Episode Transcript
Transcripts are displayed as originally observed. Some content, including advertisements may have changed.
Use Ctrl + F to search
0:05
Can you measure pedophilia in
0:07
a brain scan? Can you measure
0:09
a lie from somebody's blood pressure?
0:12
And how should a judge in court
0:14
who's not an expert in science decide
0:17
these things? What does any of this have
0:19
to do with President Ronald Reagan or
0:22
antisocial personality disorder
0:24
or how the television show CSI
0:27
has impacted courtrooms. Welcome
0:32
to Inner Cosmos with me David Eagleman.
0:35
I'm a neuroscientist and an author at
0:37
Stanford and I've spent my career at
0:39
the intersection of our brains
0:41
and our lives. In today's episode
0:44
is about an aspect of
0:47
the intersection between brains and
0:49
the legal system, and it's a tricky
0:51
one. The
0:57
question is when neuroscience
1:00
techniques are allowed
1:03
in courts, When should they be allowed?
1:05
What bars need to be passed
1:08
for a technology to be accepted.
1:11
So let's start on March thirtieth,
1:14
nineteen eighty one, when the President
1:16
of the United States, Ronald Reagan, has
1:18
been delivering a speech and
1:21
afterwards, he and his team are returning
1:23
to his limousine and he gives
1:25
a two big arm wave to the
1:28
crowd and suddenly there
1:30
are gunshots ringing out and
1:33
everyone's diving, and President
1:35
Reagan is hit with a ricochet off
1:37
his limousine, and Press Secretary
1:39
James Brady falls and
1:41
Secret Service agent Tim McCarthy falls,
1:44
and a DC Police officer named Thomas
1:47
del Haunty is also wounded, and
1:49
the President arrives at the
1:51
emergency room in critical condition
1:54
and almost dies. And for
1:56
those of you who weren't alive in nineteen eighty
1:58
one, or for whom this has re seated
2:00
in memory, just try to picture
2:03
the horror that this entailed. Now
2:06
you may remember that the gunman, John
2:08
Hinckley, had a deep psychosis.
2:11
He was divorced from reality,
2:13
and he believed that if he shot
2:16
the president, he would win the
2:18
love of the actress Jody Foster.
2:21
There's a lot to say about this case, and
2:23
in episodes thirty six and thirty seven
2:26
I talked about the insanity
2:28
defense, but here I want to zoom
2:30
in on a very particular aspect.
2:32
The thing most salient to us
2:35
today was the fact that this was
2:37
the first high profile case to use
2:39
a form of brain imaging.
2:42
Hinckley's lawyers pled not
2:45
guilty by reason of insanity, and
2:48
to support their defense, they introduced
2:50
brain imaging evidence so
2:53
his defense counsel argued that
2:56
he was schizophrenic, and they
2:58
argued they could prove this by
3:00
showing CAT scans or
3:02
CT scans. CT stands
3:05
for computer aided tomography
3:07
or computerized demography. Now,
3:09
the lawyers on both sides agreed
3:12
that cat scans had never before
3:14
been admitted as evidence
3:16
in a courtroom. Neuroimaging
3:19
was brand new at this time, So
3:22
should the judge allow this new
3:24
Fengal technology to be
3:26
accepted or not? Well,
3:29
it's not obvious. Can you really
3:31
tell if someone suffers from schizophrenia
3:33
just by looking at an anatomical
3:36
picture of the brain. It's
3:39
not obvious. So the judge decided
3:41
to dismiss the jury so that he
3:43
could hear the arguments about
3:45
whether or not the technology was
3:48
relevant and should be admitted. And
3:51
expert witness, a physician, pointed out
3:53
that Hinckley's soul
3:55
sigh, which are the valleys running along
3:58
the outside of the brain, these were
4:00
wider than average, and this
4:02
physician cited a paper suggesting
4:05
a connection between schizophrenia
4:08
and wider sulsie.
