►
From YouTube: Thinking about Motor Representation
Description
(Previous version had missing content) Jeff talks and asks a lot of questions.
B
A
A
So
it
is
related
to
it,
because
I
was
thinking
about
different
types
of
representations
and
the
vertical
columns
of
talk
of
STRs
and
their
meanings
of
them,
but
not
specifically
this
stuff.
So
I'm
gonna
start
with
it
like
a
review
of
things,
but
for
some
of
you
guys
won't
be
able
to
review.
It
might
slip
a
few
things
so
I'm
just
seeing
about
a
single
block
of
cortex
this
generic
cortex
column.
A
If
you
will-
and
we
can
say-
okay,
what's
the
goal
of
a
block
of
cortex
it
calling
for
tax
in
the
way
I
talked
about
the
gold
up
thing
is,
is
to
learn
a
model
of
the
world.
It
gets
some
input
from
some
places
it
doesn't
if
its
genetic
or
it
doesn't
want
that
input
represents.
There's
no
idea.
What
represents
it
doesn't
notice
mortality.
A
A
D
A
A
Meaning
it
predicts
whether
it
was
the
next
input
and
therefore
it
must
have
some
copy
of
the
motor
command.
That's
the
Ephrem.
Ec
is
efference
copy.
That's
a
nice
term
that
was
calm
promoter
command,
so
it
must
know
something
about
the
movement.
Otherwise
it
could
predict.
What's
going
to
happen,
we.
D
A
Where
that
shows
up
here
is
very
clear,
with
the
input
show
that
the
input
comes
from
something
and
goes
to
the
thalamus
in
this
context
of
a
fork
of
all
the
places
too,
but
this
Green
King,
the
border
efference
copy,
is
kind
of
a
mystery
when
he
spent
a
lot
of
time.
Talking
about
that,
we
do
know
a
lot
about
the
output
of
this
thing
and
the
goal
the
goal
this
comes,
not
just
for
a
model,
could
what's
the
point
of
disciplining
a
model.
A
The
purpose
of
the
end
is
to
use
that
model
to
create
behaviors,
and
so
once
you
have
a
model
of
the
world,
not
only
can
predict
the
future,
we
could
say:
oh,
how
do
I
get
to
a
particular
state
of
that
model?
How
do
I
get
to
the
particular
position
in
space
and
actually
I'm
a
distinction
between
location
and
state,
because
things
are
stakes?
I
would
turn
around
my
phone.
My
cell
phone
as
a
state
I
want
to
get
some
state.
C
A
Here's
a
5,
then
it
sells,
come
out
and
they
go
into
the
top
of
this
column-
something
like
that,
but
we
never
really
see
that
either
because
in
the
primary
sensory
cortex
now
that
that
would
apply
the
input
from
Loretta
is
a
motor
command
and
we've
actually
so
I'm
thinking
about
that,
and
could
it
possibly
be
viewed
as
a
motor
command
and
so
I'm
getting
closer
to
that
problem
here,
but
I
didn't
want
to
dress
it
over
here
here.
There's.
A
A
A
A
A
Getting
sensory
inputs
regularly,
so
this
is
the
general
idea
and
I
think
you
could
say
general
approach.
The
column
is
that
what
are
this
model
and
from
that
create
behaviors
that
that
allow
you
to
get
to
something
good
reach,
a
goal
which
is
some
location,
some
state
in
the
world
and
of
course
you
have
lots
of
columns
doing
this.
A
A
A
A
Later
find
out,
that's
that's!
Basically,
these
layer,
five
cells,
that
project
to
some
someplace
else
in
the
brain
that
is
motor
related,
so
they
never
drive
muscles
directly.
They're,
always
they're,
always
projecting
some
other
area
that
drives
the
muscles
and
the
cortex.
We
we
can
say
that
it's
going
to
be
another
pharmacist,
different
representation,
because
everything
that
rain
is
as
far
as
the
Duggars
are
presentation,
we
can
say
that
I've
always
said:
I
believe
that
that
these
cells
don't
know
what
the
projected
terror
they
don't
know
how
to
do
this.
