►
From YouTube: Numenta Research Meeting, April 1, 2019
Description
Numenta Research Meeting - neuroscience / artificial intelligence (AI) / neocortex oscillations https://gist.github.com/rhyolight/59dcd4f5810a00b001697abd70452411 -- Watch live at https://www.twitch.tv/rhyolight_
A
A
B
C
So
it's
gonna
be
a
short
meeting
because
subitize
not
here
and
the
topic
I
have-
is
not
very
interesting.
Well
interested.
So
when
when
floor
is
here
on
Thursday,
we
reviewed
this
paper.
The
working
memory,
2.0
paper
and
I
was
finishing
his
topic,
and
but
it
was
built
around
these
camera
frequency
members
and
other
duties
and
the
gamma
frequencies
for
refreshing
this
temporary
memory
and
then,
but
when
reading
this
paper,
I
found
a
whole
bunch
of
what
look
like
very
interesting
references
to
other
oscillatory
things
and
so
there's
a
whole.
C
You
know
because
in
my
mind
the
oscillations
have
always
been
this
sort
of,
like
weird
people
studied
them
for
like
going
back.
You
know
40
years
ago
because
they
found
these
oscillations.
Just
using
me,
you
know
scalp
recordings,
you
know,
but
it
was
never
like
what
are
they
for
and
so
there's
always
been
this
one
hypothesis
that
they
could
be
just
there
to
synchronize
spikes
so
that
neuron
work
better,
but
maybe
they
do
something
else
too.
So
here
he
was
with
Lauren
came
in
with
a
very
concrete
example
of
doing
something.
C
Is
using
this,
the
gamma
frequencies
for
recording
short-term
memories,
I
thought
it
very
compelling,
and
then
there
was
all
these
other
papers
and
became
clear
that
they've
been
a
lot
of
research
done
on
oscillations
in
the
last
10-15
years,
and
I
hadn't
really
looked
at
any
of
that
stuff,
and
so
they
looked
interesting.
The
title,
which
is
really
interesting,
so
I,
went
and
printed
a
bunch
of
those
papers
and
I
read
them
this
weekend.
I
read
several
of
them.
C
I
was
a
little
disappointed
over
all
that
is,
they
sounded
really
primer
trained
in
the
ranging,
but
not
as
interesting
as
I
thought
they
would
be,
and
so
here's
here's
one.
This
is
a
2010
paper
in
neurobiology
dynamics
of
active
sensing,
perceptual
selection
and
even
here
in
the
introduction
they
talked
about
active
sensing,
which
is
you
know,
moving
and
there's
going
so
much
role
of
frequencies
and
oscillations
in
active
sensing,
but
they
say.
Oh
it's.
C
Obviously
that's
dissenting
in,
like
the
things
like
the
bat
you
know
which
sends
out
its
ping,
but
you
know
inconsistent
contrast.
Biological
sensors,
like
the
eyes
and
fingertips,
have
traditionally
been
thought
of
as
passive
sensors
and
I'm,
like
really
even
2010.
People
are
thinking,
like
you
know,
eyes
and
fingers
its
passive.
So
it's
like
okay,
so
they're,
just
like
moving
beyond
that
they're
trying
to
show
no.
Those
are
really
active.
Sensors.
Okay,
this
paper
here
this
one's
called
the
sensual
cycles
from
cell
press.
This
was
2016
paper
by
roofing
ban
ruling.
C
It
was
a
review
which
is
always
nice
and
in
this
paper
is
called
perceptual
cycles.
You
get
the
first
to
get
to
the
city
I'll
just
share
with
you
guys
took
this
small.
You
memory
come
around.
It's
like
this
figure
here
figure.
One
he's
basically,
there's
two
sort
of
basic
ideas
of
people
are
thinking
about.
One
is
on
the
left,
which
is
a
rhythmic
perception,
meaning
like
you're,
representing
something,
but
at
the
peak
of
the
oscillations.
C
More
neurons
are
firing
than
other
times
and
therefore
synchronize
with
them,
but
there's
not
really
doing
anything
else
other
than
that.
That's
just
like
the
plumbing
idea,
like
we're
just
trying
to
get
the
cells
apart.
At
the
same
time,
the
author's
idea
is
actually
no
that
it's
not
like.
Those
neurons
are
just
being
getting
the
force,
there's
actually
period
of
time
when
one
one
thing
is
being
done,
improve
the
time
where
something
else
might
be
done.
