WEBVTT

00:00:00.080 --> 00:00:01.670
The following
content is provided

00:00:01.670 --> 00:00:03.820
under a Creative
Commons license.

00:00:03.820 --> 00:00:06.550
Your support will help MIT
OpenCourseWare continue

00:00:06.550 --> 00:00:10.160
to offer high quality
educational resources for free.

00:00:10.160 --> 00:00:12.700
To make a donation, or to
view additional materials

00:00:12.700 --> 00:00:16.620
from hundreds of MIT courses,
visit MIT OpenCourseWare

00:00:16.620 --> 00:00:17.275
at ocw.mit.edu.

00:00:26.367 --> 00:00:26.950
PROFESSOR: OK.

00:00:26.950 --> 00:00:28.150
I guess we'll get started.

00:00:29.560 --> 00:00:41.812
Last time, we were talking
about auditory cortex,

00:00:41.812 --> 00:00:47.100
and the tonotopic fields
in auditory cortex,

00:00:47.100 --> 00:00:48.785
the non-tonotopic fields.

00:00:53.040 --> 00:00:56.930
Any questions about that first
lecture on auditory cortex?

00:00:56.930 --> 00:01:01.740
We're going to continue
on cortex today, and talk

00:01:01.740 --> 00:01:07.665
about some areas of cortex in
a specialized mammal, the bat.

00:01:09.020 --> 00:01:10.440
It's where a lot
of excellent work

00:01:10.440 --> 00:01:13.190
has been done on auditory
cortex that really

00:01:13.190 --> 00:01:16.500
shows-- very nicely--
how neurons respond,

00:01:16.500 --> 00:01:21.630
at least, to the selective
stimuli that are emitted

00:01:21.630 --> 00:01:23.750
and listened to by the bat.

00:01:23.750 --> 00:01:26.456
So we'll be talking
about bat echolocation.

00:01:28.810 --> 00:01:32.850
We'll start out by defining
the different groups of bats,

00:01:32.850 --> 00:01:36.340
and talk about who
discovered bat echolocation.

00:01:37.420 --> 00:01:40.330
The discovery was made
just a few miles from here.

00:01:42.200 --> 00:01:47.110
We'll talk about what the
signals a bat's look like,

00:01:47.110 --> 00:01:49.183
in terms of what they look
like on a spectrogram.

00:01:51.820 --> 00:01:54.420
Then we'll talk about
the specializations

00:01:54.420 --> 00:01:58.300
for processing the emitted
pulse and the return

00:01:58.300 --> 00:02:01.510
echo in several bat
cortical fields.

00:02:03.530 --> 00:02:05.520
In the second half
of today's lecture,

00:02:05.520 --> 00:02:08.610
we'll be revisiting
speech sounds.

00:02:08.610 --> 00:02:10.750
We had a little bit of
that at the very beginning

00:02:10.750 --> 00:02:11.450
of my lectures.

00:02:12.590 --> 00:02:14.293
We'll talk about
speech spectrograms.

00:02:15.910 --> 00:02:19.030
And then we'll talk about
cortical processing of speech

00:02:19.030 --> 00:02:22.290
and language, especially
in the human, where

00:02:22.290 --> 00:02:26.445
we have a lot that is known
about processing of language.

00:02:28.780 --> 00:02:29.280
OK.

00:02:29.280 --> 00:02:31.710
So we'll start out
with bat echolocation.

00:02:31.710 --> 00:02:33.470
These are some pretty
pictures of bats.

00:02:34.990 --> 00:02:37.540
Oh, I also have
some announcements

00:02:37.540 --> 00:02:38.820
now that everybody's here.

00:02:38.820 --> 00:02:44.685
So, on Wednesday's class,
meet at the Massachusetts Eye

00:02:44.685 --> 00:02:46.499
and Ear Infirmary, if
you haven't already

00:02:46.499 --> 00:02:47.165
gotten an email.

00:02:48.620 --> 00:02:54.376
So, for a lab tour, we meet at
the Massachusetts Eye and Ear

00:02:54.376 --> 00:02:54.875
Infirmary.

00:02:56.090 --> 00:02:59.370
And there are directions
to get there from here.

00:02:59.370 --> 00:03:03.050
You just get on the Red Line,
going inbound, toward Boston.

00:03:03.050 --> 00:03:05.670
Get off one stop later,
at the Charles stop.

00:03:06.730 --> 00:03:10.310
And then you're going to the
Massachusetts Eye and Ear

00:03:10.310 --> 00:03:10.930
Infirmary.

00:03:10.930 --> 00:03:12.970
So a lot of people get
that confused, of course,

00:03:12.970 --> 00:03:16.161
with the big behemoth right
next door , Mass General.

00:03:16.161 --> 00:03:16.661
So Mass.

00:03:16.661 --> 00:03:18.792
Eye and Ear's a
different building.

00:03:18.792 --> 00:03:19.875
It's a different hospital.

00:03:19.875 --> 00:03:21.530
But But it's clearly marked.

00:03:21.530 --> 00:03:24.980
The directions are on the
website, so just follow them.

00:03:24.980 --> 00:03:30.040
So the lab tour will be within
the regular class period,

00:03:30.040 --> 00:03:31.810
so 2:35 to 4:00.

00:03:31.810 --> 00:03:33.900
We're not going to go
beyond that because people

00:03:33.900 --> 00:03:36.450
have commitments after that.

00:03:36.450 --> 00:03:39.700
And we'll-- depending on
how many people show up,

00:03:39.700 --> 00:03:42.430
it's likely we'll divide
into groups and cycle through

00:03:42.430 --> 00:03:45.870
several demonstrations that I
have prepared for you there.

00:03:47.180 --> 00:03:47.680
OK.

00:03:47.680 --> 00:03:48.950
So questions about that?

00:03:48.950 --> 00:03:50.415
So we meet at Mass.

00:03:50.415 --> 00:03:51.525
Eye and Ear on Wednesday.

00:03:53.810 --> 00:03:56.030
At that time, the
assignments are due.

00:03:56.030 --> 00:03:59.720
So we talked a little bit about
the assignment a few weeks ago,

00:03:59.720 --> 00:04:01.400
when we talked about
the Jeffers model.

00:04:04.160 --> 00:04:07.190
And you can send me the
assignments by email

00:04:07.190 --> 00:04:09.515
or give me a typed
printed version.

00:04:11.170 --> 00:04:14.450
And the idea is that I'll
look them over and then

00:04:14.450 --> 00:04:16.820
hand them back to you
at the review session.

00:04:16.820 --> 00:04:18.820
And we'll talk about
the assignments ,

00:04:18.820 --> 00:04:21.200
and what I consider
the right answers.

00:04:21.200 --> 00:04:24.375
So we did a little switch
for the review sessions.

00:04:25.410 --> 00:04:27.630
So we should put
this on the website,

00:04:27.630 --> 00:04:28.840
if it isn't already there.

00:04:28.840 --> 00:04:32.920
But next week, Monday, we have
two review sessions scheduled.

00:04:32.920 --> 00:04:36.620
The one on Monday will
now be the one on vision.

00:04:36.620 --> 00:04:40.010
And so Doctor Schiller is
going to come and review

00:04:40.010 --> 00:04:42.930
the vision part of
the course on Monday.

00:04:42.930 --> 00:04:45.690
And then I'll be back
a week from Wednesday.

00:04:45.690 --> 00:04:47.880
And we'll do the
audition review.

00:04:47.880 --> 00:04:50.350
And we'll return your
assignments then, ?

00:04:50.350 --> 00:04:50.850
OK?

00:04:50.850 --> 00:04:52.450
So that's what's
happening next week.

00:04:54.230 --> 00:04:55.490
Any questions on that?

00:04:58.130 --> 00:04:58.630
OK.

00:04:58.630 --> 00:05:00.932
So here are the nice
pictures of bats.

00:05:00.932 --> 00:05:02.015
They're beautiful animals.

00:05:05.310 --> 00:05:07.850
They have specializations,
of course, for hearing.

00:05:07.850 --> 00:05:12.110
They have large pinnae, right?

00:05:12.110 --> 00:05:15.010
Much, much larger
than other animals,

00:05:15.010 --> 00:05:16.405
especially for their size.

00:05:18.440 --> 00:05:21.670
They have very small eyes.

00:05:21.670 --> 00:05:24.365
And their visual systems
are not well developed.

00:05:26.420 --> 00:05:27.710
They of course have wings.

00:05:27.710 --> 00:05:29.095
So these are flying animals.

00:05:30.620 --> 00:05:33.310
And many of them
have noseleaves.

00:05:34.780 --> 00:05:38.350
So here's a nose
cartilage that's

00:05:38.350 --> 00:05:43.540
very well developed because
these animals emit sound.

00:05:43.540 --> 00:05:45.710
Their echolocation
pulse is emitted.

00:05:45.710 --> 00:05:47.714
And some of the sound
comes out of the mouth.

00:05:47.714 --> 00:05:49.255
But some of it comes
out of the nose.

00:05:50.400 --> 00:05:55.400
And this noseleaf tends
to focus the sound forward

00:05:55.400 --> 00:05:58.760
because that's where the bat
is interested at detecting

00:05:58.760 --> 00:06:03.290
some kind of a target,
like the insect prey

00:06:03.290 --> 00:06:04.880
that most of these bats eat.

00:06:04.880 --> 00:06:09.520
So I should backtrack
and say that we're really

00:06:09.520 --> 00:06:14.605
talking about three
types of bats.

00:06:16.460 --> 00:06:19.790
We're talking about echolocating
bats, of which there

00:06:19.790 --> 00:06:24.110
are two varieties that I'll
tell you about it in a minute.

00:06:24.110 --> 00:06:30.035
And we're also not going to talk
about non-echolocating bats.

00:06:37.070 --> 00:06:40.060
And sometimes, these
non-echolocating bats

00:06:40.060 --> 00:06:43.985
are called fruit-eating bats.

00:06:46.710 --> 00:06:48.395
They are also flying mammals.

00:06:49.990 --> 00:06:51.990
But they have big eyes.

00:06:51.990 --> 00:06:54.740
They have relatively
small pinnae.

00:06:54.740 --> 00:06:58.210
And they navigate around
like birds and other mammals,

00:06:58.210 --> 00:06:59.870
using their visual system.

00:06:59.870 --> 00:07:01.050
So they don't echolocate.

00:07:03.630 --> 00:07:05.380
So you have
non-echolocating bats

00:07:05.380 --> 00:07:07.280
that we're not
going to talk about.

00:07:07.280 --> 00:07:10.350
We have echolocating bats
that we will talk about.

00:07:11.920 --> 00:07:14.380
It's starting to get confusing
with all these groups.

00:07:14.380 --> 00:07:18.630
In fact, bats are a very
successful group of mammals.

00:07:18.630 --> 00:07:20.780
Supposedly, there
are more species

00:07:20.780 --> 00:07:24.180
of bats than all other
mammals combined.

00:07:24.180 --> 00:07:27.530
It's and amazingly
successful group of mammals.

00:07:27.530 --> 00:07:32.010
And mostly because
echolocation has opened up

00:07:32.010 --> 00:07:34.490
a whole new vista for bats.

00:07:34.490 --> 00:07:38.720
Not only can they fly around,
but they can do so at night--

00:07:38.720 --> 00:07:41.650
in total darkness--
and find prey,

00:07:41.650 --> 00:07:43.910
their targets, their insects.

00:07:43.910 --> 00:07:46.700
So instead of
being fruit eating,

00:07:46.700 --> 00:07:48.753
these echolocating
bats are carnivorous.

00:07:52.260 --> 00:07:55.410
Most of them eat insects
that they catch on the wing.

00:07:55.410 --> 00:07:57.510
So flying insects.

00:07:57.510 --> 00:08:00.430
But there are
gleaning bats that eat

00:08:00.430 --> 00:08:01.955
insects on the forest floor.

00:08:03.200 --> 00:08:08.220
There are vampire bats
that cut little holes

00:08:08.220 --> 00:08:12.470
in the top of mammals and lap
up the blood that comes out.

00:08:12.470 --> 00:08:15.790
There are fish eating
bats that eat fish.

00:08:15.790 --> 00:08:18.615
There are a whole
variety types of bats.

00:08:18.615 --> 00:08:21.270
But most of them eat insects
that they catch on the wing.

00:08:22.410 --> 00:08:24.560
And we'll have a
demonstration of that.

00:08:24.560 --> 00:08:26.490
So these are insect
eating bats here.

00:08:29.240 --> 00:08:32.830
This one I want to
point out its name for.

00:08:32.830 --> 00:08:36.830
This one is called-- in the
middle left, right here--

00:08:36.830 --> 00:08:38.620
megaderma lyra.

00:08:38.620 --> 00:08:40.220
So mega means big.

00:08:41.260 --> 00:08:43.120
Derma means skin, ?

00:08:43.120 --> 00:08:43.620
right?

00:08:43.620 --> 00:08:45.805
It's so named because
of it's big skin here.

00:08:47.160 --> 00:08:50.400
And lyra refers to
lyrical, or musical,

00:08:50.400 --> 00:08:52.250
or something that sings, OK?

00:08:52.250 --> 00:08:53.450
So these bats are singing.

00:08:55.340 --> 00:08:58.380
Let's look at the types
of singing that they do.

00:09:00.290 --> 00:09:07.240
And this display
shows the two types

00:09:07.240 --> 00:09:09.250
of signals that are
emitted by the two

00:09:09.250 --> 00:09:12.120
big groups of echolocating bats.

00:09:12.120 --> 00:09:14.950
The first I want to start
with is the simplest.

00:09:14.950 --> 00:09:16.555
It's called an FM bat.

00:09:17.790 --> 00:09:22.740
An FM-- you have
an FM radio that

00:09:22.740 --> 00:09:35.010
stands for frequency
modulated, or Fm.

00:09:37.310 --> 00:09:37.860
OK?

00:09:37.860 --> 00:09:47.420
And in this graph of the FM
bat's echolocating signal--

00:09:47.420 --> 00:09:50.670
this graph is called
a spectrogram--

00:09:50.670 --> 00:09:52.911
and it plots the
frequency on the y-axis

00:09:52.911 --> 00:09:57.760
and as a function of
time on the x-axis.

00:09:57.760 --> 00:10:01.970
And this echolocating
pulse is the thing

00:10:01.970 --> 00:10:03.229
that the bat is emitting.

00:10:03.229 --> 00:10:04.395
It's producing and emitting.

00:10:05.440 --> 00:10:08.070
And the reason it's
frequency modulated is it

00:10:08.070 --> 00:10:11.420
starts at a high frequency
and modulates down

00:10:11.420 --> 00:10:12.315
to a lower frequency.

00:10:13.490 --> 00:10:15.070
And here's another one.

00:10:15.070 --> 00:10:16.730
And here's another one.

