►
From YouTube: Numenta Research Meeting - Apr 10, 2019
Description
Broadcasted live on Twitch -- Watch live at https://www.twitch.tv/rhyolight_
A
D
C
And
even
the
evenings,
because
they're
really
great
so
I
did
get
I
feel
a
little
bad
turning
it
down.
It's
like
I
put
people
is
that
on
to
be
included
and
I,
don't
mind
I'm,
like
I,
said:
I
blown
lost
two
times
now,
two
years
in
a
row
and
I
got
a
little
bit
bad
about
that.
I
just
pointed
how
we
streaming
now
or
the
taps
right
now,
so
you
know
I
didn't
know
if
it
was
something
that
wanted
to
do.
You
ever
think.
Maybe
soon
you
talk
to
toured
or
something
like
that.
I.
B
D
So
I
guess
I'll
give
a
short
trip
report
on
the
conference
that
I
was
at
last
week.
So
this
is
a
is
the
Gordon
conference
on
dendrites
molecule
structure
and
function,
so
it
was
a
so
these
forming
conferences
are
mostly
invitation-only.
This
is
a
little
bit
larger
than
I
thought
it
would
be.
I
was
expecting
something
more
like
Banbury
side.
If
you
have
a
bit
bigger
this
was
about
so
that
would
have
put
at
like
50
people.
This
was
more
like
150.
D
C
D
C
C
D
I
sitting
yeah,
but
the
first
few
days
were
really
very,
very
low
level,
transcriptome
matrix
and
molecular
stuff
and
genetic
signaling
and
RNA
and
all
sorts
of
said,
which
I
didn't
really
get
a
lot
of
it.
The
first
day
and
first
yeah
I,
was
kind
of
wondering
hook.
What
what
a
minute
there
was
a
mistake
to
come
here.
Not,
but
it's.
C
D
D
Pure
science
disease
shows
up
as
it
doesn't
a
lot
of
these
conferences,
but
it
was
not
a
main
theme,
but
they
mention
things
like
if
it
was
this
one
or
not.
But
there's
in
the
synapse
apparently
there's
like
four
hundred
different
molecules
involved
in
the
whole
process
of
signaling
and
plasticity,
and
it's
a
lot
of
it's
triggered
by
calcium
and
then
there's
like
this
whole
chain
of
thing.
And
then
you
know
someone
studies,
three
of
them
into
someone
else,
studies
another
four.
It's
really
really
low-level.
D
So
this
is
all
about
different,
so
one
there's
stuff
that
about
plasticity
happening
at
different
time
scales
and
then
there's
homeostatic
plasticity,
which
is
how
set
of
resources
are
conserved
within
than
within
a
neighborhood
of
the
few
synapses
went
past
the
city,
of
course,
and
I
thought
that
was
interesting
because
it
kind
of
connects
to
a
lot
of
what
we
some
of
the
rules
that
we
have
in
our
temporal
memory.
So
this
is
one.
D
C
D
B
D
B
C
D
D
D
D
C
D
D
D
C
C
D
D
D
B
D
B
B
C
C
B
C
D
So
they
will
do
they
might
just
so.
The
people
studying
plasticity
at
the
level
of
spines.
They
don't
really
engage
in
very
rarely
do
they
engage
in
behavioural.
It's
really
very,
very
local,
a
lot
of
its
in
vitro.
You
know
they
can
just
barely
do
some
of
this
in
vivo
no
yeah,
but
here
they're,
actually
stimulating
spines
and
seeing
what
happened
that
the
others
find.
C
B
D
D
B
C
D
B
D
C
B
D
C
B
C
B
C
B
C
D
C
About
it
outlets
like
a
passive,
a
neuron,
you
say:
okay,
the
neuron
is
a
five
thousand
synapses
and
let's
say
almost
all
of
them
are
on
the
distal
dendrites
someplace.
So
in
how
many
different
patterns
you
can
recognize
it
well.