4:10
So the assertion was, if you have
4:12
schizophrenia, you can see
4:14
that just by looking at a cat scan
4:17
of the brain. So this doctor
4:19
said, quote the fact that
4:21
one third of schizophrenic participants
4:24
in the study had these widened
4:26
sulsie whereas in normals probably
4:28
less than one out of fifty have them. That
4:31
is a very powerful fact. But
4:33
the prosecution rebutted
4:36
this. They said, no way, it has not been
4:38
proven that a cat scan
4:40
can aid in the diagnosis
4:42
of schizophrenia, and therefore this
4:45
evidence should not be presented
4:47
to the jury. In other words, they
4:49
argued the technology should be excluded
4:52
from the courtroom because it was not yet ready
4:54
for prime time. The judge
4:57
listened to the arguments and
4:59
he finally decided that he would not
5:01
admit the cat scan. Then
5:04
nine days later he heard more expert testimony
5:07
and he confirmed that he would not take the
5:09
cat scan. And then he
5:11
changed his mind and reported he would
5:13
take the cat scan. Okay, So what
5:16
is this back and forth illustrate? It illustrates
5:19
the difficulty for a judge to
5:21
decide what makes meaningful
5:24
evidence and what does not. In
5:26
the end, Hinckley was found
5:29
not guilty by reason of insanity, although
5:31
that had little or nothing to do with the cat
5:33
scan. But this high profile case
5:36
is just one of hundreds
5:38
where this question comes up, should
5:40
neuroimaging be allowed in
5:42
the courtroom. There's no single
5:45
answer to this question, and in part that's
5:47
because there are many different guyses in
5:49
which it comes up. And so that's what we're
5:51
going to talk about today. We're going to talk
5:53
about how any technology gets
5:56
into courtrooms. So
5:58
to motivate this, imagine
6:01
that we start seeing advertisements
6:03
for a new Silicon Valley
6:05
company that has developed a new mind
6:08
reading technology. They call
6:10
this the Palo Alto three
6:12
thousand, and they strap it to your head
6:15
and they measure some brain waves and
6:17
pass that through a large language
6:19
model, and they print out in words
6:22
to a screen what you are thinking.
6:24
So you might be thinking about wanting
6:26
a hot dog with pickles. In this machine
6:29
will print to the screen, I want
6:31
a hot dog with pickles. Now
6:33
this is totally made up, but pretend it's
6:35
true. In a few years that said
6:38
company launches, and
6:40
let's say the technology looks pretty good.
6:42
It captures the gist of what
6:45
you are thinking about. Now, the question
6:47
is should this be admissible
6:50
in a court of law. Let's
6:53
imagine that someone puts it on
6:55
and states under oath that on April
6:58
twenty fifth, he was getting dinner with his family,
7:00
but suddenly the screen prints out I
7:03
committed the crime. Now, how do you
7:05
know whether to believe that or not? The
7:08
company is started by a
7:10
handful of young people who dropped out of college,
7:13
and they claim to be experts in neuroscience,
7:15
But how do you know whether it really
7:19
works? And especially in a high
7:21
stakes situation, should you accept
7:23
this in a court of law or
7:25
not? Well, some people, in order
7:27
to judge the quality of the technology, they
7:29
ask, well, are they charging for this technology?
7:32
But that's not a meaningful measure.
7:34
Of course they're charging. They can't develop
7:37
new technologies for free, anymore
7:39
than you would expect Apple to not
7:41
charge for their laptop. But the fact
7:44
that they're charging certainly doesn't rule
7:46
in or out anything about its efficacy.
7:49
So how do you know whether the technology
7:52
is efficacious? Can it be used in a
7:54
court of law? How do you know whether
7:57
it works and provides what the legal
7:59
system calls probative value,
8:02
which means can it do what it's supposed
8:04
to do? Can it provide sufficiently
8:06
useful evidence to prove
8:09
something in a trial? So
8:11
this is what we're going to talk about today.
8:14
Most of the time we don't realize that
8:16
new technologies always have to be assessed
8:19
by courtrooms to know whether
8:22
they should be accepted or rejected.
8:24
And some get in and then we take that as background
8:27
furniture, and others never make it. And
8:30
what we're going to see today is how and
8:32
why. So fast forward
8:34
some decades from the Hinckley trial.
8:36
Where are we now? What is
8:38
allowed in the courtroom? Well, we have
8:41
more sophisticated technologies
8:43
to image the brain. Now, for example,
8:46
we can get a picture of the
8:48
brain in an MRII
8:50
scan. Magnetic resonance imaging MRI
8:53
gives you a snapshot
8:55
of what the brain of a person looks like.
8:58
You're not seeing activity there,
9:00
You're just seeing the anatomy. Think
9:02
of this an analogy to the way you would
9:04
look at someone's skeleton with an X ray.
9:06
You can't see anything moving around what
9:08
you see as a snapshot. So
9:11
with MRI you can hope to see
9:13
abnormalities like a tumor
9:16
or evidence of a stroke, or the
9:18
consequences of a traumatic brain
9:20
injury. Now, I've been
9:22
called by many defense lawyers over
9:24
the years who say, I have a client
9:26
who's going up for trial. Can you take
9:28
a brain scan and see if you
9:30
can find something wrong with their brain,
9:33
so this can serve as a mitigating
9:35
factor. But I always tell them the same
9:37
thing. If you find something wrong with your
9:39
client's brain, that can serve
9:41
as a double edged sword. The duray
9:44
might think, Okay, I'm convinced there's something different
9:46
about this man's brain. But this
9:48
presumably means he'll be predisposed
9:51
to committing this kind of crime again,
9:54
so we should probably lock him up for a longer
9:56
time. So a defense lawyer has
9:58
to utilize this argument
10:00
with care. In any case, what
10:03
MRI gives you is an anatomical
10:05
snapshot. And now I want to tell
10:07
you about the next level of technology called
10:10
fMRI. Where the F stands
10:12
for fancy MRI. Okay, I'm kidding.