It's
like.
C
C
C
This
is
this
paper
here.
This
is
a
2010
paper,
Bushman
and
Miller.
Miller
was
on
that
working
memory
paper
and
it's
called
shifting
the
spotlight
attention
to
evidence
for
discrete
computations
and
cognition.
So
this
is
like
going
beyond
the
one.
I
just
talked
there
with
just
saying,
like
maybe
during
that
during
attention
that
the
tension
is
tied
to
these
different
oscillations
and
there
might
be
different
computations
going
on
at
the
points
and
the
attentional
cycle,
and
one
of
the
ideas
there
is
that
is,
is
like.
Oh,
what
I'm
attending
to
something
the
attentional
aspect.
C
That's
happened
in
one
part
of
the
cycle
and
then
maybe
I'm
doing
something
that's
not
attended.
On
the
other
aspect.
The
cycle
which
we
talked
about
is
maybe
a
really
cool
idea
for
how
we
could
do
a
displacement
that
you
might
actually
be
alternating
between
an
intended
object
and
an
unattended
object.
Intended
object,
an
unattended
object
from
every
single
cycle,
and
therefore
you
could
you
could
calculate
you
know.
Over
and
over
again
you
can
be
calculating
the
displacements
as
you're
doing
that
and.
C
They're
arguing
that
you
can
hold
to
multiple
objects
in
memory.
At
the
same
time,
the
tended
objects,
an
unattended
object.
So
I'm
not
gonna,
go
through
all
the
details
of
this,
but
in
some
sense,
they're
arguing
for
that.
What
we
found
in
the
hippocampus
with
the
pre-session,
something
equivalent
would
be
likely
happening
in
the
cortex
and
but.
A
C
They
guess
they
don't
get
too
far.
I
mean
is
what
purpose
might
be
discrete
computation
serve?
You
know
we
rename.
The
kind
of
you
know
sock.
It's
not
like
a
very
well
form.
Theory,
I'm,
leavin,
I
added
more
to
it
than
they
say
here.
It's
basically
well,
you
could
be
doing
multiple
things
at
the
same
time
and
they
don't
speculate
what
those
might
be,
and
they
may
also
point
out
here
that,
although
the
proposing
this
is
common
throughout
the
New,
York
Court
jacket,
so
you
don't,
they
should
also
say
it
doesn't
require
all
computations.
C
Example
they
gave
it's
difficult
to
follow.
It
would
require
much
closer
reading
and
but
I
took
I
just
accepted
their
conclusion.
There's
a
conclusion
is
that
we
have
these
discrete
time
frames
that
we
can
calculate
different
things
at
different
points
of
time
and
I
thought
that
that
is
a
whole
new
perspective
that
we've
kind
of
been
working
away
there
by
accepting
the
procession
in
the
hippocampus
and
now
there's
a
bunch
of
people
arguing
the
same
thing
happening
throughout
the
neocortex,
even
if
they
don't
have
to
put
theories
for.
C
C
Here's
one
sentence:
it
says
this
is
a
this
is
by
Viking
coin.
I
know
mining
these
people
because
it's
a
different
field,
dynamic
interplay
with
the
frontal
parietal
Network
underlies
rhythmic
spatial
attention,
says
converging.
Evidence
indicates
environmental
sampling,
whether
through
covert
or
over
mechanisms,
as
a
fundamental
rhythmic
process,
with
a
sampling
rate
and
the
theta
band.
So.
C
These
these
cycles
are
occurring
between
different
parts
of
the
neocortex
and
that
attention
is
occurring
on
the
cycles
they
have
these
bright.
They
have
these
very
specific
attentional
paradigms
that
monkeys
can
do
just
a
that.
They
test
for
and
it's
pretty
damn
complicated,
but
in
the
end,
you're,
basically
just
trying
to
show
that
many
of
these
operations
are
tied
to
rhythm,
that
the
secod
initiate
at
a
certain
point,
in
a
rhythmic
cycle
and
in
a
whole
bunch
of
things
that
are
tied
to
these
rhythms,
but
neither
what
a
little
disappointed
was.
C
C
This
stuff
is
happening,
and
these
are
active
senses.
There's
these
cycles
going
on
the
tension
is
tied
to
sizable
movements
are
tied
to
these
cycles.
Maybe
there's
multiple
things
we
represent
at
the
same
time
during
these
cycles.
