►
From YouTube: DevoWorm #18: Intro to GSoC 2022: Digital Microspheres and D-GNNs. Interfacing DevBio with Theory
Description
Introduction to GSoC 2022: Digital Microspheres and D-GNNs. Student introductions. Community period and associated resources. Benchmarks for ball microscopy. Discussion about the state of Developmental Biology and the interface between topical inclusion and theory. Question-driven developmental vs. unified theories and theory-building. Attendees: Susan Crawford-Young, Karan Lohaan, Harikrishna Pillai, Wataru Kawakami, Richard Gordon, and Bradly Alicea
B
C
B
Well,
welcome
to
the
meeting
we
have
a
big
announcement
this
week
was
the
google
summer
code
student
announcements,
so
we
got
four
in
this
group
and
yeah.
So
we
had
the
two
people
who
applied
to
do
the
digital
microspheres
and
the
two
people
applied
to
do
the
gnns.
D
Okay,
gotta
get
a
whole
bunch,
more
pictures,
yeah
yeah.
I
was
thinking
of
like
those
bb
guns
and
getting
a
little
one
of
those
and
then
just
putting
a
whole
bunch
of
marker
thoughts
on
it.
D
B
Yeah
you
mean
to
get
some
sort
of
like
baseline
or
benchmark
data
for
the
scope
or.
D
C
Plants
yeah
10
years
ago,
10
years
12
years
ago,
I
bought
a
software
for
making
a
an
outline
model
of
the
world.
B
B
C
B
Right
right,
yeah,
it's
usually
yeah,
so
that's
yeah,
we'll
think
about
how
to
do
that.
B
So
I
guess
I
wanted
to
go
through
each
person
and
it
doesn't
look
like
jihan
is
here
right
now,
but
the
three
people
are
here:
if
you
could
introduce
yourselves
and
maybe
talk
a
little
bit
about
your
project
and
what
you
hope
to
accomplish
this
this
this
session
or
this
summer,
so
I
think
karana
has
to
go
if
he
could
go
first.
E
F
F
So
introduction
wise,
I'm
currently
doing
my
bachelor's
degree
from
amrita
university.
My
course
objects
are
electrical
and
computer
engineering
like
we
have
computer
architecture,
it's
like
a
wide
array
of
subjects
where
we
get
into
the
hardware
abstraction
layers
to
the
software
side
as
well.
So
apart
from
that
introduction,
okay
wait.
I
think,
there's
a
lot
of
disturbances.
F
F
So
the
approach
related
to
the
digital
microsphere
like
I
had
given
you
know
two
three
presentations
before
so
the
idea
is
pretty
much
still
the
similar
idea.
You
know
I'll,
be
taking
I'll,
be
extracting
each
embryo
image
from
the
sample
data
set
and
from
that
I'll
be
okay.
F
B
I
heard
he's
gonna
write
it
in
the
chat,
so
I'm
gonna
read
what
he
says:
yeah
so
he's
presented
on
this
a
couple
times
before,
and
we've
been
working
on
this
project
a
little
bit.
You
know
getting
getting
a
sense
of
how
to
approach
the
project.
I
think
that
was
a
smart
move
for,
and
I
know
hari
krishna
has
done
the
same,
just
to
get
a
sense
of
how
to
do
it.
B
You
know,
I
don't
think
we
have
everything
worked
out,
but
I
think
one
of
the
things
I'd
like
to
do
is
to
be
complimentary
here
and
pick
his
agreed
on
this,
where
you
know
krishna
and
koran
work
together
when
you
can
and
make
sure
that
you're
not
overlapping,
you
know.
So
you
don't
do
the
same
work
twice.
Basically,
yeah.
B
Yeah
yeah
well
yeah.
We
want
something
at
the
end
that
you
know
is,
but
you
know
they're
different
right.
There
are
a
lot
of
different
things
that
can
be
sort
of
not
reproduced.
B
Okay,
well,
karan
is
typing
that
out
hurry
krishna.
Would
you
like
to
go
and
introduce
yourself.
G
G
My
goal
this
summer
will
be
to
create
a
computational
tool
which
will
basically
map
the
2d
images
of
the
android
onto
a
sphere,
and
also
I
have
to
take
care
of
the
overlappings
so
yeah
till
now.
I
think
I've
made
a
good
progress
and,
by.
G
B
Yeah
yeah,
that's
good
thanks
for
that.
Let's
see
if
I
can
pull
up.
Okay,
karen
still
hasn't
gotten
in
the
description
but
yeah
so
yeah
I've
seen
some
of
your
work
that
you've
done
you've
done
some
pretty
good
work.
So
far
and
again,
you
know
we'll
have
to
try
to
figure
out
how
to
not
overlap
in
places
where
we
don't
need
to.
So
that's
good
and
then
wataru
hello.
E
Yeah,
thank
you.
I'm
a
50
years
student
at
kyoto
university
and
I
am
currently
pursuing
a
medical
doctor
and
I
belong
to
a
laboratory
in
the
graduate
school
of
informatics
in
kyoto,
university
and.
E
I
am
developing
methods
for
visual
reconstruction
of
what
is
seen
by
subjects
by
inferring.
The
features
of
the
hidden
layers
of
dnns
from
fmri
data
function,
functional
mri
and
I
I
will
develop
in
this
summer.
I
will
make
some
gnns
that
resemble
actual
biological
networks
throughout
development,
but
I
I
have
decided
the
the
details,
so
I
have
to
start
with
the
searching
for
some
interesting
theory
or
yeah,
something.
B
B
B
So
we
have
data
that
you
know
we
use
a
lot
of
secondary
data,
a
lot
of
microscopy
data,
so
the
digital
microspheres
project
is
using.
You
know
a
certain
data
set
and
you
may
or
may
not
find
that
useful.
You
may
find
other
types
of
data
useful.
In
the
past,
we've
used
a
lot
of
data
from
c
elegans
development,
so
this
is
where
we're
looking
at
cells
and
defining
cells
in
their
sort
of
their
space
and
then
extracting
those
data
through
segmentation
or
other
types
of
techniques.
B
So
you
know
this
is
something
that
we
have.
A
lot
of
secondary
data
sets
sort
of
marked
that
are,
you
know.
We
know
they're
good
to
work
with
they're
sort
of
good
benchmarks
for
this
sort
of
thing,
and
then
that
might
help
you,
you
know
kind
of
get
get
into
the
data
and
start
working
out.
You
know
something
concrete
in
terms
of
you
know
how
you
might
build
these
networks
or
these
embeddings
again
like
with
gia
hong.
B
I
know
that
I
think
I
don't
know
if
you
to
have
talked
yet,
but
I
want
to
make
sure
that
again
you
don't
overlap
in
key
areas,
so
I
don't
think
gia.
Hang
is
proposed
anything
like
really
close
to
what
you
proposed.
Although
you
didn't
you
know,
you
didn't
give
the
like
full
details.
B
So
that's
something
we'll
have
to
work
out.
I
just
want
to
make
sure
that
you
know
you
don't
do
the
same
thing
exactly
the
same
thing
and
that's
something
you
know.
Maybe
you
could
work
on
different
parts
of
this
or
different
data
sets.
I
want
to
talk
to
him
more
about
that
before
we
start
getting
into
really
deeply
into
this,
but
welcome
aboard
to
the
group
here.
