English-Video.net comment policy

The comment field is common to all languages

Let's write in your language and use "Google Translate" together

Please refer to informative community guidelines on TED.com

TEDGlobal 2010

Gero Miesenboeck: Re-engineering the brain

Filmed
Views 615,372

In the quest to map the brain, many scientists have attempted the incredibly daunting task of recording the activity of each neuron. Gero Miesenboeck works backward -- manipulating specific neurons to figure out exactly what they do, through a series of stunning experiments that reengineer the way fruit flies percieve light.

- Optogeneticist
Using light and a little genetic engineering -- optogenetics -- Gero Miesenboeck has developed a way to control how living nerve cells work, and advanced understanding of how the brain controls behavior. Full bio

I have a doppelganger.
00:15
(Laughter)
00:18
Dr. Gero is a brilliant
00:21
but slightly mad scientist
00:24
in the "Dragonball Z: Android Saga."
00:26
If you look very carefully,
00:29
you see that his skull has been replaced
00:31
with a transparent Plexiglas dome
00:34
so that the workings of his brain can be observed
00:36
and also controlled with light.
00:39
That's exactly what I do --
00:42
optical mind control.
00:44
(Laughter)
00:46
But in contrast to my evil twin
00:48
who lusts after world domination,
00:50
my motives are not sinister.
00:53
I control the brain
00:56
in order to understand how it works.
00:58
Now wait a minute, you may say,
01:00
how can you go straight to controlling the brain
01:02
without understanding it first?
01:05
Isn't that putting the cart before the horse?
01:07
Many neuroscientists agree with this view
01:11
and think that understanding will come
01:14
from more detailed observation and analysis.
01:17
They say, "If we could record the activity of our neurons,
01:20
we would understand the brain."
01:24
But think for a moment what that means.
01:27
Even if we could measure
01:30
what every cell is doing at all times,
01:32
we would still have to make sense
01:34
of the recorded activity patterns,
01:36
and that's so difficult,
01:38
chances are we'll understand these patterns
01:40
just as little as the brains that produce them.
01:42
Take a look at what brain activity might look like.
01:45
In this simulation, each black dot
01:48
is one nerve cell.
01:50
The dot is visible
01:52
whenever a cell fires an electrical impulse.
01:54
There's 10,000 neurons here.
01:56
So you're looking at roughly one percent
01:58
of the brain of a cockroach.
02:00
Your brains are about 100 million times
02:04
more complicated.
02:07
Somewhere, in a pattern like this,
02:09
is you,
02:11
your perceptions,
02:13
your emotions, your memories,
02:15
your plans for the future.
02:18
But we don't know where,
02:20
since we don't know how to read the pattern.
02:22
We don't understand the code used by the brain.
02:25
To make progress,
02:28
we need to break the code.
02:30
But how?
02:32
An experienced code-breaker will tell you
02:35
that in order to figure out what the symbols in a code mean,
02:37
it's essential to be able to play with them,
02:40
to rearrange them at will.
02:43
So in this situation too,
02:45
to decode the information
02:47
contained in patterns like this,
02:49
watching alone won't do.
02:51
We need to rearrange the pattern.
02:53
In other words,
02:55
instead of recording the activity of neurons,
02:57
we need to control it.
02:59
It's not essential that we can control
03:01
the activity of all neurons in the brain, just some.
03:03
The more targeted our interventions, the better.
03:06
And I'll show you in a moment
03:08
how we can achieve the necessary precision.
03:10
And since I'm realistic, rather than grandiose,
03:13
I don't claim that the ability to control the function of the nervous system
03:16
will at once unravel all its mysteries.
03:19
But we'll certainly learn a lot.
03:22
Now, I'm by no means
03:27
the first person to realize
03:29
how powerful a tool intervention is.
03:31
The history of attempts
03:34
to tinker with the function of the nervous system
03:36
is long and illustrious.
03:38
It dates back at least 200 years,
03:40
to Galvani's famous experiments
03:43
in the late 18th century and beyond.
03:45
Galvani showed that a frog's legs twitched
03:49
when he connected the lumbar nerve
03:52
to a source of electrical current.
03:54
This experiment revealed the first, and perhaps most fundamental,
03:57
nugget of the neural code:
04:00
that information is written in the form
04:02
of electrical impulses.
04:04
Galvani's approach
04:08
of probing the nervous system with electrodes
04:10
has remained state-of-the-art until today,
04:12
despite a number of drawbacks.
04:15
Sticking wires into the brain is obviously rather crude.
04:18
It's hard to do in animals that run around,
04:21
and there is a physical limit
04:23
to the number of wires
04:25
that can be inserted simultaneously.
04:27
So around the turn of the last century,
04:30
I started to think,
04:32
"Wouldn't it be wonderful if one could take this logic
04:34
and turn it upside down?"
04:37
So instead of inserting a wire
04:39
into one spot of the brain,
04:41
re-engineer the brain itself
04:44
so that some of its neural elements
04:46
become responsive to diffusely broadcast signals
04:49
such as a flash of light.
04:52
Such an approach would literally, in a flash of light,
04:55
overcome many of the obstacles to discovery.
04:58
First, it's clearly a non-invasive,
05:01
wireless form of communication.
05:04
And second, just as in a radio broadcast,
05:07
you can communicate with many receivers at once.
05:09
You don't need to know where these receivers are,
05:12
and it doesn't matter if these receivers move --
05:15
just think of the stereo in your car.
05:17
It gets even better,
05:20
for it turns out that we can fabricate the receivers
05:23
out of materials that are encoded in DNA.
05:26
So each nerve cell
05:29
with the right genetic makeup
05:31