I
like
allocate
20
synapses
per
pattern,
then,
and
I
have
five
thousand
that's
what
250
patterns
and
that
you
could
recognize.
But.
C
Well,
assuming
that
all
the
segments
were
learning,
what's
the
maximum,
the
cell
can
learn
right.
You
know
I
five
thousand
synapses
twenty
per
pattern,
recognizing
if
they
were
all
perfectly
orthogonal
to
each
other
than
if
202
patterns.
If
that's,
if
multiples
of
those
I've
had
multiple
groups
of
cells
ever
doing
the
same
pattern,
then
I
fear
we
have
that
by
putting
represented,
couldn't
recognize
as
many
patterns
right.
B
C
C
B
C
C
C
Well
as
a
gym,
because
my
business
is
historically
in
the
cortex,
the
numbers
have
been
more
in
the
sort
of
the
three
to
ten
three
to
five
thousand
range.
But
then
lately
people
have
started
saying
ten,
thousands,
but
I,
don't
think
know
if
they
have
any
basis
for
that
and
then
in
the
thirty
thousand
number
was
the
hippocampus
right.
Maybe
you.
B
C
Those
in
the
cortex
I
don't
know
once
someone
telling
me
that
in
v1
stellate
cells
in
layer,
four
that
there
are
50
thousand
thousand
synapses
on
those
cells
and
then
I
went
back
later
than
asking
people
bad
news,
although
that's
not
true
so
I,
don't
know
where
that
I
don't
know
where
that
came
from
and
it's
it's
an
important
number
I
mean
from
a
theory
point
of
view.
We
could
accommodate
lots
and
lots
of
these.
C
D
Also
depends
a
lot
on
species
I
think
because
one
of
the
speakers
mark
carnet
from
MIT.
He
was
his.
He
was
basically
showing
examples
of
human
verified,
pyramidal
cells
and
mice,
pyramidal
cells
at
birth
and
rats,
and
they
were
a
completely
different
in
size
and
stuff.
The
humans
ones
are
much
larger
well,.
C
B
B
C
C
B
C
D
B
C
D
Domain,
it
comes
about
laborer,
yeah
yeah.
This
is
just
an
actual
actual
activity
at
night.
So
here's
an
example
of
like
you
get
some
stimulus
and
then
later
on,
you
get
the
reward
and
they
delay
the
reward
by
three
different.
You
know
zero
milliseconds
delay
to
5500
and
they
show
that
this
this
bump
an
apical
cuffed
shift
along
with
when
it
thinks
you're
going
to
get
the
reward
so
that.
C
D
C
C
D
C
B
C
D
C
D
D
D
C
D
Spikes,
a
bunch
of
people
had
trouble
finding
them
and
then
the
next
day
there
are
a
couple
of
people
who
said:
no,
we
do
see
them
and
I
spoke
with
Jackie,
Schiller
and
Matthew
Larkin
and
stuff
about
this,
and
they
said
that
cows.
This
is
like
a
constant
thing
for
them.
For
years,
people
have
been
saying
this.
This
debate.
B
D
C
D
Joe
they're,
like
oh
yeah,
you
can
do
whatever
you
control
all
the
membrane.
You
can
do
what
everyone,
but
does
it
actually
happen
in
Bebo
and
then
basically,
there
take
Matthew,
Matthew
and
Jackie's.
Take
is
that
these
techniques
would
not
pick
them
up
anyway,
because
they're,
calcium,
signaling,
they've,
slow
time
constants
there.
You
need
to
pick
up
something
within
30
microns,
which
and
there's
so
many
of
these
global
events
going
on
with
the
calcium
spikes
and
back
action
potential
that
those
signals.
D
C
C
C
D
C
D
Was
a
polite
public
debate,
homeless
people
would
say
no
they're,
not
there,
and
then
someone
else
would
say
they
are
there
so
but
I
think
in
general
we
have
to
be
very
careful
about
calcium
imaging
ending
with
our
level
of
stuff
is
at
timing.
So
it's
a
it's
very
noisy
signal.
