people
to
work
on
the
axolotl
atlas
or
this.
This
three-dimensional
sphere
that
you
can
map
these
images
onto
so
you
can
explore
the
surface
of
the
embryo,
and
this
is
something
that's
definitely
in
progress.
A
This
analysis
of
shapes
of
polygonal
archaea
and
shape
droplets,
so
this
is
analyzing
cell
shapes.
This
is
something
I
think
my
knock
is
engaged
with.
This
is
best
working
on
normal
cognition
paper,
we'll
talk
about
that
in
a
bit
and
then
that's
all
for
the
in
progress.
The
action
items
are
maybe
a
little
bit
less
in
progress
than
in
progress,
but
we
have
a
number
of
things
here.
We
have
the
this
network's
paper,
updating
the
website,
which
is
actually
sort
of
a
continual
thing,
revising
the
a
ns
bnn's
paper.
A
A
So
we
did
this.
We
talked
about
this
last
year
at
nm3,
which
was
last
october
or
november.
I
believe
so.
This
is
actually
something
we
can
follow
up
on.
I
think
the
talk
would
have
to
be
what
we
could
take
the
original
talk
and
modify
the
slides
a
bit
based
on
what
we've
learned
over
the
past
year.
So
we
in
the
past
year
we've
written
a
paper
on
this.
A
preprint
we've
also
done
some
additional
work
in
the
in
the
meetings
on
discussing
you
know
this.
A
A
A
So
I
mean
you
know
this
is
thinking
about
really
early
earth
really
early
life
and
the
conditions
that
led
to
maybe
like
the
development
of
embryos
or
what
a
common
ancestral
embryo
would
look
like.
This
is
interesting.
This
is
a
little
bit
speculative.
We
don't
have
a
lot
of
data
for
this,
but
it's
nevertheless
something
we
can.
Someone
wants
to
grab
onto
it.
You
can
more
issues.
B
A
Which
has
never
really
been
a
strong
project,
but
it's
been
a
theme
there
are.
There
are
a
number
of
things
in
the
meetings
that
we
talk
about
all
the
time
with
respect
to
complexity,
theory-
and
you
know
different
things
where
we
have
an
open
question
of
like
we
have
this
complex
process.
How
do
we
put
numbers
on
it?
How
do
we
put
like
measurements
on
it?
Can
we
build
a
simulation
around
it
and
if
so,
how
do
we
do
this?
A
A
I
don't
really
have
a
I'm,
not
really
we're
not
really
pursuing
it
right
now,
but
we're
still
thinking
about
it
like
tutorials
for
youtube,
or
this
morphogenesis
and
deep
learning
paper,
creating
an
embryo
model
and
blender
these
sorts
of
things
that
are
sort
of
we
perennially
talk
about
them,
and
so
these
are
things
that
we
want
to
kind
of
focus
on.
I
think
this
is
going
to
be
off
the
radar
actually,
okay.
A
A
I
think
this
goes
here,
but
this
is
this
kindle
book
idea,
but
then
there
are
other
things
here,
like
virtual
developmental
worlds.
I
think
we
talked
about
that
last
week
and
that
relates
to
these
these
issues
here,
number
38
and
number
six
we're
trying
to
create
a
visualization
for
a
docker
container,
but
also
for
more
general
outreach
and
for
just
you
know,
one
of
the
things
that
we
can
offer
to
the
public.
A
So
these
are
all
things
that
are
to
do
now.
If
you
want
to
update
this
list,
if
you
want
to
create
issues
or
if
you
want
to
address
issues,
go
to
the
devorm
github
repository
or
the
the
diva
worm
organization,
I
think
it's
actually
not
an
organization
but
you'll
find
github.comworm,
go
to
the
group
meetings
repository
and
then
go
to
the
projects,
and
this
is
a
project.
The
project
board,
major
tasks
for
2020
2021.
B
A
Just
you
know:
if
you
have
permissions,
you
know
permissions.
