15:26
TEDxBoston 2013

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

Filmed:

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. (Filmed at TEDxBoston.)

- Neuroscientist
When Steve Ramirez published his latest study in Science, it caused a media frenzy. Why? Because the paper was on implanting false memories in the brains of mice. Full bio

- Neuroscientist
In his groundbreaking work, Xu Liu investigated how to activate and deactivate specific memories in mice. Full bio

Steve Ramirez: My first year of grad school,
00:12
I found myself in my bedroom
00:13
eating lots of Ben & Jerry's
00:15
watching some trashy TV
00:18
and maybe, maybe listening to Taylor Swift.
00:19
I had just gone through a breakup.
00:23
(Laughter)
00:24
So for the longest time, all I would do
00:26
is recall the memory of this person over and over again,
00:28
wishing that I could get rid of that gut-wrenching,
00:32
visceral "blah" feeling.
00:34
Now, as it turns out, I'm a neuroscientist,
00:37
so I knew that the memory of that person
00:39
and the awful, emotional undertones that color in that memory,
00:42
are largely mediated by separate brain systems.
00:45
And so I thought, what if we could go into the brain
00:47
and edit out that nauseating feeling
00:50
but while keeping the memory of that person intact?
00:52
Then I realized, maybe that's a little bit lofty for now.
00:55
So what if we could start off by going into the brain
00:57
and just finding a single memory to begin with?
01:00
Could we jump-start that memory back to life,
01:03
maybe even play with the contents of that memory?
01:05
All that said, there is one person in the entire world right now
01:09
that I really hope is not watching this talk.
01:11
(Laughter)
01:13
So there is a catch. There is a catch.
01:17
These ideas probably remind you of "Total Recall,"
01:20
"Eternal Sunshine of the Spotless Mind,"
01:23
or of "Inception."
01:25
But the movie stars that we work with
01:27
are the celebrities of the lab.
01:28
Xu Liu: Test mice.
01:30
(Laughter)
01:32
As neuroscientists, we work in the lab with mice
01:33
trying to understand how memory works.
01:36
And today, we hope to convince you that now
01:40
we are actually able to activate a memory in the brain
01:42
at the speed of light.
01:45
To do this, there's only two simple steps to follow.
01:48
First, you find and label a memory in the brain,
01:51
and then you activate it with a switch.
01:54
As simple as that.
01:58
(Laughter)
01:59
SR: Are you convinced?
02:01
So, turns out finding a memory in the brain isn't all that easy.
02:03
XL: Indeed. This is way more difficult than, let's say,
02:07
finding a needle in a haystack,
02:09
because at least, you know, the needle is still something
02:12
you can physically put your fingers on.
02:14
But memory is not.
02:17
And also, there's way more cells in your brain
02:19
than the number of straws in a typical haystack.
02:22
So yeah, this task does seem to be daunting.
02:27
But luckily, we got help from the brain itself.
02:30
It turned out that all we need to do is basically
02:34
to let the brain form a memory,
02:36
and then the brain will tell us which cells are involved
02:38
in that particular memory.
02:42
SR: So what was going on in my brain
02:44
while I was recalling the memory of an ex?
02:46
If you were to just completely ignore human ethics for a second
02:48
and slice up my brain right now,
02:51
you would see that there was an amazing number
02:52
of brain regions that were active while recalling that memory.
02:54
Now one brain region that would be robustly active
02:57
in particular is called the hippocampus,
03:00
which for decades has been implicated in processing
03:02
the kinds of memories that we hold near and dear,
03:05
which also makes it an ideal target to go into
03:07
and to try and find and maybe reactivate a memory.
03:10
XL: When you zoom in into the hippocampus,
03:13
of course you will see lots of cells,
03:15
but we are able to find which cells are involved
03:17
in a particular memory,
03:20
because whenever a cell is active,
03:22
like when it's forming a memory,
03:24
it will also leave a footprint that will later allow us to know
03:26
these cells are recently active.
03:30
SR: So the same way that building lights at night
03:32
let you know that somebody's probably working there at any given moment,
03:34
in a very real sense, there are biological sensors
03:37
within a cell that are turned on
03:41
only when that cell was just working.
03:43
They're sort of biological windows that light up
03:45
to let us know that that cell was just active.
03:47
XL: So we clipped part of this sensor,
03:49
and attached that to a switch to control the cells,
03:51
and we packed this switch into an engineered virus
03:55
and injected that into the brain of the mice.
03:59
So whenever a memory is being formed,
04:01
any active cells for that memory
04:04
will also have this switch installed.
04:06
SR: So here is what the hippocampus looks like
04:09
after forming a fear memory, for example.
04:10
The sea of blue that you see here
04:13
are densely packed brain cells,
04:15
but the green brain cells,
04:17
the green brain cells are the ones that are holding on
04:19
to a specific fear memory.
04:21
So you are looking at the crystallization
04:23
of the fleeting formation of fear.
04:25
You're actually looking at the cross-section of a memory right now.
04:27
XL: Now, for the switch we have been talking about,
04:31
ideally, the switch has to act really fast.
04:33
It shouldn't take minutes or hours to work.
04:36
It should act at the speed of the brain, in milliseconds.
04:39
SR: So what do you think, Xu?
04:43
Could we use, let's say, pharmacological drugs
04:44
to activate or inactivate brain cells?
04:47
XL: Nah. Drugs are pretty messy. They spread everywhere.
04:49
And also it takes them forever to act on cells.
04:53
So it will not allow us to control a memory in real time.
04:56
So Steve, how about let's zap the brain with electricity?
05:00
SR: So electricity is pretty fast,
05:04
but we probably wouldn't be able to target it
05:06
to just the specific cells that hold onto a memory,
05:08
