►
From YouTube: DevoWorm (2021, Meeting 12): Mathematics of DevoWorm, Paleodevelopment, Wiring Neural Circuits
Description
DevoWorm Onboarding Guide, Mathematics of DevoWorm, Project Board and Open-source Maintanence, Neural Cellular Automata, Paleontology of Development, and Wiring up Neural Circuits. Attendees: R Tharun Gowda, Abhishek Tiwari, Jesse Parent, Yash Vadi, Krishna Katyal, Bradly Alicea, Mayukh Deb, Mainak Deb, Susan Crawford-Young, Vrutik Rabadia, Zura Isakadze, and Muhammed Abdullah
A
B
B
B
B
A
A
A
A
A
D
A
Says,
plus
one
for
inspired:
ai
bio-inspired,
ai,
so
yeah
zura
yeah.
We
we
do
a
fair
amount
of
stuff
with
bio-inspired
ai
we're
interested
in
that
we're
kind
of
you
know
kind
of
it's
a
theme
in
the
group
and
then
there's
another
group
that
we
have
that
does
more
work
on
in
this
area.
So
we
can
talk
about
that
later.
If
you're
interested
in
participating
in
that
group
as
well.
A
Yeah
yeah,
so
so
welcome
to
the
meeting
today.
I
don't
know
why
my
camera
is
not
working.
It's
keep
sending
me
a
message
failed
to
access
your
camera,
although
I'm
looking
at
it
right
now
and
the
lights
on,
so
I
don't
know
why
the
the
platform
here
isn't
letting
me
access
my
camera,
but
I
might
be
able
to
share
my
screen
if
I
can
share
my
screen.
That
will
be
just
as
good.
A
First
of
all,
I'd
like
to
thank
everyone.
Who's
who's
been
communicating
with
mayok
and
myself
on
gsoc
applications
such
google
summer
of
code
applications.
A
number
of
you
have
sent
your
applications
for
us
to
read
over
or
I've
read
them
over.
I
don't
know
about
my
hook,
but
and
suggested
you
know
how
to
make
it
how
to
improve
the
how
to
improve
the
application
to
make
it
better
and
then,
of
course,
we
talked
about
the
timeline.
A
I
think
the
timeline
is
the
most
important
part
of
the
application.
So
those
are
you
applying
really
to
any
kind
of
grant,
or
you
know
something
where
you
have
to
propose
a
project.
I
think
it's
always
a
good
idea
to
really
kind
of
work
out
your
timeline
of
deliverables
and
like
get
that
down
and
for
grants.
A
So
that's
I
mean
I
was
gonna
put
that
into
the
onboarding
guide,
but
I
think
that's
probably
okay,
I
don't
I
don't
know
how
to
put
it
in
there
without
making
it
a
big
long,
a
demo
or
or
tutorial.
So
I'm
just
gonna
say
that
in
the
meeting
and
just
tell
people
that
so,
let's
see
if
I
can
share
my
screen
here,
I
think
I
can
actually.
A
Okay.
Can
everyone
see
my
screen
here?
Yeah,
okay,
good,
so
first
thing,
I'd
like
to
talk
about
is
diva
learn.
So
this
is
the
this
is
the
github
organization
that
we
have
and
if
you're
applying
to
project
3.1
you'll
be
contributing
to
this
repository.
Diva
learn,
and
I
see
that
rudick,
for
example,
has
made
a
recent
pull
request
here
and
a
number
of
you've
made
pull
requests
in
this
repository.
A
So
I
don't
there's
one
open
right
now,
but
we
you
know,
that's
we
always
have
people
making
we've
had
maybe
about
10
to
15
pull
requests
this
season.
So
that's
good.
I
don't
know
if
I
can
get
to
the
pull
accepted,
pull
requests
or
not,
but
that's
okay,
because
I
just
wanted
to
point
out
that
there
have
been
a
lot
of
contributions
to
this
divo
learn
platform,
and
this
is
the
platform
that
we're
talking
about
here.
A
So
it's
it's
a
nice
incentive.
But
I
think
this
is
the
divo
learn
sort
of
tutorial
here.
So,
if
you're
interested
in
project
3.1,
you
should
go
to
this
repository
and
you
know,
look
it
over
make
sure
that
you're
all
right.
So
my
yoke
has
made
a
link
here
so
yeah.
He
already
accepted
the
pull
request.
Thank
you.
A
Well:
okay,
anyways
that
I
don't
know
where
the
list
of
contributors
is,
I
think,
actually
you
can
see
it
on
the
front
page
here
you
have
your
14
contributors
in
this
repo.
So
actually
I
think
it's
in
the
organization,
but
it's
been
mostly
this
repo
and
then
we
also
have
other
things
going
on
in
this
platform.
So
this
platform
was
meant
to
be
like
something
that
incorporated
the
software,
which
is
a
pre-trained
model
for
microscopy
data.
Then
we
have
other
types
of
things
like
the
devozu,
which
is
a
list
of
model
organisms.
A
This
isn't
quite
ready
yet,
but
it's
being
flashed
out.
We
also
have
some
other
repos,
like
c
elegans
divalern,
which
is
the
web
app
to
support
the
cl
against
part
of
the
diva
learn
library.
So
there
is
a
library
of
models
that
we
use
as
well.
It's
associated
with
the
diva
worm,
ai
web
interface,
that
I
I've
been
telling
people
about,
and
so
we're
trying
to
put
this
all
together.
A
It's
still
kind
of
not
quite
you
know
in
a
in
a
place
where
we
have
like
you
know,
I
mean
we're
trying
to
promote
it,
but
it's
like
it's
still
a
work
in
progress
and
that's
true
of
most
open
source
projects
as
you'll
find.
A
lot
of
them
are
sort
of
built
gradually
and
they
have
releases,
and
you
know
just
they
get
worked
out
now.
Sometimes
they
do
a
formal
release
for
an
open
source
project,
but
this
in
this
case
we're
not.
You
know,
we'd
like
to
get
things
out
and
get
things
built.
A
So
it's
you
know
it's.
We
also.
You
know
we
have
the
inter
web
interface
for
a
lot
of
this
through
the
diva
learn
ai,
which
I've
put
the
link
in
the
slack
for
that
and
that's
a
series
of-
and
some
people
have
asked
me
questions
about
that
in
terms
of
you
know
what
kinds
of
models
are
you
accepting
for
your
model?
A
Library-
and
the
answer
is,
you
know
it's
mostly
machine
learning
models,
mostly
for
microscopy
data
to
segment
and
analyze
microscopy
data,
but
we're
also
interested
in
other
types
of
models
such
as
simulations
cellular,
automata
or
other
types
of
models
like
that,
and
so
we
have
we're
building
other
repositories
in
here
to
accept
some
of
those
models
as
well.