It's
not
a
very
reliable
signal
and
to
the
timing.
Issues
are
really
quite
published
or
slow.
If
it's
too
slow
it's
and
sent
me
200,
millisecond
kind
of
time,
constant.
So.
B
B
B
D
C
C
D
C
D
Becomes
a
place
around
so
that
could
happen
how
interesting
yeah
and
then
they
showed
that
this
ramp
up
was
dependent
on
the
running
speed,
which
is
what
you
would
expect
from.
You
know
if
it's
sensory
driven,
so
that
was
kind
of
so
that
was
nice,
but
then
the
same
person
also
said:
there's
no
local
vendor!
That's
right!
So
there's.
D
Apparently,
this
is
critical
for
what
they
call
cognitive
flexibility,
which
is
kind
of
what
we
might
call
continuous
learning
like
unlearning,
something
and
relearning
something
else
like
if
you've
learned
something-
and
now
you
have
just
now-
a
signal
means
you
have
to
respond
to
something
else.
Then
what
do
you
have
to
before
then?
D
D
B
C
About
other
modulators
and
a
couple
decades
ago,
they
didn't
describe
that.
They
basically
described
that
the
neuromodulation
released
globally,
because
the
axon
spread
correctly-
and
there
was
no
there's-
no
sense
that
that
was
a
local
release.
I
kind
of
somehow
remember
the
serotonin
might
have
been
always
associated
with
a
local
release,
but
not
so
I
was
acting
because
I
couldn't
really
remember,
but.
B
D
B
D
C
D
That
is
it
our
learning
rules,
a
lot
more
flexibility
and
it's
kind
of
interesting
to
think
about
machine
learning.
The
people
don't
model
this
at
all
other
than
just
general.
Reinforcement,
learning
but
I
think
there's
sort
of
all
this
kind
of
more
local
ways
of
tuning
plasticity,
and
maybe
one
side
of
the
brain
could
tell.
C
D
D
B
D
D
B
C
That's
good,
okay,
yeah!
All
right!
Is
it
me
today
what
else
have
anything
to
talk
about?
No
I
I
wanted
to
talk
about
this.
Try
to
make
it
concrete
this
one
thing
as
we
can
do
and
I
don't
know
how
we
do
it
and
and
I
think
it's
a
it's.
A
good
way
of
sort
of
resolving
the
issue.
I've
been
talking
a
lot
about
these
sort
of
radial
and
linear
metric
spaces,
and-
and
these
relates
to
the
issues
we
plugged
markets.
C
The
past
about
you
know,
is
being
at
some
distance
from
the
object
and
solving
all
these
sort
of
problems
of
triangulation
and
so
on.
So
I'm
just
going
to
talk
about
this
one
problem,
I'm
just
going
to
try
to
articulate
it
clearly
because
I'll
talk
about
before,
but
nothing
of
a
particulate,
it
is
clueless
I,
think
I
kind
of
ticket
it
now.
So
it
relates
to
sitting
in
this
room
and
I'm.
Just
gonna
write
a
system.
C
And
the
idea
is,
you
have
your
in
some
room
and
this
is
of
course
a
metaphor
for
an
object
to
so
we
just
talk
about,
like
you
know,
place
cells
at
RIT
cells,
but
it's
the
same.
Basic
problems
occur
with
neocortical,
stuff
and
I
got
this
room,
I
see
and
I
sit
down,
someplace
and
and
I
see.
There's
some
there's
some
things
in
this
room.
C
C
C
If
I
went
from
here
to
here,
but
I
know
this,
but
I've
never
moved
and
I
never
built
around
walked
around
this
room.
I've
talked
about
before
that.
You
know
you
could
you
could
imagine
yourself
looking
at
different
objects
at
different
positions
at
different
orientations
and
that
you
could
learn
to
represent
this
position
by
objects
at
different,
different
radial
metric
locations
relative
to
that
this
is
a
way
of
learning.