This
is
happening
in
in
your
project
and
all
the
much
more
detailed
parts
of
the
paper
seem
to
be
not
illuminating,
deeper
insights
but
more
sort
of
supporting
those
conclusions.
So
I
think
it's
it's
interesting
and
it
opens
up
the
whole
a
series
of
possibilities
for
us
and
there's
a
ton
of
more
literature
out
there.
Everyone
buried
them.
C
A
C
A
So
general
question
I
have
on
oscillations
and
I'm,
not
sure
if
this
is
known
that
mostly
I've
mostly
only
write
about
oscillations
like
by
reading
other
things,
that
mention
them
and
then
I
read
a
little
bit,
but
I
haven't
dived
headfirst
into
them.
So
what
confuses
me
with
them?
A
little
bit
is:
are
they
a
cause
or
are
they
an
effect,
because
one
way
you
could
get
an
oscillation
in
local
field
potentials
or
you
a
noticeable
oscillation
is
suppose
there
is
a
large
population
of
cells.
A
That's
firing
at
40
Hertz,
like
suppose,
suppose,
a
working
memory
is
held
by
at
certain
times.
At
certain
times
periodically
there
are
cells
firing
at
40
Hertz
that
could
cause
the
oscillation
that
you
detect
I
mean
the.
If
you
well
I
mean
you're
just
measuring
local
field
potential,
yeah
I'm.
Definitely
a
large
population
of
cells
start
spiking,
it
40
Hertz,
but
why
do
they
own
it?
The.
C
A
Don't
know
it
okay,
explanation
a
so
when
you're
holding
in
a
memory
or
when
you
need
to
refresh
our
memory
Allah
to
cause
it
to
stay.
You
know
when
you
need
to
cause
a
bunch
of
cells
to
burst
in
order
to
preserve
a
memory,
your
brain,
somehow
Ram
cells,
gamma
yeah.
It
causes
basically
every
cell
to
be
receiving
kind
of
the
oscillating
input
that
causes
every
cell
to
fire
more
often
yeah
and
that's
what
gamma
is
something
that
the
brain
can
ramp
up
and
down.
Yeah
explanation
be:
when
you
some
neural
mechanism
exist,
they.
C
C
A
C
Think
the
attitude
I
mean
interesting
question:
the
attitude
in
these
papers-
I
it
comes
across
as
they're
all
thinking
the
former.
It's
like
the
oscillations
are
a
means
by
which
the
brain
decides
I'm,
going
to
return
to
something.
I
now
establish
a
an
oscillatory
frequency
or
I'm
going
to
these
two
regions
and
your
cortex
are
come
talk
to
each
other
and
that's
the
way
they
describe
it.
C
It's
interesting
question:
maybe
it's
more
of
an
epoch
bomb
and
that
comes
back
for
other
reasons,
I
think
that's
what
you're
saying
yeah
yeah
I
think
they
thing
for
me,
though
you
know
all
is
I
could,
from
an
information
point
of
view,
it
was
never
clear
to
me.
Oscillations
were
essential
if
they're,
really
just
providing
this
sort
of
synchrony
I
mean
you
know.
Okay,
we
need
to
get
the
neurons
fire
together,
that's
part
of
plumbing,
but
the
fact
that
they
right
now
we've
got
these
multiple
things
going
on
in
the
same
phase.
C
That's
a
very
interesting
computational
problem
issue
that
introduces
today.
We've
probably
we
were
gonna
work,
the
way
they
are
ready.
We
in
our
sequence
memory
boards
talked
about
hey.
You
know
that
we
could
be
doing
and
inferring
on
the
two
cycles
to
do
phases
and
that
would
solve
a
lot
of
problems.
Oh
so
that
was
like.
That
was
a
big
insight
and
now
thinking
about
displays
of
themselves.
C
This
could
really
play
a
big
role
in
that
kind
of
thing,
but
also
we
can
start
thinking
about
how
do
I
get
to
representations
alternating
back
and
forth
and
that
I
can
compute
between
them.
Things
like
that,
so
that
also
when
we
start
thinking
about
multiple
things
happening
on
the
same
octave
phases
is
like
wow.
That's
like
a
whole
other
information
processing,
realm
I,
don't
know
how
that
relates
to
your
question.
C
Somebody
talked
about
that.
You
know
there's
some
theories
and
they're
just
like.