A
They
have
to
learn
how
to
control
the
subcortical
thing.
So
this
is
the
generic
learning
system.
Is
it
okay,
I
haven't
said
nothing
about
how
to
socially
linked
to
these
other
cells,
maybe
motor
behavior,
so
that
I
can
control
them.
So
that
needs
to
be
learned.
This
is
not
like.
Oh,
this
cell
goes
to
this,
and
if
so,
does
that
thing
it's
like
I,
just
a
bunch
of
self-projected
some
other
budget
and
somehow
have
to
control
them.
A
So,
in
general,
every
region
of
cortex
for
economic,
be
self
projective,
some
sort
of
a
motor
area.
That,
then,
does
that
draw
behaviors
and
one
of
the
questions
that
I've
always
worried
about
or
thought
about
is,
and
we
don't
really
have
an
answer
to
this
is
what
is
the
form
of
this
motor
STR
Highway?
Think
about
that.
B
A
To
control
something,
but
what's
how
I
understand
them,
how
it's
like?
What
are
these
bits
mean?
What
about
five
cells
represent
inspected
me,
so
that
is
like
what
is
the
form
of
the
motor
STR?
What
do
the
bits
these
other
five
cells
mean?
Can
I
put
some
kind
of
meeting
on
and
how
do
they
associate
linked
to
the
sub
part
of
the
motors?
It
said
this
whole
link
here
is
kind
of
a
misrepresentation
and
we
was
much
poor
formed
idea
in
my
mind,
at
the
input.
A
But
think
about
that,
so
that
was
I
was
thinking
about
this
problem
and
that's
what
this
this
discussion
about
now
I'm
going
to
go
through
the
spaceship
pool
of
any
kind
of
hypothesis.
This
is
not
new.
This
is
like
okay,
we
have
an
input.
However,
you
thought
was
coming
into
here
and
I
think
that's
a
clue
to
the
how
we
were
going
to
output
here
so
far,
so
good
okay.
So
this
is.
B
C
B
A
A
A
A
A
A
A
In
a
good
mini-comic,
complex,
meaning
that
that's
not
truly,
the
the
cortex
learns
to
represent
these-
that
the
cell
becomes
active
at
different
positions.
So
there's
a
spatial
invariance
that
simple
cell
and
then
there's
a
complex
out,
and
so
there's
a
lot
of
hypotheses
about
how
these
self
sponsors
are
learned
because
they
don't
seem
to
be
in
it's
not
being
passed
in
from
the
input
space.
It
seems
to
be
determined
mostly
the
most
people
believe
it's
determined
within
the
column
or
within
the
cortex
that
it
learns
to
represent
cells
that
represent
these.
They
respond
over.
F
A
A
simple
cell
and
to
the
inputs
and
the
inputs
to
this
typically
are
like
coming
from
the
red.
Now
there
would
be,
as
these
cells
here
might
respondent.
You
like,
if
there's
a
point,
that's
on
surrounded
by
negative
and
that's
makes
that
sound,
and
so
what
you
can
do
is
to
get
one
beat
what
it
looks
like
when
you
find
a
simple
cell
on
the
cortex,
which
is
often
they
seem
in.
A
Like
metaphor,
it
looks
like
they
have
a
bunch
of
these
lined
up
literally
so
that
Oh,
when
there's
a
simple
cell
will
say
when
there's
a
bunch
of
these
guys
lined
up
like
this.
That
represents
then
I
become
active.
So
that's
like
a
feature,
but
it
notice
here.
If
I,
if
I
move
the
inputs
over
a
little
bit,
a
simple
cell
stops
working
stops
responding.
It
has
me
this
position
and.
G
A
F
A
H
A
Visually
can't
tell
they
just
observe
this
and
that's
the
distance
observation
experimental
observation.
So
there
was
some.
You
simple
sounds
most
of
cells.
If
you
go
across
again
in
a
very
calm,
you
basically
get
all
cells,
the
many-colored
every
basically
based
on
the
same
feature,
but
then
most
of
the
cells
of
the
mini-com.
