Follow US:

Practice English Speaking&Listening with: "Sex, Lies, and Brain Scans: How fMRI Reveals What Really Goes on in Our Minds" | Talks at Google

(0)
Difficulty: 0

BARBARA SAHAKIAN: Thank you.

It's a real pleasure to be here and to speak

to you about my book which I wrote with Julia Gottwald,

as you heard.

I'm just going to tell you a little bit about how

I got into this area because I started

really getting into the area of neuroethics some years ago.

And it was really because a lot of my work

is on measuring cognitive function,

measuring cognition and so understanding

the parts of the brain, the areas of the brain,

and the neurotransmitters that neuromodulate attention,

learning, memory, and planning.

And I do that within a clinical realm

so with patients who have problems in cognition.

And those include neuropsychiatric patients,

patients with Alzheimer's disease,

and also patients with traumatic brain injury.

And so you have to clinically evaluate

what they have problems with, and what their strengths are,

and that sort of thing, but also in terms

of trying to understand what the brain does

and what different areas of the brain do

and the neural networks involved combined

with how these different chemicals in the brain,

like dopamine or adrenalin and serotonin,

neuromodulate those areas.

So a lot of my work is actually geared towards improving

cognitive function.

And I've done that really working for many years.

So originally, I set up some memory clinics with colleagues

to look at the early stages of Alzheimer's disease.

Now it's called mild cognitive impairment,

but looking at very mild Alzheimer's patients,

trying to get them into the clinic

so that we could look at new drug treatments that

were becoming available from the pharmaceutical industry.

So the cholinesterase inhibitor drugs,

drugs like Aricept and donepezil which are now NICE approved

for the use in Alzheimer's patients

to improve cognitive function, I actually worked on those drugs

and published in The Lancet many years ago.

In the process of that, I saw that while we could improve

attention very well in these patients with mild Alzheimer's

disease, I couldn't really get a good signal, a good improvement

in what's called episodic memory, which

is the kind of memory we use every day.

So ever since then, I've still been

looking for good treatments, pharmacological treatments,

for episodic memory.

But I do a lot of work.

I have a psychopharmacology laboratory

in Cambridge, University of Cambridge.

And I do a lot of work trying to enhance cognitive function

in patients with these disorders such as Alzheimer's disease,

ADHD, schizophrenia, depression.

And I also work with healthy normal people

to see how, in the healthy brain,

different chemicals change ability to attend, or plan,

problem solve, what we call executive functions

and also the more posterior functions dealt

with by the hippocampus, such as episodic memory.

Now in the course of doing that work

and publishing on cognitive enhancing drugs

in different areas, it became clear to me

that healthy people were using a lot of these drugs.

So the drugs that I was examining

such as methylphenidate, which is also

known as Ritalin the most common drug treatment for Attention

Deficit Hyperactivity Disorder ADHD,

was being used by healthy people.

And also, a drug I commonly used in my laboratory, modafinil,

is being used quite widely by healthy people.

So then I became very interested in the ethical aspects of that.

And I got involved in that.

And so my first book was called "Bad Moves."

And it's all about the use of these cognitive

enhancing drugs by healthy people.

And then this is the second book which

deals with some of the interesting ethical issues

and societal issues that are raised by this.

So I hope I get a chance to come back to you sometime

and talk about the work on cognitive

enhancing drugs and also the work we're doing with apps now.

They're sometimes called brain training apps.

I call them cognitive training apps

to improve cognitive function.

We do them as games.

I don't know if you know the games company in London

called Peak.

But we've technology transferred some of the games

to the company called Peak.

And we're also doing some work with artificial intelligence

where we're looking at how we can combine some

of the features of the human brain into the AI brain,

so to speak, so that we have a more empathetic.

Empathy is involved in the health care

that we'll be using with AI in the future.

So that's sort of what I do, scopes it quite well.

And so now I'll talk about the book itself.

So I became very interested in how are some of these studies

that they're being done with fRMI,

how do they affect society.

Have people thought about the impact

that it might be having on society or some of the ethical

issues involved?

So I decided to write about that.

So all of you will know about ethics.

And neuroethics is simply the study

of the ethical, legal, and social questions

that arise when scientific findings about the brain

are actually carried into medical practice,

legal interpretations, and health and social policy.

And basically, in the brain, I was asked one time

to comment on a film where they seemed to suggest that we only

use 10% of our brain.

And I said no, that's not true.

We're using all of our brain all of the time.