10:15
It stands for functional magnetic
10:17
resonance imaging fMRI. And
10:20
this is because it's telling you about the
10:22
function of the brain. It's
10:25
measuring blood flow to show
10:27
you where the activity in the brain just
10:30
was. This works because when brain
10:32
cells are active, they consume energy,
10:35
and the blood flow to that specific
10:37
region needs to increase so
10:39
that you can bring fresh oxygenated
10:41
blood to the area to restore
10:43
the used energy. So in fMRI,
10:46
we see where the new oxygenated
10:49
blood is going and we say, aha,
10:51
there must have just been some activity
10:53
there a few seconds ago. So
10:56
that's the difference between an anatomical
10:58
snapshot or a functional picture
11:00
of what's going on. Now. Part
11:03
of the reason that you can use the static
11:05
snapshot the MRI in court
11:08
is because it's generally seen as
11:10
hard science. This is the guy's brain.
11:13
But when we're talking about fMRI,
11:16
what we're looking at is the activity in the brain, and
11:18
we're generally asking something about the
11:20
person's mental state, and
11:22
can that be the same kind
11:25
of hard science. On the one hand,
11:27
it's a clear question with a clear answer
11:29
if someone has a stroke or a brain tumor.
11:32
But this isn't the case if you
11:34
want to pose a question like did
11:36
this defendant intend to
11:39
kill the victim? fMRI doesn't
11:41
and can't give you clear answers
11:44
like that to questions that are useful
11:46
for the legal system. So we're going to dig
11:48
into this now. So
12:03
first let's start with the question of
12:05
whether fMRI has
12:08
been used in courts. The answer is yes,
12:10
But the technology can be used in different
12:12
ways. It doesn't always have to involve an
12:15
individual's brain, but can sometimes
12:17
be about brains in general.
12:20
So let me give you an example. There was
12:22
a murder case in Missouri where
12:24
a young man named Christopher Simmons
12:27
broke into the home of a woman named
12:29
Shirley Crook. He covered
12:32
her eyes and mouth with duct tape,
12:34
he bound her hands together, and
12:36
then he drove her to a state park and
12:39
threw her off a bridge
12:41
to her death. Now, this was a
12:43
premeditated crime, and the evidence
12:46
was overwhelming, and he admitted to the murder.
12:48
So the judge and jury handed down
12:51
a death sentence. But
12:53
there was a complication. Christopher
12:55
Simmons was only seventeen years
12:57
old at the time he committed the crime.
13:00
And so this case spun all the
13:02
way up to the United States Supreme Court,
13:05
and the question was can you
13:07
execute someone who was under
13:09
the age of eighteen when they committed
13:11
the crime? After all, the argument
13:14
goes, adolescence is characterized
13:16
by poor decision making, and young
13:18
people should have the chance to grow
13:20
up into a different life. Well,
13:23
one of the things that happened at his trial
13:26
is that the Supreme Court considered
13:28
fMRI evidence. Now this wasn't
13:30
from Simmons's brain in particular, but
13:33
from adolescence in general.
13:36
The study compared young people
13:38
and adults performing the same cognitive
13:41
tasks, and what the researchers found,
13:44
not surprisingly, is that young brains
13:46
are not doing precisely the same thing as
13:48
older brains. There are measurable
13:51
differences. A juvenile's
13:53
brain just isn't the same thing as an
13:55
adult. So the Supreme Court
13:57
justices saw this evidence,
14:00
considered it, and presumably this is part
14:02
of what led the court to conclude
14:04
that it is unconstitutional to
14:07
execute someone for a crime who is a
14:09
minor. Now that's an example
14:12
of fMRI making it into the
14:14
court. It's been used in this way
14:16
to compare groups of people,
14:18
juveniles versus adults in this case.
14:21
But things get a little trickier
14:24
when you're trying to say something about an
14:26
individual's brain, the brain of
14:29
the one guy standing in front of the bench.
14:32
So what can we and can we not
14:34
say with the technology? So let's zoom
14:36
in on a few examples. Many
14:39
researchers and legal minds have been asking
14:41
whether one can use brain imaging
14:43
to diagnose whether someone has antisocial
14:47
personality disorder, which is a condition
14:49
in which a person has a long term
14:52
pattern of manipulating, exploiting,
14:54
and violating other people people
14:56
with antisocial personality
14:58
disorder or a SPD. They'll
15:01
commit crimes, they'll flaunt rules,
15:04
they'll act impulsively and aggressively,
15:06
they'll lie and cheat and steal. Now,
15:10
this is a condition that is massively
15:12
overrepresented in the prison population.
