WEBVTT

00:00:00.780 --> 00:00:04.631
>>Behind every innovation
is an inspired scientist.

00:00:04.631 --> 00:00:08.891
And behind every PBS program
is a viewer like you!

00:00:08.891 --> 00:00:11.152
Get behind your PBS station.

00:00:11.922 --> 00:00:15.070
Your support makes a difference!

00:00:15.530 --> 00:00:18.550
>>Hello and welcome to
Scientific American Frontiers,

00:00:18.550 --> 00:00:19.880
I'm Alan Alda.

00:00:20.451 --> 00:00:24.039
we all like to think that we
understand our own minds.

00:00:24.039 --> 00:00:27.958
That we carefully weigh the pros and cons
when we make a decision.

00:00:27.958 --> 00:00:29.699
And that after we've made it,

00:00:29.699 --> 00:00:33.870
even if we have cause to regret it, we
know why we decided the way we did.

00:00:33.870 --> 00:00:37.923
Oh sure we know some choices
are more emotional than rational,

00:00:37.923 --> 00:00:40.285
but even then, we think
we're conscious

00:00:40.285 --> 00:00:44.357
of all the conflicting arguments that run
through our minds as we make a choice.

00:00:45.371 --> 00:00:49.868
Well, in this program we'll find out
how utterly deluded we are,

00:00:49.868 --> 00:00:55.650
as we join in experiments revealing just how
sneaky and underhanded our brains can be.

00:00:56.493 --> 00:01:02.992
[music]

00:01:04.159 --> 00:01:06.508
>>I've always thought
of myself as a feminist,

00:01:06.508 --> 00:01:09.950
so I'm pretty sure I know how I'll
do in this test of my reaction

00:01:09.950 --> 00:01:11.748
to women in the work place.

00:01:11.748 --> 00:01:13.498
I am ready to begin.

00:01:13.498 --> 00:01:19.258
Women, in fact like Mahzarin Banaji, who's
a professor here at Harvard University.

00:01:19.258 --> 00:01:22.499
The test is called the
"implicit association test"

00:01:22.499 --> 00:01:24.568
and it begins simply enough.

00:01:24.568 --> 00:01:26.657
I have to pair the word in the center,


00:01:26.657 --> 00:01:30.340
with one of the words above by
pressing the "e" key with my left hand

00:01:30.340 --> 00:01:32.695
or the "i" key with my right.

00:01:32.695 --> 00:01:34.875
Well, that could be either way.

00:01:34.875 --> 00:01:37.405
Mahzarin has told me to
do this as fast as I can

00:01:37.405 --> 00:01:42.146
because it's the time I take to make the
associations that's critical to the test.

00:01:44.280 --> 00:01:48.560
Now the target words have changed,
but the task remains the same;

00:01:48.560 --> 00:01:52.824
to quickly decide whether the
new words belong to the left or the right.

00:01:56.004 --> 00:01:58.286
But things are about
to get trickier.

00:01:59.884 --> 00:02:03.043
>>It's the same thing except now any
one of these four will show up,

00:02:03.043 --> 00:02:05.295
and when it's "career" or "male"
you'll press the left key,

00:02:05.295 --> 00:02:08.315
when it's "family" or "female" you'll
press the right key.

00:02:08.825 --> 00:02:11.464
>>"Career" or "male"
>>Yes.

00:02:11.464 --> 00:02:14.945
>>So now the categories are
described by two words,

00:02:14.945 --> 00:02:17.883
making it harder to decide
where the new words belong.

00:02:18.783 --> 00:02:23.273
But since historically, "male" and
"career" have gone together....

00:02:23.273 --> 00:02:27.394
This is like, reinforcing this stereotype.
>>Yeah, that's it exactly.

00:02:27.954 --> 00:02:32.595
>>The point of the test is to discover
if lurking beneath my feminist convictions,

00:02:32.595 --> 00:02:36.866
I actually harbor a hidden bias
against women in the work place

00:02:36.866 --> 00:02:41.665
based on all the associations between
"man" and "career" and "women" and "family"

00:02:41.665 --> 00:02:44.159
that bathed the culture
I grew up in.