will spontaneously produce a receiver
05:33
that allows us to control its function.
05:36
I hope you'll appreciate
05:39
the beautiful simplicity
05:41
of this concept.
05:43
There's no high-tech gizmos here,
05:45
just biology revealed through biology.
05:47
Now let's take a look at these miraculous receivers up close.
05:51
As we zoom in on one of these purple neurons,
05:54
we see that its outer membrane
05:57
is studded with microscopic pores.
05:59
Pores like these conduct electrical current
06:01
and are responsible
06:03
for all the communication in the nervous system.
06:05
But these pores here are special.
06:07
They are coupled to light receptors
06:09
similar to the ones in your eyes.
06:11
Whenever a flash of light hits the receptor,
06:14
the pore opens, an electrical current is switched on,
06:16
and the neuron fires electrical impulses.
06:19
Because the light-activated pore is encoded in DNA,
06:22
we can achieve incredible precision.
06:25
This is because,
06:28
although each cell in our bodies
06:30
contains the same set of genes,
06:32
different mixes of genes get turned on and off
06:34
in different cells.
06:36
You can exploit this to make sure
06:38
that only some neurons
06:40
contain our light-activated pore and others don't.
06:42
So in this cartoon, the bluish white cell
06:45
in the upper-left corner
06:47
does not respond to light
06:49
because it lacks the light-activated pore.
06:51
The approach works so well
06:54
that we can write purely artificial messages
06:56
directly to the brain.
06:58
In this example, each electrical impulse,
07:00
each deflection on the trace,
07:02
is caused by a brief pulse of light.
07:05
And the approach, of course, also works
07:08
in moving, behaving animals.
07:10
This is the first ever such experiment,
07:13
sort of the optical equivalent of Galvani's.
07:15
It was done six or seven years ago
07:18
by my then graduate student, Susana Lima.
07:20
Susana had engineered the fruit fly on the left
07:23
so that just two out of the 200,000 cells in its brain
07:26
expressed the light-activated pore.
07:30
You're familiar with these cells
07:33
because they are the ones that frustrate you
07:35
when you try to swat the fly.
07:37
They trained the escape reflex that makes the fly jump into the air
07:39
and fly away whenever you move your hand in position.
07:42
And you can see here that the flash of light has exactly the same effect.
07:46
The animal jumps, it spreads its wings, it vibrates them,
07:49
but it can't actually take off
07:52
because the fly is sandwiched between two glass plates.
07:54
Now to make sure that this was no reaction of the fly
07:58
to a flash it could see,
08:00
Susana did a simple
08:03
but brutally effective experiment.
08:05
She cut the heads off of her flies.
08:07
These headless bodies can live for about a day,
08:11
but they don't do much.
08:14
They just stand around
08:16
and groom excessively.
08:19
So it seems that the only trait that survives decapitation is vanity.
08:22
(Laughter)
08:25
Anyway, as you'll see in a moment,
08:30
Susana was able to turn on the flight motor
08:32
of what's the equivalent of the spinal cord of these flies
08:35
and get some of the headless bodies
08:38
to actually take off and fly away.
08:40
They didn't get very far, obviously.
08:47
Since we took these first steps,
08:50
the field of optogenetics has exploded.
08:52
And there are now hundreds of labs
08:55
using these approaches.
08:57
And we've come a long way
08:59
since Galvani's and Susana's first successes
09:01
in making animals twitch or jump.
09:04
We can now actually interfere with their psychology
09:06
in rather profound ways,
09:09
as I'll show you in my last example,
09:11
which is directed at a familiar question.
09:13
Life is a string of choices
09:16
creating a constant pressure to decide what to do next.
09:19
We cope with this pressure by having brains,
09:23
and within our brains, decision-making centers
09:26
that I've called here the "Actor."
09:29
The Actor implements a policy that takes into account
09:33
the state of the environment
09:36
and the context in which we operate.
09:38
Our actions change the environment, or context,
09:41
and these changes are then fed back into the decision loop.
09:44
Now to put some neurobiological meat
09:48
on this abstract model,
09:51
we constructed a simple one-dimensional world
09:53
for our favorite subject, fruit flies.
09:55
Each chamber in these two vertical stacks
09:58
contains one fly.
10:00
The left and the right halves of the chamber
10:02
are filled with two different odors,
10:05
and a security camera watches
10:07
as the flies pace up and down between them.
10:09
Here's some such CCTV footage.
10:12
Whenever a fly reaches the midpoint of the chamber
10:14
where the two odor streams meet,
10:17
it has to make a decision.
10:19
It has to decide whether to turn around
10:21
and stay in the same odor,
10:23
or whether to cross the midline
10:25
and try something new.
10:27
These decisions are clearly a reflection
10:29
of the Actor's policy.
10:32
Now for an intelligent being like our fly,
10:36
this policy is not written in stone
10:39
but rather changes as the animal learns from experience.
10:42
We can incorporate such an element
10:45
of adaptive intelligence into our model
10:47
by assuming that the fly's brain
10:50
contains not only an Actor,
10:52
but a different group of cells,
10:54
a "Critic," that provides a running commentary
10:56
on the Actor's choices.
10:59
You can think of this nagging inner voice
11:01
as sort of the brain's equivalent
11:04
of the Catholic Church,
11:06
if you're an Austrian like me,
11:08
or the super-ego, if you're Freudian,
11:11
or your mother, if you're Jewish.
11:14
(Laughter)
11:16
Now obviously,
11:20
the Critic is a key ingredient
11:22
in what makes us intelligent.
11:25
So we set out to identify
11:27
the cells in the fly's brain
11:29
that played the role of the Critic.
11:31
And the logic of our experiment was simple.