Let
me
know,
request
permissions
and
I
can
let
you
in
or
you
can
send
me
a
message
in
I
guess
in
github
or
by
email
or,
however,
that
you
want
to
add
an
issue,
and
I
can
add
it
in
so
it's
there
are
a
lot
of
ways.
You
can
do
this,
okay,
so
we'll
move
on.
Then
I
just
wanted
to
assess
where
we
were.
We
don't
really
have.
We
have
a
submissions
document,
but
we
really
don't
have
anything
on
that
right
now.
A
A
C
C
A
C
A
C
A
A
Yeah,
okay,
so
this
is
the
basilaria
paper
again
we're
pretty
close
to
the
end.
I
wasn't
sure
yeah,
so
it
needs
to
be
like
12
000
words.
I
guess
I
wasn't
sure
exactly
what
the
next
steps
were
on
this
and
I
think
I
might
wait
until
I
go
ahead
and
submit
it
to
the
people
who
want
it
and
then
you
know
we
can
see
what
they
request.
A
So
there's
a
lot
of
stuff
in
this
paper
on
different
methods,
so,
instead
of
a
cornucopia
methods,
the
idea,
basically,
is
that
there's
this
organism
vascularity
it's
a
diatom
which
means
that
it
has
sort
of
like
an
algae,
but
it
also
has
the
properties
like
it
can
move
around.
It
has
a
lot
of
motility
and
there
are
a
lot
of
interesting
behaviors
that
they
perform.
So
there
are
a
lot
of
different
types
of
diatoms.
A
There
are
round
diatoms
in
these
long
rod
like
diatoms
and
the
route
long
rod
like
diatoms
are
the
ones
that
we're
talking
about
here,
so
they're
eukaryotic
algae
that
have
chlorophyll
and
silicate
cell
walls.
So
the
cell
walls
are
very
rigid,
but
they
can
also
move
around
in
their
environment,
and
so
this
is
it's.
You
know
it's
very
small.
It's
like
80
microns
by
10
microns,
which
is
very
small.
You
need
a
microscope
to
see
that
they're,
even
smaller
than
c
elegans.
A
So
there's
a
lot
of
things
going
on
here
we
were
gonna,
do
use
some
data
to
fill
in
the
gaps
here.
We're
gonna
do
like
we
did
a
paper
two
years
ago.
That
was
on
a
formal
data
analysis
of
the
movement,
and
so
that
was
pretty
well
received.
It
was
just
published
as
a
book
chapter
recently,
so
the
you
know
we
have
the
data
set.
A
We
have
like
a
data
set
of
movement,
we
have
some
analyses,
deep
learning
and
bio
mechanics,
but
this
is
more
speculative
and
this
is
like
you
know
talking
about-
maybe
the
potential
for
what
they
call
non-neuronal
cognition,
so
non-neuronal
cognition
is
this
idea
that
these
diatoms
will
behave
in
a
way
that's
sort
of
reminiscent
of
a
cognitive
processing
type
set
of
behaviors,
so
they
don't
have
a
brain.
Obviously
they
just
have
their
own
physiology
and
they
generate
thrust
and
they
interact
with
one
another.
A
But
the
idea
is
that
they
have
this
sort
of
collective
behavior
and
they
have
this
sort
of
like
sensor.
They
can
take
in
sensory
information
and
respond
to
it.
So
it's
in
that
way
it's
kind
of
like
a
psychophysical
nature
of
cognition,
and
so
a
lot
of
this
is
you
know
we
talk
about
a
lot
of
different
models
for
this
in
the
paper
we
talk
about
different
psychophysical
models.
A
We
talk
about
different
concepts
in
in
neuroscience
that
might
apply
to
some
of
the
things
going
on
here
and
then
we
have
this
model
of
a
copg
or
a
collective
pattern,
generator
which
is
like
there's
this
idea
in
actually
in
c
elegans.