Oh
okay,
if
you
have
an
image
or
a
cell
and
a
positive
and
excited
ourselves
and
hooked
up
a
certain
way
to
get
oscillation,
but
mostly
I,
think
these
papers
were
just
sort
of
trying
to
make
an
argument
that
that
they
are
occurring
in
the
neocortex
and
that
they're
there
they
mean
they're,
meaningful
and
related
to
things
that
we
know
that
animals.
C
Certainly
animals
can
do
yeah,
yeah,
there's
a
whole
slew
of
litter
dramas
and
again,
my
reading
of
it
today
and
this
weekend,
sort
of
implies
that
they
spent
a
lot
of
time.
Reading
and
I
don't
want
to
get
the
sliding
for
mental
knowledge.
Young
boy,
if
I'm
reading
a
person
like
the
August
last
year,
and
it's
still
talking
about
sort
of
basics-
it's
like!
Oh,
it's
moving
into
their
projects,
but
it's
not
really
super
fun.
Super
well
understood
yet,
and.
A
I'm
guessing
a
lot
of
this
is
just
the
fact
that
our
measurement
tools
we
have
oscillations
are
really
convenient
thing
to
measure
you
just
like
you're,
using
EEG
or
any
of
these
types
of
these
types
of
recording
tools.
And
then
you
just
what
perform
a
Fourier
decomposition
and
like
there.
You
have
the
signal,
so
it's
kind
of
an
easy
thing
to
measure,
and
that
would
cause.
C
What
they're
trying
to
do
now,
though,
to
understand
all
this
other
stuff,
so
you
have
to
look
at
single
cell
or
multi
cell
report.
It
yeah,
because
if
you
look,
it
didn't
phase
it
like.
What's
the
population
code
at
this
point
what's
population
at
this
point,
so
all
of
them,
these
studies,
I
think
I.
Think
I
should
go
back.
Nobody
all
over
or
most
some
were
like
monkey
macaque
multi
in
a
recording
and
maybe
similar
as
human
I
can
multi
unit.
C
C
C
B
C
C
A
B
Though
I
was
trying
to
get
off
thanks,
okay,
close
all
right
there,
you
go
so
that's
that's
a
research
meeting
that
was
pretty
quick,
they're,
usually
a
little
bit
longer
than
that.
But
it
really
depends.
It's
all
you
guys
can
hear
me.
Okay,
it
really
depends
on
who's,
got
topics
and
who
wants
talk
about
what?
Sometimes
these
things
will
go
on
two
hours,
they'll
come
back
after
lunch
and
continue
it
or
it'll
go
into
lunch,
so
it
really
just
depends.
So
this
was
just
like
a
one
sample
and
to
them
into
research
meetings
and
I'll.
B
B
So
thanks
for
joining
it,
does
anybody
have
any
questions
about
what
you
just
saw
I'm
happy
to
stick
around
and
answer
questions
is
just.
What's
spell
it
right?
If
this
helps
anybody,
we
just
are
done
with
the
meeting.
Now
it
was,
they
were
talking
about
oscillations
in
neocortex.
There
wasn't
a
whole
lot
of
mention
of
thalamus
six.
It
was
a
it
was
sort
of
assumed
and
I
thought.
It
was
interesting.
The
idea
of
our
oscillations
cause
or
effect
I
think
they
have
to
be
both
right.
B
The
oscillations
have
to
be
caused
by
something,
and
then
they
probably
cause
other
things.
I
mean
that's
it's
a
part
of
the
compute
process
in
the
brain,
so
so
yeah
I,
think
I.
Think
the
general
idea
is
that
gamma
oscillations
in
the
neocortex
are
are
suspected
to
be
caused
by
the
thalamus.
Although
I
mean
there's
still
a
lot
to
be
learned,
the
hard
thing
is
experimenting.
B
B
I
got
some
followers.
If
you
guys
want
to
read
somebody:
let's
do
it.
Let's
see
who's
who's
streaming
right
now,
who
haven't
I
raided,
yet
this
code
enjoy
guy.
That
was
this.
Is
it's
this
guy's
birthday
we
could
read
it.
Let's
do
code
enjoy.
I
watch.
I
watched
him
a
little
bit
the
other
day.
Oh
he's
playing
a
game:
I'm,
not
I.
Don't
usually
raid
games
who's,
this
guy
CG
CG
games,
oh
yeah
I
was
watching
him
earlier
I
enjoyed
that
I
say:
let's
trade
him!