This
point
is
in
the
same
way:
we've
learned
objects
in
the
XYZ
coordinates
of
the
frame.
This
is
the
way
of
learning
this
point
by
there's.
C
A
integration
component
of
this
I
can
think
if
I
come
over
here
and
come
back
I'm
at
the
same
point
again,
and
that
I
could
learn
that
these
are
objects
at
different
points
in
this
radial
space,
and
so
I
can
put
the
idea
that
I
could
learn
a
place
well
coded
for
this,
and
that
place
code
would
be
stable
over
changes
in
orientation.
So
I've
talked
about
that.
So
that's
like
that's
how
you
might
learn
a
place
else.
C
You
could
just
look
around
and
observe
things
at
different
positions
here,
whether
it's
in
three
dimension
or
not-
and
you
know
that
point
so
that's
point
of
this
problem-
it
seems,
but
he
didn't
explain
how
it
is
that
I
can
now
come
into
the
room
at
a
different
point
and
I.
Don't
I
curse
it
I'm
in
a
different
position
and
I
would
have
a
different
sort
of
object.
Radial
objects
around
me
now,
but
but
I
know
I'm
in
the
same
room
and
I
actually
know
my
location
so
somehow
way.
C
And
what
tells
me
what
this
tells
me
is
that
I'm
able
to
learn
the
complete
structure
of
the,
including
the
the
two-dimensional
grid
cell
structure
of
this
room
by
just
sitting
in
one
spot
and
radially,
observing
it
and
so
I,
don't
actually
need
to
move
around
the
room.
In
this
case
I
can
just
I
can
learn
the
structure
we
already
talked
about.
Oh
I
have
to
learn
them
with
my
finger.
C
I
have
to
move
my
finger
of
the
objects
to
learn,
or
this
is
true,
because
I
can't
observe
the
object
with
my
finger
from
a
distance,
but
with
vision.
I
can
we
talked
about
the
stick
problem
and
I
can
learn
things
with
their
hands?
Oh,
so
we
wait.
We
first
had
this
notion
that
hey,
if
you
just
think
about
the
finger
I
have
to
move
around.
If
this,
if
I
was
think
about
fingers
of
coffee
cups,
I
have
to
be
moving
around
this
object
of
tips.
C
I've
heard
all
these
features,
but
with
vision,
vision,
I,
don't
need
to
do
that.
I
can
sit
here
in
and
just
sitting
here.
I
can
build
the
complete
model
of
this
room.
The
you
know
the
two
dimensional
groups
I'll
model,
this
room
and
I-
might
conclude
that,
because,
when
I
come
into
the
room
from
a
different
position,
I
wrote
the
same
model.
The
room
I
actually
know
my
location
room,
so
the
reference
thing
that
I
used
here
is
the
same
as
a
reference
frame
over
here
and
talk
about
the
radio
reference
frame.
C
Right,
so
this
is
a
puzzling
thing
and
and
what
it
just,
but
it
initially
brings
up
this
very
interesting
question
how
it
is
that
I
can
learn.
I
can
go
to
this
position
to
learning
what
we
think
of
as
the
green
cell
model
of
us
of
an
object
or
a
grid
cell
model
of
us
of
a
space
that
you
know.
We
we
struggled
in
the
past
saying.
C
Just
learn
the
structure
of
something
by
just
really
observing
it,
and
then
that
learns
the
grid
cell,
XY
and
z
XY
model
of
the
world.
So
somehow
this
has
to
occur.
Marcus
I,
don't
know
if
you've
thought
about
this
or
not
yeah,
but
it's
related
things.
We've
talked
on
the
past,
I
figure.
You
probably
have
thought
about
it
and
I
want
to
give
your
observation
as
a
moment,
but
it's
almost
like
saying
we
have
to
go
between
this
sort
of
body.