You
find
these
a
lot
they're
central
pattern,
generators
and
they're
generated
by
networks
of
neurons.
A
A
Okay,
it's
not
working,
so
I'm
not
going
to
get
that.
Let
me
see
if
I
could
replug
it.
Okay,
there
we
go
all
right,
so
a
central
pattern.
Generator
is
like
a
network
of
neurons,
we'll
just
take
two
neurons
here
for
as
an
example,
and
these
two
neurons
will
generate
pulses
in
a
way
that
creates
an
oscillation,
so
the
oscillation
will
be
like
a
a
sine
wave
like
this
all
right.
So
the
idea
is
that
this,
this
neuron
fires,
then
this
neuron
fires,
this
neuron
fires,
and
it
gets
you.
A
It
gives
you
this
oscillation
and
this
this
generates
a
collective
pattern.
There
might
be
a
bigger
network
of
neurons,
you
might
have
like
add
a
third
or
fourth
neuron
in,
and
you
know
you
have
these
networks
that
are
synchronized
so
that
they
either
fire
in
sequence,
where
they
fire
together
and
they
create
these
sine
waves
or
some
sort
of
pattern
that
allows
you
to
have
this
generate
this
rhythmic
motion,
sometimes.
A
You
know
generating
movements
like
swimming
movements
or,
like
limb
patterns
like
in
some
amphibians,
you
know
they're
in
their
spinal
cords.
They'll
have
these
central
pattern,
generators
that
generate,
like
you,
know
their
legs
moving
back
and
forth,
so
they're,
generating
like
locomotion,
they're,
some
organisms
that
have
these
in
their
spinal
cord,
where
they're
generating
tail
tail
or
fin
movements
that
oscillate.
A
A
You
have
these
rapid,
like
cells,
that
aren't
don't
have
any
neural
activity,
they
don't
have
any
like.
Well,
they
maybe
have
some
limited
electrical
activity,
but
it's
certainly
not
like
an
action
potential
that
you'd
see
in
the
neurons
up
here.
So
you
know,
this
is
something
that
is
generated
by
these
cells
moving
against
one
another.
In
you
know
it's
like.
A
The
parallel
is
the
movement
output
and
some
of
the
activities
within
the
colony
versus
in
the
neural
network.
So
we
have
a
special
class
of
pattern.
Generator
called
a
collective
pattern
generator,
which
is
something
we
talk
about
in
the
paper.
So
that's
what
we're
doing
in
this
paper.
It's
it's
it's!
A
So
this
is
something
that
you
see
across
a
lot
of
different
systems
that
don't
have
nervous
systems,
but
it's
nevertheless
they
exhibit
this
sort
of
cognition,
and
so
one
of
the
problems
with
a
neural
cognition
is
it's
very
much
a
an
analogy.
It's
like
okay,
it's
a
nice
analogy,
but
people
anthropomorphize
things
all
the
time.
They
say
they
behave
like
humans
and
you
know
obviously
they're.
You
know
like
paramecia,
don't
behave
like
humans.
They
don't
have
the
same
brains.
You
know
they
don't
have
intentions
and
motivations
like
humans.
At
least
we
don't
think
so.
A
A
I
think
if
in
akshay's
case,
if
he's
interested
in
doing
some
modeling
on
this,
I
would
just
have
him
read
this
paper
like
I'm,
going
to
submit
it
and
then
I'll
have
yeah.
Actually,
all
of
you
read
the
paper
and
then
you
know
maybe
think
about
like
where
you
could
contribute
in
terms
of
a
model
like.
There
are
a
lot
of
things
here.
There's
like
heavy
intelligence,
there's
like
some
simple
psychophysical
equations,
there's
like
the
free
energy
principles.
A
A
B
A
We
don't
put
it
in
the
main
paper
we
can
still
have
like
you
know
you
could
be
an
author,
you
could
find
a
way
for
you
to
contribute
to
it
without
you
know,
but
then
the
model
we'd
have
it
available.