We
also
have
a
series
of
data
science,
demos
where
people
have
also
contributed
to
this,
so
this
is
actually
okay.
This
is
by
repo.
A
A
A
Oh
well,
we
have
the
this,
so
this
is
the
onboarding
guide.
I've
told
people
about
and
I'll
put
this
link
in
the
chat
as
well.
So
this
is
our
onboarding
guide,
and
this
is
not
just
for
google
summer
or
code,
although
it's
sort
of
organized
like
that
right
now,
but
if
you're
interested
in
joining
diva
worm
and
getting
involved.
A
This
is
a
good
primer
to
sort
of
give
you
some
context
about
where
we
are
so,
you
know
we're
in
the
open
room
foundation
you're
in
the
open
arm
slack
when
you
come
into
the
slack
and
we're
a
couple
of
channels
in
the
open
worm,
foundation
slack,
but
we're
also
sort
of
a
semi-independent
group.
We
have
a
website,
we
have
a
github
organization
and
we
have
a
youtube
channel.
A
A
A
We
had
another
tutorial
and
I
can't
remember
what
it
was,
but
if
you're
interested
in
learning
more
about
openworm,
I
would
go
there
or
into
these
supplemental
materials
and
figshare,
which
is
basically
all
those
presentations
plus
some
side,
references
as
well,
and
then
you
know,
there's
an
introduction,
open
source,
there's
information
about
the
gsoc
project.
So
if
you're
interested
in
the
gsat
projects,
you
can
learn
more
here.
A
This
is
a
good
set
of
first
issues,
so
we
still
need
to
work
on
the
good
set
of
first
issues,
good
first
issues,
and
it
might
just
be
a
matter
of
putting
these
up
in
the
onboarding
guide.
I
wanted
to
make
it
so
that
people
could
go
and
grab
a
first
issue
and
then
potentially
get
started,
and
this
would
be
maybe
something
beyond
just
g
suck,
but
I
think
you
know
people
are
always
looking
for
things
to
do
and
we
don't
really
know
how
to
guide
them
directly.
A
It
also
has
a
little
bit
of
information
on
model
organism
biology,
so
those
of
you
working
on
project
3.1.
These
are
some
basic
resources
for
learning
about
c
elegans.
From
this
side
you
know
idea
about
unlocking
the
secrets
of
the
brain,
so
you
know
a
big
overview
of
why
c
elegans
is
important
to
some.
You
know
examples
of
micrographs
of
the
brain
and
then
some
links
to
some
basic
references
on
biology
and
microscopy.
A
A
They're
they're,
very
ubiquitous
around
the
world
likes
much
like
c
elegans,
actually
but
they're,
a
very
different
type
of
organism.
So
I
would
look
at
these
references
if,
if
you
want
to
get
some
background
for
your
proposals
and
then
finally,
if
you're
applying
to
3.3
there
is
this
information
about
axle
bottles.
So
I
have
an
introduction
to
axolotls
and
an
introduction
to
axolotl
development,
and
I
don't
know
if
susan
or
someone
else
who
is
biologically
oriented
wants
to
add
to
this
in
terms
of
tutorials.
A
You
can
send
me
information,
or
you
know,
links-
and
I
can
put
them
in
here
because
I
don't
know
you
know.
Ideally
I'd
like
to
be
able
to
have
a
couple
more
model
organisms
in
here.
It's
not
a
me
of
immediate
need,
but
we
you
know
if
we
want
to
sort
of
get
people
involved
in
some
of
these
other
types
of
projects
that
would
be
nice,
then.
Finally,
we
have
biological
data
sets.
A
Some
of
you
asked
me
about
biological
data
sets
devozu,
which
has
a
lot
of
different
data,
sets
not
only
from
c
elegans
diatoms
but
from
axolotls,
and
actually
I
don't
think
we
have
well.
We
have
the
axolotl
dataset,
but
I
don't
usually
make
it
public.
Then
we
have
like
zebrafish
and
drosophila
and
even
spiders.
So
there
are
a
lot
of
different
model
organisms
to
work
with.
A
A
Analysis,
so
you
can
read
that
over
those
are
like
once
you
get
the
data,
then
what
can
we
do
with
it?
What
are
why
is
it
interesting
to
to
build
a
machine
learning
model
to
extract
data
points,
because
that's
what
some
of
the
gsoc
projects
are,
and
I
want
to
make
sure?
Actually
I
want
to
add
to
that
a
little
bit
more,
but
to
make
it
salient
to
people
why
you
want
to
do
this,
so
I
would
like
to
okay.
So
we
have
any
questions
about
that
or
oh.
A
A
I
didn't
see
that
before
so
we
have
riddick,
we
have
my
minok,
we
have
shruti,
we
have
throne,
we
have
up
taha,
we
have
ravi
jainal,
so
we
have
a
lot
of
different
people
contributing
to
this
jesse
parent,
a
lot
of
different
people
contributing
different
types
of,
and
you
know
it
doesn't
have
to
be
a
huge
commit.
It
can
just
be
a
very
small
commit
and
you
can
go
ahead
and
make
the
commit,
and
it's
usually
mainoku's,
myocus
accepting
the
pull
requests
be
patient.
A
You
know
you
might
have
to
wait
a
day
or
two,
but
we'll
get
to
you
and
accept
the
pull
request.
If,
if
there's
something
wrong
with
your,
you
know,
if
there's
some
stylistic
problem
or
some
alignment
problem,
you
know
then
we'll
we'll
let
you
know
where
maya
is
been
very
good
at
like
spotting
errors
and
the
pull
requests.
So
thank
you
mayor
for
that,
and
we
have
you
know
it's.
A
I
think
we're
doing
pretty
well
in
terms
of
keeping
the
quality
of
code
up
because
that's
been
a
problem
in
a
lot
of
open
source
projects
is
accepting
substandard
code,
and
I
have
an
article
on
that.
I
don't
have
it
with
me,
but
well
let
it
be
known
that
that
can
be
a
problem,
and
so
you
want
to
be
careful
about
that.
So
good.
Any
questions
at
this
point.
A
Okay,
I
want
to
thank
jesse
parent
for
going
through
the
the
issues
board
and
and
sorting
it
out.
We
haven't
like
talked
about
our
issues
board
in
a
while,
but
we
have
an
issues
board
for
the
larger
diva
one
group,
so
diva
learn
is
like
a
small
subset
of
what
we
do
here.
We
do
a
lot
of
other
things.
We
do
things
with
submitting
projects.
We
do
a
lot
of
stuff
on
sort
of
the
boundary
of
computational,
biology
and
biology,
and
you
know
ai
and
data
science
and
machine
learning.
A
So
we're
doing
we
have
our
fingers
in
a
lot
of
pies,
and
so
it's
critical
that
we
have
these
boards
for
reviewing
topics
and
items.