00:10:16.730 --> 00:10:19.340
Now, if there's a
target out there,

00:10:19.340 --> 00:10:24.640
some distance from the bat,
this pulse that goes out

00:10:24.640 --> 00:10:26.210
will be reflected
off the target.

00:10:26.210 --> 00:10:30.250
And then it will come back to
the bat in the form of an echo

00:10:30.250 --> 00:10:31.310
sometime later.

00:10:32.740 --> 00:10:33.240
OK?

00:10:33.240 --> 00:10:36.800
So the bat-- in this
case the FM bat--

00:10:36.800 --> 00:10:39.710
gets information, number
one, if there's an echo,

00:10:39.710 --> 00:10:41.600
there's a target out there.

00:10:41.600 --> 00:10:46.470
And number two, the time
between the pulse and the echo

00:10:46.470 --> 00:10:49.440
is an indication of the distance
the target is from the bat,

00:10:49.440 --> 00:10:49.940
right?

00:10:49.940 --> 00:10:54.070
Because the sound has to go
from the bat, to the target,

00:10:54.070 --> 00:10:56.055
and then from the
target back to the bat.

00:10:57.290 --> 00:11:01.250
And we know that the
sound velocity in air

00:11:01.250 --> 00:11:05.060
is about 340 meters per second.

00:11:09.590 --> 00:11:12.000
So knowing that
velocity, and knowing

00:11:12.000 --> 00:11:16.410
the time between the pulse and
the echo, we-- and the bat--

00:11:16.410 --> 00:11:18.600
can get information
of how far away

00:11:18.600 --> 00:11:20.305
the target is from the bat.

00:11:22.270 --> 00:11:25.550
Now, a couple of things
I want to comment here

00:11:25.550 --> 00:11:26.460
on this spectrogram.

00:11:26.460 --> 00:11:28.920
Number one, the frequency axis.

00:11:28.920 --> 00:11:32.130
You might not be able to see
it, but it starts at zero

00:11:32.130 --> 00:11:34.550
and then quickly
jumps to 30 kilohertz.

00:11:36.090 --> 00:11:38.760
And then there's
60, 90, and 120.

00:11:38.760 --> 00:11:42.040
So those are very
high sound frequency.

00:11:42.040 --> 00:11:47.380
And so, in the early days
and to a certain extent

00:11:47.380 --> 00:11:50.327
still, those frequencies
would be called ultrasonic.

00:11:57.420 --> 00:11:59.820
And there's no real
good reason for that,

00:11:59.820 --> 00:12:01.840
except that we're humans.

00:12:01.840 --> 00:12:05.810
And everything is important
with respect to humans.

00:12:05.810 --> 00:12:09.500
And our upper limit of
frequency, as you well

00:12:09.500 --> 00:12:12.430
know-- if you're a
very young human--

00:12:12.430 --> 00:12:14.640
ends at about 20 kilohertz.

00:12:14.640 --> 00:12:18.309
And most of us, who are in
our middle or older ages,

00:12:18.309 --> 00:12:20.100
aren't hearing anything
above 10 kilohertz.

00:12:21.240 --> 00:12:24.630
So all these frequencies
emitted by the bat,

00:12:24.630 --> 00:12:27.372
and the echoes coming
back, are beyond the range

00:12:27.372 --> 00:12:28.080
of human hearing.

00:12:29.640 --> 00:12:31.810
And in that sense,
they're ultrasonic.

00:12:31.810 --> 00:12:34.465
So you can't go out and say, oh.

00:12:34.465 --> 00:12:37.080
I heard a bat, or
at least in terms

00:12:37.080 --> 00:12:41.200
of the echolocating signals
that the bats are emitting.

00:12:41.200 --> 00:12:43.600
There are some
sounds that bats emit

00:12:43.600 --> 00:12:46.160
that are communication sounds.

00:12:46.160 --> 00:12:47.970
And those are in
the human frequency

00:12:47.970 --> 00:12:49.410
range to a certain extent.

00:12:49.410 --> 00:12:51.570
But almost all the
echolocating signals

00:12:51.570 --> 00:12:55.610
are well above the upper
limit of our hearing range.

00:12:56.690 --> 00:12:59.640
Now, another thing that you
can see from this spectrogram

00:12:59.640 --> 00:13:03.120
is that there's a big
delay between this pulse

00:13:03.120 --> 00:13:05.150
and this echo, OK?

00:13:05.150 --> 00:13:06.780
The target is pretty far away.

00:13:07.830 --> 00:13:10.810
Generally, bats
head toward targets.

00:13:10.810 --> 00:13:12.310
And this has been
shown many times

00:13:12.310 --> 00:13:15.290
in behavioral experiments,
especially if they're hungry.

00:13:16.110 --> 00:13:16.610
OK?

00:13:16.610 --> 00:13:19.260
And as the bat gets closer
and closer to the target,

00:13:19.260 --> 00:13:23.824
obviously the time between the
pulse and the returning echo

00:13:23.824 --> 00:13:24.365
gets shorter.

00:13:26.390 --> 00:13:30.720
And, as you can probably
see from this spectrogram,

00:13:30.720 --> 00:13:34.910
there are a lot more
pulses emitted per time

00:13:34.910 --> 00:13:38.070
when the bat gets
close to the target

00:13:38.070 --> 00:13:40.869
because the bat is interested
in getting a lot of information

00:13:40.869 --> 00:13:41.660
when it gets close.

00:13:42.850 --> 00:13:46.300
Another reason that the bat
doesn't emit very many pulses

00:13:46.300 --> 00:13:47.840
when it's far away
from the target

00:13:47.840 --> 00:13:51.750
is, if you emitted a pulse
before the echo returned,

00:13:51.750 --> 00:13:55.125
you could get confused
between the outgoing pulse

00:13:55.125 --> 00:13:56.340
and the returning echo.

00:13:57.410 --> 00:14:00.870
So typically, bats tend to
increase their pulse rate

00:14:00.870 --> 00:14:03.740
a lot more as they get
closer to their target.

00:14:03.740 --> 00:14:04.240
OK?

00:14:04.240 --> 00:14:07.290
And I'm going to show you
a demonstration of that.

00:14:07.290 --> 00:14:09.790
I'm sure we'll
convince you of that.

00:14:09.790 --> 00:14:14.730
So this is the type of bat
we have here in New England.

00:14:14.730 --> 00:14:18.350
Examples of this are a little
brown bat or the big brown bat.

00:14:19.630 --> 00:14:23.400
And if anybody has seen
bats-- have you guys seen bats

00:14:23.400 --> 00:14:25.011
flying around at night?

00:14:25.011 --> 00:14:25.510
Yeah?

00:14:25.510 --> 00:14:26.700
Where do you see them?

00:14:27.790 --> 00:14:28.610
In your bedroom?

00:14:28.610 --> 00:14:33.270
Underneath-- [LAUGHTER] it's
where I've seen some recently,

00:14:33.270 --> 00:14:34.270
which is a little scary.

00:14:36.600 --> 00:14:39.780
Well, this view graph
says it hunts in open air.

00:14:39.780 --> 00:14:40.960
Where has anybody seen them?

00:14:40.960 --> 00:14:43.720
I sometimes see them
if I'm out canoeing

00:14:43.720 --> 00:14:47.020
on a lake at night,
or in the evening.

00:14:47.020 --> 00:14:48.060
Any other places?

00:14:48.060 --> 00:14:49.685
On golf courses, for example.

00:14:51.170 --> 00:14:54.720
And those are all
sensible, if you will,

00:14:54.720 --> 00:14:57.600
places for the bat to hunt
because they're very open

00:14:57.600 --> 00:14:58.100
situations.

00:14:59.230 --> 00:15:02.460
And there isn't a whole bunch
of clutter, if you will,

00:15:02.460 --> 00:15:05.170
that will return
echoes to the bat.

00:15:05.170 --> 00:15:07.610
If there's a moth or a
mosquito out there-- it

00:15:07.610 --> 00:15:10.520
might be the only
thing out there

00:15:10.520 --> 00:15:12.890
above the surface of
the lake, and that's

00:15:12.890 --> 00:15:15.690
very interesting to the bat
because it has one target.

00:15:15.690 --> 00:15:18.920
It doesn't have a million leaves
of the forest, if you will,

00:15:18.920 --> 00:15:20.390
to get confused.

00:15:20.390 --> 00:15:22.780
It gets one echo, it knows
there's one target out there.

00:15:22.780 --> 00:15:24.710
It goes and swoops out there.

00:15:24.710 --> 00:15:28.430
And it possibly eats the
target, if it's an insect.

00:15:30.111 --> 00:15:30.610
OK.

00:15:35.900 --> 00:15:38.020
Let me give you
some other examples

00:15:38.020 --> 00:15:40.060
of some information on FM bats.

00:15:40.060 --> 00:15:42.640
First, I want to
point your attention

00:15:42.640 --> 00:15:46.490
to who discovered
echolocating bats.

00:15:46.490 --> 00:15:47.830
And I have his book here.

00:15:48.940 --> 00:15:51.730
This is Donald Griffin's book.

00:15:51.730 --> 00:15:53.255
It's called Listening
In the Dark.

00:15:54.800 --> 00:15:59.015
And I'll write his
name on the board.

00:16:07.500 --> 00:16:10.972
And I'd like to do just a
short reading from his book.

00:16:10.972 --> 00:16:12.510
Then I'll pass it around.

00:16:14.640 --> 00:16:16.420
So this is the
section where he talks

00:16:16.420 --> 00:16:21.520
about this discovery of
bats' ultrasonic sounds.

00:16:27.240 --> 00:16:31.280
So he writes, "during my
undergraduate years at Harvard

00:16:31.280 --> 00:16:35.520
College, when I was actively
engaged in banding bats

00:16:35.520 --> 00:16:38.800
to study their
migration--" So bats here,

00:16:38.800 --> 00:16:42.290
about this time of year, start
flying south like birds do.

00:16:44.130 --> 00:16:46.970
He says, "I was familiar only
with the generally held view

00:16:46.970 --> 00:16:51.860
that bats felt with their wings
the proximity of obstacles."

00:16:51.860 --> 00:16:55.110
that was how people thought
they navigated around, by touch.

00:17:00.580 --> 00:17:02.580
"Several friends suggested
that I experiment

00:17:02.580 --> 00:17:05.720
with the ability of my
bats to avoid obstacles."

00:17:05.720 --> 00:17:07.460
Blah, blah, blah.

00:17:07.460 --> 00:17:12.140
"I decided that I should contact
a professor in the Harvard

00:17:12.140 --> 00:17:13.790
physics department.

00:17:13.790 --> 00:17:17.560
That was Professor GW Pierce,
inventor of the Pierce circuit

00:17:17.560 --> 00:17:19.930
for the stabilization of
radio frequency oscillator.

00:17:22.130 --> 00:17:25.270
Pierce had developed almost
the only apparatus then

00:17:25.270 --> 00:17:28.069
in existence that could
detect and generate

00:17:28.069 --> 00:17:32.660
a wide range of sounds
lying above the audio range.

00:17:32.660 --> 00:17:38.570
That is from 20,000 to almost
100,000 cycles per second.

00:17:38.570 --> 00:17:42.800
With some trepidation, I
approached Professor Pierce

00:17:42.800 --> 00:17:46.520
in the winter of 1938
with the suggestion

00:17:46.520 --> 00:17:48.960
that we use his apparatus
to listen to my bats.

00:17:50.470 --> 00:17:52.960
I found him eager to
try the experiment,

00:17:52.960 --> 00:17:54.430
particularly since
he was already

00:17:54.430 --> 00:17:56.790
engaged in extensive studies
of the high frequency

00:17:56.790 --> 00:17:57.880
sounds of insects.

00:17:59.970 --> 00:18:03.690
When I first brought
a cage full of bats,

00:18:03.690 --> 00:18:07.550
myotis lucifugus--" OK,
that's the little brown bat--

00:18:07.550 --> 00:18:10.130
"to Pierce's lab
and held the cage

00:18:10.130 --> 00:18:12.340
in front of the
parabolic horn, we

00:18:12.340 --> 00:18:16.520
were surprised and delighted to
hear a medley of raucous noises

00:18:16.520 --> 00:18:18.270
from the loudspeaker."

00:18:18.270 --> 00:18:22.850
So Griffin, as an
undergraduate, discovered

00:18:22.850 --> 00:18:27.520
that bats emitted
ultrasonic stimuli, OK?

00:18:27.520 --> 00:18:32.745
And he went on to pursue a
lifetime of research on bats.

00:18:33.920 --> 00:18:35.910
While he was still
an undergraduate,

00:18:35.910 --> 00:18:38.900
he designed some
experiments to see

00:18:38.900 --> 00:18:43.110
if the bats could use this
echolocation to avoid objects.

00:18:43.110 --> 00:18:46.940
And so he took a room,
turned out all the lights

00:18:46.940 --> 00:18:49.740
so the bats could
only use other senses.

00:18:49.740 --> 00:18:51.930
And he noticed that
they would fly around.

00:18:51.930 --> 00:18:56.210
And he didn't have a very
extensive equipment budget,

00:18:56.210 --> 00:19:01.480
so for objects, he went to the
store that sold piano wire.

00:19:01.480 --> 00:19:03.700
And he strung piano
wire from the ceiling

00:19:03.700 --> 00:19:05.770
to the floor of the room.

00:19:05.770 --> 00:19:07.690
And he let the bats fly around.

00:19:07.690 --> 00:19:10.730
And he knew that if something
touched the piano wire,

00:19:10.730 --> 00:19:15.280
he could hear a little
sound of the wire vibrating.

00:19:15.280 --> 00:19:18.480
And he had to go
down in diameter,

00:19:18.480 --> 00:19:21.760
to piano wire that was
the thickness of the width

00:19:21.760 --> 00:19:25.000
of a human hair,
before the bats finally

00:19:25.000 --> 00:19:26.950
started touching the wire.

00:19:26.950 --> 00:19:30.900
They could detect objects
even sub-millimeter in size.

00:19:30.900 --> 00:19:34.740
So their sense of echolocation
was really well developed,

00:19:34.740 --> 00:19:36.090
and very good.

00:19:36.090 --> 00:19:37.855
It could detect
very tiny targets.

00:19:39.370 --> 00:19:42.665
So that was one of his first,
and foremost, experiments.

00:19:44.437 --> 00:19:45.770
Now, I have some demonstrations.

00:19:47.240 --> 00:19:52.530
And some of them are from
Griffin's original work.

00:19:52.530 --> 00:20:00.540
And so, in these
demonstrations, you

00:20:00.540 --> 00:20:02.890
have a movie of the bat flying.

00:20:03.990 --> 00:20:08.940
And the target, this
time, is a small food item

00:20:08.940 --> 00:20:09.775
that's thrown up.

00:20:09.775 --> 00:20:10.775
I think it's a mealworm.

00:20:11.810 --> 00:20:14.580
So the investigator
throws up the mealworm.

00:20:14.580 --> 00:20:16.450
And the bat catches it.