So,
like
you
know,
have
you
ever
seen
papers
where
they
have
a
paper
and
then
they
have
some
supplemental
model
online
somewhere.
C
A
A
A
A
A
A
B
A
A
A
These
with
the
group
see
what
they
thought.
This
is
something
that
so
I've
been
showing
this
mathematics
of
diva
worm
thing
last
couple
meetings
and
it's
not
not
a
paper,
but
this
is
like
something
that
the
group
has
been
working
on
towards
a
realistic
biophysical
simulation
of
c
elegans
and
hasn't
really
been
updated
in
a
long
long
time.
A
We
don't
really
have
it
as
a
preprint,
but
this
is
something
gives
you
an
idea
of
what
is
going
on
with
the
open
worm
project.
So
the
idea
is
that
these
are
the
biophysical
foundations,
basically
what
they're
trying
to
do
in
the
broader
open
room
project
as
they're,
trying
to
simulate
the
movement
and
the
nervous
system
and
the
coupling
between
the
nervous
system,
movement
and
adult
worms.
So
our
goal
is
a
little
bit
different
in
this
group.
A
We're
interested
in
development
and
so
development
requires
a
whole
bunch
of
other
models,
because
you
know
you
have
like
nervous
system
models,
your
movement
models
but
development's
quite
a
different
process
than
that.
It's
like
how
does
that
thing
get
there,
and
so
we
have
a
lot
of
different
models
that
we
use
that
are
different
than
this,
but
what
they're
doing
here
is
they're
laying
out
a
lot
of
these
models
for
people
to
read
about
they're.
A
You
know,
there's
software,
their
simulation,
you
can
see
it
do
cool
things,
but
you
need
to
know
the
mathematics
or
the
theory
behind
a
lot
of
this.
So
this
wasn't
well.
I
guess
this
is
maybe
a
finished
draft,
but,
like
the
idea
is
that
you
have
you
know
a
couple
of
components.
You
have
the
basic
equations
of
motion,
so
you
have
nervous
system
simulation.
A
A
Model
there's
a
single
compartment
model,
there's
the
multi-compartment
model,
and
so
these
are
ways
you
can
model
neural
activity
or
how
to
generate
action
potentials
within
the
cell,
and
you
know
you
because
you
don't
know
how,
for
example,
you
know
you
might
have,
you
might
have
a
very
realistic
model
with
all
the
parts
that
you
need,
but
you
don't
know
how
it
actually
does
its.
You
know
how
it
actually
proceeds
with
its
with
its
process,
so
to
speak,
and
so
these
are
actually
from
the
literature
on.
A
A
So
if
you've
seen
the
stuff
with
that,
cybernetic
has
done
it's
a
it's
a
group
within
open
worm.
They
have
these
biophysical
models
where
they
have
like
a
worm
in
on
a
surface,
and
the
surface
could
be
like
a
gel
or
something
and
they
simulate
the
movement
against
that
surface,
and
they
also
look
at
hydrodynamics
where
the
worm
is
in
in
a
liquid
environment,
which
it
usually
is,
and
so
they
use
these
particle
simulations
that
have
like
hydrodynamic
properties.
A
So
they
have
like
different
drag
forces
and-
and
you
know,
the
worm
is
generating
forces
against
this,
and
it's
sort
of
you
know
gives
you
a
model
of
what
is
sort
of
the
path
of
least
resistance
for
worm
for
the
worm.
So
you
know,
why
does
the
worm
move
in
certain
ways
or
see
elegance
as
a
couple
of
modes
of
movement?
A
Are
those
like
the
most
efficient
modes
of
movement
given
its
physical
constraints,
or
you
know
that
that's
what
we
can
figure
out
with
this,
and
so
there
are
a
couple
things
the
way
out
there
different
algorithms
and
then
extending
this
to
modeling
biological
tissues.