and we'd probably fry the brain.
05:10
XL: Oh. That's true. So it looks like, hmm,
05:12
indeed we need to find a better way
05:15
to impact the brain at the speed of light.
05:18
SR: So it just so happens that light travels at the speed of light.
05:21
So maybe we could activate or inactive memories
05:26
by just using light --
05:30
XL: That's pretty fast.
05:31
SR: -- and because normally brain cells
05:33
don't respond to pulses of light,
05:35
so those that would respond to pulses of light
05:36
are those that contain a light-sensitive switch.
05:38
Now to do that, first we need to trick brain cells
05:41
to respond to laser beams.
05:43
XL: Yep. You heard it right.
05:44
We are trying to shoot lasers into the brain.
05:45
(Laughter)
05:47
SR: And the technique that lets us do that is optogenetics.
05:49
Optogenetics gave us this light switch that we can use
05:52
to turn brain cells on or off,
05:56
and the name of that switch is channelrhodopsin,
05:57
seen here as these green dots attached to this brain cell.
06:00
You can think of channelrhodopsin as a sort of light-sensitive switch
06:02
that can be artificially installed in brain cells
06:05
so that now we can use that switch
06:08
to activate or inactivate the brain cell simply by clicking it,
06:10
and in this case we click it on with pulses of light.
06:13
XL: So we attach this light-sensitive switch of channelrhodopsin
06:15
to the sensor we've been talking about
06:19
and inject this into the brain.
06:21
So whenever a memory is being formed,
06:24
any active cell for that particular memory
06:27
will also have this light-sensitive switch installed in it
06:29
so that we can control these cells
06:33
by the flipping of a laser just like this one you see.
06:35
SR: So let's put all of this to the test now.
06:39
What we can do is we can take our mice
06:42
and then we can put them in a box that looks exactly like this box here,
06:44
and then we can give them a very mild foot shock
06:47
so that they form a fear memory of this box.
06:50
They learn that something bad happened here.
06:52
Now with our system, the cells that are active
06:54
in the hippocampus in the making of this memory,
06:56
only those cells will now contain channelrhodopsin.
06:59
XL: When you are as small as a mouse,
07:02
it feels as if the whole world is trying to get you.
07:05
So your best response of defense
07:08
is trying to be undetected.
07:10
Whenever a mouse is in fear,
07:13
it will show this very typical behavior
07:15
by staying at one corner of the box,
07:16
trying to not move any part of its body,
07:18
and this posture is called freezing.
07:21
So if a mouse remembers that something bad happened in this box,
07:24
and when we put them back into the same box,
07:29
it will basically show freezing
07:31
because it doesn't want to be detected
07:33
by any potential threats in this box.
07:35
SR: So you can think of freezing as,
07:38
you're walking down the street minding your own business,
07:40
and then out of nowhere you almost run into
07:42
an ex-girlfriend or ex-boyfriend,
07:43
and now those terrifying two seconds
07:46
where you start thinking, "What do I do? Do I say hi?
07:48
Do I shake their hand? Do I turn around and run away?
07:50
Do I sit here and pretend like I don't exist?"
07:51
Those kinds of fleeting thoughts that physically incapacitate you,
07:53
that temporarily give you that deer-in-headlights look.
07:56
XL: However, if you put the mouse in a completely different
07:59
new box, like the next one,
08:02
it will not be afraid of this box
08:05
because there's no reason that it will be afraid of this new environment.
08:08
But what if we put the mouse in this new box
08:12
but at the same time, we activate the fear memory
08:15
using lasers just like we did before?
08:19
Are we going to bring back the fear memory
08:22
for the first box into this completely new environment?
08:25
SR: All right, and here's the million-dollar experiment.
08:29
Now to bring back to life the memory of that day,
08:31
I remember that the Red Sox had just won,
08:34
it was a green spring day,
08:36
perfect for going up and down the river
08:38
and then maybe going to the North End
08:40
to get some cannolis, #justsaying.
08:43
Now Xu and I, on the other hand,
08:45
were in a completely windowless black room
08:48
not making any ocular movement that even remotely resembles an eye blink
08:51
because our eyes were fixed onto a computer screen.
08:54
We were looking at this mouse here trying to activate a memory
08:57
for the first time using our technique.
08:59
XL: And this is what we saw.
09:01
When we first put the mouse into this box,
09:04
it's exploring, sniffing around, walking around,
09:06
minding its own business,
09:09
because actually by nature,
09:11
mice are pretty curious animals.
09:13
They want to know, what's going on in this new box?
09:15
It's interesting.
09:17
But the moment we turned on the laser, like you see now,
09:19
all of a sudden the mouse entered this freezing mode.
09:22
It stayed here and tried not to move any part of its body.
09:25
Clearly it's freezing.
09:30
So indeed, it looks like we are able to bring back
09:31
the fear memory for the first box
09:34
in this completely new environment.
09:36
While watching this, Steve and I
09:39
are as shocked as the mouse itself.
09:41
(Laughter)
09:44
So after the experiment, the two of us just left the room
09:45
without saying anything.
09:48
After a kind of long, awkward period of time,
09:50
Steve broke the silence.
09:53
SR: "Did that just work?"
09:55
XL: "Yes," I said. "Indeed it worked!"
09:58
We're really excited about this.
10:01
And then we published our findings
10:03
in the journal Nature.
10:05
Ever since the publication of our work,
10:07
we've been receiving numerous comments
10:10
from all over the Internet.
10:12
Maybe we can take a look at some of those.
10:14
["OMGGGGG FINALLY... so much more to come, virtual reality, neural manipulation, visual dream emulation... neural coding, 'writing and re-writing of memories', mental illnesses. Ahhh the future is awesome"]
10:18
SR: So the first thing that you'll notice is that people
10:20