D
A
A
Different--You,
so
some
cells
seem
to
prefer
it
all.
They
go.
This
way,
I
become
really
active,
but
I
don't
respond
at
all
pickle
that
way
and
there's
an
unselfie
show
the
lines
that
way.
I
respond
very
strongly,
but
I
don't
respond
at
all
going
this
way.
So
the
majority
of
the
cells
in
a
quarter
calm
a
meeting
column
all
of
the
same
sort
of
simple
representation,
but
then
they
differentiate
by
movement
over
over
through
space,
and
so
again
this
is
been
studied.
A
A
lot
empirically
now
on
my
thought
about
this
always
has
been
like
oh
well,
these
movements
here.
Why
would
why
would
the
cortex?
Why
would
mostly
cells
be
directional,
sensitive,
joyful,
sensitive
and
you
might
say,
oh
well,
because
I'm
observing
a
bird
fly
by
and
I
need
to
represent.
You
know
the
movement
going
one
way
or
the
other
way
something
like
that,
but
I'm
gonna
make
it
Alton
hypothesis
today
about
why
they
look
like
this.
A
So,
let's
do
some
that
led
to
this
question,
I
say:
okay,
most
of
those
may
come
our
complex
positioning
variant
most
of
the
complex
cells,
a
motion-sensitive.
Why
are
they
most?
Why
would
that
be
a
dominant
representation
and
throughout
the
different
layers?
Here
and
again,
it
could
have
been
just
like.
Oh
I
need
to
represent
movement
in
the
world,
but
there's
another
possibility.
The
other
possibility
is
number
there's
two
reasons
that
things
changed.
A
Be
willing
finger
choppers,
oh
my
god.
If
I
move
my
finger
one
way
or
the
other,
then
these
then
I
would
I
would
see.
I
would
see
these
these.
These
changes
like
this,
and
so
these
changes
could
be
representing
this.
These
this,
the
drunken
sensitivity,
because
we're
representing
essentially
movement
of
the
thing
again
think
about
finger
or
not
moving.
But
it's
basically
saying
this
pattern
is
changing,
because
the
thing
below
me
is
moving,
it's
it's!
A
It's
sensor
in
the
world,
it's
moving
in
space,
so
it's
my
fingers
moving
on
for
this
thing
like
this
and
and
I
would
want
to
differentiate
between
moving
one
way
or
the
other
way.
Now
in
the
visual
cortex,
the
the
predominant
breaking
up
in
accomplished
by
orientation
movement
is
basically
movement
perpendicular
to
that
orientation.
A
So,
obviously,
if
I
have
a
mini
column,
that's
representing
a
vertical
line
and
the
movement
of
the
eyes
such
that
I'm
moving
this
way,
College
of
the
Rhode
Island.
There's
not
much
change.
Really
this
way,
then,
that
things
that
go
across
so
you
can
think
of
this
is
sort
of
these
changes
are
movements
perpendicular
to
the
slice
of
the
mini
colony,
so
I
got
a
feature
and
now,
once
I
have
a
feature,
I've
been
driving
it
to
position.
A
A
So
this
is
what
I'm
saying
right.
This
is
this
half-baked
idea:
each
mini-com
is
a
plane
or
we
can
call
slice
or
some
unknown
name
for
that
of
the
input
space.
It's
easy
to
think
of
it
as
applying.
If
we're
not
limiting
ourselves
to
binary
the
complex
spells,
are
responding
to
perpendicular
shift
so
that
plane
they
mean
in
a
Cell.
A
A
A
So
that
was
it
one
of
motor
sense
themselves,
most
references
caused
by
motor
movement.
Then
a
single
mini
column
represents
a
play.
An
input
space
causes
that
most
insensitive
cells,
many
newcomers
and
perfect
different
wounds
of
their
plan,
and
that
feels
the
beginning
of
an
idea
about
how
to
think
about
motor
outputs.
It's
like
okay,
now
I
can
associate
cells
in
layer
five
with
the
behavior
down
years
and
I'm
an
SDR
in
there
v,
where
I
can
associate.