So
they
pick
an
algorithm
that
they
think
is
effective.
A
The
smooth
particle
hydrodynamic
algorithm
and
then
they
apply
it
to
the
model
and
then
they
get
results,
and
this
just
doesn't
worry
about
the
results.
It
just
lays
out
the
method,
and
then
this
boyle
cone
model
of
neuromechanical
coupling,
which
is
published
in
the
literature,
says
netocon
and
boyle.
A
A
So
this
is
a
nice
set
of
models
and
then
putting
it
all
together,
which
is
a
page
which
is
kind
of
pulling
together
all
of
these
models
into
one
simulation
environment
and
then
references.
So
you
know
this
is
just
like
this
is
laid
out
like
a
math
paper.
Almost
where
you
have
the
methods
you
have
the
equations.
A
A
So
I
just
wanted
to
show
that,
for
you
know,
thinking
about
this
mathematics
of
evil
worm
and
where
that
would
go,
but
also
this
is
very
good
if
you
really
want
to
understand
what's
going
on
in
open
world,
because
I
think
they're
you
know
not,
there
isn't
a
lot
of
like
transparency
there
in
terms
of
like.
If
someone
wanted
to
know
you
know
how
do
you
know
that
a
worm
neuron
does
this?
A
A
A
This
includes
the
onset
of
synchronization
in
a
network
of
coupled
oscillators
or
the
emergence
of
an
epidemic
state
within
a
population.
So
we
talked
about
coupled
oscillators
before
there
are
these
neurons
that
are
connected
in
a
network.
So
this
these
two
neurons
are
coupled.
They
have
this
internal
process
that
generates
action.
Potentials
then
there's
this
output,
which
is
you
know
it's
like
a
rhythmic.
You
know
regular,
regular
pattern,
and
that
gives
us
this.
A
You
know,
so
you
can
synchronize
these
oscillators
or
coupled
in
this
way
to
generate
this
patterned
output
or
the
emergence
of
an
epidemic
state.
And,
of
course,
we
know
from
the
pandemic
that
when
there's
a
lot
of
transmission
of
a
virus,
you
can
end
up
if
the
transmission
rate
is
high
enough
with
an
epidemic
or
even
a
pandemic,
where
you
can't
control
the
spread
of
things
and
they
remain
in
the
population.
You
can't
knock
them
down.
B
A
A
A
As
this
sac
starts
to
glow
and
turn
off
the
all
the
fireflies
in
a
certain
area
will
us
will
synchronize
their
flashing,
and
so
this
is
something
that
happens.
Sort
of
you
know
it's
hard
to
say
that
they
coordinate
it
closely.
It
just
happens
over
time
with
some
limited
interactions
and
then
percolation
is
where
you
have
something
that
runs
through
another
medium
like
water
running
through
rocks.
A
So
if
you
have
coffee
grounds
at
the
beginning
of
your
coffee,
it's
very
loose
and
grainy,
but
then,
when
you're
done
after
you've
run
the
water
through
it,
it's
very
sticky
and
it's
you
know,
sort
of
almost
like
a
solid,
and
so
this
is
this
percolation
transition
or
water
going
through.
It
will
change
the
sort
of
the
physical
state
of
it.
So
they
they.
A
Is
a
general
property
of
a
lot
of
systems,
so
this
is
a
lot
of
statistical
mechanics,
I'm
not
going
to
get
into
this.
But
there
are
these.
You
know
there
are
these,
so
you
have
what
they
call
critical
order.
Disorder
transitions
as
a
system
parameter
is
varied
and,
generally
speaking,
if
you
vary
the
parameters
of
a
system.
A
So
if
you
vary
the
parameters
of
some
some
system
amongst
the
fireflies
as
they're
synchronizing
they're
flashing
or
you
vary
the
parameters
of
water
as
it's
going
through
coffee,
you
can
get
these
kind
of
different
transitions
and
statistical
mechanics
that
could
be
a
continuous
or
discontinuous
transition
order.