And even at night when we're sleeping,

we're using our brain.

Now one thing that's very helpful

is that fMRI is an indirect measure of neuronal activity

because it measures blood flow.

So essentially the way it works is

that when we're using a part of our brain,

so when we're doing planning and problem solving,

what we call the higher cognitive functions,

executive functions, which involve

frontal areas of the brain, the blood flows

to that area of the brain.

So we can actually detect that in the scanner.

And then again, we talked about episodic memory

which is your everyday sort of memory.

It's kind of trying to remember where

you left your car in a multi-story car park

when you parked it there several hours ago

or trying to remember where you left

your mobile phone in the house.

That's episodic memory.

And we can see that there's activation in areas

like the hippocampus when we asked people

to do these memory tasks.

So it's a great way to be able to interrogate the systems

in the brain that are active when we're doing

particular cognitive functions.

And so as you know, MRI, we just get

a sort of snapshot of the brain and then

we superimpose these blood flow images onto that

so we can see both structure and function when

we're doing these studies.

And that's really how it all works.

And I think we don't think too much

about what an amazing invention all this is.

Because before we couldn't really

look at the brain in great detail.

And to find out about the brain, we

had to get patients who had brain damage in different areas

due to like tumors which were removed or car accidents

or whatever they had damaged.

And we had to infer what that bit of the brain

did from what they couldn't do anymore.

But now we have these amazing ways just

to see, in a healthy normal person lying in a scanner, when

we asked them to do a task, which areas of the brain

are involved in that task.

So it's really a marvelous, marvelous invention.

And it's funny because I'm sure you've

seen all these beautiful pictures of the brain.

And they're done so nicely now you

can see the connections and everything else and then

beautiful colors.

And when you show that to somebody,

they think of it as a very hard fact.

Because there's a picture of the brain.

This must be true.

But what we don't realize is all the maths and the analysis

that goes into that.

And that that can be--

we can actually change that by depending

on where we want to threshold things or manipulate things.

So it isn't quite so hard and fast as some people think.

And just even trying to make sure you can replicate

something or you can compare scanners,

so if you want to do studies with multiple people

in different sites, you have to actually

make sure the scanners are all aligned properly.

And that's quite a job to do.

So I think people need to be a bit more

skeptical about some of those, especially the general public.

And the other thing is that there's

an association between what you see in the image

and what maybe the behavior is.

So we can't say that this in the brain has caused this.

It's just a relationship.

So an example might be if somebody is manic

and is doing a lot of active things and rushing around,

is the change that we see in the brain,--

which is different maybe from a healthy brain, somebody

who doesn't have any mania--

is that caused by the running around or is it

the brain causes the running around.

So we can't tell which is the cause and which is the effect.

It's just an association, a relationship between the two.

And that's important, too.

So I just picked out a few of the interesting studies

to go through just to talk about some of the ethical

and sort of societal issues that might come up in thinking

about some of these studies.

But there are many others presented in the book.

And I'll talk about reading intentions,

which is quite interesting, and the unconscious racial biases.

There's been a lot about unconscious biases

recently in the news.

And we'll also talk about some of the use

in forensic and in neuromarketing.

So we can measure a clear cut racial bias

in the old fashioned way, by just asking people questions

such as, discrimination against blacks

is no longer a problem in the United States.

And if you strongly agree with that,

it tends to be that you have a higher racism score.

And so, there's the clear cut racial biases that we

might see in some groups.

And that can be measured this way.

But the trouble is, of course, that this

requires that you give a subjective answer

to the questions.

And of course, it's very easy for you

to lie if you're doing something where

you think it might not be socially appropriate

or politically correct to answer in the way that you believe.

So these can be very easily manipulated.

So these are just sort of straightforward conscious

racial biases when you're asking questions like this.

In an interesting study by Liz Phelps

she looked at the conscious racial biases

through that method.

And she also looked at implicit racial biases

through two methods.

And one was putting people in a fMRI scanner

and measuring their reaction to white and black faces.

And to hide the purpose behind the study, what

she did was she's going to ask them

to remember whether they've seen the face before or not.

So she didn't say, you know, focus on the fact

that these are white and black faces.

She just said, I want you to remember whether you've

seen this face before or not.

And they were white people lying in the scanner.

And so what happens is that, as you'll be seeing,

is that you get very strong activation

to black faces in the scanner if you're a white person.

And the stronger that activation, the more

your racial bias, basically.

And the thing is that we have--

the amygdala is, as you probably realize,

is very important for fear and emotion.