15:15
But biologically it's not obvious
15:18
what it's about. There's no single gene
15:20
here, and there's not a single environmental
15:22
factor. It's a complicated combination.
15:25
And the legal system often cares to know
15:28
whether someone has ASPD
15:31
or not. And so researchers started
15:34
to wonder a long time ago, could you use
15:36
brain imaging to determine
15:38
in some clear categorical
15:40
way, does this person have
15:43
ASPD or not. So, in
15:45
one study, researchers highlighted
15:48
the brain regions that had high
15:50
probability of being anatomically
15:53
different between people with
15:55
ASPD and those without. And you can
15:57
look in the cortex what's called the
15:59
gray matter, or below the core
16:01
text what's called the white matter, and you can
16:03
measure these small anatomical differences
16:06
between those with and without. So the question
16:08
arose, can you use this technology
16:11
in court as a diagnostic tool
16:13
to say that this person has
16:16
ASPD or not? Now
16:19
do you see any problems with this off the top of your
16:21
head about whether this technology
16:23
can be used. The problem is
16:25
that all the scientific results come about
16:28
from examining groups
16:30
of people, like fifty people in each
16:32
group, and the question is whether
16:35
these group differences are strong
16:37
enough to tell you about individual
16:40
differences. So this is known as the
16:42
group to individual problem. In
16:44
other words, you have data from
16:47
groups of people that can be distinguished
16:49
on average, but you're trying
16:51
to say something about this individual.
16:54
It would be like making an accurate
16:56
statement that men on average
16:59
are taught than women, and
17:01
then asking whether some individual,
17:04
like a tall woman, could be categorized
17:06
as a man because her height clocks
17:09
in at the average mail. The legal system
17:11
is well aware of this grouped
17:13
individual problem, and so as
17:16
technologies are introduced, the
17:18
justice system always needs to ask how
17:21
specific is this technology and how sensitive
17:24
is it? Is it good enough for individual
17:26
diagnosis. Brain imaging studies
17:29
generally just give us group average
17:31
results, and the question is whether
17:34
it tells us enough or anything
17:36
about the person who's standing in front
17:38
of the bench right now. Now.
17:41
The idea of bringing functional
17:44
brain imaging to bear on questions
17:46
of criminal behavior is an old one,
17:48
and this grouped individual problem
17:51
is just as old. For example, there
17:53
was a study in nineteen ninety seven where
17:55
researchers image the brain of
17:58
normal participants and murderers,
18:01
and they found, on average, there was less
18:03
activity in the frontal lobes
18:05
in murderers. So you
18:08
look at the activity in the front of the brain
18:10
behind the forehead and you say, hey, on average,
18:12
there's less going on here in the murderer
18:15
group. But you can't use this on an
18:17
individual. You can't say, oh, this
18:19
person has less activity, so he must
18:21
have been the murderer. In other words, it
18:23
has no power in a court of
18:26
law. You still face the problem of
18:28
trying to say anything about
18:30
an individual from a group average.
18:33
And so it's for reasons like this that
18:35
brain imaging on individuals
18:37
has not gotten very far in courtrooms.
18:40
Let me give one more example. Another research
18:42
group used brain imaging fMRI
18:45
to see if they could identify pedophiles.
18:49
They found twenty four pedophiles
18:51
and thirty four controls, and they
18:53
showed them images of naked
18:56
men and women and boys and girls. And
18:58
what they found is that they could on average
19:02
separate the participants who were pedophiles
19:05
from the participants who were not. In
19:07
other words, the pedophilic brain shows
19:09
a subtly different signature of
19:12
brain activity than the non
19:14
pedophilic brain when shown these
19:16
pictures. It turns out that heterosexual
19:19
versus homosexual seems to be distinguishable
19:21
as well. So you might think that
19:23
sounds quite useful for the legal system,
19:26
but when scientists and legal
19:29
scholars take a closer look, it's not
19:31
as clear. The first question is
19:33
what are these brain signals actually
19:36
measuring. The assumption is
19:38
that it's measuring a state of arousal,
19:40
like sexual attraction, but what else
19:43
might be going on? Well,
19:45
the difference in brain signals could be driven
19:47
by a stress response
19:49
or an anxiety response by the
19:52
pedophilic participants who know they're
19:54
being measured. Or perhaps
19:57
what you're seeing is a measure of disgust
19:59
by the non pedophilic group who
20:01
knows the purpose of the study and doesn't
20:04
like gazing at pictures of children in
20:06
this context. Or what if the
20:08
pedophilic participants were just
20:10
slightly more likely to avert their
20:12
eyes because of shame or
20:15
not wanting to get measured. That would
20:17
cause a statistical difference
20:19
in the brain signals and could, in
20:21
theory, explain the results. So
20:24
there are lots of things that
20:26
could yield this brain imaging
20:29
result of a difference between the two groups,
20:31
beyond the hypothesis that it's
20:33
just measuring arousal, So stress,
20:36
anxiety, discuss shame, all
20:38
these things might be what's getting
20:40
measured here. And part of why this matters
20:42
is because there are many brain imaging
20:45
measures where it turns out
20:47
it's easy to manipulate the results.