Change
at
the
transition
point
we're
talking
about
first
order
phase
transitions,
which
is
the
thing
that
people
are
really
excited
about
in
complex
systems,
research
because
they
lead
to
this
explosion,
explosive
change
of
systems,
properties,
and
so
this
for
a
wide
variety
of
complex
systems.
A
A
A
The
reason
I
picked
this
paper,
I
think
it
does
apply
there,
there's
some
nice
developmental
applications
for
this,
and
we've
talked
about
this
off
and
on
in
the
group-
and
I
don't
know-
maybe
not
recently,
so
maybe
it
doesn't
make
a
lot
of
sense
to
people
now,
but
the
idea
is,
you
know
that
there
are
these
symmetry
breaking
events
in
development,
so
in
an
embryo
you'll
have
symmetry
breaking.
You
also
have
semi,
which
is
like
where
you
have
symmetry.
You
have
a
right
and
a
left.
A
You
have,
you
know
things
that
are
undifferentiated
and
then
you
have
suddenly.
You
have
this
breaking
of
the
symmetry.
Where
say
like
two
cells,
one
on
the
right,
one
on
the
left
are
different
sizes
or
they
one
forms
a
tissue
and
the
other
one
doesn't
or
you
know
the
tissues
are
differentiated
within
the
embryo
or
you
know,
there's
some
difference
in
the
different
parts,
so
there's
a
symmetry
breaking,
and
this
is
one
of
the
things
when
you
see
the
symmetry
breaking.
A
A
Actually,
quite
broadly
relevant
and
one
of
the
arguments
about
the
or
one
of
the
criticisms
of
this
approach
is,
you
know,
can
you
take
like
mathematical
equations
and
apply
them
to
such
a
broad
range
of
systems
and
say
that
that's
a
universal
property,
some
people
say
no
well,
obviously,
if
you're
a
statistical,
mechanics
person,
you
say
yes,
but
this
applies,
I
think,
more
broadly,
to
machine
learning,
where
we
have
machine
learning,
algorithms,
that
you
know
we
apply
to
every
type
of
data,
and
maybe
that's
not
the
best
strategy.
Maybe
that's
not
something
that.
B
A
Picking
up
on
these
universal
properties,
or
are
we
just
kind
of
applying
it
broadly
and
not
really
describing
the
system
evenly?
You
know
one
system,
we
might
describe
very
well
another
system,
we
might
not
describe
it
very
well,
but
in
their
defense,
of
course
you
know
they
do
differentiate
between
different
types
of
systems,
so
they
say
well,
you
know
you
can
use
this
on
a
wide
range
of
systems,
natural
mechanical,
but
it
has
to
have
certain
properties.
So
that's
what
they're
laying
out
here.
A
Next
paper
I'm
going
to
talk
about
is
this
towards
a
physical
understanding
of
developmental
patterning.
I
think
this
kind
of
ties
in
with
the
other
paper
now
the
other
paper
was
hardcore
physics.
This
is
more
hardcore.
I
guess
it's
in
a
biology
journal
developmental
biology
journal.
Oh
it's
in
nature,
but
it's
you
know
it's
it's!
A
It's
not
physics,
so
there's
just
towards
the
physical
understanding
of
developmental
patterning,
so
the
abstract
is
the
temporal
coordination
of
events
at
cellular
and
tissue
scales
is
essential
for
the
proper
development
of
organisms
and
involves
cell
intrinsic
processes,
which
means
processes
within
the
cell.
That
could
be
gene
expression.
That
could
be
protein
synthesis,
just
things
that
happen
in
the
cell
and
that
can
be
coupled
by
local
cellular
signaling
and
instructed
by
global
signaling.
A
A
The
timing
and
structure
of
these
patterns
depending
on
organism
develops.
