A mouse. A laser beam. A manipulated memory | Steve Ramirez and Xu Liu | TEDxBoston

Can we edit the content of our memories? It’s a sci-fi-tinged question that Steve Ramirez and Xu Liu are asking in their lab at MIT. Essentially, the pair shoot a laser beam into the brain of a living mouse to activate and manipulate its memory. In this unexpectedly amusing talk they share not only how, but -- more importantly -- why they do this.

Steve Ramirez:
My first year of grad school,

I found myself in my bedroom

eating lots of Ben & Jerry's

watching some trashy TV

and maybe, maybe listening
to Taylor Swift.

I had just gone through a breakup.

(Laughter)

So for the longest time, all I would do

is recall the memory of this
person over and over again,

wishing that I could get
rid of that gut-wrenching,

visceral "blah" feeling.

Now, as it turns out,
I'm a neuroscientist,

so I knew that the memory of that person

and the awful, emotional undertones
that color in that memory,

are largely mediated
by separate brain systems.

And so I thought, what if we could
go into the brain

and edit out that nauseating feeling

but while keeping the memory
of that person intact?

Then I realized, maybe
that's a little bit lofty for now.

So what if we could start off
by going into the brain

and just finding a single
memory to begin with?

Could we jump-start
that memory back to life,

maybe even play with the contents
of that memory?

All that said, there is one person
in the entire world right now

that I really hope is not
watching this talk.

(Laughter)

So, there is a catch.
There is a catch.

These ideas probably remind
you of "Total Recall,"

"Eternal Sunshine of the Spotless Mind,"

or of "Inception."

But the movie stars that we work with

are the celebrities of the lab.

Xu Liu: Test mice.

(Laughter)

As neuroscientists,
we work in the lab with mice

trying to understand how memory works.

And today, we hope
to convince you that now

we are actually able to activate
a memory in the brain

at the speed of light.

To do this, there's only two simple
steps to follow.

First, you find and label
a memory in the brain,

and then you activate it with a switch.

As simple as that.

(Laughter)

SR: Are you convinced?

So, turns out finding a memory
in the brain isn't all that easy.

XL: Indeed. This is way more
difficult than, let's say,

finding a needle in a haystack,

because at least, you know,
the needle is still something

you can physically put your fingers on.

But memory is not.

And also, there's way
more cells in your brain

than the number of straws
in a typical haystack.

So yeah, this task does
seem to be daunting.

But luckily, we got help
from the brain itself.

It turned out that all we need
to do is basically

to let the brain form a memory,

and then the brain will tell us
which cells are involved

in that particular memory.

SR: So what was going on in my brain

while I was recalling the memory of an ex?

If you were to just completely
ignore human ethics for a second

and slice up my brain right now,

you would see that there
was an amazing number

of brain regions that were active
while recalling that memory.

Now one brain region
that would be robustly active

in particular is called the hippocampus,

which for decades
has been implicated in processing

the kinds of memories
that we hold near and dear,

which also makes it
an ideal target to go into

and to try and find and maybe
reactivate a memory.

XL: When you zoom in into the hippocampus,

of course you will see lots of cells,

but we are able to find
which cells are involved

in a particular memory,

because whenever a cell is active,

like when it's forming a memory,

it will also leave a footprint
that will later allow us to know

these cells are recently active.

SR: So the same way
that building lights at night

let you know that somebody's probably
working there at any given moment,

in a very real sense, there
are biological sensors

within a cell that are turned on

only when that cell was just working.

They're sort of biological
windows that light up

to let us know that that cell
was just active.

XL: So we clipped part of this sensor,

and attached that to a switch
to control the cells,

and we packed this switch
into an engineered virus

and injected that into the brain
of the mice.

So whenever a memory is being formed,

any active cells for that memory

will also have this switch installed.

SR: So here is what the hippocampus
looks like

after forming a fear memory, for example.

The sea of blue that you see here

are densely packed brain cells,

but the green brain cells,

the green brain cells
are the ones that are holding on

to a specific fear memory.

So you are looking at the crystallization

of the fleeting formation of fear.

You're actually looking
at the cross-section

of a memory right now.

XL: Now, for the switch
we have been talking about,

ideally, the switch
has to act really fast.

It shouldn't take minutes
or hours to work.

It should act at the speed
of the brain, in milliseconds.

SR: So what do you think, Xu?

Could we use, let's say,
pharmacological drugs

to activate or inactivate brain cells?

XL: Nah. Drugs are pretty messy.
They spread everywhere.

And also it takes them
forever to act on cells.