C
A
C
A
A
You
how
do
or
not
work
I'm
now
going
to
tell
you
what
to
do
I'm
going
to
tell
you.
You
should
be
doing
this
in
motion,
one
for
followed
by
which
is
26,
probably
frozen
932.
So
it's
like
I
need
to
this
is
when
I
go
back
to
here.
I
said
the
motor
output
must
be
associated
in
link
between
the
Apostles
and
the
subtitle
motor
errors,
because
first
it
doesn't
know
what
to
do
so.
A
Model
of
what
you've
just
done,
I've
seen
I've
seen
how
you
move
and
I've
seen
how
the
world
works.
I
put
them
on
here
and
that
model
by
the
way.
If
you
want
to
get
back
to
the
point
a
in
the
world,
I
have
to
do
that.
I'm,
just
gonna
general
series
of
behaviors
that
are
novel,
that
you've
never
have
just
sequence:
you've,
never
seen
before
I'm
going
to
create
a
novel
sequence
of
behaviors
you've
already
exhibited,
but
I'm
going
to
do
some
higher
level
of
all
where
the.
A
A
G
G
A
A
She
says
whether
she
wants
to
do
something
just
think
about
every
single
thing:
she
does
okay.
So
when
you
think
about
that
way,
the
cerebellum
is
involved
in
these
fine
motor
control
systems,
very
fine
motor
control.
They
smooth
things
out
to
do
small
movements,
a
blink,
the
eyes
of
the
rabbit
they
do
all
kinds
of
motor
behaviors,
but
they're
small,
and
not
very
they
don't
represent
a
model.
The
world
intermediate
functional.
If
you
will
and
without
it
the
cortex
has
a
hard
time
controlling
things,
because
it
has
to
be
more
direct.
A
A
But
it's
all
very
simple:
the
circuitry.
They
are
simple.
You
can
understand
how
it
operates
at
the
North.
Pole
now
prove
theories
by
binocular
those
cells
of
doing
its
operations.
This
decayed
that's
a
lot
of
it,
but
it's
not
very
sophisticated.
So
is
you
can
pay
over
like
a
smoothing
function
or
it's?
You
know
intermediate
low-level
Baker's,
but
nothing
really
interesting.
B
D
A
A
A
A
A
A
C
A
A
Or
you
trip
up
on
something
this
time.
We
noticed
it
right
away
this
time
tomorrow
said:
hey,
that's
born
into
my
vision.
If
I,
if
I
look
in
the
court
issues
directly,
my
eyes
move
around
some
object,
so
kind
of
expect
to
see
this
over
there
over
there
and
then
door
opens
aside
room.
Everyone
has
turned
to
the
door
ever
goes,
and
that
is
not
happening.
Appear,
that's
happening
down
here.
Then.
The
question
is
okay,
so
that's
pretty
good,
there's
no
door!
Something
was
changing.
It
moved
I
move
over
here
the
court.
A
A
D
A
A
With
my
model
say:
oh
yeah,
okay,
I've
decided
to
represent
movement
to
the
left
by
this
SDR
player,
five
cells,
and
you
are
sending
tape,
a
signal
that
says
that's
what
you're
doing
fine
I
can
associate
your
signal
with
my
signal,
but
I'm
not
using
this,
to
learn
the
model
I'm
using
this.
It's
inching
that
with
this
idea
that
these
mostly
directed
cells,
are
away
independently
learning
how
to
represent
behaviors
with
not
knowing
the
efference
copy.
Lord
man,
just
by
observation
of
the
many
columns
and
how
they
move.
A
B
A
Unless
I'm
not
saying
you
don't
have
it
I'm,
not
saying
it
I
guess
maybe
I'm.
What
I
am
saying
is
here's
I
thought
this
is?
How
did
you
learn
them
out
of
the
world
without
it?
How
do
I
incorporate
it?
It
would
be
different
and,
and
now
I
have
at
least
a
way
of
thinking
about,
like
five
cells
to
say,
yeah,
there
are
subsets
of
many
columns,
they
represent
what
that
slice
is
doing
in
which
way
it's
moving
and
that's
a
good
representation
of
a
motivator.