So
this
determines
how
its
developmental
trajectory,
if
you
want
to
think
about
a
simple
model
of
development,
it
would
be
that
it
would
be
these
compartments
with
a
larger
like
field,
their
signals,
their
processes,
internal
to
these
cells
and
then
that's
development.
If
you
get
different
outcomes,
it's
because
you
have
different
signals.
A
You
have
different
things
going
on
inside
the
cell,
so
traditional
developmental
genetic
methods
have
revealed
the
complex
molecular
circuits
regulating
these
processes,
but
are
limited
in
their
ability
to
predict
and
understand
the
emergent
spatiotemporal
dynamics.
So
now
we
add
another
aspect
which
is
the
spatio-temporal
dynamics,
and
you
can't
really
predict
that
using
like
molecular
biology.
In
other
words,
if
you
go
into
one
of
those
compartments
and
you
probe
those
endogenous
or
those
internal
mechanisms,
it
doesn't
tell
you
a
lot
about
the
broader
embryo.
A
It
just
tells
you
about
what's
going
on
in
that
cell,
so
we
need
to
have
approaches
for
physics
to
capture
a
lot
of
these
things
that
are
going
on
across
multiple
cells,
and
so
this
signaling
is
one
aspect,
but
there's
something
more
to
it.
There's
a
physics
that
is
involved
in
like
how
the
signaling
gradients
are
set
up,
and
things
like
that
over
space
and
over
time,
so
combined
with
advances
in
imaging
and
computational
power.
Such
approaches
aim
to
provide
insight
into
timing
and
patterning
and
developmental
or
developing
systems.
A
A
Now
that's
quite
a
bit
different
than
what
we
saw
in
the
last
paper
where
transitions
are
these
things
that
are
different
states,
and
then
you
want
to
capture
like
the
thing
that
sort
of
pushes
it
to
a
new
state.
So
you
know
you
have
these.
You
characterize
the
system,
you
characterize
these
phase
transitions
and
then
that
tells
you
what
you
need
to
know
about
the
system
in
this
case.
A
B
A
They're
also
saying
that
that's
an
incomplete
description,
so
these
people
are
critiquing
that
view,
then,
of
course
you
can
couple
genes
and
their
regulation
into
a
gene
regulatory
network.
So
each
of
these
component
compartments
have
a
grn
or
a
gene
regulatory
network
that
is
informative
and
gives
you
that
same
output
that
you
would
get
for
a
a
set
of
equations
that
would
describe
a
phase
transition.
A
So,
however,
it
has
been
challenging
to
understand
how
these
components
give
rise
to
the
dynamics
of
patterning
in
part,
because
a
list
of
components
is
very
likely
far
from
complete
and
the
interactions
between
them
remain
uncertain.
So
when
we
build
a
genetic
regulatory
network,
we
usually
pick
certain
genes.
A
We
don't
also
don't
understand
fully
how
there
are
other
downstream
processes
that
affect
say
that
the
synthesis
of
proteins
or
the
synthesis
of
signals
that
travel
outside
the
cell-
and
there
are
a
lot
of
things
we
don't
have-
that
are
really
precise.
We
don't
really
have
precise
knowledge
of
not
so
much
the
processes,
but
like
the
the
numbers
that
you
put
on
them.
A
But
these
other
processes,
like
developmental
you,
know,
phase
transitions
or
changes
in
state
symmetry
breaking
those
all
occur
at
very
different
time
scales
like
on
the
time
scale
of
days,
and
so
this
is
a
big
difference
in
time
scale.
So
this
is
something
that
needs
to
be
filled
with.
There
needs
to
be
some
link
between
the
two
time
scales.
A
A
So
if
a
genetic
oscillator
is
driving
this
process
of
differentiation,
it
looks
different
than
if
you're
getting
signaling
across
cells.
So
this
wavefront
is
where
you
have
a
signaling
dependent
interface.
This
arrow-
and
this
tells
you
you
know
this
kind
of
mediates-
this
kind
of
change,
where
it's
there's
a
striping
that
goes
on
these
two
different
stripes
of
colored.