So
we
have
all
these
github
issues,
and
some
of
them
are
a
lot
of
them.
Aren't
related
in
this
board
to
specific
code
issues
so
like
in
the
diva
learn
board.
You'll
find
issues
that
relate
to
some
code
issue,
something
in
the
in
the
code
base
in
the
diva
worm
discrete
meetings
board.
A
These
are
issues
that
are
more
general
that
are
like
things
that
we
need
to
do
concepts
even
and
we'll
split
those
into
different
issues
later
on.
So
we
have
a
number
of
things
that
are
in
progress,
finished
action
items,
so
we
have
these.
Actually,
this
action
item
is
finished,
but
we
have
a
lot
of
things
that
are
sort
of
outstanding.
A
We
have
a
lot
of
papers
like
this
morphogenesis
and
deep
learning
paper
that
I
talked
about
last
week
opens
open
worm
abstract
for
international
c
elegans
meeting
that's
actually
been
submitted,
but
it's
we're
waiting
to
hear
back
for
it
from
for
it.
Some
of
these
things
are
find
a
place
to
submit
different
abstracts
make
network
science.
Submission
is
finished,
this
bessillary
non-neuronal
cognition
paper.
This
is
something
that
we'll
be
turning
to
in
the
next
couple
weeks.
A
E
A
A
A
There
we
go
that,
that's
something
that,
like
that'll,
make
it
clear
that
that's
not
like
just
getting
the
data
that
we
already
have
so
so
yeah.
I
look
forward
to
seeing
that
this
work
on
ann
and
bnn's
extended
abstract.
I
think
that's
finished
where
we
we
submitted
this
to
the
a
live
conference,
and
I
don't
know
what,
if
it's
been
accepted
or
not,
but
we
will
probably
work
on
like
editing,
you
know
making
it
better
making.
You
know
there
will
be
opportunities
to
improve
upon
it.
A
So
I
want
to
I
want
to
say
it's
finished,
but
it
will
probably
come
back
on
our
radar.
Christian
and
myself.
We
were
working
on
these
conference
presentations
I'll
talk
about
that
in
a
little
bit
update
people,
oh
updating
the
divaware
ml
lectures
so
last
year.
Of
course,
we
did
the
steve
worm
ml
series
and
I
still
want
to
update
these,
and
I
kind
of
was
doing
this
at
the
end
of
last
year
and
then
I
stopped
and
I
want
to
get
back
into
it,
but
I've
been
very
busy.
A
A
That's
a
lot
of
embryo
animations
and
segmentation.
That
was
something
that
I
was
kind
of,
hoping
that
we
could
get
through
this.
You
know
in
this
gsoc
period,
but
I
haven't.
I
don't
know
if
we've
done
much
with
that,
but
we'll
we'll
make
that
like
it'll
put
that
on
hold
for
now
and
then
see
if
we
get
some
applicants
for
that,
how
do
we
do
this
updates
on
axolotl
data
and
analysis
in
the
action
items
column
recruit?
People
is
diva
learning
contributors,
we're
kind
of
doing
that
through
g
suck.
A
So
that's
you
know
probably
in
progress
this
demon
bibliography.
So
I
I've
had
some.
I
think
jesse
and
dick
were
talking
about
doing
this
and
maintaining
it.
So
I
don't
know
if
we
have
any
updates
on
that,
but
I'm
gonna
assume
that
we're
still
working
on
that
anyways
we're
trying
to
make
a
bibliography
out
of
are
different.
A
We're
also.
You
know
if
you
have
anything
that
you
want
to
make
a
youtube
tutorial
for,
like
I
showed
that
there
we
have
tutorials
for
data
science,
like
you
know,
digital
notebooks,
for
data
science,
but
if
you
want
to
make
a
youtube
tutorial,
maybe
on
a
paper
or
on
some
other
technique
that
you
think
is
important
to
go
over
on
youtube.
Like
you
know,
record
yourself
talking
about
a
paper
or
about
you
know,
prepare
some
powerpoint
slides
and
go
through
them
and
describe
something
for
a
wider
audience.
A
We
can
put
it
up
on
our
youtube
channel
and
we
can
you
know
that
would
be
something
that
would
be
perhaps
get
a
lot
of
views.
It
would
be
an
important
reference
for
some
people.
So
that's
something
to
think
about
this
breed
evil
and
paper
from
jaw
submission.
That's
sort
of
I'm
going
to
put
this
as
an
action
item,
because
I
keep
glossing
over
it
every
time.
I
get
some
free
time
and
I
I
want
to
get
to
this,
but
I
haven't
been
able.
A
We
submitted
a
paper,
the
journal
of
open
source
science
on
evilern,
but
they
didn't
think
it
an
important
enough
reference
right
now
so
a
resource.
So
I
think
you
know
we're
going
to
create
a
bigger
paper
that
would
end
up
being
like
something
like
a
a
preprint
or
a
technical
paper
and
then
maybe
publish
it
somewhere
in
the
future.
A
The
openworm
foundation
is
creating
a
docker
file
with
all
of
the
major
projects
in
it
and
they
want
to
have
something
you
can
execute
an
executable
file
that
you
can
execute
and
show
sort
of
a
demonstration
of
the
area
that
you're
working
in
so
for
diva,
warm.
That
would
be
like
some
sort
of
model
of
an
embryo.
A
I
would
imagine,
and
so
we're
looking
you
know
I
think
bourgeois
was
working
on
this
for
a
while
and
then
he
dropped
it,
and
I
don't
know
where
we
are
with
that,
but
I
would
like
to
maybe
continue
with
that
in
some
capacity
people
are
interested
in
contributing.
Let
me
know,
I
think
this
is
pretty
much
off
the
radar,
and
this
is
axel,
auto
montaging.
A
This
is
contingent
upon
the
pro
the
gsoc
project,
so
I
think
that's
a
good
review
of
our
major
tasks.
Thank
you
once
again
to
jesse
for
making
more
sense
of
that
than
it
was
last
week.
It
looks
a
lot
cleaner
now
and
you're
going
through
some
of
those
issues,
but
we'll
I
want
to.
I
haven't
gone
through
that
board
in
a
while.
So
I
wanted
to
kind
of
give
an
update
on
that.
A
We
have
someone
in
the
chat
here.
Okay,
myoke
says
hours,
I'm
working
on
improving
diva
at
around
mid
april.
Right
around
the
time
when
the
gsoc
student
application
period
ends.
Everyone
is
welcome
to
help
me
in
improving
diva
learn
regardless
of
g
suck.
So,
yes
thank
you
mayor
for
that
statement.
If
you
want
to
contribute
to,
can
you
contribute
to
anything
in
our
group
without
being
accepted
to
gsoc?
The
answer
is
yes,
if
you
feel
that
that's
something
that
will
help
you
in
your
studies
or
your
career,
that
would
be
fine.