11:33
We thought if we could use our optical remote control
11:36
to activate the cells of the Critic,
11:39
we should be able, artificially, to nag the Actor
11:42
into changing its policy.
11:45
In other words,
11:47
the fly should learn from mistakes
11:49
that it thought it had made
11:51
but, in reality, it had not made.
11:53
So we bred flies
11:56
whose brains were more or less randomly peppered
11:58
with cells that were light addressable.
12:01
And then we took these flies
12:03
and allowed them to make choices.
12:05
And whenever they made one of the two choices,
12:07
chose one odor,
12:09
in this case the blue one over the orange one,
12:11
we switched on the lights.
12:13
If the Critic was among the optically activated cells,
12:15
the result of this intervention
12:18
should be a change in policy.
12:20
The fly should learn to avoid
12:23
the optically reinforced odor.
12:25
Here's what happened in two instances:
12:27
We're comparing two strains of flies,
12:30
each of them having
12:33
about 100 light-addressable cells in their brains,
12:35
shown here in green on the left and on the right.
12:37
What's common among these groups of cells
12:40
is that they all produce the neurotransmitter dopamine.
12:43
But the identities of the individual
12:46
dopamine-producing neurons
12:48
are clearly largely different on the left and on the right.
12:50
Optically activating
12:53
these hundred or so cells
12:55
into two strains of flies
12:57
has dramatically different consequences.
12:59
If you look first at the behavior
13:01
of the fly on the right,
13:03
you can see that whenever it reaches the midpoint of the chamber
13:05
where the two odors meet,
13:08
it marches straight through, as it did before.
13:10
Its behavior is completely unchanged.
13:13
But the behavior of the fly on the left is very different.
13:15
Whenever it comes up to the midpoint,
13:18
it pauses,
13:21
it carefully scans the odor interface
13:23
as if it was sniffing out its environment,
13:25
and then it turns around.
13:27
This means that the policy that the Actor implements
13:29
now includes an instruction to avoid the odor
13:32
that's in the right half of the chamber.
13:34
This means that the Critic
13:37
must have spoken in that animal,
13:39
and that the Critic must be contained
13:41
among the dopamine-producing neurons on the left,
13:43
but not among the dopamine producing neurons on the right.
13:46
Through many such experiments,
13:49
we were able to narrow down
13:52
the identity of the Critic
13:54
to just 12 cells.
13:56
These 12 cells, as shown here in green,
13:58
send the output to a brain structure
14:01
called the "mushroom body,"
14:03
which is shown here in gray.
14:05
We know from our formal model
14:07
that the brain structure
14:09
at the receiving end of the Critic's commentary is the Actor.
14:11
So this anatomy suggests
14:14
that the mushroom bodies have something to do
14:16
with action choice.
14:19
Based on everything we know about the mushroom bodies,
14:21
this makes perfect sense.
14:23
In fact, it makes so much sense
14:25
that we can construct an electronic toy circuit
14:27
that simulates the behavior of the fly.
14:30
In this electronic toy circuit,
14:33
the mushroom body neurons are symbolized
14:36
by the vertical bank of blue LEDs
14:38
in the center of the board.
14:41
These LED's are wired to sensors
14:44
that detect the presence of odorous molecules in the air.
14:46
Each odor activates a different combination of sensors,
14:50
which in turn activates
14:53
a different odor detector in the mushroom body.
14:55
So the pilot in the cockpit of the fly,
14:58
the Actor,
15:00
can tell which odor is present
15:02
simply by looking at which of the blue LEDs lights up.
15:04
What the Actor does with this information
15:09
depends on its policy,
15:11
which is stored in the strengths of the connection,
15:13
between the odor detectors
15:15
and the motors
15:17
that power the fly's evasive actions.
15:19
If the connection is weak, the motors will stay off
15:22
and the fly will continue straight on its course.
15:24
If the connection is strong, the motors will turn on
15:27
and the fly will initiate a turn.
15:30
Now consider a situation
15:33
in which the motors stay off,
15:35
the fly continues on its path
15:37
and it suffers some painful consequence
15:40
such as getting zapped.
15:42
In a situation like this,
15:44
we would expect the Critic to speak up
15:46
and to tell the Actor
15:48
to change its policy.
15:50
We have created such a situation, artificially,
15:52
by turning on the critic with a flash of light.
15:55
That caused a strengthening of the connections
15:58
between the currently active odor detector
16:01
and the motors.
16:04
So the next time
16:06
the fly finds itself facing the same odor again,
16:08
the connection is strong enough to turn on the motors
16:11
and to trigger an evasive maneuver.
16:14
I don't know about you,
16:19
but I find it exhilarating to see
16:22
how vague psychological notions
16:25
evaporate and give rise
16:28
to a physical, mechanistic understanding of the mind,
16:30
even if it's the mind of the fly.
16:33
This is one piece of good news.
16:36
The other piece of good news,
16:39
for a scientist at least,
16:41
is that much remains to be discovered.
16:43
In the experiments I told you about,
16:46
we have lifted the identity of the Critic,
16:48
but we still have no idea
16:51
how the Critic does its job.
16:53
Come to think of it, knowing when you're wrong
16:55
without a teacher, or your mother, telling you,
16:57
is a very hard problem.
17:00
There are some ideas in computer science
17:02
and in artificial intelligence
17:04
as to how this might be done,
17:06
but we still haven't solved
17:08
a single example
17:10
of how intelligent behavior
17:12
springs from the physical interactions
17:15
in living matter.
17:17
I think we'll get there in the not too distant future.
17:19
Thank you.
17:22
(Applause)
17:24