A
C
B
C
A
Maybe
yeah
it's
a
purpose.
They
didn't
want
to
talk
about
multiple
columns
here,
because
if
you
think
about
what
what
comes
in
a
hierarchy-
but
you
want
to
be
before
I
met-
come
up
with
something
if
I
think
about
multiple
columns
as
not
hierarchical,
but
parallel
that
different.
You
know
a
thousand
branch
apotheosis
they're
parallel
at
some
point
and
representing
different
slices,
the
size
of
the
input
space
I
might
come
up
with
a
different
thing
and
there's
a
huge.
We
still
don't
understand
how
layer
5
cells,
how
do
they
become
the
input
next
column,
but.
C
A
So,
let's
be
careful
so
you're
right,
so
a
single!
This
isn't
one
of
the
mysteries
of
movement.
So
if
your
single
column,
like
remember,
saying
when
tip
my
finger
and
it
could
represent
the
motion
that
finger
in
its
space,
but
then
when
I
actually
do
something,
that's
this
whole
complex,
arbitrary
changes.
I
was
avoiding
that
question
at
the
moment.
B
A
Motion
sensitive
cells
are,
that
is
like,
therefore,
out
the
cortical
column,
and
why
would
it
be
that
way?
Well,
it
could
be.
They
are
representing
behaviors
that
could
be
the
representation
of
movement
and
the
whole
system
is
centrally
motor,
so
I
have
sensory
endowment
I
have
a
SD
operative
represents
that
sensory
input
under
this
motor
of
position
motor
command.
A
C
B
A
C
A
But
the
same
thing:
our
homes,
the
whole
theory,
the
thousand
brains
there
in
the
common
theory
says
the
world
is
a
complex
structure
and
you
cannot
observe
it
all
at
once
and
therefore
to
learn
a
model
wall.
You
have
to
move
through
the
world
and
if
what
you're
doing
is
feed-forward
of
stuff
you're,
not
really
a
model
classification
of
pattern
classification,
but
to
learn
the
structure
of
the
world
and
not
move
through
the
world,
you
have
to
move.
A
It's
like
I
can't
learn
a
model
of
Redwood
City.
If,
by
sitting
in
this
room,
it's
impossible,
I
have
to
walk
around
the
city.
So
the
same
thing
with
a
cat.
All
I
see
these
things
flying
past
me
I'm.
How
am
I
going
to
know?
What's
going
on?
It's
like
I
can't
associate
my
behaviors
with
you
know:
I
can't
map
out
my
input
on
some
metric
space.
So
we
need
to
have
that.
It's
but
I,
just
I
thought
there's
something
would
really.
A
And
it's
basically
sort
of
integrating.
If
we
think
about
those
motion,
sensitive
cells,
not
as
like
the
world
is
moving
on
the
time.
It's
mostly
because
I'm
moving,
then
those
are
sensory,
motor
representations
and
that
and
then
it's
intriguing
to
think
like.
Why
would
I
want
to
represent
motion
and
orthogonal
movement,
so
the
primary
slices
I've
decided
it's
like
it's
like.
It
somehow
feels
right.
B
A
D
B
A
B
A
Of
the
feature
representing
movement,
where
the
other
one
is
a
subset
of
the
feature,
it's
a
the
feature
in
context
here
is
a
feature
in
motion
context.
How
about
it's
a
feature
in
motion
context
where
our
temporal
memory
is
the
feature
in
pure
time-based
context,
so
it's
again
subset
of
cells
would
become
active
in
the
mini
column,
representing
this
particular
behavior
at
this
moment
of
time.
E
I
drop
one
interpretation
of
please
just
this
picture
of
moving
orthogonal
to
of
I
just
wanted.
Please
don't
there's
one
easy
way
to
dry
and
I
am
50/50
I,
whether
this
is
what
you
mean
so
like
first,
just
like
we're
talking
about
like
a
single
mini
column,
and
you
like
some
set
of
input
bits
so
and
let's
say
this
particular
mini
column,
say
it's
featured
just
to
keep.
This
simple
is
going
to
be
these
two
bins.