C
D
C
D
At
the
wrinkles,
oh
well,
you
could
you
could
paint
it
and
then
dry
it,
and
then
the
paint
would
crack.
Maybe
I'll
try
stuff.
I
have
some
glass
beads
I'll,
take
a
look,
or
else
I
could
get
one
of
these
bb
gun
things
and
melt
it
a
bit.
B
D
I
still
haven't
been
able.
The
people
on
kijiji
are
trying
to
sell
me
salamanders
for
350
bucks,
so
I
don't
need
it
to
be
a
whatever
golden
thingy
that
you
have.
It
just
needs
to
be
black
wild
type,
something
or
other
it
doesn't
have
to
be
a
specialty
female
axolotl,
but
anyways
I'll
keep
trying
kijiji.
D
That
might
produce
a
texture
yeah,
I'm
just
I
don't
think
they
make
ball
bearings
that
small.
C
B
B
After
this,
the
second
part
is
based
on
using
the
embryo
image
and
the
contour
of
that
embryo
image
to
create
a
3d
point
cloud
by
rotating
the
embryo,
outlines
or
contours
about
the
axis
of
rotation
and
then
after
generating
the
3d
model.
From
this
3d
point
cloud
I'll,
be
applying
cylindrical
projection
techniques
to
project
the
image
onto
my
3d
generated
model.
B
So
this
part
will
require
stitching
all
eight
embryo
images
together
into
one
cohesive,
equirec,
rectangular
image.
Then
I
will
then
project
onto
my
model,
so
you're,
building
this
you're,
taking
these
images,
you're
stitching
them
into
this
cohesive
image
and
then
you're
sort
of
warping
it
into
the
model.
C
B
C
Make
an
outline
suggestion:
yeah,
okay.
I
once
made
a
3d
point
cloud
and
I
viewed
it
in
two
directions
about
six
degrees
apart
and
that
made
a
very
nice
stereo
pair.
You
could
actually
see
it
in
three
dimensions:
yeah,
okay,
so
it
might
be
a
useful
tool
to
to
make
a
stereo
pair
and
look
at
these
3d
point
clouds.
B
Yeah,
okay
yeah
they
have
yeah,
they
have.
I
think
they
have
tools
to
like
optimize
the
distance
between
the
images,
so
the
overlap
that
you
need
to
have
a
coherent
stereo.
B
Yeah,
that
would
be
good
yeah.
I
think
that
sounds
like
a
good
project
description.
So
again,
this
is
something
that
we'll
be
working
on
over
the
course
of
the
summer
and
again
we
have
these
two
groups,
the
digital
microspheres
and
the
gnns,
and
so
want
you
to.
I
mean
we'll,
have
these
meetings
once
a
week
mondays
and
then
we'll
try
to
catch
up
on
slack
over
the
course
of
the
week.
So
I
want
people
to
discuss
things
between
yourselves
I'll.
Try
to
answer
questions.
B
I'm
gonna
in
the
next
few
weeks,
I'm
gonna
be
giving
you
these
materials
in
terms
of
data,
or
you
know
things
that
you
might
need
to
to
sort
of
get
going
and
yeah.
So
I
hope
you
know
if
you
have
any
questions,
let
me
know
it's
don't
be
afraid
to
ask,
or
you
can
ask
susan
or
dick
here
about
some
of
the
especially
for
the
digital
microspheres,
because
this
is
something
that
they've
been
working
on.
I
think
for
a
while
this
idea,
so.
B
Yeah,
so
so
the
next
few
weeks
for
the
students
is,
is
going
to
be
a
community
period
and
that's
built
into
the
program
and
so
check
your
schedule
that
they
have
on
the
website
they
kind
of
meant
they.
They
talk
about
the
community
period,
which
is,
I
think,
two
to
three
weeks,
maybe
even
a
little
bit
longer
this
year,
but
I'm
not
sure,
and
then
you
start
your
projects
in
june
and
then
you
go
through.
You
can
go,
you
know,
go
through
to
september
or
go
through.
B
B
I
think
12
weeks
at
maybe
20
to
30
hours
a
week
doing
these
sorts
of
things
and
you'll
get
your
stipend
and
that,
but
you
can
extend
that
time
out
until
I
think
november.
So
if
there
you
have
exams
or
something
you
can
extend
it
for
a
cup
by
a
couple
weeks,
you'll
have
to
consult
the
schedule
to
figure
out
exactly
how
you
want
to
do
that,
and
then
you
know
I
made
you
in
your
proposals.
B
B
It
tells
you
what
you
need
to
you
know
catch
up
on
and
you
know
where
you
need
to
go
next,
but
the
other
thing
it
does
is
it
allows
you
to
move
things
around
accordingly.
So
if
you
miss
one
week
because
you
have
exams,
you
can
pick
up,
you
might
have
to
shift
things
around,
but
at
least
you
know
where
you
are
so.
B
The
other
thing
is,
you
know
when
you
do
these
weekly
meetings,
you're
going
to
be
you're,
going
to
be
presenting
something
at
every
meeting.
So
maybe
not
next
week,
but
definitely
when
we
start
the
coding
period.
I
want
to
have
a
report
once
a
week.
Each
person
will
go
around
and
you
know
you
can
tell
us
what
you've
done.
Maybe
ideally
I'd
like
to
see
like
a
screen
share
of
some
things
that
you've
done.
You
know
it
could
be
like
a
a
notebook,
a
collab
notebook.
B
It
could
be
like
some
slides
it
could
be.
You
know
some
visualization
just
something
to
show
that
you
know
you're
progressing
and
that
you
know
if
you
have,
especially
if
you
have
questions
or
if
you
have
a
problem
that
arises,
that's
a
good
time
to
to
sort
of
bring
that
up
and
we
can
help
you
debug
things.
B
You
know
maybe
not
like
some
really
big
problem,
but
just
something
that
you
know
if
you're
having
a
problem,
you
need
advice
on
it.
That
would
be
a
good
time
to
do
that
and
I
just
want
to
make
sure
that
people
do
this
every
week
because
it's
you
know
it
just
helps
you
keep
on
on
task
and
it
keeps
you
going.
So,
that's
that's
really
what
I
want
to
do,
because
you
don't
have
a
lot
of
time
to
you,
know
kind
of
figure
out
what
you
want
to
do.
B
You
have
to
get
into
this
hit
the
ground
running.
So
I
know
whataro
is
you
know,
he's
trying
to
make
sure
that
he
has
a
grasp
on
things
and
energia
hong
had
a
more
detailed
plan,
but
still
I
want
to
make
sure
that,
especially
for
the
gnns
project
that
we're
going
in
the
right
direction
on
that,
it's
a
that.
That
project
is
a
little
hard
to
sort
of
grasp,
mainly
because
you
know
we
haven't
really
done
much
work
on
it
before.
B
So
that's
something
we'll
just
have
to
kind
of
figure
out
as
we
go
along.
We
also
want
to
have
like
some
really
nice
targets.
You
know
so
we
know
kind
of
what
we're
doing
week
by
week
and
that
we
make
progress
on
that
yeah.
Three
steps
forward.
Two
steps
back
is
progress.
So
yeah
we'll
see,
I
mean
don't
worry
about
like
if
you
get
stuck
on
something
you
know,
and
I
I
think
I've
told
people
this
in
the
application
period.