So it's important in anxiety and in fear.

And so it comes with--

when we consider in groups and out groups,

if somebody is in the out group, then the activation

to the amygdala will be stronger.

Now the other way she did it was looking to see how long it

took you to classify words.

Some of them were positive like wonderful.

And some of them were negative like horrible.

And you could either classify them as a good word

and that was associated with sort of white and good.

Or you could classify them differently.

And that was that black was associated

with good and wonderful.

So you were told to classify them as quickly as possible.

Now if you could classify them just as well whether it

was white and good or black and good,

then you'd wouldn't have any bias at all.

But if you were slower to classify with the black

and good, that indicates an implicit bias.

Now these are unconscious biases.

And we all have them.

I mean, it's not just race that you can have a bias to.

But you can have a bias to gender, or something else,

that you just feel that is that way.

And of course, as educated people,

we try to overcome our biases and we can usually act on them

and educate ourselves and try to change the way that we

think about something.

But if you don't know about the bias, if it's implicit,

it's harder to do, of course.

So it's quite useful to be aware of biases that we have

so we can try to improve ourselves

as people and in society.

So what she found was that when subjects viewed black and white

faces in the scanner and there was activation

and the amygdala was measured, and then

they completed this test of implicit racial bias

and explicit racial bias.

And the results showed a significant correlation

of the amygdala activation, as you can see on the graph there,

with the implicit bias.

But no correlation with the amygdala activation

and the explicit racial bias.

And that's probably because you could conceal your views

on that test.

But you can't conceal them in the implicit condition.

And that's just shown in these graphs

here where you can see on the left hand side,

you've got the explicit bias.

But here on the right hand side, you have the implicit bias.

And it's directly related to the activation in the amygdala.

So the question is, what do we do about these

in place racial biases?

I mean, is it important that we know about them?

So maybe judges or people who come in trials,

if they have these biases, should they

be allowed to judge people and so forth?

So what should we do as a group about these biases?

Should we just train for them?

I know at the University of Cambridge,

we do have training on racial bias.

And maybe that's one of the useful ways forward for this.

So this is a study by Ganis.

And it's quite interesting.

And it talks about lying.

And so they had the truth's condition.

And they had a spontaneous lying condition.

And then they had the memorized lying condition.

And in the memorized lying condition,

you would sort of a lie.

And the spontaneous one, you generated your own lie.

And then the truth, you told the truth.

And they asked things about, what

did you do during the summer or where were you born

or where did you grow up.

So there was sort of definite facts that you could have.

And you were asked to sometimes tell the truth,

sometimes give a spontaneous lie,

and sometimes give a memorized scenario.

Now interestingly, your brain doesn't

have to work as hard if you're telling the truth.

So it seems that it's a more natural thing

to outwith what the truth is.

And part of this is that you can imagine

that if you have to either give a spontaneous lie

or a memorized lie, you first have to inhibit the truth

and then tell the lie.

So it's a little bit more difficult.

And we know that front part of the brain

is often involved in trying to inhibit and exert

what we call top down cognitive control over our emotions

or what we're going to do, our actions.

And so we have to do that first before we come out with it.

And interestingly, there's been studies

where they've looked at lying across the lifespan

and it seems that people in their--

sort of young adults and middle aged people

are much better at lying than are children

or older, very older adults.

And again, this is probably because, in children, you

probably know that frontal lobe development,

our brains don't fully develop till age 24 or 25 years of age.

And in women, it's about three years earlier.

But that development, the frontal part

of the brain, the frontal lobes are the last area

of the brain to develop.

So it may well be that we haven't

got that ability as easily to inhibit the truth

and then tell a lie.

And at the older end of the spectrum,

it may be that obviously we have some degree of nerve

degeneration at that stage.

And it may be that it's harder, also, to do so.

So anyway what was found is that you

can see the different activations here

for the different conditions, three different conditions that

are shown.

And the sort of ventral medial prefrontal cortex

is activated during the lies and it

seems that you can even tell a little bit

the difference between the spontaneous lie

and the memorized lie.

Although, for the most part they show similar activations

but in a couple of areas of the brain that

are slightly different, like frontal pole, BA 10.

There's differences in the anterior cingulate.

So we can see some differences.

But the truth looks very different

from the lying condition.

So it may be that we can, obviously--

and companies are starting to use these procedures

for telling whether or not somebody is lying.

So you can imagine that perhaps in future you

put your partner in a scanner and ask,

does he really love me?