00:02:44.159 --> 00:02:49.009
"Family or "male", "career" or "female".
Okay, now you're testing me.

00:02:49.199 --> 00:02:53.600
The implicit association test
is designed to ferret out any bias

00:02:53.600 --> 00:02:58.389
by seeing if it takes me fractionally longer
to figure out where a word belongs

00:02:58.389 --> 00:03:02.989
when the pairing of the target words, in
this case: "family" and "male" together

00:03:02.989 --> 00:03:08.578
and "career" and "female", is slightly
more difficult for my brain to except.

00:03:08.578 --> 00:03:12.939
Corporation - "family"
or...[breaks off in laughter]

00:03:12.939 --> 00:03:16.379
"brother", "management"...

00:03:17.780 --> 00:03:19.251
and we're done.

00:03:20.872 --> 00:03:24.502
>>Slight. I could tell he has a slight.
This is very good.

00:03:24.502 --> 00:03:26.952
You are...so let's see,
you're showing a slight,

00:03:26.952 --> 00:03:29.551
automatic association
between male and career,

00:03:29.551 --> 00:03:34.903
and the...only 12% of the
population that takes this test

00:03:34.903 --> 00:03:38.102
shows this bias. What you're
seeing is that you're,

00:03:38.102 --> 00:03:41.071
you're showing a much smaller bias
then what many other people show.

00:03:41.071 --> 00:03:44.320
I'm up here. I make a strong
association between

00:03:44.320 --> 00:03:47.772
"male" and "career" and "female" and
"family" even though that's not what I

00:03:47.772 --> 00:03:49.232
consciously express.

00:03:49.232 --> 00:03:50.673
>>Even though that's
not what you live, huh?

00:03:50.673 --> 00:03:52.331
>>And that's not what I live,
but in my world.

00:03:52.331 --> 00:03:54.781
And I don't know, I'd be
interested to know. Maybe

00:03:54.781 --> 00:03:58.680
many years of working in feminist
causes made you show this lesson.

00:03:58.680 --> 00:04:01.442
>>So all those years of working
in feminist causes

00:04:01.442 --> 00:04:06.490
didn't manage to totally eliminate my lingering
association between "female" and "family"

00:04:06.490 --> 00:04:09.542
and "male" and "career".

00:04:09.542 --> 00:04:12.990
What's astonishing though is that
Mahzarin is far worse,

00:04:12.990 --> 00:04:16.463
and she's not only enjoying
a very successful career-

00:04:16.463 --> 00:04:18.233
she designed the test!

00:04:18.894 --> 00:04:21.854
Another of its designers
is Brian Nosek,

00:04:21.854 --> 00:04:24.473
who personally developed
the test he's taking now,

00:04:24.473 --> 00:04:27.594
intending to reveal hidden
racial prejudice.

00:04:27.594 --> 00:04:29.864
>>These are the same faces
you've seen many, many times.

00:04:29.864 --> 00:04:33.633
>>Many times, many times. In fact I helped
create the faces, so...

00:04:33.633 --> 00:04:36.974
>>The heart of this test is to see whether
it's easier for Brian to match words

00:04:36.974 --> 00:04:40.743
or pictures to the pairing
of African American with bad,

00:04:40.743 --> 00:04:43.282
and European American with good.

00:04:43.742 --> 00:04:45.934
Then when the pairings
are reversed,

00:04:45.934 --> 00:04:50.424
European Americans with bad and
African Americans with good.

00:04:50.985 --> 00:04:55.445
[quiet]

00:04:55.445 --> 00:04:57.533
>>I can predict.
>>Yeah [chuckle]

00:04:57.533 --> 00:04:58.583
What do you think?

00:04:58.583 --> 00:05:00.904
>>Moderate. Moderate to strong.

00:05:00.904 --> 00:05:03.455
>>Yeah, I'll say moderate as well.
>>Yeah.