B
If
you
get
stuck
on
something,
don't
spend
like
four
weeks
trying
to
figure
it
out,
you
know:
ask
around
you
know
there
are
ways
to
find
out
on
like
stack
exchange
and
things
like
that
about.
You
know
simple
problems,
but
if
you're
having
some
more
fundamental
issue,
please
let
me
know
because
I'll
you
know
we
could.
We
can
clear
it
up
pretty
quickly.
I
think
if
we
talk
amongst
ourselves,
so
that's
that's
something
we
want
to
make
sure
we're
active
about.
B
D
I
had
a
brief
some
information
about
the
images
they
look
like
they're
ovals,
because
they're
being
taken
on
an
angle,
so
they
round
peppercorn
will
probably
look
slightly
oval
and
my
suggestion
is
to
just
stretch
it
back
to
being
a
sphere
or
circle.
C
C
C
C
D
D
It's
okay,
that
I'll
get
some
more
images
with
that.
Like
I
said
it's
set
up
for
three
millimeter
peppercorn,
I
can
surely
find
some
other
things
to
image,
and
then
there
could
be
a
variety.
B
Okay,
yeah,
that's
good
yeah,
so
I
also
wanted
to
point
out
what
taro
and
karan
had
to
go,
but
we,
I
have
some
other
things.
So
another
thing
for
the
community
period
is
we're
going
to
have.
I
want
people
to
also
get
to
know
the
open
worm
community
so
I'll
try
to
facilitate
some
interactions
with
the
rest
of
the
open
world
community.
Well,
you
know
they're
all
if
you're
in
the
slack
there
are
a
lot
of
different
channels
that
exist,
so
it's
not
just
the
diva
worm
channel.
It's
also.
C
Can
we
wake
up
steve
larsen
and
have
him
a
very
early
learning
session.
B
I
don't
know
he's
pretty
busy
these
days,
but
actually
he's
got
some
talks
online
I'll
put
together
a
list
of
things
that
we
have
he's
he's
done.
Some
talks
on
open
worm
and
then
that'll
be.
You
know,
you
can
view
those
and
you
get
a
bit
a
bigger
perspective
on
the
open
worm
community,
but
there
are
also
other
channels
that
you
can.
You
know
I'd
like
to
encourage
people,
maybe
to
check
out
maybe
get
into
some
conversations
on.
B
So
one
of
the
things
that
open
worm
has
is
a
data
set
on
movement
of
c
elegans.
So
this
is
where
you
have
worm
movement
and
they're
using
computer
vision
to
segment.
Some
of
those
movements
into
you
know
distinct
sort
of
identities
so
like
in
c
elegans.
They
have
these
stereotype
movements
that
you
can
classify.
B
You
know
they're,
not
ambiguous,
they
they're
reproducible
from
worm
to
worm
and
you
can
easily
see
them
when
you
look
in
a
culture.
So
you
know
they've
tried
to
work
out
all
those
movements,
and
so
that's
that's
something
that
maybe
is
a
good
source
of
data
for
some
people
for
the
gnn
people.
That
might
be
a
place
to
go
and
look
at
some.
B
You
know.
Maybe
it's
not
really
embryos,
but
it's
you
know
it's
enough
data
to
sort
of
benchmark
your
algorithm
and
say:
oh,
this
is
something
that's
structured.
Maybe
we
can
create
an
embedding
from
this,
so
I
want
to
facilitate
some
interactions
there
as
well.
B
B
So
I
have
a
set
of
readings
here
that
I've
collected
over
the
years
and
these
are
on
different
topics
in
open
source.
So
we
have
this
set
of
open
source
books
that
I
have
so
I
have
this
book
on
producing
open
source
software,
and
this
is
a
nice
book
on
like
the
production
process
behind
open
source
software.
So
this
goes
through
a
lot
of
different
things
with
respect
to
you
know
if
you're
creating
a
program
and
you're
licensing
it
as
an
open
source
product.
You
know
what
goes
into
that
there's.
B
A
lot
of
you
know
a
lot
of
sort
of
there's
production.
There's
up,
you
know
keeping
your
software
up
to
date,
so
there
are
a
lot
of
things
in
there.
That
are
things
you
have
to
think
about,
and
this
is
especially
useful
if
you
want
to
go
beyond
the
summer
of
code,
and
you
want
to
think
about
you
know
what
do
you
want
to
do
with
your
software
in
years
past?
B
B
It
can
either
sit
there
and
you
know,
collect
moss
or
like
it's
digital,
moss
or
yeah,
or
you
can
improve
upon
it,
and
so
I'm
not
saying
that
you
have
to
be
burdened
with
improving
it,
but
there
is
this
aspect
of
getting
people
involved
to
you
know
commit
make
commits
to
the
software,
and
there
are
other
types
of
things
that
you
can
do.
B
B
We
also
have
some
other
readings
in
here
on
open
data
on
new
types
of
institutions.
This
has
to
do
with,
like.
H
B
H
B
We
usually
think
about
attribution
so
this
you
know
it
involves,
there's
an
emphasis
on
sharing
an
emphasis
on
collaboration
instead
of
you
know,
putting
things
behind
a
wall
of
of
you
know
in
an
intellectual
property
wall,
the
intellectual
property
is
it's
managed
differently,
and
so
this
kind
of
goes
through.
Some
of
that.
B
I'm
going
to
add
to
this
repository
and
I'm
going
to
make
it
public
here
during
the
community
period,
and
I
hope
people
can
maybe
we'll
go
over
next
week,
one
or
two
of
these
resources
and
we'll
kind
of
get
an
idea
of
what
it
means
to
work
open
and
what
it
means
to
build
something
open
source
and
how
we
facilitate
that
yeah.
H
B
B
I
have
a
repository
here:
gsoc
2022.,
this
repository
is
two
sub
repositories
or
subdirectories
dgns
and
digital
microsphere,
and
so
I
want
the
students
to
I'm
gonna
invite
them
to
join
these
as
appropriate,
so
hari
krishnan
koran
will
be
in
digital
microsphere
and
motaro
and
jiahun
will
be
in
dgnns,
and
so
those
two
places
will
be
where
all
their
work
is
pushed.
B
So
this
is
going
to
be.
This
is
labeled
by
project.
So
this
is,
I
think,
hari
krishna's
project.
This
one
is
karan's
project,
that's
digital
microsphere
and
then
in
dgnns
we
have
devo
graph,
which
is
jiahung,
and
then
this
is
wataru.
Gnn's
developmental
networks
I'll
send
out
an
email
to
make
this
clear.
You
know
how
this
works
out
and
where
I
expect
people
to
push
it'll
basically
be.
B
You
know
where
I'll
invite
the
students
into
this
space
as
collaborators
and
then
they'll
be
able
to
push
their
work,
and
I
want
to
basically
have
people
update
their
work,
maybe
once
a
week
they
can
create
a
fork
of
this
repository
and
update
and
then
push
back
to
this
main
repository,
maybe
once
or
twice
a
week,
maybe
once
every
two
weeks,
depending
on
how
much
is
being
done
and
how
much
needs
to
be
committed,
but
basically
one
one
force
using
github
as
a
main
place
to
store
code,
because
I
think
you
know
we
don't
want
to
store
code
on
our
local
machine,
because
if
the
local
machine
crashes,
you
lose
your
work
so
be
sure
to
use
github
for
everything
be
sure
to
push
at
least
here
for
your
local
fork.