Or has he gone out with somebody else, or something like that.

And we'll be able to tell this or whether you're

telling the truth in court, and so forth.

So one of the issues, of course, is false memories.

And I'm sure we all have some of these.

Often what happens is that, as you're growing up,

your families will tell you different stories about when

you were little, this used to happen

or one time you did this.

And you incorporate that as a memory,

sometimes, if you're told it enough or you hear it enough.

And you almost believe that it was something

that you have as a memory.

But it's actually been put into your mind,

so to speak, because you've heard about it so much that you

begin to believe that this is what went on

and this is your memory.

So it's possible that people can have false memories.

And we know that if you go to a scene of a crime

and you ask a number of people about what's happened,

particularly because that's often an emotional memory

and people's emotions are very high

and they're obviously in a state of stress,

you'll get different answers about what actually went on.

And they won't always accord with each other

as to exactly what went first, what went second, who did what.

So it's possible that we could find out that people,

we're picking up memories that aren't actually there.

Now some of you will know the moral dilemma, the trolley

problem, because it's quite a famous one.

And so I'll just ask you about it.

So on the left hand side, you can see

that there's a person there.

Imagine it's you with the lever.

And the trolley's coming down the track.

And it's out of control.

And it can either smash into these five people.

Or you can flick that lever.

And it will turn onto the other track

and hit that other person.

So how many in the audience would flick the lever in order

to--

yeah.

Well, usually it's almost 100%.

But this group seems conflicted.

So that's very interesting.

I'll have to ask afterwards about this and why.

So usually, most people will say that they

would do that because they would save five lives

at the expense of one life.

Now in this problem over on the right hand side,

you have a rather large person standing on a bridge.

And you've got the five people there.

And you can either let the five people

die with this trolley whizzing out of control.

Or you can push this person off the bridge

into the track to stop the trolley.

So under those conditions, how many people are

going to give a shove to this--

yeah.

So rationally, it is still the right thing

to do because you're saving five lives at the expense of one

life.

But very few people will say that they would do that.

Because it requires an action on your part

to actually push this person yourself onto the bridge.

So most people won't do that.

And it's very interesting because drugs will sometimes

alter what people will do under these conditions.

Serotonin, the SSRIs like Prozac,

tend to be pro-moral drugs or pro

considering other people's empathy type drugs.

So they sometimes can alter this type of behavior in people.

So I'll talk about a study of neural correlates of morality.

And here you can see some slides.

You're lying in a scanner, imagine,

and these slides are coming up at you.

And some of them are neutral.

I haven't shown a neutral one.

But it might just be a scene of something.

And some of them are very--

they're both unpleasant, these ones, as you can see.

But this one has a moral connotation.

Because there's a knife being held

to this throat of this lady.

And the other one has no moral connotation because it's just

a really scary dog.

And so you're looking at these.

And we know that neural correlates of morality

are often associated, actually, with the pro social areas

of the brain.

And those include the TPJ as it's listed up there,

which is the Temporoparietal Junction, and also

the ventral medial prefrontal cortex.

And that's a very important area.

These areas regularly come up when

you do experiments looking at social behavior

and social interactions and empathy and things like that.

So what if your brain doesn't show

this pattern of activation?

You know, do we decide that you're a psychopath?

Or what do we do about that?

And in this study by Harenski, what they did

was they actually looked at prisoners

and some were psychopaths as determined

by the fHare Psychopathy Scale.

And others were non-psychopaths.

And they actually looked at how they

responded to those moral, non-moral, and neutral pictures

that you saw.

And what you can see is that offenders who are psychopaths

do not show the same pattern of activation

as offenders who are non-psychopaths.

So they don't respond to that moral condition in the same way

at all.

We've had several discussions on the radio recently

in the news about what should we be doing with people

who are incarcerated.

People are trying to put them to programs, check

if they're safe to let out.

And in the future, will we be doing this by brain scan

or by what?

And if you show an abnormal activation pattern,

should you be incarcerated before you commit a crime?

And I think "Minority Report" was

a film that had to do with that type of scenario.

So we need to think about, well, what

do we do with these findings from these studies.

Or should we be doing anything?

We all know that we have impulses and urges.

And we frequently use our top down cognitive control

over our actions not to do the impulsive thing.

I was recently at the Karolinska Institute

examining a PhD thesis.

And it was on alcohol use disorder.

And the point was made that--

let's just ask, how many in this room

have had alcoholic drink in the past month?

Yeah.