A
We're
we'll
absolutely
willing
to
work
with
you,
but
right
now
we're
doing
the
g
suck
student
applications.
So
the
application
period
starts
today
and
it
goes
through
mid-april,
and
so
that's
a
good,
but
a
good
opportunity
for
people
to
get
some
experience
in
coding.
In
that
so
jesse
would
like
to
speak.
F
Yeah,
I'm
actually
using
the
hand
function
for
once
so
one
thing,
because
I
I'm
looking
at
my
notes
from
last
week
and
looking
at
the
board-
and
I
don't
know
this
one
was
on
the
board
or
not
so
I
might
miss
it
little
connection
error,
but
there
were
a
few
things
from
that.
I
took
notes
off
or
stuff
to
do.
One
of
them
was
the
boring
building
stuff
is
that
in
there?
F
A
A
This
I
don't
know
if
it
might
even
be
a
book
at
some
point
where
we're
looking
at
this
period
of
evolution
called
the
boring
billion
it's
between
the
time
you
have
like
the
establishment
of
life
and,
like
you
know,
complex
multicellular
life.
So
you
get
like
a
long
period
where
there's
very
little
going
on
in
evolution,
then,
all
of
a
sudden
you
get
like
five
or
six
kingdoms
that
arise
at
once,
or
you
know
within
about
half
a
million
years,
and
so
this
is
like.
Why
would
that
happen?
A
There
are
key
oxygenation
events
that
sort
of
prime
the
pump
for
this
explosion
of
diversity
in
life,
but
like
this
period,
where
there's
nothing
going
on,
we
don't
really
know
beyond.
Maybe
like
some,
you
know
ecological
factors
why
that
would
be.
I
mean
you
know
it.
So
you
know
it's
it's
a
pretty
open
area
because
there
isn't
a
lot
of
good
data.
You
know
it's
like.
We
have
a
lot
of
we're
doing
a
lot
of
looking
at
the
paleontological
data.
We're
looking
at
you
know
we're
considering
diff
we're
creating
visualizations.
A
There's
there's
a
lot
going
on
there
and
it's
not
like
strictly
developmental
and
it's
not
strictly
c
elegans.
But
it
is.
You
know
it's
an
interesting
topic,
I
think,
to
kind
of
get
involved.
You
know
started,
you
know,
get
people
thinking
about,
maybe
like
evolution
or
you
know,
simulating
long
time
periods.
F
F
A
A
It's
like
a
network
approach
to
looking
at
like
organization
of
multicellularity,
so
using
a
mathematical
tool
called
euler
paths
to
determine
like
if
a
comedy
you
know
a
colony
of
cells
is
coherent
or
modular,
or
something
like
that
and
then-
and
I
gave
a
presentation
on
that
earlier-
this-
I
think
last
year
to
the
group
and
trying
to
get
that
back
up
and
running.
But
it's
you
know
something
that
probably
requires.
A
Maybe
some
more
thought
I
was
gonna
submit
a
paper
to
one
of
the
network
conferences
this
year,
but
I
don't
know
if
that's
doable
and
then
the
mathematics
of
diva
worm
is
something
that
someone
in
the
open
worm.
Earp
has
solicited
it's
it's
basically,
and
I
have
it
here.
Actually
let
me
go
to
the
I
have
it
open.
So
this
is
the
mathematics
of
diva
worm,
and
this
is
just
kind
of
a
first
pass
approximation
at
this.
A
Then
you
have
this
angle
of
movement
or
angle
of
division,
where,
when
the
cell
is
born
and
it
moves
away
from
its
origin
point
from
the
where
the
parent
was
in
the
embryo,
it
moves
to
a
new
position
and
they
move
around
a
lot
anyway.
So
you
can
characterize
that
with
an
angle,
a
phase
angle,
and
so
that's
one
you
know
equation.
A
Then
you
have
this
von
neumann
neighborhood,
which
can
characterize
cells
behaving
in
parallel,
they're
doing
a
lot
of
signaling
and
of
course
we
can
model
morphogenesis
using
a
cellular
automata.
So
we
have
these
these
that
this
is
the
basic
unit
or
kernel
of
that
type
of
cellular
arrangement,
the
von
neumann
neighborhood.
This
is
just
the
radius
of
the
neighborhood
here
and
then
you
know
we
have
these
data
structures
for
looking
at
developmental
function
like
node,
attaching
complex
networks.
A
F
A
F
A
A
A
A
A
There's
this
evolution
conference
at
the
end
of
april.
That
is,
we
have
a
couple
potential
things
that
euler
paths
for
life
might
be
an
abstract
there,
but
it's
not
really
for
that
audience,
I'm
trying
to
make
it
maybe
relevant
to
that
audience,
but
I
want
to
make
sure
that
you
know
wherever
it
gets
submitted.
It
doesn't
get
lost
in
the
shuffle
so
and
then
krishna
had
a
another
submission
he
wanted
to
make.
A
But
I
haven't
heard
an
update
on
this
recently,
so
we'll
keep
thinking
about
this
and
working
on
it,
and
hopefully
we
can
make
this
deadline.
There's
this
evil
learned
paper,
which
I
mentioned
before.
It's
doesn't
really
have
a
deadline.
It's
just
a
pre-print
right
now
we
have
this
growth
form
and
theory
of
deep
learning.
This
is
a
paper
where
we
consider
deep
learning,
as
maybe
a
way
for
to
just
maybe
describe
developmental
biology
or
contribute
to
development
developmental
biology
beyond
just
kind
of
like
you
know,
segmenting
images.
A
Okay,
okay,
so
verdict
said
I
am
still
working
on
something
and
improving
it
I'll
go
for
it
later.
So
that's
very
good.
Let's
go
back
to
the
submissions,
so
that's
for
this
some
people,
you
know
there
might
be
people
who
are
interested
in
contributing
to
that
who
haven't
seen
this
before
so
there's
this
bacillary
and
non-neuronal
cognition
paper.
So
this
is
a
paper
we're
going
to
be
putting
into
the
mathematics
of
diatoms
book,
and
this
is
april
30th.
A
A
A
We
don't
know
yet
what
track
it's
going
to
be
accepted
into,
but
this
is
the
conference
that
all
the
big
c
elegans
people
go
to,
and
it's
like
every
two
years-
and
so
you
know
it's
nice
to
get
exposure
there,
that
the
larger
c
elegans
community,
it's
a
lot
of
people
who
work
on
the
biology
of
c
elegans,
like
you
know
they
use
c
elegans
and
all
sorts
of
different
biological
systems.
It's
a
model
organism
for
you
know
starvation
for
aging
for
different,
is
you
know,
motor
dis,
motor
neuron
diseases?
A
E
E
A
A
Mute
please
whoever
that
is,
thank
you,
so
that
would
you
know
that's
going
to
be
you
know.