B
And
then,
when
your
fork
is
ready
to
be
updated,
I
mean
you
know,
maybe
once
a
week
make
it
habit
of
it
push
once
a
week
and
label
the
pull
request
as
like
you
know,
I've
done
this,
and
sometimes
there
won't
be
anything
in
the
push.
Much
at
all.
Maybe
a
couple
files,
sometimes
you'll
have
a
lot
of
stuff,
but
I
just
want
to
get
an
update
through
github
and
to
make
sure
that
we're
you
know
keeping
on
track.
B
I
think
github
updates,
you
know
updated,
pushes
to
the
main
repository,
I
think,
are
a
good
way
to
enforce
progress
in
keeping
things
moving
and
it
you
know
it
gives
you
the
sort
of
this
account
of
what
you've
done.
So
I
made
some
changes
to
this
file.
I've
added
this
file
and
you
can
see
the
progress
before
your
eyes,
so
that's
good,
so
I'll
send
out
those
invites
this
week.
B
I
know
hari
krishnan
is
the
only
one
here
is
ng
sock,
so,
but
I'm
gonna
we'll
have
this
available
later
and
I'll
see.
B
Well,
it
does
it
can,
but
it's
like
the
thing
about
github
is,
you
can
have
like
a
version
of
it
where
you
can
download
a
desktop
version
and
have
everything
local
and
then
have
everything
up
there.
You
know
in
the
cloud,
so
you
can
you
know
things
can
be
synchronized
when
it
goes
down.
It
just
goes
down
and
it
comes
back
up
and
yeah
you're
going
to
say
something
right.
Krishna.
B
Okay,
yeah
so
yeah.
I
think
this
will
be
good,
a
good
way
to
do
things
and
I'm
looking
forward
to
starting
off.
You
know
getting
getting
going
on
this.
B
Like
I
said
this
first
couple
weeks
is
the
community
period,
so
I
want
to
make
sure
that
people
are,
you
know,
getting
to
know
their
community,
and
that
means
you
know
being
in
the
meetings
and
you
can
watch
the
meetings
afterwards,
but
you
know
having
some
slack
interactions
and
just
be
learning
a
little
bit
more
about
like
open
worm-
and
I
know
like
even
like
krishna
and
koran
have
been
in
the
meetings
here,
but
I
don't
know
how
much
they're
familiar
with
open
worm
overall.
B
B
So
then
again,
I
don't
want
this
first
week
first
week
or
so
to
be
overwhelming,
because
you're
going
to
get
a
lot
of
information,
but
just
just
kind
of
you
know
keep
those
emails
available
to
yourself
and
then
kind
of
you
know
go
through
them
and
use
them
as
references
for
later,
because
there's
going
to
be
a
lot
of
detail,
you're
going
to
have
to
absorb
to
get
started,
but
once
you
get
started
it
shouldn't
be
too
bad.
B
In
previous
years,
I've
tried
to
get
people
to
do
presentations
of
their
work
like
at
the
beginning
of
the
session,
and
I
think
I
mentioned
that
in
the
introductory
email
is
that
I'm
kind
of
wanting
people
to
do
a
very
short
presentation-
and
I
think
hari,
krishna
and
koran
have
kind
of
done
that
already,
but
I
want
to
have
like
I
want
everyone
to
kind
of
get
together,
maybe
about
like
a
10-minute
slideshow
of
what
their
project's
about
and
the
reason
I
want
to
do.
B
So
you
know
this
is
like
before
and
after
you
know,
you
get
to
see
what
worked
and
what
didn't
and
then
you
get
to
see
like
the
finished
product.
I
think
it'll
be
useful
for
people
to
you
know
kind
of
for
future
years
to
see
and
for
yourself
to
see
your
progress
so,
okay,
so
that's
it
yeah
go
ahead.
C
Yeah
bradley
on
this
harry
krishna,
I
had
some
discussion
about
what
doesn't
work.
C
Okay-
and
I
think
if
you
invest
some
time
in
something
that
doesn't
work,
it's
worthwhile
when
you
write
it
up
to
include
that
right,
because
that
keeps
other
people
from
making
the
same.
C
I
would
call
a
mistake,
the
same
path
that
didn't
work
right:
okay,
for
example,
he
tried
simulated
annealing
for
the
overlap
problem
and
for
some
reason
it
didn't
work,
and
it
would
be
nice
to
see
it
work
right
up
a
short
write-up,
but
why
it
didn't
you
know
what
what
happened
and
is
there
any
explanation
for
why.
B
B
Yeah,
that's
that's
definitely
true.
I
think
you
know
what
what
what
didn't
work,
what
worked
and
kind
of
like
you
know,
you're
making
this
progress
towards.
So
you
probably
proposed
I'm
gonna.
Take
the
number
of
you
know,
potential
approaches
and
then
you
know
you've
tried
a
couple
of
them.
Maybe
some
work,
some
didn't
so
that's
good.
You
know
for
because,
usually
in
in
our
projects,
we
usually
encourage
people
to
think
about
like
what
the
potential
solution
is.
B
They
don't
hand
them
the
solution
and
say
here
you
go,
you
know,
so
it's
it's
useful
to
think
like.
Well,
you
know
to
start
off
by
thinking.
Well,
what's
most
likely
to
work,
and
then
you
know
it's
like.
No
one
really
knows
that
answer
you
just
have
to
kind
of
guess
from
what
you've
been
exposed
to.
B
So
I
might
pick
something
that
is
a
thing
that
I'm
familiar
with,
and
there
may
be
solutions
that
I
have
no
idea
exist
and
they
might
be
better
fits
for
it.
So
that's
something
you
just
have
to
learn
by
trial
and
error,
but
it's
also
good
to
know
kind
of
what
people
have
done
in
the
past,
because
it's
like
okay,
at
least
I
know
now
this
new
method,
the
pros
and
cons
and
so
forth.
B
Okay,
okay,
so
I
think
where
we
want
to
go
next,
I
think
I'm
going
to
go
to
some
of
the
stuff
that
we
have.
I
have
some
things
in
the
presentation
bin
or
in
the
paper
bin
that
I
want
to
talk
about
a
little
bit.
B
B
So
there's
this
idea
that,
like
developmental
biology,
is
all
washed
up
and
that
there's
not
much
more
to
discover
and
then
other
people
think
I
not
my
words
and
then
there
are
other
people
who
think
well,
there's
a
vast
frontier
of
things
to
talk
about.
So
you
know
there
have
been
a
couple
papers
that
have
come
out.
This
is
one
of
them.
This
is
developmental
biology,
is
poised
to
discover
altogether
new
principles
in
biology
in
the
21st
century,
and
this
was
published
in
developmental
biology
this
year.
B
So
this
is
something
that
is
just
kind
of
a
you
know,
sort
of
a
recent
conversation,
but
something
people
were
talking
about.
So
the
abstract
reads:
in
the
20th
century:
developmental
biology,
spearheaded
a
revolution
in
our
understanding
of
complex
biological
problems.
B
Its
success
rests
in
great
part
on
a
truly
unique
approach
that
has
recruited
a
diversity
of
systems
and
research
organisms
rather
than
focusing
on
isolated
cells
and
molecules.