▲Back to top

About the speaker:

Gero Miesenboeck - Optogeneticist
Using light and a little genetic engineering -- optogenetics -- Gero Miesenboeck has developed a way to control how living nerve cells work, and advanced understanding of how the brain controls behavior.

Why you should listen

Gero Miesenboeck is pioneering the field of optogenetics: genetically modifying nerve cells to respond to light. By flashing light at a modified neuron in a living nervous system, Miesenboeck and his collaborators can mimic a brain impulse -- and then study what happens next. Optogenetics will allow ever more precise experiments on living brains, allowing us to gather better evidence on how electrical impulses on tissue translate into actual behavior and thoughts.

In one experiment, done at Yale, he and his team engineered fruit flies to be light-sensitive in the neural area responsible for escape response. Then the flies were beheaded; fruit flies can live for a day without their heads, but they don't move. When the modified cells were flashed with light, though, the headless flies flew. Miesenboeck had successfully simulated an order from a brain that wasn't even there anymore.

Miesenboeck's current research at Oxford's growing department of neurobiology focuses on the nerve cell networks that underpin what animals perceive, remember and do. In one recent experiment, he used optogenetics to implant an unpleasant memory in a fruit fly, causing it to "remember" to avoid a certain odor as it traveled around. He and his team were able, in fact, to find the fly's specific 12-neuron brain circuit that govern memory formation.

More profile about the speaker
Gero Miesenboeck | Speaker | TED.com