20:50
So let's say you are a pedophile
20:52
who doesn't want to be labeled as such. Can
20:54
you purposely move your eyes
20:57
whenever you see a picture of the children,
20:59
and that messes up the ability
21:02
of the scanner to measure something. If
21:04
something can be faked or messed
21:06
up, then the technology is useless.
21:09
But let's say, for argument's sake, that you have a technology
21:11
that can't be faked or manipulated,
21:14
and that allows us to move on to the second point.
21:16
Let's say you don't even care what's getting measured,
21:19
like stress or anxiety or whatever. All
21:21
you care to know is whether there is a
21:23
neural signature that can distinguish
21:25
the pedophiles from the non pedophiles, irrespective
21:28
of what is causing that signal. Well,
21:30
there's also a legal problem here, which is
21:32
that it's not illegal for a person
21:34
to be attracted to children. It is
21:36
only illegal if they act
21:38
on that. All that's illegal is
21:41
whether you have committed a crime
21:43
or not, not whether you are attracted
21:45
to children. So you can think about whatever
21:48
attracts you ostriches or
21:50
jello or whatever, as long as you
21:52
don't commit an illegal act.
21:56
So whether you're talking about ASPD
21:58
or murderers or pedophile while, you'll
22:00
see that measuring something that
22:03
matters for a court of law isn't
22:05
as straightforward as it might
22:07
have originally seemed. So
22:10
now let's return to the Palo Alto three
22:12
thousand. The question is just
22:14
because the company claims
22:16
that it functions, well, how do
22:18
you know whether or not to admit it into
22:21
the courtroom? After all, remember
22:23
what I said about the John Hinckley case,
22:25
how cat scans were admitted into
22:27
the court to argue that he had schizophrenia.
22:30
Well, it's now known that wide
22:32
and salsa in the brain have no relationship
22:35
to schizophrenia. There are other
22:37
better anatomical signatures that we have
22:40
now, like thinner cortices in the
22:42
frontal and temporal lobes and shrunkened
22:44
thalamuses. But it turned out
22:47
that the idea of white and salsa I
22:49
just didn't hold up. Now, there was nothing fraudulent
22:52
going on with the claim. It was just a new
22:54
technology at the time and they
22:56
were doing the best they could with small sample
22:59
sizes. But it turned out the
23:01
theory of white and sul sight was
23:03
scientifically unsound. Remember
23:05
how I mentioned that the judge went back
23:07
and forth several times about
23:09
the issue of whether to accept Hinckley's
23:12
cat scan into the courtroom. That's exactly
23:14
the right thing that should have happened. Not
23:17
all claims are going to be correct just because
23:19
a scientist says so. Despite
23:22
best efforts, science can often be
23:25
incorrect, and that is the
23:27
importance of the scientific method.
23:30
It's always knocking down its own walls.
23:33
So what is a court
23:36
to do about all this? Well,
23:38
let's say that someone wants to introduce
23:40
the Palo Alto three thousand into
23:43
a court case, and you are the
23:45
judge. You have expertise
23:47
in the legal system, but you don't know the details
23:50
of what's possible in neuroscience
23:53
and large language models, and you have
23:55
questions about whether this technology
23:58
should be admitted, questions about whether
24:00
it can accurately read people's
24:02
thoughts, So how do you decide
24:04
whether it should or should not be admitted.
24:21
So let's step back to nineteen twenty three.
24:23
There was a man named mister Fry who
24:26
said that he had developed a lie
24:28
detection technology and it
24:30
relied on a measure of your blood
24:32
pressure, and he wanted to introduce this
24:34
into a court case the
24:37
way you might want to get the Palo Alto three
24:39
thousand into a case. But it turned
24:41
out that mister Fry's claims were
24:43
not widely accepted by anyone
24:45
else in the scientific community, and
24:48
so on those grounds, the
24:50
court decided not to admit
24:53
it into the courtroom. What they said
24:55
was, look, we'll accept
24:58
expert testimony that comes from
25:00
well recognized science, but if there's some new
25:02
technology, it has to be
25:04
sufficiently established so that it's
25:07
gained general acceptance
25:10
in the field in which it belongs. In
25:12
other words, if other experts
25:14
in the field don't believe that
25:17
mister Fry's systolic blood
25:19
measurement is actually good at detecting
25:21
lies, then you can't admit it
25:23
as evidence in the court. And
25:25
that case set the bar for what came
25:27
to be known as the Fry standard,
25:30
which is that technologies need to
25:33
be generally accepted by other experts
25:35
in the field before they can be
25:37
admitted into the courtroom. So under
25:39
the Fry standard, the court would work
25:41
to determine whether the Palo
25:44
Alto three thousand has met
25:46
the general acceptance of the scientific
25:48
community. If science experts
25:51
around the world say, I've never
25:53
heard of this Palo Alto three thousand, I
25:55
don't think that it can actually work, then
25:57
you, as the judge, can
26:00
glued it from admissibility. So
26:02
the court solves the problem by
26:04
deferring to the expertise of other
26:07
people in the field. But this
26:09
isn't the only way to make that decision.