So
a
lot
of
biology
has
been
very
much
focused
on
specific
genes
or
specific
proteins,
or
you
know
specific
cells
or
cell
populations.
B
If
you
think
of
neuroscience,
you
know
you
don't
think
in
terms
of
model
organisms,
so
much
you
think
in
terms
of
circuits,
where
you
think
in
terms
of
different
type
in
insects,
they
have
different
types
of
neurons
that
do
specific
things.
So
people
study
those
just
as
like
a
topic.
So
these
are.
You
know
this
has
been
the
focus
in
developmental
biology.
By
contrast,
you
think
about
like
different
model
organisms
like
zebrafish
or
like
axolotl.
B
That's
why
we
talk
about
these
different
species,
because
those
are
the
comparators
like
you
get
a
you
look
at
the
process
of
development
in
those
model
organisms.
Now,
that's
not
to
say
that
there
hasn't
been
work
on
say
like
the
molecules
of
development
or
the
cells
of
development.
There
have
been,
you
know,
specific
cells
that
are
or
cell
populations
that
are
interesting
in
development,
but
this
the
they're
kind
of
making
this
distinction
between
the
two
approaches
and
then
well
developmental
biology,
has
also
employed
a
wide
variety
of
technological
and
intellectual
approaches.
B
But
what
will
developmental
biology
contribute
to
this
century?
Advances
in
technology
and
instrumentation
are
presently
moving
at
neck
breaking.
I
guess
I
mean
breakneck
speed
and
herald
the
advent
of
an
age
of
technological
wonders
in
which
previously
inaccessible
biology
is
now
tangibly
within
our
grasp,
so
they
give
the
example
of
single
cell
rna-seq.
B
B
So
you
know
we
have
genome
sequences
about
what's
in
the
genome,
what
the
genes
are,
what
the
regulatory
elements
are,
and
you
take
these
little
fragments
of
rna
that
you
find
in
the
cell
and
you
can
sequence
those
at
like
you
know.
Maybe
a
hundred
base
pairs
and
then
you
can
match
them
up
to
the
source
in
the
genome.
B
So
there's
a
lot
of
computational
work
involved
in
that,
but
basically
it
gives
you
some
candidate
regions
where
that
transcript
came
from,
and
you
know
you
can
say
things
about
its
function
and
things
like
that.
So
this
is
a
nice
technique
and
living
in
the
single
cell,
of
course,
in
development
is
good
because
for
in
the
case
of
c
elegans,
a
lot
of
the
cells
are
very
unique
in
terms
of
their
function.
So
you
have
a
cell.
That's
you
know
a
cell
that
is
born
of
like
a
precursor
cell.
B
They
divide,
then
they
differentiate
into
their
adult
form
and
they
retain
their
sort
of
identity
as
a
single
functional
type
throughout
development.
It's
different
than
most
other
parts
of
development,
but
having
the
single
cell
approach
in
c.
Elegans
is
nice
because
you
can
see
what's
being
expressed
as
the
cell
goes
through
its
divisions
and
then
it
becomes
differentiated.
B
So
that's
you
know.
That's
some,
but
of
course
single
cell
rna-seq
is
as
problematic
in
some
ways
because
it's
very
hard
to
understand
what
the
data
mean.
So
you
know
what
does
it
mean
to
have
a
bunch
of
genes
upregulated
as
they
say,
meaning
that
there
are
more
copies
of
the
rna?
What
does
it
mean
to
have
the
genes
downregulated,
which
means
there
are
fewer
copies
of
the
rna?
Well,
you
know
it's
hard
to
know
exactly
what
that
means.
B
These
are
things
that
we
think
about
the
21st
century,
but
so
this
technology,
single
cell
rna-seq,
has
revealed
novel
transient
cell
states
in
both
the
stem
and
differentiated
cells
that
are
specified.
You
know
the
in
the
organism.
This
defines
changes
in
gene
expression,
frequency
during
something
like
regeneration
or
some
other
process.
So
you
know
again,
this
is
the
ideal
case,
but
in
a
lot
of
cases
you
don't
necessarily
know
what
some
fluctuation
in
rna
means.
It's
just
kind
of
a
there's.
A
lot
of
you
know:
noise
at
different
scales.
B
You
have
environmental
stress,
you
have
you
know
you
have
to
characterize
what's
going
on
in
that
cell
type,
so
that's
always
the
challenge
there,
but
that's
one
tool
and
then,
of
course,
we
have
genome-wide
epigenetic
analyses
which
are
the
analysis
of
say,
like
methylation,
which
is
this
process
that
drives
sort
of
it
mediates,
gene
expression.
B
You
have
these
methyl
groups
that
sit
on
top
of
a
certain
parts
of
a
gene
and
they
facilitate
transcription
of
parts
of
the
gene.
So
a
methyl
group
will
sit
on
top
of
different
parts
of
a
gene
and
they'll
either
be
opened
or
closed
and
when
they're
open,
they
enable
certain
transcripts
to
be
expressed
and
when
they're
closed,
they
disable
those
transcripts
to
be
expressed.
And
so
there
are
a
lot
they're.
B
You
know
so
they've
been
able
to
define
different
switches
in
genomes
so
like,
for
example,
you
can
you
do
this
sort
of
methylation
analysis
where
you
identify
all
these
methyl
groups
and
then
you
look
at
their
comp,
their
state,
whether
they're,
open
or
closed,
and
that
can
be
mapped,
say
to
some
phenotypic
outcome
or
it
can
be
mapped
to
what
the
gene
expression
is
doing.
B
So
you
know
there
are
ways
you
can
look
at
those
you
can
look
at
sort
of
a
methylation
analysis,
bring
it
to
bear
on
gene
expression,
and
it
tells
you
something
about
some
of
these,
what
they
call
epigenetic
changes
to
the
genome.
So
that's
another
tool,
that's
useful
and
then
so.
These
are
all
useful
sort
of.
Ultimately
in
telling
us
things
about,
you
know,
development
and
differentiation
and
other
sort
of
processes
that
involve
cells,
maybe
cell
populations
or
single
cells.
B
Now
the
problem
with
looking
at
these
things
in
cell
populations
is
that
you're
doing
a
lot
of
averaging
and
average
behavior.
So
this
is
a
little
bit.
You
know
this
is
another
tricky
aspect
of
it
is
what
you
know,
which
is
why
it's
you
know
interesting,
but
it's
you
know
we're
still
kind
of
thinking
about
the
implications
of
it.
B
So
these
circumstances,
combined
with
our
disciplines,
diversity
of
experimental
and
intellectual
approaches,
offer
unimaginable
opportunities
for
developmental
biologists,
not
only
to
discover
new
biology,
but
also
to
reveal
entirely
new
principles
of
biology.
B
B
This
top
panel
is
the
flatworm,
so
this
is
a
model
organism
that
has
all
the
cells
in
this
flat.
Worm
are
toady
potent
meaning
if
you
take
any
one
cell
and
you
put
it
in
a
new
location,
you
know
put
it
like
the
culture
of
these
worms
and
dishes.
So
if
you
take
one
cell
out
of
this
flatworm
and
you
put
it
in
a
new
dish,
it
can
grow
an
entirely
new
flatworm.