00:20:16.450 --> 00:20:20.700
And on the audio
part of the track,

00:20:20.700 --> 00:20:23.160
you'll hear some popping.

00:20:23.160 --> 00:20:27.970
And I think that's a stroboscope
that's illuminating the image.

00:20:27.970 --> 00:20:30.530
You will also hear
a little chirp.

00:20:30.530 --> 00:20:32.590
And they're pretty
high frequency.

00:20:32.590 --> 00:20:36.010
But that is the bat
echolocating pulse going out.

00:20:36.010 --> 00:20:39.560
It's detected by
microphone, and transformed

00:20:39.560 --> 00:20:42.470
from the high frequencies
down into lower frequencies

00:20:42.470 --> 00:20:45.680
in your audio range,
so you can hear it.

00:20:45.680 --> 00:20:46.730
OK?

00:20:46.730 --> 00:20:49.950
And notice that,
number one, when

00:20:49.950 --> 00:20:53.010
the bat gets close to the
target, the chirps increase

00:20:53.010 --> 00:20:54.340
in frequency.

00:20:54.340 --> 00:20:56.350
And number two,
when the bat eats

00:20:56.350 --> 00:20:58.630
the target, the
chirps stop, right?

00:20:58.630 --> 00:21:02.690
Because unlike you or me,
they can't talk and eat

00:21:02.690 --> 00:21:03.690
at the same time.

00:21:09.010 --> 00:21:13.845
So those chirps are the
bat echolocating pulse.

00:21:15.485 --> 00:21:16.110
Here's the bat.

00:21:18.540 --> 00:21:20.235
Here's the target coming up.

00:21:25.620 --> 00:21:28.530
In that case, as
sometimes happens,

00:21:28.530 --> 00:21:30.950
the bat missed the target.

00:21:30.950 --> 00:21:32.040
So it's falling away.

00:21:43.600 --> 00:21:45.160
Here's the target coming up.

00:21:46.760 --> 00:21:50.090
In this case, the bat caught
it in the tip of its wing.

00:21:50.090 --> 00:21:52.120
And it brings the wing
in toward its mouth.

00:21:52.120 --> 00:21:54.090
And it eats the target.

00:21:54.090 --> 00:21:56.870
And it starts pulsing
again after it's

00:21:56.870 --> 00:21:57.960
swallowed the target.

00:22:08.115 --> 00:22:10.240
And there, I think it caught
it right in its mouth,

00:22:10.240 --> 00:22:11.750
without having to use its wings.

00:22:20.850 --> 00:22:21.350
OK.

00:22:21.350 --> 00:22:27.000
So these are some other films
that I won't go through.

00:22:27.000 --> 00:22:28.890
But they were some
experiments that Griffin

00:22:28.890 --> 00:22:34.610
did, testing the ability of
bats-- and certain species--

00:22:34.610 --> 00:22:36.730
to actually catch fish.

00:22:36.730 --> 00:22:40.690
And he was mystified about
this because-- as we've

00:22:40.690 --> 00:22:42.400
talked about--
sound and air, when

00:22:42.400 --> 00:22:46.510
it comes to a fluid boundary,
mostly reflects off.

00:22:46.510 --> 00:22:50.290
So it seemed unimaginable
that the bat echolocating

00:22:50.290 --> 00:22:52.770
pulse could go under
the water and would

00:22:52.770 --> 00:22:54.190
be reflected of the fish.

00:22:54.190 --> 00:22:57.650
So what he figured out
later is that, when

00:22:57.650 --> 00:23:00.510
there is a smooth
surface of the water,

00:23:00.510 --> 00:23:04.090
the bats did not
seem interested.

00:23:04.090 --> 00:23:06.060
But when the fish
came up and rippled

00:23:06.060 --> 00:23:09.370
the surface of the water,
that was what the bats we're

00:23:09.370 --> 00:23:12.000
actually detecting,
the little ripples

00:23:12.000 --> 00:23:14.515
on the surface of the water--
just as you see them visually.

00:23:18.120 --> 00:23:22.030
There's some videos of
bats catching fish here.

00:23:22.030 --> 00:23:23.250
Bats hanging out.

00:23:34.930 --> 00:23:36.960
Bats flying in rooms.

00:23:36.960 --> 00:23:37.960
Sorry about that.

00:23:41.330 --> 00:23:49.020
Now this last part of the demo
is from more modern experiments

00:23:49.020 --> 00:23:53.170
in which the bat
approaches a target.

00:23:53.170 --> 00:23:56.040
In this case, the target
is somewhere here.

00:23:56.040 --> 00:23:57.250
And it's tethered.

00:23:57.250 --> 00:23:58.285
The target is fixed.

00:23:59.620 --> 00:24:02.910
This is a spectrogram of
the bat echolocating pulse.

00:24:04.020 --> 00:24:08.420
And this is the bat
flying, in slow motion,

00:24:08.420 --> 00:24:09.345
to catch the target.

00:24:11.670 --> 00:24:13.460
And you can see the
tremendous increase

00:24:13.460 --> 00:24:15.450
in pulse repetition rate.

00:24:15.450 --> 00:24:17.850
As it eats the target,
it stops vocalizing.

00:24:19.190 --> 00:24:22.320
And now it starts again.

00:24:22.320 --> 00:24:25.610
And this'll be
repeated at least once.

00:24:25.610 --> 00:24:27.050
So here's another run.

00:24:27.050 --> 00:24:29.795
Here's the bat
spectrogram down here.

00:24:32.220 --> 00:24:33.830
And here's the bat coming in.

00:24:33.830 --> 00:24:35.350
Here is the target, right there.

00:24:57.820 --> 00:24:59.080
I guess I'd better stop that.

00:25:01.901 --> 00:25:02.400
OK.

00:25:02.400 --> 00:25:16.260
And so that second demo is
from Doctor Cynthia Moss,

00:25:16.260 --> 00:25:17.430
who used to be at Harvard.

00:25:17.430 --> 00:25:20.070
And now she's at the
University of Maryland.

00:25:20.070 --> 00:25:22.810
And she does extensive
work on bat echolocation.

00:25:27.000 --> 00:25:30.460
And so I think it clearly
shows the increase

00:25:30.460 --> 00:25:33.510
in pulse repetition rate as the
bat gets close to the target.

00:25:35.250 --> 00:25:36.310
Any questions on that?

00:25:38.050 --> 00:25:38.550
OK.

00:25:38.550 --> 00:25:43.770
So those were all the first
kind of bat I talked about,

00:25:43.770 --> 00:25:44.860
the FM bat.

00:25:48.170 --> 00:25:51.700
And now, let's get into
the second group of bats.

00:25:51.700 --> 00:25:57.491
And this group is
called CFFM bats.

00:25:57.491 --> 00:25:57.990
OK.

00:26:00.020 --> 00:26:03.390
And these are completely
different species of bats.

00:26:03.390 --> 00:26:08.440
They are new world CFFM bats
and old world CFFM bats.

00:26:08.440 --> 00:26:10.475
It probably has
evolved several times.

00:26:12.180 --> 00:26:17.160
The echolocating pulse and
echo are completely different,

00:26:17.160 --> 00:26:19.160
compared to the FM bat.

00:26:19.160 --> 00:26:22.100
So in the case of the
mustache bat, which

00:26:22.100 --> 00:26:26.110
is an example of CFFM bat,
this is the spectrogram.

00:26:26.110 --> 00:26:30.420
Again, frequency on the
y-axis and time on the x-axis.

00:26:32.340 --> 00:26:37.250
But in this case, instead of
the pulse sweeping downward,

00:26:37.250 --> 00:26:39.900
or a frequency modulated
very downward-- almost

00:26:39.900 --> 00:26:45.870
like a chirp-- instead, the
pulse is a constant frequency.

00:26:45.870 --> 00:26:49.340
So the CF stands for
constant frequency.

00:26:49.340 --> 00:26:51.360
That just means the
frequency is staying

00:26:51.360 --> 00:26:53.900
constant as a function of time.

00:26:53.900 --> 00:26:55.830
And that's this
flat section here.

00:26:58.230 --> 00:27:02.070
In the case of bat sounds,
just like human speech sounds,

00:27:02.070 --> 00:27:03.195
there are many harmonics.

00:27:04.730 --> 00:27:07.860
There's the first harmonic.

00:27:07.860 --> 00:27:09.050
That's called CF1.

00:27:10.930 --> 00:27:15.740
There's the second harmonic,
an octave above, CF2.

00:27:17.450 --> 00:27:19.340
There's a third harmonic, CF3.

00:27:20.750 --> 00:27:23.000
And there's a fourth
harmonic, CF4.

00:27:25.490 --> 00:27:30.190
And it's conventional on this
kind of spectrogram display

00:27:30.190 --> 00:27:33.150
to illustrate the sounds
that have the most

00:27:33.150 --> 00:27:36.840
energy with the
boldest marking here.

00:27:36.840 --> 00:27:42.560
So CF2 is the one that has the
highest sound pressure level.

00:27:42.560 --> 00:27:44.960
And so that's the darkest here.

00:27:44.960 --> 00:27:50.470
These other ones, especially
CF1 and CF4, are lower in level.

00:27:50.470 --> 00:27:51.970
They're not as intense.

00:27:51.970 --> 00:27:54.750
And so they're not as
black on this display.

00:27:54.750 --> 00:27:59.240
And in this display of the FM
bat, the pulse is very intense.

00:27:59.240 --> 00:28:02.510
And the echo is, of
course, much reduced.

00:28:03.520 --> 00:28:06.340
Echolocating pulses
can be 110 dB,

00:28:06.340 --> 00:28:09.920
if you measure them
right at the bat's mouth.

00:28:09.920 --> 00:28:11.810
They can be very intense.

00:28:11.810 --> 00:28:14.660
And, of course, the bat
contracts its middle ear

00:28:14.660 --> 00:28:18.410
muscles to prevent those
kinds of intense stimuli

00:28:18.410 --> 00:28:20.870
from damaging its own ears.

00:28:20.870 --> 00:28:24.410
And then it relaxes the
muscles when the echo comes by,

00:28:24.410 --> 00:28:26.985
and its hearing is fine.

00:28:28.070 --> 00:28:28.570
OK.

00:28:28.570 --> 00:28:34.890
So at the end of the CF portion,
this echo-locating pulse.

00:28:34.890 --> 00:28:37.390
There is a little FM sweep.

00:28:37.390 --> 00:28:37.890
OK.

00:28:37.890 --> 00:28:39.780
So you can appreciate,
maybe, that there's

00:28:39.780 --> 00:28:45.670
an FM1, a little FM2
sweep, a little FM3

00:28:45.670 --> 00:28:48.370
sweep, and a little Fm4 sweep.

00:28:50.050 --> 00:28:57.030
And it's thought that this
bat uses the FM of the pulse,

00:28:57.030 --> 00:28:59.860
and the FM of the
return echo, to get

00:28:59.860 --> 00:29:03.290
a measure of the distance that
the target is from the bat.

00:29:04.261 --> 00:29:04.760
OK.

00:29:04.760 --> 00:29:07.664
It if comes back
in 10 milliseconds,

00:29:07.664 --> 00:29:09.080
and you know the
sound of velocity

00:29:09.080 --> 00:29:12.450
is 340 meters per second,
you can figure out

00:29:12.450 --> 00:29:13.660
how close that is.

00:29:15.020 --> 00:29:16.940
And this bat can
do that, as well.

00:29:16.940 --> 00:29:19.340
Now, what's going on
with this CF part?

00:29:21.200 --> 00:29:25.940
Well, as you can see
on the spectrogram,

00:29:25.940 --> 00:29:29.670
the CF of the pulse
is not exactly

00:29:29.670 --> 00:29:32.420
the same as the CF of the echo.

00:29:33.850 --> 00:29:34.570
OK.

00:29:34.570 --> 00:29:38.100
The echo has shifted up to be
a little higher in frequency

00:29:38.100 --> 00:29:38.790
in each case.

00:29:40.020 --> 00:29:41.520
Now, how can
frequencies be shifted?

00:29:42.780 --> 00:29:45.260
Well, it has to do
with the Doppler shift.

00:30:12.380 --> 00:30:12.880
OK.

00:30:12.880 --> 00:30:17.410
So this is a shift in, in
this case, sound frequency.

00:30:24.120 --> 00:30:26.955
But you can have a Doppler
shift for any kind of wave.

00:30:28.040 --> 00:30:31.300
For example, you can have a
Doppler shift for light waves.

00:30:32.840 --> 00:30:35.010
If you've studied
the Big Bang theory

00:30:35.010 --> 00:30:38.370
of the origin of the universe,
there's a big explosion, right?

00:30:38.370 --> 00:30:42.620
Everything exploded out and
is moving far away from us.

00:30:42.620 --> 00:30:45.230
So you look at the light
coming from a star that's

00:30:45.230 --> 00:30:47.400
moving away from you.

00:30:47.400 --> 00:30:52.400
It's actually shifted a little
bit to longer wavelengths,

00:30:52.400 --> 00:30:55.070
toward the more reddish
hues, because it's

00:30:55.070 --> 00:30:56.200
moving away from you.

00:30:57.880 --> 00:31:02.260
So Doppler shifts have to
do with wave sources that

00:31:02.260 --> 00:31:04.440
are moving relative
to the receiver,

00:31:04.440 --> 00:31:07.550
or the receiver moving
relative to the emitter.

00:31:09.120 --> 00:31:12.590
Another example of a Doppler
shift, this time for sound,

00:31:12.590 --> 00:31:16.770
would be if you were in the
grandstand of a race track,

00:31:16.770 --> 00:31:18.970
and the race cars were
going around a big oval.

00:31:20.150 --> 00:31:20.650
OK.

00:31:20.650 --> 00:31:22.316
And you hear the sound
of their engines.

00:31:23.330 --> 00:31:27.930
As the race car comes toward
you, along the straight away,

00:31:27.930 --> 00:31:29.620
it sounds like it's
higher in pitch

00:31:29.620 --> 00:31:31.340
because it's moving toward you.

00:31:32.930 --> 00:31:37.910
As it passes you and then
starts to move away from you,

00:31:37.910 --> 00:31:39.710
it sounds like it's
lower in pitch.

00:31:39.710 --> 00:31:46.320
So the thing you'd
hear would be [BUZZING]

00:31:46.320 --> 00:31:49.660
as each race car went by you.

00:31:49.660 --> 00:31:54.630
So, as the race car is here, and
you're the observer listening

00:31:54.630 --> 00:31:58.320
here, it emits-- let's
say-- a pulse of sound.

00:32:00.461 --> 00:32:00.960
OK.

00:32:00.960 --> 00:32:03.210
This might be the peak
of the wave front,

00:32:03.210 --> 00:32:08.010
if it were just a
sinusoid, let's say.

00:32:09.450 --> 00:32:12.700
Now, by the time a race
car coming toward you

00:32:12.700 --> 00:32:16.940
has emitted the next peak,
the race car's actually moved.