C
Centric
framework
like
this
is
that
this
distance
from
my
body-
and
this
is
that
this
distance
from
body-
and
this
is
the
distance
of
my
body-
this
is
this
distance,
an
orientation
from
my
body
and
go
to
that
being
that
this
is
that
this
location,
relative,
the
room
and
that's
at
this
location
remember
this
location
was
good
and
you
have
to
do
the
orientation
student
you'd
like
to
observe
the
orientation
of
the
object
melts
in
my
body.
Then
I
have
to
translate
that
to
the
orientation
of
the
object
relative
to
the
room.
A
C
A
So
the
way
I've
approached
this
in
the
past
is
that
you
have
here
a
B,
C
and
D
are
like
child
objects.
And
yes,
and
when
you,
when
you're,
when
you're
right
here,
you
see
you
look
we'll
just
go
in
order,
you
see
a
what
you
want
to
represent
is
like
you
have
a
hero.
It
seems
like
down
twice
to
signifies
that
this.
C
A
A
B
C
A
C
But
you
flip
that
around
and
say
what's
this
orientation
to
me
yeah
if
I,
don't
think
of
as
a
we're
calm
again,
it's
like.
Oh
here's.
Here's
where
two
ways
to
look
at
this!
This
is
where
I
am
relative.
The
object
yeah
in
my
orientations
on
or
it
is
where
the
object
is
drawn
to
me
in
its
orientation.
C
A
So
so,
where
I'm
going
with
this
is
either
and
different
cell
populations
are
either
order
simultaneously,
you
want
to
say:
okay,
now,
here's
where
I
am
and
my
and
my
location,
space
of
seed,
yeah
and
and
at
this
point
you
will
have
turned
so
you
look
here.
Suppose
you
are
able
to
see
that
that
will
also
turn
this
that
this
will
also
turn.
Then
you'll
be
looking
that
way.
Yeah.
C
A
A
A
C
Historic,
okay,
so
it
could
be
a
transitionary
displacement.
We've
talked
about
many
times
like
I
could
have
the
displacement
this
way
to
a
yeah
and
I
switch
it
to
this
way
to
be
in
that,
and
if
we
do
the
displacement
on
the
moment,
the
moment
transition
I
got
what
I
need,
so
that
wouldn't
mean
I
wouldn't
have
to
keep
this
around
okay.
C
C
C
C
B
C
Yeah,
because
there
is,
you
know
it's
hard
for
me
to
think
of
it.
That
way,
it's
easy
for
me
to
think
about
it.
Who
am
I
again
going
back
flip
again,
where
I
know,
you've
got
it
in
your
head.
One
way
trouble
was
think
of
it.
That
way,
if
I
think
about
it,
even
though
they're
equivalent,
if
I,
think
about
it
more
like
okay,
where
are
where
astrologist
many
words
this
relative
to
me,
then
I
can
easily
imagine
building
up.
C
It's
like
all
the
words
make
sense,
but
you're
saying
by
doing
this
by
including,
like
okay
I'm
going
the
displacements,
the
displacements
portunity
two
objects
placements
in
these
two
objects.
I
can
learn
with
this
place
between
these
two
objects.
Yeah.
So
that's
what
you're
saying
right
yeah,
but
it's
it's
not
what's
the
actual
step
to
do
that,
I've
got!
C
You
know
this
displacement
is
between
here
here
and
then
here
here
and
you're,
saying
it
doesn't
matter
I'm
going
to
end
up
with
this
display
the
same
doesn't
matter
where
I
am
they
made
this
observation
rather
I,
don't
I
like
that?
I
don't
see
the
mechanism
for
it.
What
is
the
mechanism
would
make
that
displacement?
The
same.
Is
it
just
naturally
come
out
of
this
basement,
so
it.
A
C
C
C
A
A
Okay,
what
this
requires
is
that
you
need
to
be
able
to.
You
want
me
to
room
you
pop
yourself
down
here.
You
look
at
a
and
you
need
to
be
able
to
recognize
where
you
are
in
a
is
location
space
you
need,
and
that
requires
this.