26:11
The Fry standard still is the rule
26:14
in about half the states in America,
26:16
but the rest use a different
26:18
rule to decide whether evidence
26:21
should be admitted, and this is called
26:23
the Dowbert standard. So in
26:25
nineteen ninety three there was a lawsuit from
26:27
a man named Jason Dalbert.
26:30
He was born with severe birth
26:32
defects and his parents brought
26:34
suit against Merrill Dow the pharmaceutical
26:37
company, and they said these severe
26:40
birth defects were caused by the
26:42
medication that the mother was on called
26:44
Bendicton. So the
26:46
pharmaceutical company said the birth
26:49
defects were not caused by this medication,
26:51
and it went to federal court and
26:53
Dalbert said, look, here
26:56
are animal studies showing that
26:58
this drug is related to birth defects.
27:00
And the pharmaceutical companies expert
27:02
witnesses got up and said, look, this
27:05
is not generally accept in the field because
27:07
these are just animal studies and there's no
27:09
conclusive evidence that
27:11
shows the link between these and humans.
27:14
So if you're the judge, how
27:16
do you know how to arbitrate this? It's
27:19
difficult. Right, here's some science from
27:21
the laboratories and here's the pharmaceutical
27:23
company saying it's not generally accepted in the
27:25
field that this causes birth effects. So
27:28
what do you do? Well, what happened
27:31
is the case was decided in favor of the pharmaceutical
27:33
company. So Dalbert took it on appeal
27:36
to the Ninth Circuit and the Ninth Circuit judges
27:38
also awarded this to the pharmaceutical company.
27:41
So Dowbert brought this case to the Supreme
27:43
Court, and the Supreme Court analyzed
27:45
this carefully, and what came out of this was
27:48
a new standard for when evidence
27:50
should be admissible, and that's known
27:52
as the Dalbert standard, and
27:55
the Doalbert standard says, look,
27:57
you accept expert testimony about,
28:00
for example, these labrat studies if
28:02
it will help the jury to understand the evidence
28:05
better or determine the fact
28:07
in issue. In other words, it doesn't
28:09
demand general acceptance
28:12
in the community. Under the Dalbert standard,
28:14
the key is just whether some piece
28:16
of evidence is relevant and reliable.
28:19
Now, the key is that the Fry standard
28:22
made the scientific community the gatekeeper,
28:25
but the Dalbert standard makes the
28:28
judge the gatekeeper. The judge gets
28:30
to say from the beginning that they'll
28:32
evaluate this and ask is
28:34
this evidence relevant and reliable?
28:37
Does it pass my bar for that? So,
28:39
regarding this hypothetical palo alto
28:41
three thousand, the judge might ask
28:44
has the technique been tested in actual
28:47
field conditions as opposed to just in a
28:49
laboratory. Have there been any papers
28:51
on the palab alter three thousand that were
28:53
published in peer reviewed journals? What
28:56
does the rate of error? Do
28:58
standards exist for controlling the
29:00
operation of the machine, and so on? These
29:02
are often difficult questions. It's not always
29:04
easy for a judge to make a decision
29:07
about whether or not to accept a new
29:09
technology. But this gives a pathway
29:12
where the judge is the gatekeeper.
29:14
So let's imagine for a moment
29:16
that the Palo Alto three thousand
29:19
passes the standard for admissibility.
29:22
Is there any reason why the technology
29:25
might still be excluded from
29:27
the courtroom. There is one reason.
29:29
Let's say that you're the defence lawyer
29:31
and you say, Gosh, this thing is
29:34
so stunning that it's going to
29:36
prejudice the jury because they're going
29:38
to look at this fancy technology, and
29:41
even in the absence of really good
29:43
evidence, they'll say, Wow, this
29:45
guy seems guilty. Let's send him to
29:47
the electric chair without considering
29:49
the other points. So
29:52
to prevent that from happening, there's a special
29:54
rule called Federal Rules of Evidence
29:56
four h three, and this just says
29:59
you you should exclude evidence if
30:02
what you can learn from it is substantially
30:05
outweighed by the risk of undue
30:07
prejudice. In other words, does
30:10
it sway the jurors more
30:12
than it should. So what you'll see in
30:14
courtrooms all the time is that if a lawyer
30:16
tries to exclude a piece of evidence
30:19
from being admitted based on let's say a
30:21
Doubert objection, but the evidence
30:23
gets past that, then the lawyer
30:25
is going to take a second bite at the apple
30:28
by calling on federal rules
30:30
of Evidence four three, saying, look,
30:33
even if this is relevant and reliable,
30:35
it's going to have too much sway
30:38
on the jury. So why is this
30:40
an issue? Are there technologies that
30:42
have undue sway on
30:44
jurors? Is that a concern? It
30:47
is? And this brings us back
30:49
to fMRI. In a
30:51
court of law where jurors
30:54
are your neighbors and your community
30:56
and probably not experts in neuroscience,
30:59
a lot of people will be swayed
31:01
by a colorful brain image.