00:05:03.455 --> 00:05:06.335
>>Strong preference for white.
[exclamation of laughter]

00:05:06.335 --> 00:05:09.205
>>You've taken this
for the umpteenth time,

00:05:09.575 --> 00:05:14.055
and you still haven't caught on
to the fact that you're a little biased?

00:05:14.055 --> 00:05:15.975
[laughter]

00:05:15.975 --> 00:05:21.526
>>I think part of that is the insistence
that my conscious beliefs still matter.

00:05:21.526 --> 00:05:25.276
It isn't that-- that fact that I
keep showing these implicit biases

00:05:25.276 --> 00:05:28.577
mean that I'm a biased person so
I should just accept that and move on,

00:05:28.577 --> 00:05:32.635
it's that-- no, I don't agree with
those, I do have them

00:05:32.635 --> 00:05:35.065
and I will admit to having
those implicit biases

00:05:35.065 --> 00:05:38.667
but I'm not going to let that rule
what I would consciously want to have.

00:05:39.927 --> 00:05:43.335
>>It really is remarkable that here I am
in Cambridge, Massachusetts--

00:05:43.335 --> 00:05:45.555
perhaps the liberal
capitol of the country--

00:05:45.555 --> 00:05:49.904
with two academics who pride themselves
on their enlightened attitudes,

00:05:49.904 --> 00:05:52.246
only to discover that they
have latent within in them

00:05:52.246 --> 00:05:56.731
biases they would fervently deny
if their own test hadn't revealed them.

00:05:59.581 --> 00:06:01.569
But all may not be lost.

00:06:02.202 --> 00:06:05.794
>>We found in research, and
also in testing of myself,

00:06:05.794 --> 00:06:10.685
that if I put myself in a situation where
I think about positive black exemplars--

00:06:10.685 --> 00:06:14.402
I think about Michael Jordan
and Colin Powell and people...

00:06:14.402 --> 00:06:19.396
Martin Luther King who have had a very positive
impact and who are also African American,

00:06:19.396 --> 00:06:23.153
I show much less bias immediately
after thinking about those exemplars

00:06:23.153 --> 00:06:25.355
then if I hadn't thought
about them before.

00:06:25.355 --> 00:06:29.715
>>Even simple things, like the presence
of an African American experimenter,

00:06:29.715 --> 00:06:34.943
reduces race bias. That the presence
of a competent woman,

00:06:34.943 --> 00:06:38.085
makes women's attitudes to math
become more positive.

00:06:38.085 --> 00:06:40.575
Those kinds of studies

00:06:40.575 --> 00:06:44.515
sit, I think, in contradiction the ways
in which people like I thought about this-

00:06:44.515 --> 00:06:46.654
we thought they were learned
over long periods of time,

00:06:46.654 --> 00:06:49.143
that they were entrenched,
they were rigid, inflexible--

00:06:49.143 --> 00:06:51.224
and that does not appear
to be the case,

00:06:51.224 --> 00:06:53.534
and I think that that's where
the room for optimism is.

00:06:53.534 --> 00:06:56.805
On the one hand, I would like people
to take these data seriously

00:06:56.805 --> 00:07:00.967
when we can show the vast numbers
of people who show the bias,

00:07:00.967 --> 00:07:02.945
I think we need to contend with that.

00:07:02.945 --> 00:07:07.224
On the other hand, what this work
is showing is that it might be

00:07:07.224 --> 00:07:10.755
not that hard to shape environments
that can change these,

00:07:10.755 --> 00:07:13.878
even automatic kinds of attitudes.

00:07:15.188 --> 00:07:19.806
>>You can find out if you harbor unconscious
biases you would deny even to yourself,

00:07:19.806 --> 00:07:23.536
by logging on to the
Project Implicit website.

00:07:23.536 --> 00:07:27.868
There you'll find tests of attitudes towards
such things as age and religion,

00:07:27.868 --> 00:07:33.149
as well as lighter fair like the Harry
Potter movies versus Lord of The Rings.

00:07:33.878 --> 00:07:37.256
Give it a try. You may be in for a shock.