So it will not allow us
to control a memory in real time.

So Steve, how about let's zap
the brain with electricity?

SR: So electricity is pretty fast,

but we probably wouldn't
be able to target it

to just the specific cells
that hold onto a memory,

and we'd probably fry the brain.

XL: Oh. That's true.
So it looks like, hmm,

indeed we need to find a better way

to impact the brain at the speed of light.

SR: So it just so happens that light
travels at the speed of light.

So maybe we could activate
or inactivate memories

by just using light -

XL: That's pretty fast.

SR: - and because normally brain cells

don't respond to pulses of light,

so those that would respond
to pulses of light

are those that contain
a light-sensitive switch.

Now to do that, first we need
to trick brain cells

to respond to laser beams.

XL: Yep. You heard it right.

We are trying to shoot lasers
into the brain.

(Laughter)

SR: And the technique that lets us do that
is optogenetics.

Optogenetics gave us this
light switch that we can use

to turn brain cells on or off,

and the name of that switch
is channelrhodopsin,

seen here as these green dots
attached to this brain cell.

You can think of channelrhodopsin
as a sort of light-sensitive switch

that can be artificially
installed in brain cells

so that now we can use that switch

to activate or inactivate the brain
cell simply by clicking it,

and in this case we click it on
with pulses of light.

XL: So we attach this light-sensitive
switch of channelrhodopsin

to the sensor we've been talking about

and inject this into the brain.

So whenever a memory is being formed,

any active cell for that particular memory

will also have this light-sensitive switch
installed in it

so that we can control these cells

by the flipping of a laser
just like this one you see.

SR: So let's put all of this
to the test now.

What we can do is we can take our mice

and then we can put them in a box
that looks exactly like this one,

and then we can give them
a very mild foot shock

so that they form a fear
memory of this box.

They learn that something
bad happened here.

Now with our system,
the cells that are active

in the hippocampus
in the making of this memory,

only those cells will now
contain channelrhodopsin.

XL: When you are as small as a mouse,

it feels as if the whole
world is trying to get you.

So your best response of defense

is trying to be undetected.

Whenever a mouse is in fear,

it will show this very typical behavior

by staying at one corner of the box,

trying to not move any part of its body,

and this posture is called freezing.

So if a mouse remembers that something
bad happened in this box,

and when we put them
back into the same box,

it will basically show freezing

because it doesn't want to be detected

by any potential threats in this box.

SR: So you can think of freezing as,

you're walking down the street
minding your own business,

and then out of nowhere
you almost run into

an ex-girlfriend or ex-boyfriend,

and now those terrifying two seconds

where you start thinking,
"What do I do? Do I say hi?

Do I shake their hand?
Do I turn around and run away?

Do I sit here and pretend
like I don't exist?"

Those kinds of fleeting thoughts
that physically incapacitate you,

that temporarily give you
that deer-in-headlights look.

XL: However, if you put the mouse
in a completely different new box,

like the next one,

it will not be afraid of this box

because there's no reason that it
will be afraid of this new environment.

But what if we put
the mouse in this new box

but at the same time,
we activate the fear memory

using lasers just like we did before?

Are we going to bring back the fear memory

for the first box into this
completely new environment?

SR: All right,
and here's the million-dollar experiment.

Now to bring back to life
the memory of that day,

I remember that the Red Sox had just won,

it was a green spring day,

perfect for going up and down the river

and then maybe going to the North End

to get some cannolis, #justsaying.

Now Xu and I, on the other hand,

were in a completely windowless black room

not making any ocular movement
that even remotely resembles an eye blink

because our eyes were fixed
onto a computer screen.

We were looking at this mouse
here trying to activate a memory

for the first time using our technique.

XL: And this is what we saw.

When we first put the mouse into this box,

it's exploring, sniffing
around, walking around,

minding its own business,

because actually by nature,

mice are pretty curious animals.

They want to know,
what's going on in this new box?

It's interesting.

But the moment we turned on the laser,
like you see now,

all of a sudden the mouse
entered this freezing mode.

It stayed here and tried not to move
any part of its body.

Clearly it's freezing.

So indeed, it looks like
we are able to bring back

the fear memory for the first box

in this completely new environment.

While watching this, Steve and I

are as shocked as the mouse itself.

(Laughter)

So after the experiment,
the two of us just left the room

without saying anything.

After a kind of long,
awkward period of time,

Steve broke the silence.

SR: "Did that just work?"

XL: "Yes," I said. "Indeed it worked!"

We're really excited about this.