have really strong opinions about this kind of work.
10:22
Now I happen to completely agree with the optimism
10:25
of this first quote,
10:28
because on a scale of zero to Morgan Freeman's voice,
10:29
it happens to be one of the most evocative accolades
10:31
that I've heard come our way.
10:34
(Laughter)
10:35
But as you'll see, it's not the only opinion that's out there.
10:37
["This scares the hell out of me... What if they could do that easily in humans in a couple of years?! OH MY GOD WE'RE DOOMED"]
10:39
XL: Indeed, if we take a look at the second one,
10:41
I think we can all agree that it's, meh,
10:43
probably not as positive.
10:45
But this also reminds us that,
10:47
although we are still working with mice,
10:49
it's probably a good idea to start thinking and discussing
10:52
about the possible ethical ramifications
10:55
of memory control.
10:58
SR: Now, in the spirit of the third quote,
11:00
we want to tell you about a recent project that we've been
11:02
working on in lab that we've called Project Inception.
11:04
["They should make a movie about this. Where they plant ideas into peoples minds, so they can control them for their own personal gain. We'll call it: Inception."]
11:07
So we reasoned that now that we can reactivate a memory,
11:10
what if we do so but then begin to tinker with that memory?
11:14
Could we possibly even turn it into a false memory?
11:17
XL: So all memory is sophisticated and dynamic,
11:20
but if just for simplicity, let's imagine memory
11:24
as a movie clip.
11:27
So far what we've told you is basically we can control
11:28
this "play" button of the clip
11:31
so that we can play this video clip any time, anywhere.
11:33
But is there a possibility that we can actually get
11:37
inside the brain and edit this movie clip
11:40
so that we can make it different from the original?
11:43
Yes we can.
11:46
Turned out that all we need to do is basically
11:48
reactivate a memory using lasers just like we did before,
11:50
but at the same time, if we present new information
11:54
and allow this new information to incorporate into this old memory,
11:58
this will change the memory.
12:02
It's sort of like making a remix tape.
12:04
SR: So how do we do this?
12:08
Rather than finding a fear memory in the brain,
12:11
we can start by taking our animals,
12:13
and let's say we put them in a blue box like this blue box here
12:14
and we find the brain cells that represent that blue box
12:17
and we trick them to respond to pulses of light
12:20
exactly like we had said before.
12:22
Now the next day, we can take our animals and place them
12:24
in a red box that they've never experienced before.
12:26
We can shoot light into the brain to reactivate
12:29
the memory of the blue box.
12:31
So what would happen here if, while the animal
12:33
is recalling the memory of the blue box,
12:35
we gave it a couple of mild foot shocks?
12:36
So here we're trying to artificially make an association
12:39
between the memory of the blue box
12:42
and the foot shocks themselves.
12:44
We're just trying to connect the two.
12:45
So to test if we had done so,
12:47
we can take our animals once again
12:49
and place them back in the blue box.
12:50
Again, we had just reactivated the memory of the blue box
12:52
while the animal got a couple of mild foot shocks,
12:55
and now the animal suddenly freezes.
12:57
It's as though it's recalling being mildly shocked in this environment
12:59
even though that never actually happened.
13:02
So it formed a false memory,
13:05
because it's falsely fearing an environment
13:07
where, technically speaking,
13:09
nothing bad actually happened to it.
13:10
XL: So, so far we are only talking about
13:13
this light-controlled "on" switch.
13:15
In fact, we also have a light-controlled "off" switch,
13:18
and it's very easy to imagine that
13:21
by installing this light-controlled "off" switch,
13:23
we can also turn off a memory, any time, anywhere.
13:25
So everything we've been talking about today
13:31
is based on this philosophically charged principle of neuroscience
13:33
that the mind, with its seemingly mysterious properties,
13:38
is actually made of physical stuff that we can tinker with.
13:42
SR: And for me personally,
13:46
I see a world where we can reactivate
13:47
any kind of memory that we'd like.
13:49
I also see a world where we can erase unwanted memories.
13:51
Now, I even see a world where editing memories
13:54
is something of a reality,
13:56
because we're living in a time where it's possible
13:58
to pluck questions from the tree of science fiction
13:59
and to ground them in experimental reality.
14:02
XL: Nowadays, people in the lab
14:04
and people in other groups all over the world
14:06
are using similar methods to activate or edit memories,
14:08
whether that's old or new, positive or negative,
14:12
all sorts of memories so that we can understand
14:16
how memory works.
14:19
SR: For example, one group in our lab
14:20
was able to find the brain cells that make up a fear memory
14:22
and converted them into a pleasurable memory, just like that.
14:25
That's exactly what I mean about editing these kinds of processes.
14:28
Now one dude in lab was even able to reactivate
14:31
memories of female mice in male mice,
14:33
which rumor has it is a pleasurable experience.
14:35
XL: Indeed, we are living in a very exciting moment
14:38
where science doesn't have any arbitrary speed limits
14:42
but is only bound by our own imagination.
14:46
SR: And finally, what do we make of all this?
14:49
How do we push this technology forward?
14:51
These are the questions that should not remain
14:53
just inside the lab,
14:55
and so one goal of today's talk was to bring everybody
14:57
up to speed with the kind of stuff that's possible
14:59
in modern neuroscience,
15:02
but now, just as importantly,
15:03
to actively engage everybody in this conversation.
15:05
So let's think together as a team about what this all means
15:08
and where we can and should go from here,
15:10
because Xu and I think we all have
15:13
some really big decisions ahead of us.
15:15
Thank you.
XL: Thank you.
15:17
(Applause)
15:18