B
B
Yeah
yeah-
and
you
know
so-
there
have
been
more
there's
been
work
done
on
this
in
terms
of
bioelectricity
like
if
you
take
one
cell
and
you
put
it
in
a
new
context
and
you
can
grow
a
whole
new
individual.
How
does
that
happen?
I
mean
there
has
to
be
some
sort
of
three-dimensional
coordinate
system
here,
but
what
is
it
so?
It's
called.
B
You
know,
I
think
mike
levin's
done
a
lot
of
work
on
like
bioelectricity
and
how
that
sets
up
this
sort
of
spatial
reference
frame.
But
you
know
again,
this
is
a
problem
that
isn't
reducible
to
necessarily
to
gene
gene
expression
and
methylation
and
epigenetic
modifications,
at
least
in
the
way
that
we're
measuring
them
so
they're
in
this
paper,
they're
kind
of
proposing
that
a
lot
of
the
tools
that
we
have
are
molecular
and,
of
course
there
are
new.
There
are
other
tools
that
we
have.
We
have
ways
to
measure
physics.
B
So
there's
also
then
this
organism
here
this.
This
is
just
a
fish
with
a
fin
and
this
shows
fin
regeneration.
So
this
is
like
in
this
case
the
head
is
developing
from
sort
of
a
an
embryonic
state.
In
this
case
the
fin
has
already
been
developed,
but
it's
been
damaged
somehow,
and
so
it's
being
regenerated
and
so
they're
arguing.
Is
that
the
same
pro?
And
we
know
this
that
similar
processes
are
at
work
to
rebuild
the
channels,
not
that
cutting
off
the
head
doesn't
work
for
a
fish.
B
B
C
C
Any
rate
yeah
bradley,
I
my
answer
to
people
who
think
that
developmental
biology
is
over
is
in
the
chat.
B
Okay,
so
let's
see
yeah
so
so
yeah
dick
said
that
he
has
well.
He
has
his
paper.
Are
we
on
the
cusp
of
a
new
paradigm
for
biology,
theological,
molecular
developmental
biology
versus
janus
faced
control
of
our
biogenesis
by
differentiation
waves
yeah?
This
is
a
paper
that
you
did,
I
think
with
yeah.
So
this
is.
C
E
B
This
paper
that
we
have
here
is
just
talks
about
a
lot
of
it
talks
about
plasticity
and
development.
So
plasticity,
of
course,
is
this
other
idea
that
you
have
an
adult,
phenotype
and
you're
actually
able
to
regenerate
it
in
a
way
similar
to
development.
So
this
is,
you
know,
using
sort
of
the
same
molecular
pathways
and
some
of
the
same
mechanisms,
and
so
this
paper
is
very
much
focused
on
the
molecular
aspects
of
it.
B
You
know
they
make
these
really
colorful
figures,
so
this
one
shows
the
relationship
between
transcriptional
state
and
time
and
showing
these
fluctuations
and
transcriptional
state
and
how
that
can
affect
sort
of
the
progression
of
different
developmental
processes.
B
That
is
important
in
setting
up
a
lot
of
the
stuff
and
driving
it
forward,
and
then
you
have
these
type
of
graphs
where
you
have
rna
over
versus
time,
so
you
have
different
tissues
and
if
the
rna
for
those
that
define
those
tissue
types
are
expressed
at
certain
points
in
time,
you
get
the
correct
tissue.
If
you
don't,
then
you
get
defects
in
the
tissue,
that's
what
they're
showing
there.
B
So
that's
that
paper
and
then
there's
this
other
paper
called
the
aerotechnic
organization
of
developmental
biology,
and
so
this
kind
of
talks
about
they're
kind
of
talking
about
the
presence
or
absence
of
theory
and
development.
B
And
of
course
we
know
that
there's
theory
and
development
in
this
group,
but
some
people
you
know
are
debating
you
know
what
the
relative
value
of
it
is
so
and
this
this
book,
chapter
developmental
biology,
is
a
science
of
explaining
how
a
variety
of
interacting
processes
generate
the
heterogeneous
shapes,
size
and
structural
features
of
an
organism
as
it
develops
from
embryo
to
adult
or
more
generally,
through
its
life
cycle.
B
So,
although
it
is
commonplace
in
philosophy
to
associate
sciences
of
theories
such
that
the
individuation
of
a
science
is
dependent
on
the
constituent
theory
or
group
of
models,
so
this
would
be
like
where
you
associate
a
certain
scientific
field
with
certain
theories
or
models.
So
you
know
you
would
come
up
with.
I
think
in
physics,
you
know
you
would
have
certain
models.
If
I
said,
like
quantum
loop
gravity,
you
would
have
like
a
series
of
models
or
theories
associated
with
that,
and
you
know
it.
B
It
is
uncommon
to
find
presentations
of
developmental
biology,
making
reference
to
ethereum
theories
of
development,
for
example,
in
the
third
edition
of
essential
developmental
biology,
three
families
of
approaches
are
described
and
those
are
developmental,
genetics,
experimental
embryology
and
molecular
in
cell
biology,
so
those
are
kind
of
like
the
sort
of
the
consensus
views
of
how
to
look
at
developmental
biology,
and
that's
kind
of
you
know.
I
don't
know.
Experimental
embryology
might
include
like
a
vast
umbrella
of
approaches
or
that's
an
umbrella
for
a
lot
of
different
approaches.
B
The
same
thing
probably
holds
true
for
molecular
and
cell
biology
there's
a
lot
of
stuff
under
that
umbrella.
So
it's
not
really.
You
know
that's
kind
of
a
family
of
approaches
and
the
appendix
contains
a
catalog
of
key
molecular
components,
so
genes
transcription
factor
families,
including
factor
families,
cytoskeleton
cell
adhesion,
extracellular,
matrix
components.
B
However,
no
standard
theory
or
group
of
models
provides
a
theoretical
scaffolding
to
the
book,
nor
is
any
mentioned
so
in
that
book
they
didn't
have
any
overarching
theory,
and
so
again
we
know
that
we
have
theories
of
development,
but
the
idea
is
that's
not
something
that
they
in
the
textbooks,
at
least
they
don't
organize
it
by.
This
is
a
theory
that
unifies
everything
and
so.
C
Yeah,
you
know
you're
talking
about
this,
led
to
something
that
you
might
want
to
try
look
at
the
chat.
C
Yes,
it
seems
to
me
that
you
know
just
like
was
done
with
the
the
vehicles
thing.
One
could
start
with
networks
and
ask
what
does
it
take
a
network
to
regenerate
itself?
We
have
examples
in
some
of
the
law
firm
patterns,
okay,
because
it
won't
bother
patterns.
You
regenerate
itself.
B
C
B
C
B
B
B
I'm
not
sure
I've
seen
some
papers
where
they
kind
of
maybe
go
in
that
direction,
but
I
don't
think
anyone's
ever
established
it
in
terms
of
rules
or
in
terms
of
I
know,
people
do
a
lot
of
things
with
things
called
motifs,
so
they
look
at
motifs
of
networks,
but
this
is
largely
like
usually
something
like
gene
expression.
It
could
be
like
even
like
a
connectome
of
of
neural.
C
B
Yeah,
so
I
guess
yeah
that
just
to
finish
up
this
paper,
the
absence
of
any
references
to
a
theory
development
is
puzzling
on
its
face.