00:32:18.120 --> 00:32:19.020
OK.

00:32:19.020 --> 00:32:20.350
So the peak is here.

00:32:20.350 --> 00:32:22.620
And the next peak is emitted.

00:32:22.620 --> 00:32:23.950
And it's very close together.

00:32:25.060 --> 00:32:27.740
If the race car is
moving away from you,

00:32:27.740 --> 00:32:29.960
it emits one peak of sound.

00:32:29.960 --> 00:32:32.850
And then by the time it
emits the next peak of sound,

00:32:32.850 --> 00:32:35.130
it's moved a little
away from you.

00:32:35.130 --> 00:32:37.275
The peaks are farther apart.

00:32:38.620 --> 00:32:42.490
And we know the sound source
that has a quick oscillation

00:32:42.490 --> 00:32:43.865
sounds like a high frequency.

00:32:45.520 --> 00:32:48.700
And the sound source that
has a very slow oscillation

00:32:48.700 --> 00:32:49.875
sounds like a low frequency.

00:32:52.000 --> 00:32:57.220
So Doppler shifts are
positive, higher frequencies

00:32:57.220 --> 00:32:59.950
for objects making
sound moving toward you.

00:33:01.160 --> 00:33:05.800
And Doppler shifts are low,
or negative, in frequency.

00:33:05.800 --> 00:33:08.080
They make lower
frequencies if the object

00:33:08.080 --> 00:33:09.565
is moving away from you.

00:33:09.565 --> 00:33:11.500
It's just the physical
characteristics

00:33:11.500 --> 00:33:13.460
of sound coupled with movement.

00:33:16.450 --> 00:33:17.310
OK.

00:33:17.310 --> 00:33:23.090
In the case of these positive
Doppler shifted echoes,

00:33:23.090 --> 00:33:26.280
we know either that
the object that

00:33:26.280 --> 00:33:31.040
has been reflecting the echo
is moving toward the bat

00:33:31.040 --> 00:33:35.210
or, conversely, that
the bat has been flying

00:33:35.210 --> 00:33:38.370
toward the object that
is emitting the echo.

00:33:39.610 --> 00:33:42.107
So a positive
Doppler shift means

00:33:42.107 --> 00:33:43.565
things are getting
closer together.

00:33:44.630 --> 00:33:46.320
And a negative
Doppler shift would

00:33:46.320 --> 00:33:49.530
be things are going
farther apart.

00:33:49.530 --> 00:33:53.080
So not only does the
bat, from its FM sweep,

00:33:53.080 --> 00:33:57.140
get an indication of how far
it is away from the target.

00:33:57.140 --> 00:34:00.490
But by its Doppler
shifted CF part,

00:34:00.490 --> 00:34:05.600
it gets an idea of the
relative motion of the target.

00:34:05.600 --> 00:34:07.390
Why is that important?

00:34:07.390 --> 00:34:09.840
These types of bats,
instead of hunting

00:34:09.840 --> 00:34:12.570
in open air-- like
the FM bats we

00:34:12.570 --> 00:34:15.650
have-- these are tropical bats.

00:34:15.650 --> 00:34:18.429
And they hunt in
dense vegetation,

00:34:18.429 --> 00:34:20.670
like the tropical rain forests.

00:34:20.670 --> 00:34:24.020
And there are millions
of objects around.

00:34:24.020 --> 00:34:24.900
There are leaves.

00:34:24.900 --> 00:34:25.659
There's vines.

00:34:26.860 --> 00:34:29.190
There's lots of clutter here.

00:34:29.190 --> 00:34:32.099
What the bat is interested
in is not stationary clutter.

00:34:33.300 --> 00:34:36.670
Presumably, things that are
Doppler shifted all the same.

00:34:36.670 --> 00:34:39.650
But something that is
moving in all this clutter.

00:34:39.650 --> 00:34:41.370
It's very interested
in moving objects

00:34:41.370 --> 00:34:43.005
because that's a
life form, perhaps.

00:34:44.250 --> 00:34:47.389
And just imagine the
kind of a Doppler shift

00:34:47.389 --> 00:34:50.670
that would be made
by a moth, first

00:34:50.670 --> 00:34:53.909
beating its wing toward
you, if you were the bat.

00:34:53.909 --> 00:34:56.980
And then beating its wing
downward, and away, from you.

00:34:56.980 --> 00:34:57.480
OK.

00:34:57.480 --> 00:35:01.180
That's a very complicated
positive and negative Doppler

00:35:01.180 --> 00:35:05.430
shift that the bat would
pick up on its return echo.

00:35:05.430 --> 00:35:08.540
And that would be a very
interesting target to the bat.

00:35:08.540 --> 00:35:09.050
Yeah.

00:35:09.050 --> 00:35:12.026
AUDIENCE: And how do they
tell that the difference is

00:35:12.026 --> 00:35:14.436
when they're moving versus
if the object is [INAUDIBLE]?

00:35:14.436 --> 00:35:15.560
PROFESSOR: They don't care.

00:35:15.560 --> 00:35:19.410
All they care about is that
they might be getting thousands

00:35:19.410 --> 00:35:24.010
of Doppler shifted echoes from
the targets in front of them.

00:35:24.010 --> 00:35:27.330
Let's say they're flying toward
a whole bunch of tropical rain

00:35:27.330 --> 00:35:28.862
forest vegetation.

00:35:28.862 --> 00:35:31.320
There are going to be thousands
of positive Doppler shifted

00:35:31.320 --> 00:35:32.020
echoes.

00:35:32.020 --> 00:35:37.160
And then something is moving
away from them, or toward them,

00:35:37.160 --> 00:35:39.340
faster than the background.

00:35:39.340 --> 00:35:41.040
All they care is that.

00:35:41.040 --> 00:35:44.110
It's Doppler shifted different
relative to the background.

00:35:45.680 --> 00:35:47.820
They're just looking
for something special.

00:35:47.820 --> 00:35:51.360
That is, something that's moving
relative to the background.

00:35:54.550 --> 00:35:55.830
Any other questions on that?

00:36:00.310 --> 00:36:07.290
So you can, of course,
design sonar systems.

00:36:07.290 --> 00:36:13.100
Submarine sonar systems work by
sending out a pulse of sound,

00:36:13.100 --> 00:36:15.220
and listening for the echo.

00:36:15.220 --> 00:36:17.730
And most of the kinds
of sonar systems

00:36:17.730 --> 00:36:21.480
that we have send out a
ping, which is a frequency

00:36:21.480 --> 00:36:24.800
swept signal, and
listen for the echo.

00:36:24.800 --> 00:36:28.315
Because it's mostly interested
in the distance from a target,

00:36:28.315 --> 00:36:30.620
and whether there's
a target out there.

00:36:30.620 --> 00:36:35.180
This is a very unusual type
of echolocation, or sonar,

00:36:35.180 --> 00:36:35.680
if you will.

00:36:38.540 --> 00:36:39.900
Now why am I bringing this up?

00:36:39.900 --> 00:36:42.160
Well, it's very
interesting because the bat

00:36:42.160 --> 00:36:45.580
gets two queues
instead of just one.

00:36:45.580 --> 00:36:52.230
Also because a lot of work on
bat cortex has used CFFM bats.

00:36:52.230 --> 00:36:55.740
And one of the most popular
has been the so-called mustache

00:36:55.740 --> 00:37:00.950
bat, I believe because
it has a noseleaf that's

00:37:00.950 --> 00:37:03.920
between its upper lip--
it looks like a mustache--

00:37:03.920 --> 00:37:04.887
and its nostrils.

00:37:04.887 --> 00:37:06.220
So it's called the mustache bat.

00:37:07.310 --> 00:37:09.700
And a lot of this
work has been done

00:37:09.700 --> 00:37:14.900
by a researcher, who is
still active, at Washington

00:37:14.900 --> 00:37:16.480
University in Saint Louis.

00:37:16.480 --> 00:37:20.000
And his name is Nobua Suga.

00:37:25.770 --> 00:37:29.280
And he was the first,
really, to work successfully

00:37:29.280 --> 00:37:30.835
on the bat auditory cortex.

00:37:33.100 --> 00:37:39.680
A lot of his work comes from
the 1970s, '80s, and '90s.

00:37:39.680 --> 00:37:43.280
And before I explain this
bottom part of this slide,

00:37:43.280 --> 00:37:47.030
let me just go on and show
you the kinds of experiments

00:37:47.030 --> 00:37:47.835
that Suga did.

00:37:47.835 --> 00:37:52.020
And here's one from one of
his publications in the 1980s.

00:37:52.020 --> 00:37:57.540
So Suga's work was innovative
because he played around--

00:37:57.540 --> 00:38:01.020
well, first he rationalized, OK.

00:38:01.020 --> 00:38:04.430
I could try any sound
system I'd like to.

00:38:04.430 --> 00:38:05.990
I could try clicks.

00:38:05.990 --> 00:38:07.600
I could try pure tones.

00:38:07.600 --> 00:38:09.480
I could try noise.

00:38:09.480 --> 00:38:10.900
I could try speech.

00:38:10.900 --> 00:38:14.650
I could try-- but why don't
I try what the bat listens to

00:38:14.650 --> 00:38:16.220
over and over?

00:38:16.220 --> 00:38:17.815
It listens to a pulse.

00:38:19.170 --> 00:38:21.120
And then a little
bit later, an echo.

00:38:22.460 --> 00:38:26.470
And this turned out to be
a very, very wise choice,

00:38:26.470 --> 00:38:28.040
as we'll see in a minute.

00:38:28.040 --> 00:38:30.710
Secondly, about the
time of the 1970s

00:38:30.710 --> 00:38:34.840
and '80s, speech
researchers in human speech

00:38:34.840 --> 00:38:37.831
were using synthesized speech.

00:38:37.831 --> 00:38:40.080
Of course, we all know what
synthesized speech is now.

00:38:40.080 --> 00:38:42.170
But back then, it
was very novel.

00:38:42.170 --> 00:38:46.850
And Suga says, well, I'm
going to use synthesized bat

00:38:46.850 --> 00:38:48.630
echolocating calls.

00:38:48.630 --> 00:38:51.315
And here is an example
of a synthesized pulse.

00:38:52.920 --> 00:38:55.310
So this is for the mustache bat.

00:38:55.310 --> 00:39:03.010
And this looks like CF1-FM1,
CF2-FM2, and CF3-FM3.

00:39:04.560 --> 00:39:07.130
So he's just using
three harmonics.

00:39:07.130 --> 00:39:08.950
So one thing about
synthesized calls

00:39:08.950 --> 00:39:13.240
is you can do things like
dispense with one harmonic,

00:39:13.240 --> 00:39:18.150
if you want to, easily take
it out, and put it an echo.

00:39:18.150 --> 00:39:19.830
So here's a pulse.

00:39:19.830 --> 00:39:20.870
And here's an echo.

00:39:20.870 --> 00:39:22.480
It's a little bit
Doppler shifted.

00:39:23.510 --> 00:39:27.720
You can look at the no response
to the pulse, and to the echo.

00:39:27.720 --> 00:39:31.440
And this lower
trace is a histogram

00:39:31.440 --> 00:39:34.350
from a single neuron
in the auditory cortex

00:39:34.350 --> 00:39:37.370
of the echolocating bat to
a pulse, and to an echo.

00:39:37.370 --> 00:39:38.510
There's not much response.

00:39:40.380 --> 00:39:43.250
Suga found that when
you play a pulse

00:39:43.250 --> 00:39:47.800
and, a short time later, an
echo, you get a huge response.

00:39:47.800 --> 00:39:49.640
And that's what's
indicated here.

00:39:49.640 --> 00:39:52.330
So this is the
pulse-echo combination.

00:39:53.410 --> 00:39:55.620
And these are
synthesized stimuli.

00:39:57.520 --> 00:40:00.370
And you can read in
the original paper,

00:40:00.370 --> 00:40:02.400
it looks like it's
not given here,

00:40:02.400 --> 00:40:05.765
exactly what the delay is
between the pulse and the echo.

00:40:07.650 --> 00:40:11.590
But Suga tried various
pulse-echo delays.

00:40:11.590 --> 00:40:15.310
And he found that cortical
neurons, in many cases,

00:40:15.310 --> 00:40:19.550
were very sensitive
to the exact delay

00:40:19.550 --> 00:40:21.930
between the pulse and the echo.

00:40:21.930 --> 00:40:24.195
So they were, if you
will, delay tuned.

00:40:26.310 --> 00:40:30.850
And that's indicated
here, in the first bullet.

00:40:33.110 --> 00:40:35.360
"The neuron pictured above
responds little to--" blah,

00:40:35.360 --> 00:40:36.990
blah, blah. "--pulse
of an echo alone.

00:40:38.290 --> 00:40:41.220
But vigorously to a
pulse followed by an echo

00:40:41.220 --> 00:40:41.970
at certain delay.

00:40:41.970 --> 00:40:44.710
In this case, 9.3
milliseconds is

00:40:44.710 --> 00:40:47.060
the best delay for this neuron."

00:40:47.060 --> 00:40:48.650
So this is a delay tuned neuron.

00:40:50.900 --> 00:40:56.660
Once Suga did recordings from
different parts of the bat

00:40:56.660 --> 00:41:00.600
cortex, he found that the
best delay was actually

00:41:00.600 --> 00:41:04.270
mapped along the
surface of the cortex.

00:41:04.270 --> 00:41:09.020
So here is some of his
work from the 1990s,

00:41:09.020 --> 00:41:13.470
showing you maps of best
delays, and other properties,

00:41:13.470 --> 00:41:15.560
in the bat auditory cortex.

00:41:15.560 --> 00:41:19.740
So here is a side view
of the bat cortex.

00:41:19.740 --> 00:41:21.310
This is looking
at the left side.

00:41:21.310 --> 00:41:27.020
This is the front, where
the olfactory areas are.

00:41:27.020 --> 00:41:30.410
Way in the back would
be the occipital cortex.

00:41:30.410 --> 00:41:32.670
And our old friend,
the auditory cortex,

00:41:32.670 --> 00:41:37.750
is in the temporal region,
on the side of the brain,

00:41:37.750 --> 00:41:39.030
just like it was in the cat.

00:41:39.030 --> 00:41:40.930
And like it is in the human.

00:41:40.930 --> 00:41:41.860
And here is A1.

00:41:43.080 --> 00:41:47.220
And this rectangle
here is expanded here.

00:41:47.220 --> 00:41:49.290
And some of Suga's
maps are shown.

00:41:51.360 --> 00:41:54.810
This part right here--
the biggest part--

00:41:54.810 --> 00:42:00.290
is cortical field A1, which,
as we've seen in other animals,

00:42:00.290 --> 00:42:04.150
is the tonotopically
organized field.

00:42:04.150 --> 00:42:06.465
And it's tonotopically
organized in this bat.

00:42:07.640 --> 00:42:13.230
And these numbers and lines
are the ISO frequency laminae.