31:03
They're going to put a higher weight
31:06
on this than maybe they should, and possibly
31:08
at the cost of not weighing this evidence
31:11
appropriately in the context of
31:13
the whole case. And this is part
31:15
of the concern that some legal scholars
31:17
have, and this has come to be known as
31:20
the CSI effect. So
31:22
you remember the television show CSI.
31:24
This stood for Crime Scene Investigation,
31:27
and it's a television drama about a team of
31:30
forensic scientists and detectives
31:32
in Las Vegas who use cutting
31:34
edge scientific techniques to solve
31:37
murders. So they go around
31:39
each week and meticulously gather and analyze
31:42
evidence from crime scenes and each
31:44
episode features a complex
31:46
case with an intricate puzzle and the
31:48
CSI team has to solve this
31:51
to bring the criminals to justice. Well,
31:54
the idea with the real life
31:56
CSI effect is that jurors
31:59
come to expect what they've seen on
32:01
TV in terms of magical
32:03
machinery that does something
32:06
like you hit a button to enhance
32:08
the picture and then the computer enhances
32:10
it, and they see everything with clarity, where
32:12
the plot twist requires that the
32:15
investigator pull out some magical
32:17
technology that suddenly solves the crime,
32:20
or looking at the pedophile's brain with neuroimaging
32:22
and knowing whether he did the crime or not. So
32:25
jurors have come to expect this sort of
32:27
thing because you don't spend
32:29
all your time in a courtroom if you're not a lawyer,
32:32
and something like the television
32:34
show CSI is their only window into
32:36
that world. The problem
32:38
is that it often turns out to be a false
32:41
window, and when researchers do
32:43
studies on this, they generally find that
32:46
jurors see neuroimaging
32:48
as the truth of the matter asserted.
32:51
So we just spent a minute on
32:53
looking at the claim that you can measure pedophilia
32:56
and we noted that the brain signals might
32:59
represent that you're a pedophile, or
33:01
it might represent stress or anxiety,
33:03
or disgust or shame or averting the
33:05
eyes or all kinds of things. But that kind
33:08
of nuanced analysis doesn't usually
33:10
get done, and so neuroimaging
33:12
often comes to be interpreted by the
33:14
jury as the truth of the
33:17
matter asserted. This is what scholars
33:19
sometimes call the fallacy of
33:21
neurorealism, and the fallacy is
33:24
just that what you see in these
33:26
pretty false color images is the
33:28
truth. In other words, somebody thinks,
33:30
oh, you're capturing the moment
33:33
of pedophilia in its raw form
33:35
there whereas, of course, the truth is
33:37
that fMRI signals are not
33:40
direct proof of the experience
33:42
itself. As a side note, these
33:44
questions of bringing visual
33:47
evidence into the courtroom, they're not unique
33:49
to brain imaging. They've been around for a long time.
33:52
It goes back at least to X
33:54
rays. So when X rays got
33:56
introduced in the eighteen nineties, they
33:58
immediately started showing up in court
34:01
and everybody was absolutely blown away
34:03
by the idea of being able to
34:05
see inside of a body.
34:07
It's like magic. So what happened
34:09
over a century ago is people asked
34:11
this question of can we use this as evidence
34:14
in court? And the judge said
34:16
at the time, as long as it was scientifically
34:19
reliable, it could be introduced.
34:21
But the same questions about influence
34:24
on the jury came up, because there's a
34:26
real power to seeing
34:28
something. And of course what
34:31
we have currently with brain imaging
34:33
is even a deeper issue because
34:35
it touches on all our notions of
34:37
being human. For example,
34:39
I saw a cover of a Time
34:42
magazine a while ago and the title
34:44
read what makes us Good
34:47
or Evil? And the cover image
34:49
was a huge picture of a brain
34:51
scan, and there was a little picture
34:53
of Mahatma Gandhi with a pointer
34:56
to a part of the brain. And there was a little picture
34:58
of Adolph Hitler with a pointer to
35:01
a different part of the brain. And
35:03
in case you haven't heard my other episodes on this,
35:05
I want to make it clear there is no such thing.
35:08
You can't measure some spot in
35:10
the brain to determine whether
35:12
someone is good or evil.