Why
is
it
so
difficult
to
identify
or
isolated
constituents.
B
C
B
B
The
second
interpretation
is
a
lack
of
reference
to
theories
indicates
an
immaturity
in
the
field,
and
then
you
know
that
kind
of
goes
into
the.
So
that's
what
he'll
argue
or
the
third
option
is
what
I'll
argue
for,
which
is
that
the
lack
of
reference,
the
theory
should
be
taken
at
face
value,
so
I
mean
you
know.
C
B
C
B
B
C
B
But
yeah
it's
it's
always.
B
B
B
So
this
is
his
argument
here
and
I
don't
know
I
guess
it's
more
about,
like
whether
theory
is
going
to
be
a
guiding
force
in
the
field
or
not,
rather
than
whether
there
are
theories
or
not.
So
this
is
kind
of
a
you
know
this.
This
could
be
sort
of
a
chance,
wiggle
room
on
this,
but,
on
the
other
hand,
holding
a
developmental
biology
already
has
a
theory
or
theories
costumed
in
different
guys,
not
referred
to
as
such,
by
developmental
biologists
as
a
possible
interpretation.
B
So
you
know,
we've
talked
about
things
like
positional
information
or
differentiation,
trees
and
differentiation
waves.
Those
are
you
know.
Those
are
of
course
theoretical,
but
are
they
something
that
we
use
in
the
field
in
general
as
an
interpretive
mechanism?
Or
is
it
just
something
that
we
don't
we
kind
of
deal
with
separately
in
in
certain
cases?
B
So
this
is
kind
of
an
interesting
relationship
between
the
two
things,
let's
see
so
in
in
his
infamous
book
against
method.
Fiera
bend,
who
was
a
famous
philosopher
of
science,
argued
against
the
presumption
that
their
universal
methodological
rules
that
govern
scientific
inquiry.
B
B
The
association
of
maturity
with
a
science
having
a
central
organizing
theory
is
a
long-standing
idea,
so
you
know
the
idea
is
that
maybe
not
all
sciences
have
a
central
theory
by
which
everything
is
explained.
Maybe
you
know
having
multiple
theories
that
explain.
Certain
aspects
is
something
that
is
sort
of
the
nature
of
developmental
biology.
B
This
is,
of
course,
supposed,
as
opposed
to
evolutionary
biology
where
natural
selection
is
basically
the
mechanism
and
darwinian
evolution
is
the
consensus
view.
So
that's
basically
so
when
he
says
aerotactic
in
this
paper,
so
he
offers
an
account
of
how
problems
play
the
organizing
and
guiding
rule
typically
presumed
to
come
from
theories.
B
So
the
idea
of
aerotectic
or
pertaining
to
questioning
sort
of
frames
how
developmental
biology
is
developed,
so
developmental
biology
is
developed
around
questions
and
about
you
know
different
problems
that
come
about.
So
it's
very
different
from
like
a
field
where
you
have
a
lot
of
sort
of
overarching
questions
that
might
be
best
explained
by
a
theory
from
the
data,
and
you
know
it
has
sort
of
a
nature.
It
flows.
Naturally,
from
that
in
developmental
biology.
B
And
they
can
talk
about
this.
This
is
a
figure
here.
The
relative
frequency
of
theory,
theory
theories,
theoretical
and
article
keywords
from
different
domains
of
scientific
inquiry.
So
this
is
the
percentage
of
articles
in
this
domain
that
have
theory
in
the
title
and,
as
you
can
see,
anatomy
and
biomechanics
have
zero
percentage
of
articles
devoted
to
theory,
whereas
economics
have
almost
a
hundred
percent
engineering,
that's
almost
100
percent,
where
the
frequency
of
theory
exists.
B
I
don't
know
how
engineering
was
calculated,
but
in
case
you
see,
there's
variation
between
different
fields,
and
so
I
guess
the
point
of
this
graph
is
to
show
kind
of
how
closely
theory
is
tied
to
empiricism.
So
when
you
have
a
paper,
do
they
reference
theory
or
not?
And
so
you
know
a
lot
of
engineering
papers,
for
example,
have
a
set
of
equations
that
they
review,
and
so
that's
often
implicitly
theory,
and
so
they
mention
theory
in
the
paper.
B
There's
there's
an
engagement
with
theory.
There's
a
dialogue
and
another
field.
There
really
isn't
that
development
is
here
actually
and
that's
zero
percent.
Just
like
anatomy
and
biomechanics-
and
you
know
that's-
maybe
it's
maybe
an
artifact
of
people
not
really
engaging
with
sort
of
you-
know,
mathematical
models
or
other
types
of
theories
that
may
exist
and
so
that
there's
there's
there
are
multiple
explanations
for
this.
B
So
again
textbook
presentations
of
developmental
biology,
it's
aerotechnically
organized.
So
a
lot
of
textbooks
are
organized
around
questions
around
some
of
these
themes,
but
not
necessarily
room
theories,
and
so
this
kind
of
goes
through
the
textbook
structure.
For
this
book,
essential
developmental
biology,
which
there
are
two
versions
second
and
third
edition,
so
they
kind
of
play
out.
B
It
plays
out
the
changes
in
these
two
editions
to
show
that
basically,
the
sort
of
presenting
the
field
to
a
bunch
of
students,
you
have
to
lay
the
groundwork,
but
the
groundwork
is
generally
these
topical
areas
and
not
necessary
necessarily
theoretical
areas.
So
there
are
approaches
to
say
developmental,
genetics
or
experimental
embryology
and
they
typically
lay
out
maybe
experiments
or
techniques
and
they're
not
based
around
theory,
major
major
model
organisms.
Again,
these
are
model
organisms
where
we
know
the
biology,
but
there's
no
real
theory
around
them
and
certainly
no
theory
linking
all
of
them
together.
B
One
criticism
of
evo
devo
is
that
it's
not
really
a
theory
that
it's
just
a
collection
of
neat
findings.
They
package
it
in
this
very
superficial
explanatory
framework
and
then
they
present
it
as
sort
of
an
alternative
to
the
evolutionary
synthesis.
And,
of
course
you
know
it's,
it's
not
really
a
theory
in
the
conventional
sense.
It
doesn't
predict
necessarily,
it
may
predict
things,
but
it
doesn't
really
put
them
into
a
larger
framework,
and
so
this
is
a
a
common.
B
The
you
know
it's
a
common
criticism
of
evo
devo
and
it's
also
sort
of
a
criticism.
I
guess
of
this
approach
to
development,
where
you're
laying
out
the
facts,
you're
laying
out
the
models,
but
they're
model
organisms,
so
they
don't
really
have
any
integrative
power,
just
examples
that
you're
picking
out.
B
I
think
this
is
so
more
analytically.
The
space
of
problems
in
the
field
of
development
is
characterized
in
terms
of
five
variables:
abstraction
variety,
connectivity,
temporality
and
spatial
composition,
and
so
this
is
something
that
we
the
way
we
kind
of
view
development
and
some
of
the
different
ways
that
you
know
you
can
look
at
development.
B
The
values
given
to
these
variable
structures,
a
constellation
of
research
questions
within
the
broad
problem,
agendas
corresponding
a
generally
delineating
phenomena.
So
you
could
imagine
now,
if
you
want
to
really
connect
development
or
theory
to
developmental
biology,
you've
got
to
focus
on
these
five
different
variables
and
kind
of
bring
the
theory
to
bear
on
each
of
these.