This
whole
this'll
way
of
approach
of
approaching
this
is
child
objects
and
parent
objects,
like
the
kind
of
fundamental
approaches
like
you've
learned,
the
child
objects
really
well
and
then
you're,
then
then
you're
set
then
you're,
then
you're
really
good
at
then
you
can
handle
parent
objects
or.
C
C
B
B
C
At
the
same
time,
too
right
this
has
to
stay
active,
cuz,
I'm,
building
up
this,
this
voyager
object
and
I
have
to
anchored
it
and
continue
to
and
somehow
I
have
to
transfer
this
knowledge
here
to
that,
it's
not
on
the
stand
on
last
time,
so
you
know
I'm
wondering
if
this
can
all
be
done
in
one
column
or
if
it
requires.
You
know
the
other
way
of
looking
has
you've
got.
C
I,
don't
see
six
a
could
be
both
of
these
things.
It
couldn't
be
grid
cells
and
orientation
cells.
It's
it's
more
like
saying.
You
know.
I've
argued
that
if
I
think
about
orientation
has
its
own
metric
space,
then
the
orientation
actual
representation
of
orientation
would
be
unique
to
the
object
in
the
location
in
the
world.
It's
like
just
like
grid
cells
are
unique,
and
so,
if
I
think
about
this
is
orientation,
then
that's
not.
C
Orientation
module
it's
a
whole
bunch
of
learning,
takes
two
modules
and
therefore
it
is
it's
an
unique
it's
unique
to
all
the
world
and
therefore
I
can
I
can
predict
exactly
what
my
input
should
be
because
I
know
my
location
and
orientation.
That's
it's
one
set
of
cells
that
do
this!
There's
only
it's
it's!
You
know
it's
a
bunch
of
modules
here,
let's
say,
and
we
think
of
orientation
modules,
but
together,
they're,
unique,
so.
C
Cells
specifies
the
location,
Melser
some
object,
but
the
location
is
unique
to
the
entire
world,
in
this
case,
I'm
spire
specifying
an
orientation
relative
to
the
object,
but
it's
also
unique
to
the
location
or
locations
in
the
world.
So,
if
I'm
going
to
predict
whether
my
inputs,
gonna
be
I,
have
that
prediction
can
only
occur
at
a
specific
location
at
a
specific
orientation
in
that
specific
location,
it
has
to
do
both
of
those.
So
if.
C
This
state
of
cells
here
I
can
particularly
exactly
what
you
should
put
your
input,
your
they.
This
is
why
we
never
really
predicted
orientations,
isn't
really
work.
It's
that
you
can't
predict
the
end
of
what
your
input
is
going
to
be
just
on
your
orientation
knowledge
of
the
object.
You
have
to
know
where
you
are
in
the
object
in
your
orientation
to
it.
Then
you
can
fit
into
if
I.
If
I
saw
two
sets
of
cells
here,
both
projecting
to
lay
at
four
and
I
said:
okay,
one's
orientation,
one's
location,
maybe
I'll!
C
You
get
away
with
that.
Although
I
don't
like
that,
because
it's
always
hard
for
this
layer
cells
to
know
what
to
look
at
so
this
more
copacetic
example
is
not
to
do
that.
Four
percent,
of
example,
is
just
to
say:
no
I
got
one
set
of
cells
here,
six
a
and
if
it's
going
to
present
it's
good,
if
it's
going
to
be
predictive
of
what's
going
on
any
therefore
then
disruptive
needs
to
represent
a
specific
location
in
the
world.
C
That's
also
specific
to
the
object,
because
it's
you
know
and
as
and
it's
going
to
represent
the
orientation
at
that
particular
point
in
the
world,
and
now
we
welcome
abouts
with
a
bunch
of
small.
You
know
even
single
dimension
orientation
molecules
you
just
they
just
anchor
differently,
just
like
we
think
Ritchie
language
Utley.
So
that's
completely
consistent
with
how.
C
I
mean
orientation,
cells,
I,
hate,
direct
insult
work.
Do
they
look?