Hopefully
people
see
it
and
are
impressed
as
they
always
are,
with
a
lot
of
the
work
that
open
rooms
doing
so.
Let's
see
this
is
this
paper
which
didn't
happen,
this
embryo
networks
and
connect
them.
This
is
something
that
was
submitted
in
networks
2021,
but
you
know
it's
something
that
is
sort
of
ongoing.
It's
it's
in
the
form
of
an
abstract,
and
this
is
actually
something
that
you
know.
We
have
some
net.
A
A
So
this
talks
about
extracting
networks
from
different
subsystems
within
the
developing
organism,
and
in
this
case
it's
c
elegans.
So
you
have
this
network,
that's
like
the
embryo
network,
which
is
just
a
bunch
of
cells
that
are,
you
know
you
have
an
embryo,
that's
where
the
cells
are
dividing
and
moving
around,
but
they're
forming
these
proximity
networks.
A
A
What
is
the
sort
of
the
efficacy
of
any
you
know
signal
that
gets
sent
out
by
one
cell
what
other
cells
are
within
its
proximity,
and
so
those
networks
can
describe
that
structure
across
the
organism,
but
you
also
have
connectomes
that
are
emerging
and
so
connectomes
you
know,
cells
are
differentiating
into
neurons
and
they're,
also
making
forming
networks,
and
so
the
idea
of
this
paper
would
be.
How
do
a?
How
do
we
characterize
the
growth
of
these
networks
and
b?
How
do.
A
C
A
A
A
A
There's
also
this
kindle
book,
which
is
an
interesting
idea
posed
by
krishna
who's
here,
and
it
basically
compiling
a
lot
of
the
stuff
that
we've
done
into
a
kindle
book.
I
don't
know
if
you
know
that's
still
something
that's
on
hold,
but
if
other
people
are
interested
in
contributing
to
that,
it's
something
you
can
do.
A
We
have
this
divalern
presentation,
incf
neuroinformatics
assembly,
which
is
coming
up
in
april.
You
can
register
for
it
now.
It's
if
you
go
to
the
incf
main
page.
Incf.Org
you'll
probably
have
a
link
to
the
symposium
somewhere,
and
this
is
an
assembly
where
people
present
on
different
projects,
so
we're
gonna,
I'm
gonna,
try
to
present
on
evil,
learn
and
give
like
a
you
know.
Some
information
about.
A
You
know
to
a
broader
audience
about
what
divalert
is
trying
to
accomplish.
Maybe
some
of
the
things
we're
doing
in
this
group
a
little
bit
beyond
like
the
scope
of
the
platform,
but
this
is
the
actual
abstract
for
this.
So
this
abstract
goes
through
some
of
our
educational
efforts
in
the
group
and
those
served
as
the
precursor
of
the
stephen
learn
platform.
A
A
A
A
If
you're
interested
in
submitting
the
new
ips,
which
is
very,
very
competitive,
I
will
say,
but
if
you're
interested
in
submitting
something
a
paper,
the
deadline
is
may
19th
and
the
full
paper
is
may
26th
may
19th
deadline
is
for
abstracts.
So
if
you're
again,
I
wouldn't
hold
my
breath.
If
you
submit
a
paper,
that's
going
to
get
accepted,
but
I
you
know
if
there's
something
that
people
want
to
submit
there,
definitely.
C
A
A
A
Finally,
we
have
a
couple
of
things
that
the
society
for
developmental
biology
conference
is
happening
this
summer,
so
the
submissions
for
that
are
april.
19Th.
The
submissions
are
a
bit.
You
know
it
might
cost
like.
I
think,
fifty
dollars
to
submit
an
abstract-
I'm
not
crazy
about
that.
But
this
is
the
information
for
the
meeting
in
this
link.
A
So
that's
that's
our
update
on
that.
Finally,
I
wanted
to
go
well
actually.
I
wanted
to
talk
about
this
before
I
get
into
our
papers
and
this
is
sort
of
related
to
our
papers.
This
is
a
new
distilled
paper
on
neural
cellular
automata.
So
we've
talked
about
neural
cellular
automata
in
the
past.
We've
done
these.
You
know,
I
think
we
talked
about
the
paper
where
they
were
looking.
They
were
building
like
they
were
modeling
morphogenesis.
They
were
building
shapes
in
cellular
automata.
This
is
another
paper
from
the.
I
think.
A
C
A
A
So
what
is
a
neural
cellular
automata?
Well,
this
article
tells
you
what
that
is.
That's
actually.
This
article
is
part
of
a
the
differentiable
self-organizing
systems
thread.
So
if
you're
interested
in
this
sort
of
thing
there's
this
thread
of
papers
or
articles
that
you
can
look
at
and
these
are
different
papers
on
this-
you
know
they're
trying
to
find
topics.
I
think
so,
if
you're
interested,
you
should
check
that
out.
You
should
also
try
these
in
a
collab
notebook,
so
they
actually
have
a
link
to
the
collab
notebook
that
has
all
of
this
in
it.
A
A
A
But
what
they're
doing
here
is
they're
actually
able
to
create
patterns
using
these
cellular,
automated
neural
network
hybrids,
and
so
they
talk
about
partial
differential
equations.
They
talk
about
the
chemical
basis
of
morphogenesis,
which
is
this
paper
that
has
this
reaction
diffusion
model
or
proposes
a
reaction
of
fusion
model
amorphogenesis.
A
They
also
talk
about
the
grace
scott
reaction,
diffusion
model
and
this
interactive
atlas,
which
kind
of
goes
over
that
model
so
that
those
type
of
approaches
have
been
around
for
quite
a
while.
But
what
they're
doing
here
is
they're
describing
their
state
space
with
a
different
partial
differential
equation.
A
Then
they're
moving
to
cellular
automata,
so
they
kind
of
go
through
this
a
little
bit
and
they
show
a
lot
of
the
math.
They
show
how
these
computational
models
can
be
pattern
generators,
so
they
show
a
convenient
here
and
they
do
this.
They
have
this
neural
part
and
then
they
show
how
they
actually
can
get
this
result.
So
they
use
a
lot
of
they
use
a
lot
of
sophisticated
neural
network
processing
and
then
they
use
a
cellular
automata
to
update
the
model,
and
then
they
get
this
pattern.
A
So
if
you
go
to
that
link,
actually,
let's
see
what
they
have
there.
They
have
self-classifying
eminence
digits
self-worth.
Well,
we
have
self-organizing
textures.
This
paper
growing
neural
cellular
automata.
This
was
the
original
paper
we
visited
about
a
year,
maybe
yeah
about
a
year
ago
now,
where
we
talked
about
this
type
of
model,
but
they
didn't
give
as
much
detail
on
that
paper
as
they
do
in
this
paper.