I
think
really
to
get
people's
attention.
B
Then
scott
gilbert
put
out
a
textbook.
The
10th
edition
of
developmental
biology
and
this
a
consistent
feature
or
multiple
editions
of
this
book
is
that
developmental
biology
is
cons
constituted
by
a
set
of
core
questions.
So
you
have
these
different
questions
that
the
book
is
organized
around
the
first
being
the
question
of
differentiation.
B
So
how
can
a
single
cell
generate
so
many
different
cell
types,
the
question
of
morphogenesis?
How
can
the
cells
in
our
body
organize
themselves
into
functional
structures,
the
question
of
growth,
how
our
cells
know
when
to
stop
dividing?
How
is
cell
division
so
tightly
regulated
and
so
on
and
so
forth?
And
so
it
gives
us
disorganized
questions
rather
than
like
this
conforms
to
this.
The
prediction
of
this
theory.
B
So
that's
what
he
means
by
aerotactic
and
this
again
is
different
types
of
morphogenesis.
So
you
can
see
that
there
are
all
sorts
of
different
types
of
morphogenesis
and
really
this
is
this
brings
up
another
point,
which
is
that
you're
dealing
with
very
diverse
systems?
So
you
know
the
question
is:
isn't
this
isn't
that
there
isn't
a
theory,
but
that
the
theory
has
to
be
very
broad
to
explain
all
these
different
modes
of
of
variation
where
these
all
these
variants
of
of
a
common
set
of
processes?
B
B
You
have
positional
information
as
a
theoretical
model,
but
then,
if
you
try
to
put
this
in
the
context
of
all
these
different
things
that
you're
observing
in
nature,
you
know
is
there
one
theory
that
can
bridge
all
of
these
and
the
question
is
probably
not
yet,
but
we
have
candidates
and
of
course,
there's
no
reason
why
you
couldn't
apply
theories
to
each
of
these
different
cases
like,
for
example,
you
know
if
you
had
reaction
diffusion,
which
one
of
these
examples
would
it
explain
best
and
which
ones
would
it
fail
at,
and
so
you
know
this
is
maybe
another
road
to
another
paper
where
you
could
say
well,
you
know
there
are
all
these
different
modes
of
development
and
we
have
these
different
candidate
theories,
which
ones
fit
which
context
the
best
and
which
ones
fail.
B
And
so
you
know
you
have
the
sort
of
mosaic
of
theories
that
might
be
useful,
but
you
know
breaking
that
down
for
people,
especially
practitioners
in
the
field
which
theories
are
best
for
which
modes
of
development
and
so
forth,
which
examples,
because
a
lot
of
developmental
biology
is
driven
by
examples
that
you
might
see
under
a
microscope,
and
that
makes
sense
because,
if
you're
actually
engaging
in
the
field-
and
you
have
a
lot
of
variation,
you
might
want
to
know-
you
know
my
model
system.
How
do
I
explain
what's
going
on
there?
B
How
do
I
explain
what's
going
on
in
another
model
system,
if
you're
comparing
model
systems,
however,
it
becomes
hard
to
really
understand
that
thing
is
a
common
process.
For
example,
you
know:
we've
worked
with
c
elegans
and
zebrafish
and
xenophos
and
axolotl
in
this
group.
We've
worked
on
these
different
model
organisms,
but
they
have
very
vastly
different
developmental
periods.
B
B
So
this
is
so
yeah
this.
This
paper
is
nice
because
it
compares
some
things
on
the
ground
like
textbooks
and
how
they're
approaching
the
field
and
how
they're
kind
of
framing
the
thing
how
they're
framing
the
material
in
the
field
for
people
to
understand
its
structure,
and
so
it
doesn't.
You
know
there
isn't
a
lot
of
theory
that
gets
introduced.
I
mean
usually
in
an
evolutionary
biology
course.
You
would
learn
about
evolutionary
theory.
You
would
have
examples
of
things
in
nature.
B
You
have
examples
in
evolutionary
theory
like
the
peppered,
moth
or
other
models.
You
know
like
simple
hardy-weinberg
equations
things
like
that,
but
you
don't
really
have
this
level
of
sort
of
question
driven
presentation
where
you
have
all
these
different
subtopics
and
then
you're
asking
questions
but
you're
not
really
integrating
them
very
much.
B
Okay,
so
this
it
kind
of
goes
through
this
question
of
sort
of
how
the
heterogeneity
of
developmental
biology
is
kind
of
driving
this
sort
of
question-driven
approach.
B
So
this
is
really
kind
of
a
key
thing,
and
I
think
it's
you
know,
maybe
that's
something
that
needs
to
be
overcome.
Maybe
there
needs
to
be
a
developmental
biology
textbook
that
doesn't
focus
on
the
question
so
much,
but
maybe
focuses
on
theories,
or
maybe
we
need
to
do
a
session
where
we
kind
of
work
out
these
theories
or
we're
a
series
of
sessions
where
we
kind
of
work
out
these.
B
How
theories
connect
to
a
lot
of
what's
going
on
in
developmental
biology?
In
fact,
that's
why
I
do
a
lot
of
these
paper
presentations,
because
I
like
to
bring
up
these
topics
and
I
like
to
see
how
we
can
maybe
fit
theory
to
these
topics
or
unpack
them
in
in
a
similar
way,
so
make
connections
with
other
parts
of
development.
B
C
D
I've
noticed
that
even
even
oh,
I
forget
the
guy's
name
anyway,
they'll
say
something
and
it
looks
like
there's
making
a
statement
that
this
is
true
and
then
they'll
go
oh
well.
Maybe
I'm
not
correct
and
they
say
that
in
biology
all
the
time
and
as
an
engineer
you
take
the
paper
and
you
go
shred.
It
shred.
D
D
D
Well,
this
man
they're
called
difficult
for
me
some
days
anyway,
yep
yeah,
okay.
D
I
guess
I
can
turn
this
on.
I
don't
know
I
am
getting
fiber
optic,
it's
been
cooked
up
to
my
house,
but
it
hasn't
been
attached.
So
I
asked
when
that
might
happen.
They
said
september.
So
I
don't
know
I
have
fiber
optic,
but
it's
not
don't
do
anything.
B
C
C
Okay,
even
though
the
number
term
is
infinite,
it
does
converge,
because
the
steps
get
smaller
and
smaller.
B
B
C
C
B
C
B
B
Yeah,
like
some
sort
of
what
they
call
sort
of
a
toy
network
where
you
just
kind
of
generate
a
yeah.
C
And
it
would
fit
in
with
the
idea
of
the
snail
shell
patterns
regenerating
themselves
after
they
after
the
shell
has
been
broken,
because
they've
got
to
start
over
again
yeah
okay,
so
since
they
obviously
do
start
over
and
reproduce
the
same
pattern
they
that,
if
they
really
have
a
net
worth
basis,
then
maybe
we
could
we've
got
a
simple
model.
There
yeah.
C
D
Yeah,
just
I'll
take
pictures
of
of
gumballs
or
something
and
I'll
get
enough
trouble
over
that
so
and
I'll
just
say,
but
I
hadn't
had
the
machine.
I
just
put
a
little
thing
in
it
and
I
took
pictures
for
somebody.