But very few people have alcohol use disorder.

So the thing is, most of us are able to go have a nice time,

have a drink, relax, and then know to stop.

Because we can actually control this

with our top down, cognitive control from the frontal cortex

to say, enough is enough.

I've had enough to drink.

I realize I'm drinking too much.

So what can we do about this self-control?

And maybe we should-- are there ways--

I'm looking at ways through cognitive training

to enhance different cognitive functions.

And maybe cognitive control is something

that we can work on in that way.

So this is a very unusual study.

I can't imagine if I put it forward

to the University of Cambridge Ethics Committee what

they would have to say.

But young healthy men were asked to come into the scanner

and look at sexually erotic films.

And half the men were just told to just look at the film

and I guess enjoy it.

And the other half were told to try

to control their feelings about the film or the eroticism

of the film to see what they could do.

And what you found is that in the sexually arousing condition

where they could just look at the films and watch them,

you get the limbic areas are activated.

You're getting the amygdala, hypothalamus, and so forth,

the emotional areas of the brain.

In the inhibitory control condition,

they were able to inhibit to this.

They did not show these neural correlates of sexual arousal.

And instead, they showed activations

in these more frontal areas where you expect

to see cognitive control.

So if your scan shows a lack of self-control ability,

are you likely to commit a crime?

We do know that there is a relationship between inability

to express cognitive control and ability to have

self-control and the relationship between committing

crimes.

Now there's many other complicated issues

associated with that.

But there is a relationship.

And then some peoples say, they use the expression,

my brain made me do it.

And can you be considered responsible for your actions

if you haven't got that ability to express cognitive control

in your brain or if you're a psychopath

and you haven't got the ability to respond to moral actions?

It's also come up in the context of drugs.

We know that people who have substance abuse problems

may be--

particularly with cocaine usage or whatever--

that they start very impulsively maybe using it.

And then eventually the people, some people,

a portion of people-- not everybody

because we know, again, there are

recreational uses of these drugs who don't show substance use

disorder.

But others go on to get substance use disorder.

And at that stage, it becomes very habitual.

And initially, we can see activations

in the reward areas of the brain like the ventral tegmental area

or the nucleus accumbens.

And then later, as it progresses,

we see a more habitual pattern of behavior.

So people are originally taking it because it's reinforcing

and they're impulsively drawn into it.

And then later on, they're taking it just

to avoid the aversive part of not taking it.

So it becomes very habitual and controlled.

And so, we know that areas such as the caudate

are involved in that.

Whereas when we're doing goal oriented or goal

directed behavior, that's more, again,

associated with frontal lobe function.

So there's also a lot of interest in neuromarketing

and consumer neuroscience.

And to what extent you might be able to,

instead of polling lots and lots of people,

just be able to do a small experiment with fMRI

and determine whether somebody likes

this packaging of that packaging or they

like this chocolate or that chocolate.

And it might be a lot cheaper and easier just

to do small neuroimaging studies rather than large marketing

studies.

And of course, obviously now, people

can tell online what people are searching for and looking at.

So it's a lot easier to get out the information about what

people might want to buy.

So the idea is, can it be used for idea

for product or evaluation of the finished product or marketing

strategy.

And there's this very interesting study by Kuhn

where they had six ads.

And they tested them with 18 women in ab fMRI scan study.

And then afterwards, they looked at these 60,000 shoppers

who were recorded what they actually purchased.

So this is a really interesting and clever study

because they did the scanning to find out what

people liked in the scanner.

And then they actually monitored what all these shoppers

bought so they could see how well it related to each other.

And so the actual sales look like this.

So of the six ads, the most popular one

was the one at the top where you're looking

at this sort of chocolate bar.

And then the second most popular was

this lady who seems to be very excited about the chocolate

bar.

And then you have two people enthusing over the chocolate

bar.

And those seem to be the most popular.

They also had this very interesting one

that includes a toothbrush, which I wouldn't have thought

was terribly what you'd think of when you were

going to eat a chocolate bar.

But there you go.

And they sort of interviewed people,

just to see what the people said they liked.

And when they interviewed people to see

what they said they like, they found that this

was the order that they liked.

So I guess it was the socially acceptable thing

to say that you like the one with the toothbrush in it

because that came second.

But you can see, it doesn't accord

very well with the actual sales, that interview technique.

But in contrast, the fMRI did.

And what you see there is that it looks very similar to what

the actual sales were, for the first two,

at least, of the adverts.