And then we published our findings

in the journal Nature.

Ever since the publication of our work,

we've been receiving numerous comments

from all over the Internet.

Maybe we can take a look at some of those.

SR: So the first thing
that you'll notice is that people have

really strong opinions
about this kind of work.

Now I happen to completely
agree with the optimism

of this first quote,

because on a scale
of zero to Morgan Freeman's voice,

it happens to be
one of the most evocative accolades

that I've heard come our way.

(Laughter)

But as you'll see, it's not
the only opinion that's out there.

XL: Indeed, if we take
a look at the second one,

I think we can all agree that it's, meh,

probably not as positive.

But this also reminds us that,

although we are still working with mice,

it's probably a good idea
to start thinking and discussing

about the possible ethical ramifications

of memory control.

SR: Now, in the spirit of the third quote,

we want to tell you about a recent project

that we've been working on in lab
that we've called Project Inception.

So we reasoned that
now that we can reactivate a memory,

what if we do so but then
begin to tinker with that memory?

Could we possibly even
turn it into a false memory?

XL: So all memory
is sophisticated and dynamic,

but if just for simplicity,
let's imagine memory

as a movie clip.

So far what we've told you
is basically we can control

this "play" button of the clip

so that we can play this video clip
any time, anywhere.

But is there a possibility
that we can actually get

inside the brain and edit this movie clip

so that we can make it
different from the original?

Yes we can.

Turned out that all we need
to do is basically

reactivate a memory
using lasers just like we did before,

but at the same time,
if we present new information

and allow this new information
to incorporate into this old memory,

this will change the memory.

It's sort of like making a remix tape.

SR: So how do we do this?

Rather than finding
a fear memory in the brain,

we can start by taking our animals,

and let's say we put them in a blue box
like this blue box here

and we find the brain cells
that represent that blue box

and we trick them to respond
to pulses of light

exactly like we had said before.

Now the next day, we can take
our animals and place them

in a red box that they've never
experienced before.

We can shoot light
into the brain to reactivate

the memory of the blue box.

So what would happen here if,

while the animal is recalling
the memory of the blue box,

we gave it a couple of mild foot shocks?

So here we're trying to artificially
make an association

between the memory of the blue box

and the foot shocks themselves.

We're just trying to connect the two.

So to test if we had done so,

we can take our animals once again

and place them back in the blue box.

Again, we had just reactivated
the memory of the blue box

while the animal got a couple
of mild foot shocks,

and now the animal suddenly freezes.

It's as though it's recalling being
mildly shocked in this environment

even though that never actually happened.

So it formed a false memory,

because it's falsely
fearing an environment

where, technically speaking,

nothing bad actually happened to it.

XL: So, so far we are only talking about

this light-controlled "on" switch.

In fact, we also have
a light-controlled "off" switch,

and it's very easy to imagine

that by installing this
light-controlled "off" switch,

we can also turn off a memory,
any time, anywhere.

So everything
we've been talking about today

is based on this philosophically
charged principle of neuroscience

that the mind, with its seemingly
mysterious properties,

is actually made of physical stuff
that we can tinker with.

SR: And for me personally,

I see a world where we can reactivate

any kind of memory that we'd like.

I also see a world where we can
erase unwanted memories.

Now, I even see a world
where editing memories

is something of a reality,

because we're living in a time
where it's possible

to pluck questions from the tree
of science fiction

and to ground them
in experimental reality.

XL: Nowadays, people in the lab

and people in other groups
all over the world

are using similar methods
to activate or edit memories,

whether that's old or new,
positive or negative,

all sorts of memories
so that we can understand

how memory works.

SR: For example, one group in our lab

was able to find the brain cells
that make up a fear memory

and converted them into
a pleasurable memory, just like that.

That's exactly what I mean about editing
these kinds of processes.

Now one dude in lab
was even able to reactivate

memories of female mice in male mice,

which rumor has it
is a pleasurable experience.

XL: Indeed, we are living
in a very exciting moment

where science doesn't have
any arbitrary speed limits

but is only bound by our own imagination.

SR: And finally,
what do we make of all this?

How do we push this technology forward?

These are the questions
that should not remain

just inside the lab,

and so one goal of today's talk
was to bring everybody

up to speed with the kind
of stuff that's possible

in modern neuroscience,

but now, just as importantly,

to actively engage everybody
in this conversation.

So let's think together as a team
about what this all means

and where we can and should go from here,

because Xu and I think we all have

some really big decisions ahead of us.

Thank you.
XL: Thank you.

(Applause)