They
look
like
individual
education
cells,
like
oh
yeah,
I,
know
I'm
in
this
direction
and
this
object.
But
if
you
look
at
a
set
of
them,
they
would
all
be
anchored
differently.
They'd
all
be
changing
together,
but
they'd
all
be
done
differently,
and
so
you
would
have
known
we're
seeing
this
unless
you
looked
at
the
population,
how
you
know
the.
C
It'd
be
probably
to
meet
Tanya
struggle.
True,
you
have
to
look
at
a
bunch
of
in
grid
Direction
cells,
and
you
say:
oh
this
one
represents
this
way
and
this
way,
mr.
members,
in
this
way
in
a
different
object,
they
have
a
different
relationship
to
each
other
right.
Another
object.
This
one
could
be
this
way
and
this
one
this
one
could
be
that
way,
and
this
one
could
be
this
way.
They
all
look
like
hedgehogs
themselves
in
any
environment,
they're
all
consistently
doing
the
right
thing,
but
cinema
would
be
unique.
C
C
And
it
has
to
change
obviously,
as
I
move,
it's
weird.
It's
almost
like
I'd,
be
saying
individual
modules
here
wouldn't
be
changing
just
my
orientation
if
I'm
here,
I
just
move
this,
what
am
I
now
and
I'm
still
facing
same
direction
with
almost
a
second
of
these
cells
already
updating,
but
somehow
they
have
to
reflect
it.
It
doesn't
make
sense
to
me
anyway,
so
this
is
sort
of.
C
C
A
A
A
A
So,
like
there's
a
wonder
one
correspondence
between
these
yeah.
However,
if
you
look
at
these
in
isolation,
if
you
look
at
just
this
in
isolation
versus
just
this
in
isolation,
their
properties
are
different.
How
so
well
right
now,
like
your
location
relative
to
this
marker,
it's
changing
like
crazy
when
I
rotate.
What.
C
A
C
Some
fancy
forked,
it
does
seem
much
more.
It
does
feel
like
it.
I
told
you
even
just
feels
to
me
like
when
I
observed,
I
sit
in
the
room
here
and
done
this
sitting
in
the
room
and
look
around
with
different
objects.
It
feels
to
me
is
I'm
noticing
where
these
things
are
relative
to
me.
It
just
feels
I,
don't
feel
like
these
I'm
flipping
around
these
objects.
You
know.
C
Cells
have
to
change
based
on
a
movement
right.
That's
the
whole
idea
of
the
grid
cells
have
to
change
on
the
move
McMahon,
if
I
think
about.
If
I
am,
if
I'm
asking
where
something
is
relative
to
me.
I
have
that
movement
command.
The
movie
commands
not
turning
my
head
and,
and
that
is
easy.
I
turn
my
head
in
distinct
changes.
If
I'm
turning
my
head
and
I.
C
In
this
world,
in
this
world,
anyway,
I
think
you
know
we've
written
about
the
way
ourselves
aware
columns
as
being
object
solitude
to
me
and
the
Chloe
that's
happening.
We've
we've
talked
about
it
in
terms
of
its
in
the
paper
and
the
frameworks
paper.
I
wrote
about
it
in
terms
of
movement
right
I
like
no
I,
wanted
my
hand
from
one
place
in
my
body.
It's
relative,
my
body
to
another
place
the
reason
they
didn't
write
about
in
terms
of
objects.
It's
because
I
didn't
have
an
understood
that
wasn't.
C
I
wasn't
I
had
no
explanation
to
say.
Well,
what
is
he?
You
know?
What's
the
equivalent
an
object
in
a
body
space?
This
explanation
here
is
an
example
of
an
object
body
space.
It's
it's!
It's
like
I
brought
us
up
to
my
body.
There's
an
object
right
now,
anyway,
I'm
just
toying
on
this
thing.
I
guess.