And when they looked at the scanner, what they found

was that the areas of the brain that

were involved in reward and things like that

were activated to the adverts, the top two adverts.

And the areas of the brain that were really more

related to rational and cognitive control

were more activated in the lower down categories.

So there's also a company No Lie MRI.

And they claim that they're the first and only direct measure

of truth verification and lie detection.

And they've been trying for a long time

to get fMRI data brought into the courts.

But they haven't been very successful with that.

There has been a very interesting court case

with MRI.

And that had to do with a person who was a gentleman.

He was a very, very nice person and seemed to be fine.

And then all of a sudden, he started making advances

to his partner's daughter.

And she was a child.

And then he was also looking at pornographic material.

And so they sent him away on a course

to try to retrain him and get him to not

show these unacceptable sexual urges.

And apparently, he started bothering all the nurses

who were on the program.

So he got kicked out of that program.

And he was going to be sent to court and tried.

And he actually started to have headaches

and started to stumble and fall.

And they brought him into the hospital

and they found he actually had a tumor in his frontal lobe.

So they removed the tumor.

And after that, he went back to his normal behavior

of being quite a nice gentleman and everything else.

And then slowly over the course of time, the tumor grew back

and the behavior started again.

So they could actually show that the behavior was more or less

directly linked to the tumor growth in that case.

And that has been used.

But really fMRI has not been used yet.

Although it might be used in the future to mitigate problems,

mitigate the jail sentence or something like that.

At the present, with neuroimaging,

it's better than polygraphs.

It's about 87% correct when it's done.

But of course, you have to often give people

many trials to know that it's working properly

and things like that.

And they're claiming now that it's even higher than that.

So they're trying to get right up to 99% percent correct

using the technique.

So it will be interesting to see if they can do that.

So just a little bit of the practical limitations

and ethical implications.

So one of my PhD students was Adrian Owen.

And he's done some really excellent work

on the vegetative state patients and imaging them.

And what he did was he just sent a scan of his brain around

while he was lying in the scanner doing something.

And he asked 10 of the top neural imagers

if they knew what he was doing in the scanner.

And he essentially got about 10 different answers

about what this was all about that he

was doing in the scanner.

And what he was doing was telling a lie, he said.

So without context on an individual,

it's very difficult to be able to tell what they're actually

doing, what's going on.

Most of our studies to date our group studies.

And as you know, we're now using fMRI feedback so that we

can say, for instance, with problems like with depression

or anxiety, we can try to train our top down control

over the emotional brain.

So we're getting instant feedback.

And so we can try to regulate our brains.

So it's becoming more individualized.

But most of the studies that are in the literature

will be group studies.

So for any one person, it will be different.

And our brains are slightly different.

So for any one person, the same area

may be slightly not quite where it is for the group as a whole.

And this is something that I think is very exciting.

And I suspect that actually you may be working on things

like this here at Google.

But this is work by Jack Gallant's laboratory

where he trained people looking at movies.

And then he recorded the brain activation.

And then that was used as a training session.

And then they saw new movies.

And they had to, using that training information,

they had to figure out what they were looking at reconstructed.

And as you can see here, it's a bit fuzzy

and things don't look 100%.

I don't know if you couldn't see the pictures above which

are the films they are looking at,

whether you realized that those below are actually associated

with it.

But they sort of have some credibility to them.

And I think they're getting better and better at that.

But they did have to use a rather long training session.

And so it still requires a heck of a lot of work

before we'll be able to reconstruct what people

are looking at from fMRI.

And this was a very early study by John-Dylan Haynes,

and a very exciting one at the time.

Because although it doesn't sound

like the stuff of movies or films,

it was trying to decode what somebody was intending to do

before they did the action.

So that was rather interesting.

So it was very simple and elegant design

where you had people lying in the scanner.

And he gave them some numbers, two numbers.

And then they had a time period where they could decide,

am I going to add these two numbers or am

I going to subtract them.

And then when the response opportunity came up,

there were four opportunities.

One was if you added them, that's the answer you get.

One was if you subtracted them, that's the answer you get.

And then there were two that were incorrect, but not too far

away from the correct answers.

So during that delay between the response and the task stimuli,

you could actually measure and record

what people were intending to do.

And of course it was very--

the media was very interested in this when it first came out.

Because the idea that you could eavesdrop on thoughts

or figure out somebody's intention in advance

before they made a response was very exciting.

It still is, actually.

Although the adding and subtracting isn't too exciting.