A
A
Susan
says
the
reaction
to
fusion
model
might
be
a
mechanical
reaction,
diffusion
system.
That's
true,
and
there
are
other
ways
to
approach
this.
I
know
they're
generating
a
lot
of
patterns
with
this,
but
there
you
know,
there's
a
lot
of
life
left
in
the
reaction
of
fusion
models
that
we
have,
that
are
the
classic
models,
and
so-
and
we've
also,
you
know,
done
a
lot
of
good
pattern.
Formation
simulation
with
just
straight
up
cellular
automata
models,
so
you
don't
necessarily
need
a
neural
network
to
put
on
top
of
it.
A
A
Like
a
reaction
diffusion
model,
but
I
think
the
point
here
is
they're
trying
to
solve
the
pdes
that
form
this
reaction,
diffusion
model
with
the
neural
network
and
that's
the
kind
of
the
gold
there
in
the
in
the
in
the
model,
and
so
I'm
going
to
talk
about
some
papers
now.
So
if
you
have
to
leave
at
the
top
of
the
hour,
you
can
leave
if
you
want,
but
I'm
going
to
go
over
a
couple
more
I'm
going
to
go
over
a
couple
papers
before
we
end
our
meeting.
F
F
F
F
Living
machines
conference-
that's
on
the
submission
the
submit
then
here
this
I
actually
heard
about
this
through
this
conference.
So
through
this
event
and
oops.
So
it's
really,
I
I'm
kind
of
writing
up
a
little
article
or
report
on
it.
I've
taken
a
bunch
of
notes
on
it
and
bradley
and
another
groups
that
I'm
presented
and
danielle
would
also
add
it.
So.
F
F
They
start
from
soft
robots,
just
from
philosophy,
there's
stuff
from
cognitive
science,
yeah
sister
neuroscience
kristin,
shepard,
josh
vanguard
a
a
lot
of
surprisingly
big
names
were,
at
this
event,
brock
goldman
and
then
many
many
people,
and
so
I
would
just
say,
check
this
out
if
you're
interested
in
the
slack
and
there's
also
what's
really
cool
is
the
videos
are,
are
available
now
on
youtube.
So
so
you
can
pretty
much
go
back
and
look
at
the
discussions
on
a
lot
of
them.
F
Well,
this
there's
a
lot
of
good
stuff
here
I
prepared
that
at
some
at
things
for
my
own
combination,
futures
project
and
the
frontier
maps
and
bradley
presents,
except
about
developmental
ai,
and
so
there's
a
lot
of
good
things
here
and
there's
also
a
little
like
like
a
podcast
series
that
they're
running
adjacent
to
it.
But
it's
in
collaboration
with
the
the
soft
robotics
podcast
and
like
the.
F
A
A
You
know,
you
know
developing
systems
a
developing
embryo
which
is
going
to
become
a
body,
and
then
you
have
a
developing
brain
within
that
which
is
also
going
to
be
interacting
with
the
body
to
produce
behavior.
So
we
don't
talk
about
all
those
parts
all
the
time,
but
I
think
that's
an
important.
You
know
those
are
important
things
to
think
about
in
development.
A
A
Thank
you
for
that
jesse
and
if
you
want
to
know
more
you
can
we
can
raise
some
issue
with
our
slack
about
it.
You
know
it's
up
to
people
what
they
want
to
talk
about,
but
so
I
want
to
talk
about
some
papers.
So,
actually,
I'm
going
to
talk
about
this
folder
first,
I
decided
so
out
in
in
talking
about
this
boring
billion
work.
One
of
the
ideas
is
that
you
know
we
might
be
interested.
A
C
A
A
A
These
findings
demonstrate
that
the
evolution
of
reproductive
biology
along
birdline
archosaurs,
where
kant
was
a
complex
rather
than
linear
and
incremental
process,
and
suggests
that
some
aspects
of
non-anal
and
theropod
reproduction
were
unique
to
these
dinosaurs.
So
this
is,
I
mean
this.
Let
me
see
so
there's
a
lot
of
row,
as
I
mentioned
before,
there's
a
lot
of
really
interesting
paleontology
going
on
in
china.
They
have
some
nice
assemblages
there
of
different
things
of
flowering
plants
and.
A
This
dinosaur
era
specimens-
and
so
I
don't
know
if
there
are
any
pictures
in
here
that
we
can
see-
I
think
that
would
be
most
informative,
but
okay,
so
here's
some
pictures
here,
here's
a
dinosaur,
I
think,
sitting
laying
sitting
on
some
eggs.
I
think
these
are
the
eggs.
I'm
not
really
sure.
Let's
see
what
the
legend
says,
I
don't
have
a
legend
right
available
here,
but
this
is
the
okay.
So
this
is
the
dinosaur
sitting
on
the
eggs
here.
A
These
are
the
dinosaur
eggs
underneath
and,
of
course,
the
elegance
you
know
has
eggs
and
the
eggs
hatch,
but
the
the
parents
don't
sit
on
top
of
the
eggs
they
kind
of
leave
them
in
the
deposit
them
in
different
places
and
go
about
their
business.
In
this
case
you
have
an
organism
where
they
need
to
sit
on
their
eggs
to
make
them
warm
enough,
so
they'll
hatch,
and
so
this
is
sort
of
you
know.
A
C
A
Interesting
paper,
you
know
you
might
not,
you
might
not
understand
a
lot
of
it,
but
it's.
It
seems
like
it's
a
pretty
interesting
paper
for
learning
about
like
what
you
know
you
can
find
a
fossil
of
an
egg
and
a
fossil
of
an
embryo
on
what
those
look
like
second
paper
is
developmental
processes
and
etiacara
macro
fossils,
so
the
ediacara
biota
preserves
the
oldest
fossil
evidence
of
abundant
complex
metazoans,
which
are
basically
animals.
It's
a
group
that
incorporates
any
animal
cell
any
organism
with
an
animal
cell.
A
So
that's
what
metazone
is,
despite
their
significance,
assigning
individual
attacks
at
a
specific
phylogenetic
groups
has
proved
problematic
to
better
understand
these
forms.
We
identify
developmentally
controlled
characters
or
traits
and
representative
taxa
or
species
or
groups
or
biological
groups
from
the
white
sea
assemblage
and
compare
them
with
regulatory
tools
drawing
similar
traits
in
modern
organisms.
So
they
compare
fossils
of
modern
organisms
and
they
make
this
comparison.
A
A
That
the
genetic
pathways
for
multicellularity,
axial,
polarity,
musculature
and
a
nervous
system
were
likely
present
in
some
of
these
early
animals.
So
what
they
do
is
they
take
like
modern
organisms
and
they
look
at
the
regulatory
mechanisms
that
control
some
of
these
things.
So,
like
you
know,
axial
polarity
is
where
the
cells
form
a
head
and
a
tail
very
early
on
where
the
musculature
is,
where
the
muscles
form.