00:42:13.230 --> 00:42:14.890
So remember, the
experiment here is

00:42:14.890 --> 00:42:20.120
to go in and sample at a
specific place in the cortex,

00:42:20.120 --> 00:42:22.250
and find the
characteristic frequency

00:42:22.250 --> 00:42:23.670
for neurons in that column.

00:42:25.480 --> 00:42:28.220
And then move the
electrode a little bit.

00:42:28.220 --> 00:42:28.910
Do the same.

00:42:28.910 --> 00:42:29.890
Get the tuning curve.

00:42:29.890 --> 00:42:32.090
Get the CF for those neurons.

00:42:32.090 --> 00:42:33.140
And so on and so forth.

00:42:33.140 --> 00:42:35.470
And build up a map here.

00:42:35.470 --> 00:42:39.710
So just like in the
cat, posterior areas

00:42:39.710 --> 00:42:41.610
are tuned to low CF.

00:42:41.610 --> 00:42:44.240
Low CFs in the bat
are 20 kilohertz.

00:42:44.240 --> 00:42:46.305
We're dealing with very,
very high frequencies.

00:42:47.410 --> 00:42:51.460
As you go more rostrally, the
CFs get increasingly high.

00:42:51.460 --> 00:42:55.920
And at the very rostral end of
A1, the CF is 100 kilohertz.

00:42:55.920 --> 00:42:56.770
Extremely high.

00:42:58.200 --> 00:43:01.680
And everything looks
exactly like other mammals,

00:43:01.680 --> 00:43:06.700
except for this huge area
right in the middle of A1.

00:43:08.210 --> 00:43:10.610
And almost all of
the neurons here

00:43:10.610 --> 00:43:14.940
are tuned to between
61 and 66 kilohertz.

00:43:16.300 --> 00:43:18.260
And you should perk up
your ears a little bit

00:43:18.260 --> 00:43:22.740
because that is where
the most intense harmonic

00:43:22.740 --> 00:43:26.410
of echolocating pulse is,
right around 61 kilohertz.

00:43:28.130 --> 00:43:30.940
And a lot of the Doppler
shifted echoes are just

00:43:30.940 --> 00:43:32.560
going to be a little
bit above that.

00:43:32.560 --> 00:43:35.500
If the bat is flying
toward the target,

00:43:35.500 --> 00:43:38.090
this region goes up
then to 66 kilohertz.

00:43:39.380 --> 00:43:46.100
So then there is an expanded
region of the A1 of the bat

00:43:46.100 --> 00:43:50.690
that's tuned to a very important
frequency for the echolocating

00:43:50.690 --> 00:43:51.190
signal.

00:43:52.370 --> 00:43:57.490
And, at first, this was
called an acoustic fovea.

00:44:08.850 --> 00:44:13.360
Because if you go down into
lower nuclei of the bat

00:44:13.360 --> 00:44:17.110
pathway, and if you actually
go into the cochlea,

00:44:17.110 --> 00:44:21.470
you find an expanded region
of the cochlea devoted

00:44:21.470 --> 00:44:23.010
to these same frequencies.

00:44:23.010 --> 00:44:24.960
That is, you go along
the basilar membrane,

00:44:24.960 --> 00:44:28.991
starting at the
most apical regions,

00:44:28.991 --> 00:44:29.990
and you march down them.

00:44:29.990 --> 00:44:32.330
When you get to the
61 kilohertz place,

00:44:32.330 --> 00:44:36.920
there's a lot of cochlea
devoted to processing that area.

00:44:36.920 --> 00:44:39.860
And so the fovea,
the eyes, where

00:44:39.860 --> 00:44:45.200
you have lots of receptor cells
packed in to a certain part.

00:44:45.200 --> 00:44:47.960
And this is where you have a
lot of hair cells packed in,

00:44:47.960 --> 00:44:50.615
or expanded region of
the basilar membrane,

00:44:50.615 --> 00:44:53.880
where lots of hair
cells processing

00:44:53.880 --> 00:44:55.620
this small range of frequencies.

00:44:55.620 --> 00:44:59.880
So this is very much
different from other mammals.

00:44:59.880 --> 00:45:03.470
And the cochlea of this CFFM
bat is also very different.

00:45:05.600 --> 00:45:08.890
Now, we were talking
about delay to neurons.

00:45:08.890 --> 00:45:13.220
And Suga found a very
interesting area near A1

00:45:13.220 --> 00:45:16.800
in which there is a
mapping for best delay.

00:45:16.800 --> 00:45:18.475
And that's indicated here.

00:45:19.550 --> 00:45:26.000
And the best delays are marching
from short to longer best

00:45:26.000 --> 00:45:28.640
delays, as these
arrows go along here.

00:45:29.930 --> 00:45:34.340
Now they're marked
FM1, FM2, FM3.

00:45:34.340 --> 00:45:35.740
So what does all that mean?

00:45:39.060 --> 00:45:44.840
So Suga found that with his
synthesized pulses and echoes,

00:45:44.840 --> 00:45:47.350
he could dissect this
rather complicated

00:45:47.350 --> 00:45:52.980
pulse-echo constellation
into smaller parts.

00:45:52.980 --> 00:45:55.720
And here's an example of
a stimulus where you have,

00:45:55.720 --> 00:46:00.280
it looks like, CF2-FM2, CF3-FM3.

00:46:02.390 --> 00:46:06.810
And you have the
echo for CF1-FM1,

00:46:06.810 --> 00:46:09.180
and the echo for CF3-FM3.

00:46:09.180 --> 00:46:12.850
And that hardly gave any
response to the neurons.

00:46:12.850 --> 00:46:18.810
Here's an example where you
have only CF1-FM1 for the pulse,

00:46:18.810 --> 00:46:22.220
and only CF2-FM2 for the echo.

00:46:22.220 --> 00:46:23.725
And it gave a big response.

00:46:25.810 --> 00:46:27.160
And you could do even more.

00:46:27.160 --> 00:46:33.630
You can strip off
everything except the FM1

00:46:33.630 --> 00:46:35.365
and the echo FM2.

00:46:37.030 --> 00:46:39.840
And you get a big
response from the neuron.

00:46:39.840 --> 00:46:42.090
So this type of
neuron, then, would

00:46:42.090 --> 00:46:48.800
be called an FM1-FM2
best delay neuron.

00:46:48.800 --> 00:46:50.180
FM1 is the pulse.

00:46:50.180 --> 00:46:51.730
FM2 is the echo.

00:46:51.730 --> 00:46:53.680
And with a certain
delay between those two,

00:46:53.680 --> 00:46:57.390
you get as big a response as
with the whole constellation

00:46:57.390 --> 00:46:58.317
of pulse and echo.

00:46:59.630 --> 00:47:04.510
And those neurons were located
in this specific region, called

00:47:04.510 --> 00:47:05.010
FM1-FM2.

00:47:07.410 --> 00:47:11.150
And their delays were
mapped along this axis,

00:47:11.150 --> 00:47:16.606
with delays going from
0.4 to 18 milliseconds.

00:47:18.290 --> 00:47:20.170
And knowing the
velocity of sound,

00:47:20.170 --> 00:47:23.730
you can convert that
to a target range

00:47:23.730 --> 00:47:27.470
of between 7 and
310 centimeters.

00:47:27.470 --> 00:47:30.160
That's how far the
target was from the bat

00:47:30.160 --> 00:47:32.550
at that specific best delay.

00:47:32.550 --> 00:47:36.300
Suga also found some
other best delay regions.

00:47:36.300 --> 00:47:41.420
For example, FM2, FM3,
so on and so forth

00:47:41.420 --> 00:47:43.460
in other adjacent
parts of cortex.

00:47:43.460 --> 00:47:46.130
And this is really
beautiful work,

00:47:46.130 --> 00:47:53.300
showing specializations for
cortical neuron response.

00:47:53.300 --> 00:47:57.090
And for showing mappings
for those specializations

00:47:57.090 --> 00:47:58.880
in the bat cortex.

00:47:58.880 --> 00:48:03.270
We don't really have any data
anywhere near as beautiful

00:48:03.270 --> 00:48:08.820
on non-echolocating mammalian
cortex that show specialization

00:48:08.820 --> 00:48:14.080
for specific features of sound
stimuli as we do like this

00:48:14.080 --> 00:48:15.300
in the bat cortex.

00:48:15.300 --> 00:48:17.040
And this is really
beautiful work.

00:48:17.040 --> 00:48:20.770
This is Nobel Prize
deserving work

00:48:20.770 --> 00:48:25.530
because it really shows us
what this bat cortex is doing.

00:48:25.530 --> 00:48:27.885
It's responding to specific
features of the pulse.

00:48:29.260 --> 00:48:32.340
And the return echo delay
a specific delay later.

00:48:32.340 --> 00:48:33.718
So it's very beautiful work.

00:48:39.460 --> 00:48:42.210
Let me just mention
one other region here.

00:48:42.210 --> 00:48:46.070
Suga showed nearby, a
region where the neurons are

00:48:46.070 --> 00:48:49.450
specialized to
certain combinations

00:48:49.450 --> 00:48:53.340
of the constant
frequency of the echo.

00:48:53.340 --> 00:48:56.590
And then he showed
a Doppler shifted

00:48:56.590 --> 00:49:00.250
constant [INAUDIBLE]
pulse and CF of the echo.

00:49:00.250 --> 00:49:03.569
So this is the CFCF
region right over here.

00:49:13.730 --> 00:49:14.980
Yeah, question.

00:49:14.980 --> 00:49:17.537
AUDIENCE: In the last diagram--

00:49:17.537 --> 00:49:19.120
PROFESSOR: This one
or the one before?

00:49:19.120 --> 00:49:20.050
AUDIENCE: This one right here.

00:49:20.050 --> 00:49:20.633
PROFESSOR: OK.

00:49:20.633 --> 00:49:22.980
AUDIENCE: The last graph,
like the middle bottom.

00:49:22.980 --> 00:49:24.246
PROFESSOR: This guy?

00:49:24.246 --> 00:49:28.214
AUDIENCE: Why is that the
neuron is responding, like,

00:49:28.214 --> 00:49:33.680
before the onset frequency?

00:49:33.680 --> 00:49:35.360
PROFESSOR: I don't
know why that is.

00:49:36.380 --> 00:49:42.429
There's some clue to that, as to
why this stimulus starts here.

00:49:42.429 --> 00:49:42.970
I don't know.

00:49:45.210 --> 00:49:47.196
I don't know the answer to that.

00:49:47.196 --> 00:49:50.790
Let's see if it says
anything in the caption.

00:49:55.650 --> 00:49:56.290
I don't know.

00:49:56.290 --> 00:49:57.790
I don't know why that is.

00:49:57.790 --> 00:49:58.470
Sorry.

00:49:58.470 --> 00:50:01.720
I'll have to dig out the
paper and figure that out.

00:50:03.180 --> 00:50:04.200
Any other questions?

00:50:09.720 --> 00:50:13.710
So one thing that's gone
on after Sugo's early work

00:50:13.710 --> 00:50:17.480
on these specializations
has asked the question,

00:50:17.480 --> 00:50:21.810
is this really happening
in the auditory cortex?

00:50:21.810 --> 00:50:25.460
Or is the cortex just
merely a reflection

00:50:25.460 --> 00:50:29.770
of some beautiful processing at
a lower level of the pathway?

00:50:29.770 --> 00:50:35.140
And to a certain extent,
best delay sensitivity

00:50:35.140 --> 00:50:36.640
is found at lower levels.

00:50:36.640 --> 00:50:39.800
For example, the
inferior colliculus

00:50:39.800 --> 00:50:42.330
has some best
delay tuned neurons

00:50:42.330 --> 00:50:43.950
in the echolocating bat.

00:50:43.950 --> 00:50:47.760
So it probably has more
in the auditory cortex.

00:50:47.760 --> 00:50:50.970
But they can arise at
the inferior colliculus.

00:50:57.330 --> 00:50:57.880
OK.

00:50:57.880 --> 00:51:03.060
So that's what I wanted to
say about bat echolocation.

00:51:03.060 --> 00:51:06.760
And now I'm going to move on and
spend the last part of today's

00:51:06.760 --> 00:51:09.000
class talking about
speech sounds.

00:51:10.370 --> 00:51:13.940
So we had this particular
slide in an earlier lecture,

00:51:13.940 --> 00:51:16.930
I think the very first
lecture that I gave,

00:51:16.930 --> 00:51:18.760
talking about what
speech sounds are.

00:51:23.110 --> 00:51:28.070
So speech sounds, obviously,
are formed in humans

00:51:28.070 --> 00:51:33.600
by the vocal cords,
or vocal folds,

00:51:33.600 --> 00:51:38.170
closing and opening during
airflow from the lungs

00:51:38.170 --> 00:51:39.590
to the upper vocal tract.

00:51:41.180 --> 00:51:46.180
And this closing and
opening of the glottis

00:51:46.180 --> 00:51:48.610
gives rise to the
so-called glottal pulses.

00:51:49.690 --> 00:51:54.700
When the vocal cords are closed,
there's no airflow coming out.

00:51:54.700 --> 00:51:56.825
But when they open,
there's turbulent airflow

00:51:56.825 --> 00:51:58.125
and it makes a sound.

00:51:59.150 --> 00:52:00.180
So these are pulses.

00:52:00.180 --> 00:52:02.550
And they have a whole bunch
of different frequencies.

00:52:04.040 --> 00:52:06.930
So this is the wave form
as a function of time.

00:52:06.930 --> 00:52:09.450
Sound pressure is
a function of time.

00:52:09.450 --> 00:52:14.094
And this is the spectrum showing
the different frequencies

00:52:14.094 --> 00:52:14.760
that are formed.

00:52:14.760 --> 00:52:17.470
There's a whole bunch
of different frequencies

00:52:17.470 --> 00:52:18.575
in your glottal pulses.

00:52:18.575 --> 00:52:19.560
It's very complicated.

00:52:21.430 --> 00:52:23.320
To form different
speech sounds, you

00:52:23.320 --> 00:52:26.285
do things with your
upper vocal tract.

00:52:27.460 --> 00:52:31.860
In this case of vowels,
you position the muscles

00:52:31.860 --> 00:52:39.270
so that your upper vocal tract
forms filters that enhance

00:52:39.270 --> 00:52:43.300
and decrease some of
these frequencies.

00:52:43.300 --> 00:52:47.230
And after you apply the filter
function of the vocal tract

00:52:47.230 --> 00:52:52.380
to this glottal pulse spectrum,
you get this type of spectrum

00:52:52.380 --> 00:52:53.680
where there are certain peaks.

00:52:55.210 --> 00:52:58.870
And in the production
of a vowel,

00:52:58.870 --> 00:53:01.370
these peaks are at
different frequencies.

00:53:01.370 --> 00:53:06.070
So, for this example,
the vowel "eh" in hit,

00:53:06.070 --> 00:53:11.660
you have a very low peak, and
a couple of high peaks up here.