But again, this required that areas like in the frontal lobes

were very much involved in this.

Ventral medial prefrontal cortex was a key area

for this type of activity.

But you could actually see whether somebody

was going to subtract or add.

So as I said, one of the limitations

is that most of the studies are group studies.

So for any individual, it might not hold.

And that you have to interpret scans within a context usually.

And so it's not really visual proof.

They are associations.

And they're just snapshots of what's going on,

We're getting more and more elaborate equipment.

So when I first started doing neuroimaging studies,

we had 1.5 Tesla magnets in the scanners.

And then we got 3 Tesla magnets.

And most of my research was probably done on the 3 Tesla.

And now we have 7 Tesla.

We're still struggling to do all the maths and physics

associated with those.

And there's also problems where you

don't get particularly good images

in the orbital frontal cortex, which

is a very important area, because of the bones and things

by the eyes and whatnot.

So there's a lot of work that needs

to be done to sort of be able to work with those.

But it's also raising questions.

Because issues to do with, you often ask healthy people,

come in the scanner and do this task.

And of course they are healthy.

And they think they're healthy.

And they get in the scanner and then

you might see what they call unintentional consequence.

You might see that they actually have

some difference in their brain, some problem with their brain.

And usually that's catered for by when

they sign ethical approval to be in the study,

they're usually told that if there's

anything of significance that the radiographer finds

that they will inform your GP.

So there's usually a procedure for that

that can be dealt with.

But when we're getting up to these 17 magnets,

they are quite exciting in many ways.

Because multi-infarct dementia has to do with just

like little mini-strokes in the brain.

And they were very hard to see on a 1.5 machine.

They were clearer on the 3 machine.

But on a 7, they're going to be very, very clear.

So just even figuring out, well, when

are these mini-strokes of a nature that somebody might have

multi-infarct dementia and when is it

just the normal aging process will be quite something

to determine?

And we can only do that, really, with large normative samples.

So there's lots of interesting issues that will be coming up.

And we're now using very large databases and very

elaborate computational methodology

to analyze the combination of genetic data

with neuroimaging data with phenotyping data of the person.

And so that will all be very exciting.

I'm involved in a study now with the Medical Research Council

funded, Deep and Frequent Phenotyping Study.

And it's the largest study of looking at biomarkers

for people who have memory problems and the elderly,

and tracking them and following them up and to the point

where some of them will be developing Alzheimer's disease.

And the CANTAB test which I've co-invented

will be used in that study.

But it's just very interesting because we'll

have these very large data sets which will include data

from PET imaging, MEG imaging, fMRI imaging, MRI

imaging, lumbar puncture, and everything else.

And we'll have to put all this together.

So there's a real need for more work with these large scale

computational databases that we'll be combining.

So finally, some people have been very concerned

about DARPA and the experiments that they do in America.

And they're worried that they're going

to want to monitor everybody.

And part of which, they're probably already doing.

As far as cognitive enhancing drugs go, even in World War II

they were using amphetamines for people

who were flying planes and things like that to keep

them awake and alert at night.

And now they're using other drugs

like Ritalin or modafinil.

So they also have interests in whether or not

you can determine what somebody is going to do in advance,

whether you can tell their intention

from these techniques.

So the question is, should we attempt to predict behavior.

And what impact will this have on our legal system

in the future?

Will we be bringing fMRI studies into the courtrooms?

Or will we be just using them to mitigate and say, well,

this person doesn't have very good self-control.

So that needs to be worked on.

But obviously, they did commit this crime.

So they have to account for that.

And then what is the risk and benefit of these technologies?

Who should have access to them?

A lot of the companies don't really

have any regulatory control in terms of the ethical studies

that they do.

Whereas if I do a study at the University of Cambridge,

I have to go through an extensive procedure

where it has to be approved by an ethical committee

before I can do the experiment.

And they check everything from the consent forms

to what the study is and everything else.

So we need to think a little bit about that.

And then I mentioned the incidental findings,

the unintentional findings that you get.

Over here, because we have NHS, it might not matter.

But maybe in America, you find out

things that will affect your insurance and so forth.

So these are important.

And then finally, there are these exciting experiments.

This one was done by my PhD student Adrian Owen some years

ago after he had qualified.

And basically, he was able to show

that these people in the vegetative state

actually were able to understand what was said to them

and respond to it.

So what they basically did was to get

these people to think about-- they asked them

to think about playing tennis.

And they got activity in areas like the supplementary motor

area because they were thinking about running

around and playing tennis.