A
The
nervous
system,
of
course,
is
where
you
get
neurons
and
they're.
You
know
how
they
connect
and
you
can
actually
sample
those
cells
and
look
at
their
molecular
pathways,
and
we
know
enough,
probably
about
like
things
how
things
are
regulated
at
the
genetic
level
to
say,
maybe
something
about
how
they
worked
in
the
fossil
example.
A
So,
equally,
meaningful
is
the
absence
of
evidence
for
major
differentiation
of
macroscopic
body
units,
including
distinct
organs,
localized
sensory
machinery
or
appendages,
meaning
that
there's
this
lack
of
evidence
that
there's
this,
that
there
are
these
modules
within
the
organisms
and
this
in
this.
So
this
is
570
to
539
million
years
ago,
and
so
this
was
a
long
time
ago.
A
C
A
Rudimentary
nervous
systems,
but
probably
not-
and
so
that's
an
apparent
lack
of
heads
with
concentrated
sensory
machinery
or
ventral
nerve
cords,
and
such
taxes
supports
the
hypothesis
that
these
evolved
independently
in
desperate
glades
or
groups
of
organisms,
so
they
didn't
have
heads
in
concentrated
sensory
machinery
back
in
the
ediacara.
This
is
some.
A
You
know
something
that
evolved,
maybe
in
the
in
the
cambrian
period,
when
these
organisms
were
diversifying
into
different
groups,
and
so
this
is
interesting
because
you
know
kind
of
gives
us
an
idea
not
only
of
development
but
also
of
evolution
and
how
these
things
kind
of
emerged
and
when
they
emerged
in
this.
So
this
is
a
you
know.
This.
C
A
A
kind
of
a
different
approach,
they're
using
fossils
and
then
they're
taking
modern
organisms
and
they're
kind
of
making
a
comparison
between
the
different
things
that
they're
finding
in
this
assemblage
and
modern
organisms
are
modern
sort
of
analogs
that
they
can
use
to
understand.
What's
going
on
in
these
organisms,
they
just
leave
an
impression-
and
you
know
you
know
fossils
are
such
that
not
everything
preserves.
So
we
may
not
be
getting
a
full
picture
of
what
was
going
on
back
then,
but
they
do
give
us
a
pretty
good
picture
as
it
turns
out.
A
A
So
this
paper
is
interesting
because
it
shows
sort
of
the
development
of
teeth
and
bones
associated
with
tooth
development
in
dinosaurs,
and
so
they
show
embryonic
an
embryonic
denture
measuring
just
three
centimeters
long,
already
exhibits,
distinctive
triansteroid
character,
characters
like
a
chin
and
deep
mechanine
groove.
I
mean
these
are
again.
A
A
A
New
paper,
so
this
these
cell
papers
often
give
like
a
graphic
at
the
beginning,
a
graphical
abstract,
and
so
this
basically
summarizes
the
paper.
So
you
have
this.
This
is
an
adult
c
elegans.
This
is
the
head.
This
is
the
tail
and
here's
an
odorant,
so
this
is
a
stimulus
that
the
head
is
moving
towards,
and
this
is
you
know
this
is
responsible
for
this.
Is
this
chemosensory
circuit,
which
is
part
of
the
nervous
system?
A
So
the
worm
has
sensory
neurons
in
the
head
that
pick
up
this
odor
and
contract
the
odor
and
move
towards
it.
So
the
circuit
is
formed
in
development
and
then
persists
across
the
adult
phenotype,
and
if
you
can,
you
can
create
mutants
that
have
defects
in
this
circuit,
so
cells
cannot
grow
up
in
development
and
affect
the
formation
of
the
adult
circuit,
or
you
can
have
connections
that
don't
form.
So
there
are
other.
A
Is
bad
for
the
worm?
If,
if
the
odor
is
the
signal,
they
need
to
find
food
and
they
don't
have
vision,
they
have
some
other,
they
have
mechanosensation
and
they
have
chemo
sensation.
So
if
they
don't
have
good
chemo
sensation,
they're,
probably
in
a
lot
of
trouble
and
then
finally,
in
this
paper,
what
they
do
is
they
knock
out
part
of
the
circuit
and
then
they
enter.
They
have
a
genetically
inserted
synapse
that
they
can
put
in.
A
A
You
can
insert
a
synapse
and
electrical
synapse
to
circumvent
the
damage.
You
can
find
alternate
pathways
for
information
flow,
so
these
circuits
are
hardwired
to
specific
connectivity
pattern.
Their
alternate
pathways
that
you
can
use
and
a
weakened
signal
can
be
enhanced
due
to
a
new
lateral
left
electrical
coupling.
So
in
doing
this
regeneration
of
of
connections
between
cells
in
the
circuit,
you
can
actually
find
maybe
things
that
are
almost
as
good
as
the
original
connection,
but
you
know
maybe
something
that
was
lost
in
evolution
or
never
happened.
A
A
So
you
have
those
two
types
of
connectivity
and
the
c
elegans
connect
them.
Of
course,
as
we
know,
you
we've
defined
every
cell
and
every
cell
is,
you
know,
has
a
defined
role
from
development,
so
we
know
like
how
to
manipulate
this
thing,
pretty
well
as
a
consequence,
and
so
that's
what
they
do
in
this
paper.
They
manipulate
it.
They
they
show
the
gap
junction.
Here
they
show
some
of
the
evidence
here
for
what
they're
doing
they're
able
to
verify
these
things.
A
If
you
don't
know
much
about
how
they
do
these
knockouts
and
napkins,
they
use
a
fluorescent
element
to
find
like
different
things
that
are
expressed,
and
they
can
tell
whether
some
gene
is
being
expressed
in
a
neuron.
These
green
blotches
here
show
like
different
cells
where
a
gene
is
expressed
and
if
they
knock
it.
C
A
So
they
know
that
that's
what's
responsible
for
this,
and
so
this
again
is
another
example
of
the
connectome
with
the
cells
and
the
connections
chemical
and
electrical,
and
then
they
have
these
synthetic
electrical
synapses.
So
they
form
these
gap
junctions
between
cells
artificially
they
knock
in
a
an
element.
They
can
verify
it
with
these
green
blobs
and
then
now
they
should
observe
restored
function
in
this
circuit,
and
so
that's
exactly
what
they
observed,
and
this
is
an
example
here
of
sort
of
a
figure,
six
or
figure
seven.
A
This
is
a
famous
thing
in
biology
papers
where
they
show,
like
the
higher
level
concept,
they're
trying
to
communicate
in
this
figure,
and
so
in
figure
6.
They
have
distributed
information
flow
in
the
c
elegans
olfactory
circuit.
They
show
sensory,
neurons,
interneurons
and
then
going
down
to
motor
neurons
and
they
show
so
neural.