00:53:11.660 --> 00:53:13.036
These pacer called formants.

00:53:14.280 --> 00:53:16.410
And they're labeled by Fs.

00:53:16.410 --> 00:53:17.830
So we went over this before.

00:53:17.830 --> 00:53:20.670
So there's F1 here,
F2, and F3 here.

00:53:21.680 --> 00:53:25.800
And certain cochlear implant
processors, of course.

00:53:25.800 --> 00:53:27.790
Try to look at the
acoustic spectrum,

00:53:27.790 --> 00:53:30.965
and pick off these formants.

00:53:32.860 --> 00:53:36.310
And they present a lot
of electrical stimuli

00:53:36.310 --> 00:53:39.140
to electrodes that correspond
to them in the cochlea.

00:53:40.950 --> 00:53:43.120
This one we have a
lot of stimulation

00:53:43.120 --> 00:53:44.940
at these low
frequency electrodes,

00:53:44.940 --> 00:53:46.960
which would be apical
in the cochlear implant.

00:53:48.130 --> 00:53:50.580
And then in the
intermediate electrodes,

00:53:50.580 --> 00:53:54.440
they completely shut them
down, even if there's

00:53:54.440 --> 00:53:56.490
a little bit of
background noise.

00:53:56.490 --> 00:53:58.990
And then they would present a
lot of stimuli at the position

00:53:58.990 --> 00:54:00.660
corresponding to F2 and F3.

00:54:02.330 --> 00:54:07.010
And that's an effort to
decrease background noise, which

00:54:07.010 --> 00:54:09.260
is always a big
problem in listening

00:54:09.260 --> 00:54:10.900
to any kind of
acoustic wave form.

00:54:10.900 --> 00:54:12.830
But especially if you
have a cochlear implant.

00:54:14.750 --> 00:54:18.200
So this vowel, "ah,"
in the word "call"

00:54:18.200 --> 00:54:20.740
has two formants very
low in frequency.

00:54:20.740 --> 00:54:22.310
And one in the middle frequency.

00:54:22.310 --> 00:54:24.920
It sounds very
different, of course.

00:54:24.920 --> 00:54:27.100
And your vocal tract
position is very different.

00:54:28.440 --> 00:54:33.190
And this volume, which is
"oo," as in the word cool,

00:54:33.190 --> 00:54:37.030
has three fairly evenly
spaced formants here.

00:54:37.030 --> 00:54:40.440
Your vocal tract is yet
in a different position.

00:54:40.440 --> 00:54:43.020
And you interpret this
as yet a different vowel.

00:54:44.180 --> 00:54:50.590
Now that's a display that's
not very conventional.

00:54:50.590 --> 00:54:54.510
Much more conventional is to
look at a speech spectrogram.

00:54:54.510 --> 00:54:57.100
So this is very similar
to what we've just

00:54:57.100 --> 00:55:00.570
looked at for bat
echolocating pulses.

00:55:00.570 --> 00:55:05.560
This spectrogram is a graph
on the y-axis of frequencies.

00:55:05.560 --> 00:55:09.460
And now these are more
normal sonic, if you will,

00:55:09.460 --> 00:55:10.690
frequencies.

00:55:10.690 --> 00:55:13.820
These are well within the
human range, of course,

00:55:13.820 --> 00:55:17.265
going from 0 to 7
kilohertz on this axis.

00:55:18.740 --> 00:55:20.460
This is a time axis here.

00:55:21.730 --> 00:55:26.640
And again, the higher
in level, the darker the

00:55:26.640 --> 00:55:28.280
display in the spectrogram.

00:55:28.280 --> 00:55:31.050
So there's some really
dark bands here.

00:55:31.050 --> 00:55:34.300
And there's some very light
stuff here, and here, and here.

00:55:36.200 --> 00:55:44.515
And this is the utterance-- "Joe
took father's shoe bench out."

00:55:44.515 --> 00:55:46.420
OK, and that's what
the sound looks

00:55:46.420 --> 00:55:49.920
like when you make
that utterance.

00:55:49.920 --> 00:55:53.230
So the spectrogram plots the
frequencies of speech sounds

00:55:53.230 --> 00:55:53.790
over time.

00:55:55.290 --> 00:55:57.620
What we've talked
about-- up until now--

00:55:57.620 --> 00:56:01.731
are voiced segments,
which are mostly vowels.

00:56:01.731 --> 00:56:03.980
So in this utterance, you
have a bunch of vowels here.

00:56:03.980 --> 00:56:08.110
Here's a nice one, "ah"
in the word father's.

00:56:09.770 --> 00:56:13.750
And you can quite clearly
see there's a nice band here

00:56:13.750 --> 00:56:14.760
that would be F1.

00:56:16.700 --> 00:56:19.550
That would be about
at 500 kilohertz.

00:56:19.550 --> 00:56:23.206
And F2 is about at 1 kilohertz.

00:56:25.620 --> 00:56:28.580
F3 would be about
at 2 kilohertz.

00:56:29.910 --> 00:56:36.500
And there's a fourth
formant about 3 kilohertz.

00:56:36.500 --> 00:56:39.275
And that's the very beautiful
vowel, "ah" as in father.

00:56:41.520 --> 00:56:42.580
Here's another one.

00:56:47.190 --> 00:56:48.100
"Joe."

00:56:48.100 --> 00:56:53.430
So "oh." "Oh" is a vowel
where, in this case,

00:56:53.430 --> 00:56:56.050
there's a beautiful
stable formant here.

00:56:56.050 --> 00:56:59.070
But here is a formant
that's transitioning

00:56:59.070 --> 00:57:02.720
from higher frequency,
maybe about 1.5 kilohertz,

00:57:02.720 --> 00:57:05.520
down to below 1 kilohertz.

00:57:05.520 --> 00:57:06.195
So it's Joe.

00:57:07.800 --> 00:57:10.430
And there's a
higher formant here.

00:57:10.430 --> 00:57:11.280
OK.

00:57:11.280 --> 00:57:15.702
So those are the vowels, or
the so-called voiced segments.

00:57:15.702 --> 00:57:17.160
And voicing just
means that there's

00:57:17.160 --> 00:57:23.860
a constant outflow of sound
coming through the vocal tract.

00:57:23.860 --> 00:57:25.574
And you can make
these sounds forever.

00:57:25.574 --> 00:57:26.490
You can say, ahhhhhhh.

00:57:27.846 --> 00:57:30.770
And you can just keep
going if you want to.

00:57:30.770 --> 00:57:32.990
Of course, you don't
in normal speech.

00:57:32.990 --> 00:57:37.530
Now for consonants,
there are several types.

00:57:40.070 --> 00:57:42.890
And these are generally
called unvoiced segments.

00:57:42.890 --> 00:57:44.960
Mostly consonants are
intervals containing

00:57:44.960 --> 00:57:47.730
bands of frequencies,
swept frequencies,

00:57:47.730 --> 00:57:49.060
and silent intervals.

00:57:50.170 --> 00:57:52.900
So for example, one
of the consonants

00:57:52.900 --> 00:57:55.760
here is F in the word fathers.

00:57:57.060 --> 00:58:00.380
And right before, at the
beginning of the sound F,

00:58:00.380 --> 00:58:03.520
you go, which is
no sound, right?

00:58:03.520 --> 00:58:04.700
You close your lips.

00:58:04.700 --> 00:58:07.270
You keep sound from coming out.

00:58:07.270 --> 00:58:09.816
And you finally
go, father, right?

00:58:09.816 --> 00:58:11.440
And that's what's
happening right here.

00:58:12.710 --> 00:58:14.400
So that's a stop consonant.

00:58:14.400 --> 00:58:18.450
You stop the vocal tract
before you let it go.

00:58:19.530 --> 00:58:25.020
And the vowel T, as in took,
is another stop consonant.

00:58:25.020 --> 00:58:28.390
So you're not doing
anything at the beginning.

00:58:28.390 --> 00:58:30.710
And finally, you go, took.

00:58:30.710 --> 00:58:31.880
Right?

00:58:31.880 --> 00:58:38.630
When you let go, and
emit the sound of took,

00:58:38.630 --> 00:58:41.095
you have a very
complex frequency band.

00:58:42.230 --> 00:58:44.490
That's generally
high frequencies.

00:58:44.490 --> 00:58:47.520
2 kilohertz in this case,
up beyond 7 kilohertz.

00:58:48.780 --> 00:58:50.520
There's that explosion
of sound right

00:58:50.520 --> 00:58:55.500
at the beginning of the
consonant T, as and took.

00:58:55.500 --> 00:58:56.500
OK.

00:58:56.500 --> 00:59:00.870
Now there been a lot of
course studies on speech

00:59:00.870 --> 00:59:03.610
coding in the auditory nerve
and the cochlear nucleus.

00:59:05.750 --> 00:59:08.980
And one of the findings
shouldn't be too surprising

00:59:08.980 --> 00:59:09.480
at all.

00:59:09.480 --> 00:59:14.985
You have auditory nerve fibers
that have tuning curves, right?

00:59:14.985 --> 00:59:17.335
We've been over tuning
curves many times before.

00:59:28.200 --> 00:59:33.430
So tuning curve is a
graph of sound pressure

00:59:33.430 --> 00:59:34.390
level for a response.

00:59:35.535 --> 00:59:37.220
There's a sound frequency here.

00:59:42.070 --> 00:59:47.776
And then a 1 kilohertz CF.

00:59:47.776 --> 00:59:50.570
The CF for this tuning
curve would be 1 kilohertz,

00:59:50.570 --> 00:59:51.160
let's say.

00:59:58.530 --> 01:00:03.700
And a 10 kilohertz CF
might look like that.

01:00:05.456 --> 01:00:06.420
OK.

01:00:06.420 --> 01:00:10.550
So you can explore the responses
of 1 kilohertz auditory nerve

01:00:10.550 --> 01:00:14.270
fibers and 10 kilohertz
auditory nerve fibers

01:00:14.270 --> 01:00:15.725
to this type of stimulus.

01:00:17.810 --> 01:00:21.040
And obviously, the
1 kilohertz fibers

01:00:21.040 --> 01:00:22.255
are going to be very active.

01:00:24.390 --> 01:00:26.090
During portions
of this utterance,

01:00:26.090 --> 01:00:29.420
for example, there's
a lot of 1 kilohertz

01:00:29.420 --> 01:00:32.480
in this "oh" second
formant here.

01:00:32.480 --> 01:00:35.610
So the 1 kilohertz fiber's going
to respond like crazy there.

01:00:36.960 --> 01:00:41.260
In the vowel "ah," there's a
big 1 kilohertz band there.

01:00:41.260 --> 01:00:44.210
The 1 kilohertz fiber is
going to respond a lot there.

01:00:44.210 --> 01:00:48.060
It's not going to
respond here or here.

01:00:48.060 --> 01:00:52.790
But it's going to respond a lot
at the end of the "out" sound.

01:00:52.790 --> 01:00:54.090
OK.

01:00:54.090 --> 01:00:57.980
The 10 kilohertz fiber is
kind of out of luck, right?

01:00:57.980 --> 01:00:58.780
Its way up here.

01:00:58.780 --> 01:00:59.745
It's off the axis.

01:01:01.940 --> 01:01:08.655
But notice the tail of
this 10 kilohertz fiber.

01:01:09.880 --> 01:01:12.020
If I had drawn it a
little bit further

01:01:12.020 --> 01:01:15.700
it would be extending
past 1 kilohertz.

01:01:15.700 --> 01:01:18.720
So it's certainly going to
respond, right, in here,

01:01:18.720 --> 01:01:21.135
as long as the sound
level is high enough.

01:01:22.600 --> 01:01:26.440
If you keep the sound level
of this utterance low,

01:01:26.440 --> 01:01:30.435
down here, then the frequency
is obviously down here.

01:01:32.080 --> 01:01:34.870
That 10 kilohertz fiber
is not going to respond.

01:01:34.870 --> 01:01:38.070
But if you boost the
sound level, such

01:01:38.070 --> 01:01:41.080
that you're in the tail
of the tuning curve,

01:01:41.080 --> 01:01:45.150
this 10 kilohertz CF fiber
is going to start to respond.

01:01:45.150 --> 01:01:51.410
For example, response to these
frequencies here, this 4, 5,

01:01:51.410 --> 01:01:52.460
and 6 kilohertz.

01:01:52.460 --> 01:01:54.730
Maybe the 7
kilohertz can respond

01:01:54.730 --> 01:01:59.520
to things like the consonants,
if the sound level is high

01:01:59.520 --> 01:02:02.785
enough so that it's
within its response areas.

01:02:04.370 --> 01:02:11.970
So there's clear CF processing
of this type of speech signal

01:02:11.970 --> 01:02:14.710
at the auditory nerve and
in the cochlear nucleus.

01:02:14.710 --> 01:02:16.830
There's also phase locking.

01:02:16.830 --> 01:02:20.430
For example, these
lower frequencies

01:02:20.430 --> 01:02:23.590
are within the frequency
range where there's really

01:02:23.590 --> 01:02:26.090
good phase locking for
the auditory nerve.

01:02:27.120 --> 01:02:30.340
Remember, phase locking
falls off above 1 kilohertz.

01:02:30.340 --> 01:02:31.850
And by about 3
kilohertz, there's

01:02:31.850 --> 01:02:34.070
not much phase locking at all.

01:02:34.070 --> 01:02:37.374
But many of these voiced,
or vowel, segments

01:02:37.374 --> 01:02:38.790
are going to have
low frequencies.

01:02:38.790 --> 01:02:41.060
And they're going to
be good phase locking

01:02:41.060 --> 01:02:43.010
in the auditory nerve
or cochlear nucleus.

01:02:44.320 --> 01:02:45.880
So that's just sort of a review.

01:02:45.880 --> 01:02:49.530
Think about the auditory
nerve response to these speech

01:02:49.530 --> 01:02:52.730
signals because clearly,
the auditory nerve

01:02:52.730 --> 01:02:55.700
is going to respond
very nicely, in terms

01:02:55.700 --> 01:02:59.180
of what its CFs tell it to.

01:02:59.180 --> 01:03:01.610
Now, there's been a lot
of interesting work.

01:03:01.610 --> 01:03:05.566
Of course, I don't have time to
get into much speech processing

01:03:05.566 --> 01:03:08.390
and language representation.

01:03:08.390 --> 01:03:10.650
But I just wanted to
show you some things that

01:03:10.650 --> 01:03:13.985
relate quite nicely to what
we've just gone through

01:03:13.985 --> 01:03:14.693
for echolocation.

01:03:16.320 --> 01:03:19.650
Here are some synthesized
speech stimuli.

01:03:21.240 --> 01:03:21.740
OK.

01:03:21.740 --> 01:03:24.010
And you can do this very
nicely on your computer.

01:03:24.010 --> 01:03:27.440
This is a spectrogram of
the synthesized sound.