And then they asked them to think about moving

through their house.

And they got activation in the hippocampal formation

because that's the area that does

with the locations of places in space.

So some of you will have heard about the interesting studies

by Eleanor Maguire where she looked at taxi drivers.

And they have the knowledge, as you know, of the city.

They have to learn where everything is.

And so they have to learn the location in space of all

these places and remember it.

And she actually demonstrated that the volume

of the hippocampus was greater depending on the years

that you were a taxi driver.

So it's very interesting from that point of view.

So spatial navigation was associated

with the hippocampal formation areas,

parahippocampal I believe.

And then the other was the supplementary motor area.

And more recently, they've been trying

to use similar techniques, not the same,

to look at brain-computer interfaces

and to be able to ask these locked in patients

yes and no questions.

And they were able to find bimodal differences.

But it was only at about 70%.

So I guess 50% would have been chance.

So it wasn't as perfect as you'd like.

But it was one way to start to get them to--

they train them on a lot of yes and no questions.

And then they could ask some new questions

that had to do with yes and no, and figure out

whether they were accurate.

So these techniques are very powerful.

And they may allow us to communicate with people

that otherwise, in the hospital, we

wouldn't be able to communicate with.

So I kind of want to finish off in an upbeat way with five ways

to mental wellbeing.

And I always like to promote exercise because it's

good for your brain.

It increases neurogenesis and new brain cells

when you exercise.

And it also improves your mood and improves

your physical health.

So that's a great way for mental wellbeing.

These are based on the UK government Foresight

Project on Mental Capital and Wellbeing

that I was involved with as a lead.

And then mindfulness, many of you

may be doing mindfulness already.

But it's very helpful to be focusing and aware of things.

And it helps your mind and well being

to be attentive about the environment in a positive way.

And then keep learning, education also

is extremely good for us.

It increases our cognitive reserve.

So if anything does happen to you later in life,

you'll respond better than other people.

It.

Instills resilience, essentially.

And then connect, social communication

is very important, to have social support systems.

And then I wanted to get to this last one.

This, by the way, is the UK government Foresight Project

on Mental Capital and Wellbeing from the "Nature"

paper, Bennington et al, which I was involved in.

And it's interesting to look at because all these features

in blue here, pointing upwards, so what

improves our cognitive health, our cognition

and our mental well being.

And they're such things as good education and a good family

environment.

And then all the things pointing down in red

are things that detract or take away

from our good cognitive function and good mental wellbeing.

And there's such things as stress and substance abuse.

And we took a lifespan approach to that.

And the final thing is that, interestingly,

giving is very rewarding for the individual.

And it's a good thing to do.

If you donate to societal causes,

you get activations in the reward system,

this ventral tegmental striated network involved in dopamine.

So that's the dopamine network.

And so this is just to show you that there's actually

been studies where they've looked at romantic love.

And it activates the reward system.

So this may explain why when we first fall in love,

we're in the sort of interesting state with all this dopamine

and activation.

And the VTA and caudate nucleus are all

involved in this reward and motivation.

So this is probably what's going on.

So now we can check whether you're really

in love with somebody or not.

And interestingly, as I said, giving to charity is also good.

So especially if it's your favorite charity, not just

any charity.

But if it's your favorite charity,

people have done studies where they've

looked at you can give money to your favorite charity.

And then when you donate, you can see,

again, you'd get activations in the neural network

that's involved in both reward and also social attachments.

So as you'd expect, you'd find this social affiliative areas

of the brain as well.

So again, the ventral tegmental area

is lit up there, as you can see.

So really just to conclude, I mean we're really lucky

and it's very exciting to have this new technology.

And we're able to do more and more.

And now we're able to combine using some of these machines

with actually identifying neurotransmitter systems

and being able to work with them and find out how

they might alter our thinking.

But also in terms of targets for new drugs for treatments

for mental health are also very beneficial.

So they're going to be fantastic in helping

us to both understand the healthy brain

and also for understanding what goes wrong

when people have a psychiatric disorder

or neurological disorder and the brain is affected.

But I think it's important that we

keep discussing the implications of these studies

so that we know how they affect ethical issues

and also how they affect society.

So thank you very much.

And I hope I get another chance to come back

and talk about some of the other work

that I do on cognitive enhancement using

apps and drugs.

Thank you.

[APPLAUSE]

The Description of "Sex, Lies, and Brain Scans: How fMRI Reveals What Really Goes on in Our Minds" | Talks at Google