It
it
likes
when.
E
E
So
because,
because
right
now
there
are
four
possible
responses
for
these
two
bits.
So
they
could
be
this
as
this
or
this
yeah
and
when
we
talk
about
a
plane
in
terms
of
the
neural
networks
and
SVM's
and
everything
you're
talking
about
like
this,
this
plane
depicts
this
columns
receptive
field,
yeah.
E
B
A
I
think
of
it
as
those
cells
are
defining
a
plane,
so
both
it
want
to
be
or
on.
Then
the
plane
goes
through
that
upper
right
point,
because
I
know
this
is
different.
I'm
just
confused
by
this.
Oh
sorry,
super
there
you
go
it's
just
like
my
visualization
to
start
with
you're
saying
you
want
to
be
far
away.
E
E
A
E
A
B
A
I
would
say
that
has
been
no,
but
all
of
a
sudden
all
my
haven't
exchanged.
Well
then,
it's
not
all
my
chains
in
a
job
type
of
movement,
thing
I,
think
I
get
difficulty
when
I
think
about
the
plane
and
where
is
but
we
could
talk
about
bits,
changing
and
I
would
say.
Well
if
my
inputs
that
I'm
responding
to
all
change
at
once,
then
the
movement
was
orthogonal
to
my
my
representation.
So.
A
A
D
A
Word
plane
I
mean
that's
my
tendency
to
think
about
physical
planes
that
are
slicing.
Something
three
dimension.
If
you
think
of
a
plane
is
a,
is
a
lower
dimensional
projection,
a
higher
dimensional
space,
so
some
dimensions.
If
you
change
in
some
dimensions,
it
doesn't
your
your
mic,
my
point
the
plane.
What
I
was
thinking
about
this
and
maybe
I'll
be
using
it
incorrectly
I
was
thinking
about
this
mini-com
represents
a
plane.
What.
A
The
same
mini
column
responds
exactly
the
same
where
some
of
the
bits
that
it's
not
paying
attention
could
change.
It
doesn't
affect
it,
but
if
the
bit
Center-
and
so
that's
like
those
are
those
are
movements
orthogonal
to
the
plane
and
orthogonal
to
the
representation-
and
he
goes
any
changes
from
these
bits
here-
don't
change
this
mini
column
and
and
then
you
say:
okay
well,
those
are
that
was
our
own
movements
but
but
yeah.
But
it's
not
just
any
change
here.
Remember
the
complex
cells
were
saying
where
the
complex
all
idea
comes
along.
A
A
G
G
A
A
G
A
A
A
A
C
C
C
A
G
A
A
E
B
A
G
A
D
D
A
A
Built
into
the
system,
everything
is
this
patient
pool
or
a
temple
puller
or
basically
the
whole
idea
that
you
can
put
them
out,
I'm
going
to
say
them.
The
20
bits
on
that
dendritic
on
the
generate
excitement
aren't
doing
that.
That's
something
different
they're,
not
pooling,
multiple
as
they
are.
They
essentially
doing
spatial
pooling.
If
you
want
to
consider
it
that
way,
yeah.
B
A
At
Empire
temple
point
typically
means
that
that
one
STR
and
another
STR
that
occurs
over
time,
which
are
completely
orthogonal
STRs,
they
don't
know
overlaps
bits,
get
converted
into
the
same
output
and
the
dendritic
segments.
We
don't
have
been
doing
that
the
dendritic
segments
say
I
just
recognized
a
single
pattern
and,
and
they
don't
they
don't
pull
on
that.
Second
I,
don't
the
same.
Dungeon
segment
doesn't
more
later
represent
completely
different
STR.
A
A
Anyway,
we
can
end
you
how
many
guys
have
a
lot
to
do
really.
The
thing
I
would
need
help
with
is
really
coming
up
with
a
visual
metaphor
and
the
use
of
the
term
for
planes
or
slices
or
you
know,
is
it
a
vector
these
are.
This
is
where
I'm
confuse
I'm
thinking
about
playing
to
something
else,
and
so
maybe
that
would
be
hopeful
on
that.