01:03:27.440 --> 01:03:30.560
Frequency is on the
y-axis and time,

01:03:30.560 --> 01:03:34.880
in this case in milliseconds,
the very quick stimulus

01:03:34.880 --> 01:03:35.810
is on the x-axis.

01:03:36.860 --> 01:03:39.300
And there are several
harmonics, very much

01:03:39.300 --> 01:03:43.200
like we had for the CF
echolocating bat sound.

01:03:43.200 --> 01:03:45.320
And there's some regions
of constant frequency.

01:03:46.850 --> 01:03:48.530
And there are
clearly some regions

01:03:48.530 --> 01:03:53.340
of frequency modulation, very
much like the bat echolocating

01:03:53.340 --> 01:03:54.780
signal that we just went over.

01:03:54.780 --> 01:03:58.960
Except that now the
modulated part of the signal

01:03:58.960 --> 01:04:02.450
is in the front
instead of at the back

01:04:02.450 --> 01:04:03.550
like it was for the bat.

01:04:05.290 --> 01:04:09.230
And so right here,
on the third formant,

01:04:09.230 --> 01:04:12.430
is a very interesting transition
that's not shown in black.

01:04:12.430 --> 01:04:13.305
It's shown in white.

01:04:14.740 --> 01:04:20.910
Because in the work of Liberman
and Mattingly from the 1980s,

01:04:20.910 --> 01:04:25.400
they studied this so-called
formant transition.

01:04:25.400 --> 01:04:29.930
So the format here
is the vowel "ah."

01:04:29.930 --> 01:04:33.530
And you have these three
formants, 1, 2, and 3.

01:04:34.660 --> 01:04:38.030
And the transition
leading up into that

01:04:38.030 --> 01:04:43.110
is either the consonant
D or the consonant G.

01:04:44.950 --> 01:04:46.535
It's when it's coming down.

01:04:49.070 --> 01:04:54.310
It's the combination "da."
but when this third formant

01:04:54.310 --> 01:04:59.890
transition is instead going
up, it's the consonant "gah."

01:04:59.890 --> 01:05:03.300
completely different speech
sound, "da" versus "gah."

01:05:03.300 --> 01:05:04.640
No one would ever mistake them.

01:05:06.020 --> 01:05:06.520
Right.

01:05:06.520 --> 01:05:10.500
And so what Liberman and
Mattingly did was they

01:05:10.500 --> 01:05:15.290
varied this transition
into the third formant.

01:05:15.290 --> 01:05:24.390
Instead of just having one
like that, or one like this,

01:05:24.390 --> 01:05:26.885
they sloped it any
number of degrees.

01:05:29.740 --> 01:05:30.240
All right.

01:05:30.240 --> 01:05:39.170
And when it goes, I think
when it's falling, it's "da."

01:05:39.170 --> 01:05:42.680
And when it's rising, it's
"gah" if I'm not mistaken.

01:05:44.110 --> 01:05:47.480
So what would you expect if
it was right in the middle?

01:05:49.840 --> 01:05:51.400
Well, you could expect anything.

01:05:51.400 --> 01:05:54.660
But actually, the
observation is,

01:05:54.660 --> 01:05:58.710
as you move this
formant transition over,

01:05:58.710 --> 01:06:00.740
subjects do not report
something that's

01:06:00.740 --> 01:06:03.670
in between "gah" and "da."

01:06:03.670 --> 01:06:06.000
Instead, all of a
sudden, they quickly

01:06:06.000 --> 01:06:09.780
shift from "gah" to "da."

01:06:09.780 --> 01:06:10.500
All of a sudden.

01:06:10.500 --> 01:06:13.700
And there's a very sharp
boundary in the shift.

01:06:13.700 --> 01:06:17.353
And the subjects never report
something that's intermediate.

01:06:19.730 --> 01:06:24.130
So this is an example of
putting the speech sound, which

01:06:24.130 --> 01:06:29.900
can be modulating continuously,
into two sharply defined

01:06:29.900 --> 01:06:31.350
perceptual categories.

01:06:31.350 --> 01:06:34.720
Either "gah" or either "da."

01:06:34.720 --> 01:06:36.210
But nothing in between.

01:06:36.210 --> 01:06:38.580
No gradual slope in between.

01:06:38.580 --> 01:06:40.350
It's just one or the other.

01:06:40.350 --> 01:06:45.165
This gave rise to the idea
of categorical perception

01:06:45.165 --> 01:06:46.570
of speech sounds.

01:06:50.250 --> 01:06:53.280
The other thing they could
do is do things like this.

01:06:54.480 --> 01:07:01.430
Present the black stimuli to
one ear and the white stimuli

01:07:01.430 --> 01:07:02.290
to the other ear.

01:07:03.440 --> 01:07:06.760
And you get the perception,
then, of a speech sound.

01:07:06.760 --> 01:07:14.600
If you don't present any
formant transition at all,

01:07:14.600 --> 01:07:16.590
what would you expect to happen?

01:07:16.590 --> 01:07:19.780
Well, what actually happens is
you do hear something ambiguous

01:07:19.780 --> 01:07:22.480
if there's actually no
formant transition at all.

01:07:24.200 --> 01:07:26.780
If there's a formant
transition, in one ear

01:07:26.780 --> 01:07:29.200
and the rest of the
sound to the other ear,

01:07:29.200 --> 01:07:31.120
you hear the complete
speech sound.

01:07:31.120 --> 01:07:35.320
If you just present this formant
transition and nothing else,

01:07:35.320 --> 01:07:37.910
you just hear a little
chirp, a little speech sound.

01:07:37.910 --> 01:07:40.470
But you add that to
the rest, and you

01:07:40.470 --> 01:07:46.120
get an unambiguous or
categorical "da" or "gah."

01:07:46.120 --> 01:07:46.620
OK.

01:07:46.620 --> 01:07:50.160
These are beautiful
series of experiments

01:07:50.160 --> 01:07:56.690
by Alvin Liberman in the 1980s.

01:07:56.690 --> 01:08:01.360
Now, in cortex, the
interesting question,

01:08:01.360 --> 01:08:06.210
then, if you pull an
analogy between bat signals

01:08:06.210 --> 01:08:08.740
and human signals,
we've had spectrograms

01:08:08.740 --> 01:08:10.800
from the two which are
not that much different.

01:08:11.960 --> 01:08:15.240
The question is, do we
have specialized neurons

01:08:15.240 --> 01:08:22.029
in our cortices that are
sensitive to specific features

01:08:22.029 --> 01:08:22.979
of those signals?

01:08:22.979 --> 01:08:27.250
For example, features are things
like whether these two formants

01:08:27.250 --> 01:08:30.100
are close together,
whether there's

01:08:30.100 --> 01:08:33.954
a formant sweep going
down or going up.

01:08:33.954 --> 01:08:36.890
Do we have specific,
if you will,

01:08:36.890 --> 01:08:40.220
feature detectors
in the human cortex?

01:08:40.220 --> 01:08:42.000
We don't know that.

01:08:42.000 --> 01:08:44.090
What we do know
clearly is that there

01:08:44.090 --> 01:08:48.080
are areas that are very
selective for language

01:08:48.080 --> 01:08:51.019
and speech stimuli in
the cortex of humans.

01:08:52.200 --> 01:08:53.700
So in the cortex
of humans, we've

01:08:53.700 --> 01:08:56.420
talked about there
being a primary auditory

01:08:56.420 --> 01:08:59.220
cortex in the
temporal lobe here.

01:08:59.220 --> 01:09:02.020
And we had the little
model that showed you

01:09:02.020 --> 01:09:03.644
that there was a Heschl's gyrus.

01:09:04.689 --> 01:09:07.729
And that is the site of
primary auditory cortex, or A1,

01:09:07.729 --> 01:09:08.229
in humans.

01:09:10.310 --> 01:09:15.370
All around that region,
an area that's sometimes

01:09:15.370 --> 01:09:17.870
called perisylvian cortex.

01:09:19.229 --> 01:09:23.390
And it gets its name from
this big sylvian fissure,

01:09:23.390 --> 01:09:26.620
if you will, that divides
the temporal lobe down here

01:09:26.620 --> 01:09:30.439
from the rest of the brain,
especially the parietal lobe.

01:09:30.439 --> 01:09:32.760
All around this
perisylvian cortex

01:09:32.760 --> 01:09:35.380
is associated with
language processing.

01:09:35.380 --> 01:09:36.480
And how do we know that?

01:09:36.480 --> 01:09:37.855
Well, of course,
imaging studies.

01:09:39.229 --> 01:09:42.720
But in the beginning,
the early pathologists

01:09:42.720 --> 01:09:46.680
like Broca and Wernicke,
who studied patients

01:09:46.680 --> 01:09:51.120
who had lesions in the
cortex, mostly from strokes.

01:09:51.120 --> 01:09:52.569
But sometimes from
other injuries.

01:09:53.856 --> 01:09:58.390
It showed that lesions in
this region of the brain

01:09:58.390 --> 01:10:01.620
left patients with deficits
in language processing,

01:10:01.620 --> 01:10:04.640
especially with a
deficit called aphasia.

01:10:06.280 --> 01:10:06.780
OK?

01:10:06.780 --> 01:10:12.800
Disorders of comprehending
or producing spoken language

01:10:12.800 --> 01:10:13.865
are known as aphasia.

01:10:16.120 --> 01:10:20.670
And these aphasias are
often classified into types.

01:10:21.750 --> 01:10:26.710
If you're a neurology resident,
or you do your medical rotation

01:10:26.710 --> 01:10:29.440
in neurology, you
will see patients

01:10:29.440 --> 01:10:31.520
with so-called Broca's aphasia.

01:10:33.550 --> 01:10:38.200
And this, originally, was
brought to light by Broca.

01:10:38.200 --> 01:10:41.910
We saw such patients
with lesions

01:10:41.910 --> 01:10:43.170
in this part of the brain.

01:10:43.170 --> 01:10:45.525
That's come to be
known as Broca's area.

01:10:46.570 --> 01:10:50.180
That's part of the
frontal cortex, the lower

01:10:50.180 --> 01:10:52.740
frontal cortex,
near motor areas.

01:10:55.070 --> 01:11:00.750
And the clinical manifestation
is a major disturbance

01:11:00.750 --> 01:11:04.890
in speech production, with
sparse or halting speech.

01:11:04.890 --> 01:11:09.270
It's often misarticulated,
missing function words,

01:11:09.270 --> 01:11:10.220
and parts of words.

01:11:10.220 --> 01:11:15.960
So this is clearly a problem
with producing speech.

01:11:15.960 --> 01:11:17.700
Sometimes this is
called motor aphasia.

01:11:22.050 --> 01:11:24.850
Wernicke is another
early physician

01:11:24.850 --> 01:11:27.950
who saw patients with
damaged cortices.

01:11:30.330 --> 01:11:34.200
He saw some of them with
damage to this region,

01:11:34.200 --> 01:11:39.350
in the caudal temporal lobe
and associated parietal lobe.

01:11:39.350 --> 01:11:44.580
It's an area which has become
known as Wernicke's area.

01:11:46.520 --> 01:11:48.946
And here, the
clinical manifestation

01:11:48.946 --> 01:11:49.945
is completely different.

01:11:51.040 --> 01:11:56.060
In this case, the production
of speech is fine.

01:11:56.060 --> 01:11:58.580
But it's a major disturbance
in auditory comprehension.

01:12:00.830 --> 01:12:01.330
OK.

01:12:01.330 --> 01:12:06.130
So you ask the patient something
and they cannot understand you.

01:12:06.130 --> 01:12:08.385
But they have fluent
speech production.

01:12:08.385 --> 01:12:12.110
It's fluent speech
with maybe disturbances

01:12:12.110 --> 01:12:13.990
of the sounds and
structures of the words.

01:12:13.990 --> 01:12:17.474
But the major deficit is in
the auditory comprehension,

01:12:17.474 --> 01:12:18.432
language comprehension.

01:12:21.670 --> 01:12:26.690
So the question,
then, has always been,

01:12:26.690 --> 01:12:30.270
is this Broca's
area the motor area

01:12:30.270 --> 01:12:34.070
for production of
speech and language?

01:12:34.070 --> 01:12:37.600
And is this Wernicke's
area the area

01:12:37.600 --> 01:12:39.830
for comprehension of speech?

01:12:39.830 --> 01:12:42.670
And clearly, this is a
very simplistic idea,

01:12:42.670 --> 01:12:46.780
in what would be called
the localizational idea.

01:12:46.780 --> 01:12:48.375
I'm not sure if
that word is here.

01:12:49.920 --> 01:12:57.560
But if you're a so-called
localization proponent,

01:12:57.560 --> 01:12:59.940
you would say each
little part of the cortex

01:12:59.940 --> 01:13:01.030
has its own function.

01:13:01.030 --> 01:13:03.870
And they do that independently
of all the other areas.

01:13:03.870 --> 01:13:08.600
So the Broca's area is
involved in producing speech.

01:13:08.600 --> 01:13:12.140
And Wernicke's area is
responsible for comprehending

01:13:12.140 --> 01:13:13.120
speech.

01:13:13.120 --> 01:13:16.890
And this is clearly
from imaging studies

01:13:16.890 --> 01:13:21.210
that we know now is a very
simplistic view, and probably

01:13:21.210 --> 01:13:22.320
an incorrect view.

01:13:22.320 --> 01:13:28.030
It's more likely this
whole of perisylvian cortex

01:13:28.030 --> 01:13:29.755
contributes to
language processing.

01:13:30.770 --> 01:13:35.950
And so people who
subscribe to that theory

01:13:35.950 --> 01:13:37.530
would be called holistic.

01:13:37.530 --> 01:13:41.650
They'd have the holistic
view of processing in cortex.

01:13:44.900 --> 01:13:48.190
And we'll go over the imaging
studies in just a minute.

01:13:48.190 --> 01:13:50.760
One thing I want
to make sure to say

01:13:50.760 --> 01:13:59.490
is that language processing is
clearly a cortical phenomenon.

01:13:59.490 --> 01:14:02.200
That is, if you have
injury to the cortex

01:14:02.200 --> 01:14:06.355
in these specific areas,
you're likely to have aphasia.

01:14:07.420 --> 01:14:11.180
If you have injuries to the
brain stem, to the thalamus,

01:14:11.180 --> 01:14:14.310
you are much less likely to
have any kind of aphasia.

01:14:15.510 --> 01:14:18.240
So clearly, the
cortex is the place

01:14:18.240 --> 01:14:19.790
where language is processed.

01:14:21.530 --> 01:14:24.490
Another thing about cortex
and language processing

01:14:24.490 --> 01:14:29.600
is that it's usually lateralized
into one hemisphere or another,

01:14:29.600 --> 01:14:30.100
OK?

01:14:30.100 --> 01:14:34.160
So if you're right handed,
usually your language

01:14:34.160 --> 01:14:36.830
is processed in your
opposite hemifield,

01:14:36.830 --> 01:14:39.750
in the left cortical area.