▲Back to top

About the Speakers:

Steve Ramirez - Neuroscientist
When Steve Ramirez published his latest study in Science, it caused a media frenzy. Why? Because the paper was on implanting false memories in the brains of mice.

Why you should listen

Steve is a graduate student at MIT’s Brain and Cognitive Sciences department pursuing a Ph.D. in neuroscience. His work focuses on finding where single memories are located throughout the brain, genetically tricking the brain cells that house these memories to respond to brief pulses of light, and then using these same flickers of light to reactivate, erase and implant memories. The goals of his research are twofold: to figure out how the brain gives rise to the seemingly ephemeral process of memory, and to predict what happens when specific brain pieces breakdown to impair cognition. His work has been published in Science and covered by New Scientist, Discover, Scientific American, and Gizmodo.

Ramirez aims to be a professor who runs a lab that plucks questions from the tree of science fiction to ground them in experimental reality. He believes that a team-oriented approach to science makes research and teaching far more exciting. When he’s not tinkering with memories in the lab, Ramirez also enjoys running and cheering on every sports team in the city of Boston.

More profile about the speaker
Steve Ramirez | Speaker | TED.com
Xu Liu - Neuroscientist
In his groundbreaking work, Xu Liu investigated how to activate and deactivate specific memories in mice.

Why you should listen

During his PhD, Xu Liu studied the mechanisms of learning and memory, using fruit flies as a model system. By changing the expression of certain genes in the fly brain, he generated smart flies that can learn many times faster than their peers. Using live imaging, he also detected learning-induced changes in the brain cells and observed memory formation inside the brain with light.

After graduation, he moved to MIT and joined Dr. Susumu Tonegawa's lab as a postdoctoral associate. He continued his pursuit of memory with light there. Instead of just watching memory formation, he developed a system in mice where one can not only identify and label cells in the brain for a particular memory, but also turn these cells on and off with light to activate this memory at will. This work was published in Science and has been covered by the media worldwide. Liu passed away in February 2015.

More profile about the speaker
Xu Liu | Speaker | TED.com