C
It
strikes
me
as
I
hear
their
explanation
for
it,
but
it's
and
it
doesn't
even
deal
with
the
orientation
part
yet
so
we
don't
full
solution,
but
it
feels
to
me
that
this
there
is
a
simple
solution
to
this
there's
a
simple
way
of
saying:
how
did
I
you
know
using
the
neural
tissue,
we
see
how
it
is.
I
have
all
these
issues
we
talk
about.
How
does
I
just
do
this
very
rapidly
and
maybe
you
feel
comfortable
in
God
already
I
just
peeled
tomato
I
have.
C
D
C
C
C
Let
me
just
finish
my
thought:
I,
don't
think
that
I
now
got
this
idea
that
we
encode
distance
partially
by
these
oscillation
frequency
changes
in
the
Islamic
cortical
loop
I.
Think
there's
a
lot
to
that
idea,
and
so,
if
I'm,
if
I'm,
observing
this,
this
object
from
a
distance
and
I'm
and
I'm
gonna,
say
I'm
moving
my
eyes
over
the
object,
left
and
right
and
as
I'm
looking
at
this
draw
right
few
little
features.
That's
how
I
think
about
it.
The
radial.
C
Is
small
this
last
one
an
object
for
the
way
and
it's
large
of
an
object
is
close
and,
and
so
you
know
for
me-
I
threw
you
think
that
I
even
know
how
to
do.
That
requires
that
the
that
you
you're
going
to
you're
going
to
take
your
movement
command
and
scale
up.
So
the
ideas
that
have
I
have
an
expectation.
C
How
far
have
to
move
to
get
from
one
side
of
this
object
to
the
other
and
the
further
away
is
I
have
to
scale
down
my
movements
of
the
same
movement
vector
results
in
a
smaller
change
and
that
it's
going
to
you
know
I
think
that's
going
on
between
the
Islamic
oscillation
change
to
do
it.
So
my
point
is
that
even
just
observing
this
thing
out
here
and
knowing
where
it
is
I
have
to
I,
have
an
encoding.
I
Kaleo
gonna
have
an
encoding
of
that
distance
home
at
the
moment
as
I.
C
C
Do
that
so
that's
another
part
of
the
solution.
I
think
we
have
to
come
up
with
it's
just
somehow
that
I
have
these
orientations
of
things
and
I'm
encoding
the
distances
using
this
and
that's
gonna,
be
part
of
the
neural
I'm
working
on
the
idea
that
that's
going
to
be
part
of
the
neural
mechanism.
It's
one
way
to
think
about.
C
C
Is
that,
therefore,
I
can
build
the
know
where
to
place
it
in
the
room?
I
don't
know
anyway,
I
said
I'm
not
working
on
this
all
the
time.
Just
writing
so
much,
but
I
think
this
is
the
problem.
I
really
feel
this
like
the
core
problem.
This
is
a
some
often
is
that
you
wanted
to
find
us
the
right
problem
to
solve.
This
feels
to
me
like
this
is
for
me
the
language
it
works
to
solve
this
problem,
that
this
is
a
I
can't
if
I
can
put
my
finger
on
this
problem.
C
Okay,
how
do
I
know
this
is
going
on,
and
then
we
have
to
be
able
to
make
everything
work.
You
know
everything
the
simple
problem,
the
state
and
also
it
also
by
the
way
it
gives
you
an
interesting
pause.
It's
the
same.
What's
a
place
cell
I
argue
that
play
cells
are
are
just
a
way
of
encoding.
You
could
think
of
a
place
out.
There
is
encoding.
This
radio
object
around
you
so,
but
but
then
I've.
C
C
It
sort
of
like,
as
I
move
over
here,
I
place
of
the
purely
determined
by
my
sensory
inputs?
I
could
I
couldn't
know
that
in
advance
in
order
to
populate
the
model
just
anyway,
I
just
I
just
want
to
if
I
want
to
keep
I'm
gonna
keep
bringing
this
up
over
and
over
again,
because
I
think
we
have
to
solve
this
problem
and
I
think
that's
the
key,
the
key
to
the
research
that,
on
the
neuroscience
out
we
have
to
do
this
year.