You
know
the
cells
have
different
colors,
as
this
wave
goes
through
it
with
a
genetic
oscillator.
A
The
single
cell
level
is
the
thing,
that's
driving
change,
so
you
get
the
sort
of
mosaic
of
different
colored
cells.
You
get
striping,
but
it
looks
a
lot
different,
so
they're,
you
know,
depending
on
what
you're
claiming
is
the
mechanism
for
change.
You
know,
gives
you
different
outcomes,
and
this
is
a
genetic
timer
where
you
have
different
kinds
of
striping,
where
the
cell
just
changes
state
over
time,
so
level
of
maturity
determines
its
state,
and
so
that
can
actually
lead
to
striping
in
a
different
way,
and
so
there
are
different
mechanisms.
A
B
A
A
They
also
have
things
called
phase
waves,
which
are
fast
structures
that
that
move,
they
call
them
kinematic
structures
that
emerge
in
the
isolated
cells
of
a
single,
stable
biochemical
state.
So
these
are
things
that,
like
scroll
waves
and
things
like
that,
that
move
through
the
embryo.
So
these
are
things
where
you
have
different
levels
of
activity
for
this
gene
product
and
it
moves
through
the
embryo
like
this.
So
there's
this
wave
front
and
it's
controlled
by
the
circuit
of
genes.
A
So
this
is
you
know,
and
then
this
shows
you
kind
of
how
it
works.
It
organizes
things
at
that
global
level,
so
you're
linking
then
you're
linking
gene
expression
here
in
different
cells.
With
this
thing
that
goes
outside
the
cell
and
coordinates
the
changes
in
state
across
cells,
because
it's
going
out
into
the
cell,
you
know
into
the
extracellular
matrix
as
they
call
it
and
it's
moving
across
as
a
front.
So
you
have
these
a
lot
of
stuff
produced
over
time
by
genes.
A
So
that's
one
way
we
can
link
those
time
scales,
so
there
are
a
lot
of
different
types
of
waves
and
wave
fronts.
They
get
into
these
things
here:
different
model
organisms,
and
they
also
model
molting
and
c
elegans,
which
is
where
c
elegans
has
like
a
exoskeleton
that
it
molts
over
time.
They
just
change
their
skin.
Basically,
it's
a
cyclic
process
that
occurs
four
times
of
the
periodicity
of
8
hours,
so
these
are
very
regular
patterns.
A
These
molting
cycles
are
driven
by
some
of
these
pattern
formation
mechanisms
as
well.
So
you
can
model
a
lot
of
things
with
these.
You
can
model
genetic
timers
where
you
have.
You
know
where
there's
something
set
up
by
a
gene
expression
network
that
or
a
gene
regulatory
network
that
happens
at
regular
intervals.
You
have
these.
You
end
up
with
an
output
like
these
us
like
these
oscillatory
waves,
but
now
it's
a
gene
expression
network.
Let's
do
it
that's
responsible
for
this.
A
You
have
temporal
phase
shifts
and
spatial
phase
shifts
that
are
possible
to
model.
So
there's
a
lot
of
stuff
in
this.
That's
not
computational
as
a
paper,
but
they
kind
of
give
you
this
these
models,
these
different
state
diagrams,
and
they
show
you
how
they
map
out
in
a
graph.
So
you
can
use.
This
is
very
useful.
A
lot
of
the
stuff
in
this
paper
can
be
used
to
model
a
lot
of
different
outputs.
A
So
there
are
a
lot
of
different
things
we
can
explore
with
that.
If
people
are
interested,
so
I
think
that's
it
for
today,
I'm
going
to
I'm
not
going
to
go
through
that
whole
paper,
but
I
just
wanted
to
give
people
a
taste
of
this
sort
of
stuff.
So
surely,
did
you
have
any
questions
about
any
of
that
or.