01:14:39.750 --> 01:14:41.110
And so how is that known?

01:14:41.110 --> 01:14:45.070
Well, if you have a stroke
patient whose right handed,

01:14:45.070 --> 01:14:47.320
they have a lesion
in the left cortex.

01:14:47.320 --> 01:14:48.460
They show up with aphasia.

01:14:49.560 --> 01:14:52.370
If they have a lesion
in the right cortex,

01:14:52.370 --> 01:14:54.940
there's minimal effect on
their language functions.

01:14:55.960 --> 01:14:59.110
Another way is by the
so-called Wada test.

01:14:59.110 --> 01:15:07.376
So people who are getting ready
to have cortical neurosurgery--

01:15:07.376 --> 01:15:09.750
so why would you ever want to
have cortical neurosurgery?

01:15:27.500 --> 01:15:28.000
Anybody?

01:15:29.230 --> 01:15:35.800
So another big disease in the
cortex is epilepsy, right?

01:15:35.800 --> 01:15:38.817
Epilepsy is uncontrolled
activity, usually starting

01:15:38.817 --> 01:15:39.400
in the cortex.

01:15:40.860 --> 01:15:43.280
Of course, the first line
of attack is by medication.

01:15:44.590 --> 01:15:49.140
But some epileptic
patients have epilepsy

01:15:49.140 --> 01:15:50.940
that is not controlled
by medication.

01:15:50.940 --> 01:15:53.985
And they have seizures
every half hour.

01:15:55.120 --> 01:15:57.530
And it's pretty
much intractable.

01:15:57.530 --> 01:16:02.150
So the last line of attack,
then, by the neurosurgeons

01:16:02.150 --> 01:16:06.140
is to try to go into the cortex
and find the part of the cortex

01:16:06.140 --> 01:16:09.710
where the epileptic
focus begins.

01:16:09.710 --> 01:16:10.960
And then they'd lesion that.

01:16:10.960 --> 01:16:12.640
And this is a
successful treatment.

01:16:13.740 --> 01:16:18.740
But if the surgeon goes in and
lesions part of the language

01:16:18.740 --> 01:16:22.930
areas, you have a patient
that wakes up as an aphasic.

01:16:22.930 --> 01:16:24.440
That's not a happy patient.

01:16:24.440 --> 01:16:29.470
So the surgeons do lots of
tests before such surgery.

01:16:29.470 --> 01:16:33.980
And one is to try to figure out
which hemisphere is processing

01:16:33.980 --> 01:16:34.730
the language.

01:16:34.730 --> 01:16:36.490
So they do the Wada test.

01:16:36.490 --> 01:16:38.010
Has anybody heard
of the Wada test?

01:16:39.800 --> 01:16:40.300
OK.

01:16:40.300 --> 01:16:44.320
They take the patient, of
course, and-- if they're smart

01:16:44.320 --> 01:16:46.120
and they plan ahead,
the patient is seated.

01:16:47.670 --> 01:16:48.170
OK?

01:16:48.170 --> 01:16:51.960
And they have a carotid
artery on the left side,

01:16:51.960 --> 01:16:53.800
and a carotid artery
on the right side.

01:16:55.090 --> 01:16:58.150
And into the carotid
artery is injected

01:16:58.150 --> 01:17:00.710
a quick acting
barbiturate anesthetic.

01:17:02.580 --> 01:17:06.570
On one side, that carotid
artery feeds one hemisphere

01:17:06.570 --> 01:17:08.470
of the cortex and not the other.

01:17:09.860 --> 01:17:11.770
So the patient is seated
because the patient

01:17:11.770 --> 01:17:13.270
is likely to slump
because they're

01:17:13.270 --> 01:17:14.700
going to have some
motor problems.

01:17:14.700 --> 01:17:19.100
And the patient might fall
over if they were standing up.

01:17:19.100 --> 01:17:21.270
But in the test, the
patient is supposed

01:17:21.270 --> 01:17:25.580
to recite from reading
or from memory.

01:17:25.580 --> 01:17:27.760
And the anesthetic is injected.

01:17:27.760 --> 01:17:30.870
And as soon as that anesthetic
hits the hemisphere that's

01:17:30.870 --> 01:17:35.050
processing language, the
patient stops reciting.

01:17:35.050 --> 01:17:38.760
That's if they injected
on the correct side.

01:17:38.760 --> 01:17:40.310
If they injected
on the other side,

01:17:40.310 --> 01:17:42.760
the patient keeps
reading, keeps reciting.

01:17:42.760 --> 01:17:43.965
So that's the Wada test.

01:17:48.215 --> 01:18:08.570
Language function is
mostly in one hemisphere

01:18:08.570 --> 01:18:12.700
in right handed individuals
In left handers,

01:18:12.700 --> 01:18:14.460
things are a little
bit different.

01:18:14.460 --> 01:18:16.790
Left handed
individuals sometimes

01:18:16.790 --> 01:18:20.790
have the opposite
hemispheric dominance.

01:18:20.790 --> 01:18:22.930
Sometimes they have
language function

01:18:22.930 --> 01:18:24.110
distributed bilaterally.

01:18:25.770 --> 01:18:28.240
Sometimes they have
language function

01:18:28.240 --> 01:18:31.691
in the same side as
their handedness.

01:18:31.691 --> 01:18:32.190
OK.

01:18:32.190 --> 01:18:36.510
But for this, I'm talking
about right handers, OK?

01:18:36.510 --> 01:18:38.090
Here's some very
interesting work

01:18:38.090 --> 01:18:43.630
looking at the language
areas in people.

01:18:46.000 --> 01:18:49.480
In postmortem, material has
been done by Al Galaburda.

01:18:49.480 --> 01:18:52.425
So he's at Beth Israel
Deaconess Hospital now.

01:18:53.430 --> 01:18:58.157
And he looked at
the left hemisphere

01:18:58.157 --> 01:18:59.240
and the right hemispheres.

01:19:00.370 --> 01:19:04.610
And he looked at the very,
of course, closely associated

01:19:04.610 --> 01:19:07.800
language areas,
especially right near A1.

01:19:09.360 --> 01:19:13.770
And so here is-- you can't see
them on a side view like this--

01:19:13.770 --> 01:19:18.740
but if you cut off the top of
the cortex, which you can do.

01:19:18.740 --> 01:19:20.560
This is postmortem material.

01:19:20.560 --> 01:19:25.350
You look down on the superior
surface of the temporal lobe,

01:19:25.350 --> 01:19:27.540
you see these views here.

01:19:27.540 --> 01:19:33.720
And the area that's just caudal
to the primary auditory cortex

01:19:33.720 --> 01:19:35.150
is called the plenum temporally.

01:19:36.760 --> 01:19:41.350
And there are clearly some
left-right asymmetries

01:19:41.350 --> 01:19:45.920
in that region of the brain
that almost certainly relate

01:19:45.920 --> 01:19:48.040
to language processing
in that area.

01:19:49.481 --> 01:19:49.980
OK?

01:19:49.980 --> 01:19:53.500
So there are
anatomical asymmetries

01:19:53.500 --> 01:19:55.560
in the perisylvian regions.

01:19:55.560 --> 01:19:57.890
And so this is the first
time in our course,

01:19:57.890 --> 01:20:02.010
now, where right and left
makes a big difference.

01:20:02.010 --> 01:20:04.330
All along, we said
it doesn't matter

01:20:04.330 --> 01:20:06.505
if we simulated the right
ear or the left ear.

01:20:06.505 --> 01:20:09.860
And here, clearly, there
is a dominant hemisphere

01:20:09.860 --> 01:20:10.800
for language.

01:20:14.560 --> 01:20:17.710
Now finally,
imaging studies have

01:20:17.710 --> 01:20:21.950
shown us a great deal about
the cortical processing

01:20:21.950 --> 01:20:22.650
of language.

01:20:22.650 --> 01:20:28.060
And here's data from a pet
study, in which the subjects

01:20:28.060 --> 01:20:30.790
are listening to
language stimuli.

01:20:32.300 --> 01:20:36.020
And these happen to be
French speaking subjects.

01:20:36.020 --> 01:20:38.785
So the last condition
is a story in French.

01:20:39.950 --> 01:20:44.930
And obviously, the subjects
understood the story.

01:20:44.930 --> 01:20:47.420
They could tell you
what was going on.

01:20:47.420 --> 01:20:50.700
And these are, by the way,
right handed subjects.

01:20:50.700 --> 01:20:55.570
And here's the imaging of the
areas in the left hemisphere,

01:20:55.570 --> 01:20:58.209
which would be expected to
be the dominant hemisphere

01:20:58.209 --> 01:20:58.750
for language.

01:21:00.000 --> 01:21:01.500
And this is in the
other hemisphere,

01:21:01.500 --> 01:21:03.350
which shows a lot
less activation.

01:21:05.730 --> 01:21:09.260
The activation in the areas
where the subjects were

01:21:09.260 --> 01:21:12.690
listening to language
they understood

01:21:12.690 --> 01:21:18.190
is the superior temporal
area, superior temporal gyrus.

01:21:18.190 --> 01:21:21.372
That's including the temporal
pole here, in purple.

01:21:21.372 --> 01:21:22.830
You can't see the
yellow very well.

01:21:22.830 --> 01:21:27.060
But believe there's a
lot of activation here.

01:21:27.060 --> 01:21:33.080
This blue area, labeled
IFG inferior frontal gyrus,

01:21:33.080 --> 01:21:34.345
this is Broca's area.

01:21:36.520 --> 01:21:38.980
And why does it light up
if Broca's area is only

01:21:38.980 --> 01:21:40.370
a motor area?

01:21:40.370 --> 01:21:43.280
It's clearly involved
in motor functions.

01:21:43.280 --> 01:21:49.730
But here is imaging results from
subjects just listening, not

01:21:49.730 --> 01:21:53.000
producing language, where
Broca's area lights up.

01:21:53.000 --> 01:21:56.050
It's activated on
the dominant side,

01:21:56.050 --> 01:21:57.440
just in the listening task.

01:21:59.260 --> 01:22:03.120
Contrast that to when the
subjects were listening

01:22:03.120 --> 01:22:07.390
to a story in a language
that they did not understand,

01:22:07.390 --> 01:22:09.490
this language is called Tamil.

01:22:09.490 --> 01:22:11.170
And none of the
subjects could speak it.

01:22:11.170 --> 01:22:15.730
There's hardly any
activation in the original.

01:22:15.730 --> 01:22:19.020
There's a little bit
of yellow activation

01:22:19.020 --> 01:22:22.080
near the primary
auditory cortex.

01:22:22.080 --> 01:22:24.490
It's pretty symmetric,
left to right.

01:22:24.490 --> 01:22:26.010
And that's just
what you'd expect

01:22:26.010 --> 01:22:30.230
if you were, for example,
given pure tones or noise.

01:22:30.230 --> 01:22:35.120
This is a nice control because,
presumably, this language

01:22:35.120 --> 01:22:37.033
has about the same
frequency content.

01:22:38.450 --> 01:22:41.580
And other factors are fairly
similar between these two

01:22:41.580 --> 01:22:42.250
languages.

01:22:42.250 --> 01:22:44.510
The one difference
is the subjects

01:22:44.510 --> 01:22:48.310
were not perceptually aware
of what they were learning

01:22:48.310 --> 01:22:51.065
about in the story, in the
case of the unfamiliar.

01:22:52.870 --> 01:22:55.330
These intermediate
conditions, some

01:22:55.330 --> 01:22:59.250
of them having pseudo words
and anomalous sentences,

01:22:59.250 --> 01:23:03.230
didn't light up the language
areas to a great degree.

01:23:03.230 --> 01:23:06.870
But listening to a list, in this
case, of French words-- which

01:23:06.870 --> 01:23:11.490
the subjects were familiar
to-- again, showed activation.

01:23:11.490 --> 01:23:14.690
I'll point out to
you in Broca's area

01:23:14.690 --> 01:23:17.850
in the dominant hemisphere, in
these right handed subjects.

01:23:17.850 --> 01:23:21.780
So again, clearly,
just a listening task

01:23:21.780 --> 01:23:24.180
can light up Broca's area.

01:23:24.180 --> 01:23:27.125
And so that is a
very clear example.

01:23:28.220 --> 01:23:32.170
I'll show you that
these so-called motor

01:23:32.170 --> 01:23:34.480
areas, like Broca's
area, are involved

01:23:34.480 --> 01:23:41.150
in listening and cortex
processing of language stimuli.

01:23:41.150 --> 01:23:46.100
It's not just involved in
motor production of speech,

01:23:46.100 --> 01:23:51.680
even though what we call
clinically Broca's aphasia

01:23:51.680 --> 01:23:53.720
has a major disturbance
in speech production.

01:23:55.850 --> 01:23:56.850
OK.

01:23:56.850 --> 01:23:58.850
And one final thing
to leave you with.

01:24:00.900 --> 01:24:05.170
Broca's area tends to
light up, in this case,

01:24:05.170 --> 01:24:07.950
in fairly simple stimuli.

01:24:07.950 --> 01:24:11.160
But it tends to light
up in other studies,

01:24:11.160 --> 01:24:13.340
like in cases where
the substances have

01:24:13.340 --> 01:24:17.390
difficult grammar
or complex meaning.

01:24:17.390 --> 01:24:19.070
And so the subjects,
you can imagine,

01:24:19.070 --> 01:24:21.980
are really listening
hard and trying

01:24:21.980 --> 01:24:23.860
to figure out the meaning
of a sentence that

01:24:23.860 --> 01:24:25.260
has a complicated grammar.

01:24:25.260 --> 01:24:27.610
I think we've all
written such senses.

01:24:27.610 --> 01:24:30.270
We've all tried to read
them from other writers.

01:24:30.270 --> 01:24:34.310
And it takes a lot of brain
power, then, to decode that,

01:24:34.310 --> 01:24:36.290
and figure out the meaning.

01:24:36.290 --> 01:24:39.460
And maybe that's
what happens here.

01:24:39.460 --> 01:24:41.860
Broca's area's called
in when the task gets

01:24:41.860 --> 01:24:45.390
more difficult than just
a simple list of words.

01:24:45.390 --> 01:24:46.610
In this case, it lit up.

01:24:46.610 --> 01:24:50.020
And in other studies, it's
clearly showing more activation

01:24:50.020 --> 01:24:52.252
when the task gets
more difficult.

01:24:55.521 --> 01:24:56.020
OK.

01:24:56.020 --> 01:24:57.585
So we're out of time.

01:24:58.920 --> 01:25:00.735
I can take a question or two.

01:25:03.490 --> 01:25:06.443
And just a reminder,
class meets at Mass.

01:25:06.443 --> 01:25:09.240
Eye and Ear on Wednesday
for the lab tour.

01:25:09.240 --> 01:25:11.660
So I'll see you over there.