Information
flow
is
often
described
as
a
progressing
from
sensory
neurons
to
interneurons
to
motor
neurons.
A
A
So
it's
not
a
hierarchical
system
from
sensory
neuron
to
motor
neuron
as
a
lot
of
people
like
to
characterize
it
as
a
feed-forward
network,
but
it
also
you
have
a
lot
of
communication
between
cells
and
so
by
knocking
in
these
alternate
gap
junctions,
you
can
actually
observe
these
types
of
connections
between
cells
at
these
different
levels
and
so
a
feed
forward.
A
simple
feed
forward
network
becomes
this
really
complex
thing.
A
A
This
is
a
paper
on
retinal
developments.
This
isn't
c
elegans,
but
this
is
about
development
of
the
nervous
system
and
they
say
retinal
waves
and
visual
experience
have
been
implicated
in
the
formation
of
retinotopic
and
eye
specific
maps
throughout
the
visual
system.
So,
as
you
might
know,
in
humans
and
in
mammals
and
in
that
type
of
nervous
system,
you
have
this
retinotopic
map
that
forms
and
it
forms
generally
through
experience.
A
A
Retinal
waves
here,
so
they
don't
know
which
one
plays
a
role
in
the
development
of
the
maps
within
the
retina
so
which
one,
actually
you
think,
like
visual
experience,
forms
these
maps.
But
then,
of
course,
they
want
to
know
if
that's
actually
true,
so
we
explore
this
question
using
direction.
Selective
retinal
ganglion
cells,
which
are
organized
into
a
map
that
aligns
to
the
body
gravitational
axis
of
optic
flow,
so
it's
aligned
to
the
the
where
the
body
is
going
in
space
and
some
of
the
axes
of
what's
being
experienced.
A
So
when
you
move
around
a
room,
you
see
things
come
up
on
you.
You
know
you
see,
like
the
wall
kind
of
loom
in
front
of
you,
bigger
and
bigger
as
you
walk
towards
it
and
you
walk
away,
it
becomes
smaller
and
smaller.
All
that
information
is
called
optic
flow,
and
this
is
one
of
the
things
they're
using
in
this
experiment.
A
A
It's
some
sort
of
animal
model
in
a
dark
room
constantly,
so
that
means
that
they
can't
see
things
as
well
as
they
would
in
a
lighted
room
and
so
remarkably,
the
horizontal
component
of
the
direction
selectivity
map
is
absent
in
mice,
lacking
normal
retinal
waves,
whereas
the
vertical
component
remains
normal.
These
results
indicate
that
intrinsic
patterns
of
activity,
rather
than
extrinsic
motion
signals,
are
critical
for
the
establishment
of
direction
selectivity
maps
in
the
retina.
A
So
this
means
that
what
they
call
intrinsic
patterns
of
activity,
rather
than
these
motion
cues
from
the
environment,
are
critical
for
the
establishment
of
these
selectivity
maps.
So
they
kind
of
go
into
the
the
details
of
this.
You
know
how
this
happens.
They
do
these
experiments
in
light
and
dark.
A
A
If
you
are
exposed
in
a
rich
environment
or
a
lighted
environment,
you
get
more
things
that
you
interact
with
and
you
you
can
detect
visually,
and
so
what
they're
saying
is?
That's
not
really
the
only
aspect
to
this
there's
another
aspect
which
are
these
retinal
waves,
and
so
I'm
trying
to
find
an
example
of
what
they
mean
by
retinal
wave
exactly,
but
they
show
that
there's
some
results
that
occur
in
mice
that
you
don't
necessarily
see
in
a
model
like
a
ferret
which
is
another
model
organism
that
they
use
for
this
sort
of
thing.
A
These
findings
indicate
direction.
Selectivity
maps
are
well
established
and
eye
opening,
independent
and
visual
experience,
so
these
maps
are
actually
formed,
not
just
by
just
pure
experience,
but
there's
an
underlying
biological
sort
of
process
that
goes
on
to
form
them,
and
so
this
is
all
you
know.
A
These
papers
are
usually
not
like
terribly
synthetic
in
terms
of
like
what
the
findings
are,
and
you
know
placing
in
the
larger
context,
but
it
definitely
has
a
lot
of
detail
on
some
of
the
things
that
they've
found
in
their
experiment
and
how
it's
relevant
to
things
in
the
literature.
So
I
think
this
is
a
really
good
paper
for,
like
you
know,
I
know
jesse's
interested
in
this
and,
if
you're
interested
in
biologically
inspired
ai.
This
is
interesting.
A
F
A
E
B
D
C
Yes,
I
so,
for
example,
there
is
a
simple
environment
and
there
are
several
food
spots
and
we
would
like
to
predict
where
the
cla
ground
would
go,
which
full
source
loop.
It
will
go
to
first
or
there
are
some
obstacles
and
how
it
will
behave
depending
on
different
kind
of
obstacles
like
have
you.
Have
you
either
you
or
open
one
has
done
any
work
related
to
that,
like
modeling,
really
simple
environments
and
then
trying
to
try
to
do
some
machine
learning
agents
compare
to
the
real
ones.
A
Sometimes
they
do
movements
to
back
away
from
a
food
source
or
to
move
towards
a
food
source,
and
then
they
do
other
types
of
movements
like
that,
so
they
have
a
lot
of
that
characterized
in
in
in
that
movement
database
that
I
showed
earlier.
I
also
know
that
the
gepetto
people
are
also
interested
in
doing
work
on
different.
A
C
Yeah,
I
would
say,
for
example,
I
would
not.
Maybe
I'm
not
interested
in
like
detailed
movement
of
the
sea
eligible,
but
I'm
more
interested
in
high
level.
If
you
can
call
it
the
cognition
of
abilities
or
how
to
say,
choose
one
mate
or
another,
or
how
do
they
prefer
one
food
source
or
if
you
get
the
idea,
what
I'm
eating.
A
Yeah,
I
know
yeah
I'm
familiar
with
that
yeah.
I
don't
know
if
we're
doing
anything
like
that
currently,
but
actually
tom
portages
might
be
someone
to
talk
to.
He
sometimes
attends
this
group
and
he's
doing
some
work
on
with
agents.
He
does.
He
works
in
the
robots
channel,
sometimes
too,
and
he's
doing
a
lot
of
work
with
agents
and
simulating
behavior,
and
things
like
that,
so
he
might
actually
be
a
good
person
to
talk
to.
C
A
Yeah
yeah,
my
other
group,
were
kind
of
interested
in
this
stuff
too.
So
the
other
group
that
we
have
and
you
can
contact
jesse
parent
to
get
in
the
loop
on
this,
but
we
do
this.
We
we
have
we've
been
working
on
sort
of
topics
like
this,
like
reinforcement,
learning
and
other
things
that
are
maybe
more
interesting.