35:14
And by the way, Friedrich Nietzsche
35:16
wrote about this over a century ago, the
35:19
words good and evil don't
35:22
even represent something fundamental,
35:24
but instead these words end up getting defined
35:27
by your moment in time. What is
35:30
good right now may be seen as
35:32
evil in a century. These terms
35:34
are defined by your culture.
35:37
What you think is good might be seen as sacrilege
35:40
by another group. So the idea
35:43
that you could just measure something in the brain and
35:45
say whether the person is good or evil
35:47
really makes no sense. However,
35:50
millions of people see this kind
35:52
of Time magazine cover, and this
35:55
is why legal scholars worry that
35:57
brain images could be
35:59
persue of past the point that
36:01
they should be in the legal
36:04
argot. This is known as something having undue
36:07
influence. Brain images are influential
36:10
because they take some abstract
36:13
issue like evil intent and
36:15
seem to nail it down to
36:17
the physical. So this is why
36:20
something like Federal Rules of Evidence
36:22
four h three plays an important
36:25
role in asking whether
36:27
something has undo influence,
36:29
whether it sways people more
36:32
than it should now. At the extreme,
36:34
some people say functional brain
36:37
images should never be allowed in the
36:39
courtroom because of their influence. One
36:41
solution that a colleague of mind suggested
36:44
is that you ban the visual
36:46
aspects of brain images from the courtroom,
36:49
so you just have expert witnesses come on to
36:51
the stand and tell you what they
36:53
think is going on as best they can. But
36:55
they're verbally presenting the results, not
36:57
showing them. But these are tough issues,
36:59
right because you can show a gory
37:02
photograph from a crime scene, which
37:04
can also prejudice an entire courtroom.
37:07
Or you can show a reenactment
37:09
of a murder, but if you can't show
37:11
a brain scan, that seems like
37:13
maybe a double standard. So should you
37:15
rule out all visual images
37:18
or allow everything? And
37:20
if you heard episode nineteen, I
37:22
talked about eyewitness testimony
37:25
and how massively swaying that
37:27
is to jurors. You can have all
37:29
sorts of expert scientific testimony,
37:32
but then you have the person get up on the stand
37:35
with tears and a cracking voice and say,
37:37
I don't care what they say. I know that's the
37:39
guy. And we're all moved and influenced
37:42
by that, even though eyewitness testimony
37:44
is so deeply fallible. So
37:47
this is all just to say that the question of undue
37:50
influence always has to be
37:52
asked. Compared to what compared
37:54
to other technologies, compared
37:57
to gory photographs of the
37:59
crime scene, compared to acting
38:01
out a rape scene or a murder scene, do
38:04
those unduly sway a jury?
38:07
So I hope what you see is that These are tough issues,
38:10
perhaps tougher than you
38:12
had intuited at the beginning of the episode,
38:14
So let's wrap up. We
38:17
often think that when a new technology
38:19
comes along, like a new brain technology,
38:22
it always gives useful information,
38:24
and we might assume that courts start leveraging
38:27
it right away. But there are complexities
38:29
around this. For example, in an
38:31
earlier episode, I talked about lie detection. How
38:33
do you know when somebody is actually lying? There
38:36
are lots of technologies that try to measure
38:39
some version of this, but nothing
38:41
can simply tell you the answer because
38:43
the whole concept of a lie is
38:45
complex. Sometimes you might be
38:48
telling the truth but you're factually incorrect,
38:50
for example, because you're honestly
38:53
misremembering how something went, but you
38:55
believe your memory. Or for someone
38:57
else, they might have no associated
39:00
stress response because they just don't care
39:02
that they're lying. So when somebody comes
39:05
to the courts and says, hey, I have a
39:07
new lie detection technology,
39:09
the judge can't just say great,
39:12
bring it to the case, because the judge first
39:14
has to decide whether it should
39:16
be admitted or instead, whether
39:19
its promise will sway
39:21
the jurors more than its value.
39:24
We're all enthusiastic about the next
39:26
stages of technology and being able
39:28
to make important measures about
39:30
what's happening in the brain. But the
39:33
legal system has to be very careful
39:35
about this, whether by standards
39:37
of general acceptance in the scientific community
39:40
or by the choice of the judge's gatekeeper.
39:44
Each new technology has to be weighed
39:46
carefully for admissibility every time
39:49
before it can enter the esteemed
39:52
halls of justice.
40:00
Eagleman dot com slash podcast. For
40:02
more information and to find further
40:04
reading, send me an email at
40:06
podcast at eagleman dot com with
40:09
questions or discussion, and check
40:11
out and subscribe to Inner Cosmos
40:13
on YouTube for videos of each episode
40:16
and to leave comments Until next
40:18
time. I'm David Eagleman, and this is
40:20
Inner Cosmos.
Podchaser is the ultimate destination for podcast data, search, and discovery. Learn More