1
00:00:00,080 --> 00:00:01,670
The following
content is provided

2
00:00:01,670 --> 00:00:03,820
under a Creative
Commons license.

3
00:00:03,820 --> 00:00:06,550
Your support will help MIT
OpenCourseWare continue

4
00:00:06,550 --> 00:00:10,160
to offer high-quality
educational resources for free.

5
00:00:10,160 --> 00:00:12,700
To make a donation, or to
view additional materials

6
00:00:12,700 --> 00:00:16,620
from hundreds of MIT courses,
visit MIT OpenCourseWare

7
00:00:16,620 --> 00:00:17,275
at ocw.mit.edu.

8
00:00:25,920 --> 00:00:27,730
PROFESSOR: Last
time we were talking

9
00:00:27,730 --> 00:00:34,980
about descending auditory
pathways and the brainstem

10
00:00:34,980 --> 00:00:38,520
reflexes, which
are, in some sense,

11
00:00:38,520 --> 00:00:39,970
part of those
descending pathways.

12
00:00:41,020 --> 00:00:43,520
And especially how
they protect our sense

13
00:00:43,520 --> 00:00:44,590
of hearing from damage.

14
00:00:44,590 --> 00:00:48,440
So I had a question
about subway noise

15
00:00:48,440 --> 00:00:49,850
on the Red Line in Boston.

16
00:00:49,850 --> 00:00:52,030
And I was asking
around at my hospital

17
00:00:52,030 --> 00:00:54,300
and nobody seems to
have measured it.

18
00:00:54,300 --> 00:00:57,840
But I saw online that there
were some measurements

19
00:00:57,840 --> 00:01:00,370
of subway noise
in New York City.

20
00:01:00,370 --> 00:01:02,870
And that it can
actually be damaging

21
00:01:02,870 --> 00:01:05,099
if you get exposed to
it for long enough.

22
00:01:05,099 --> 00:01:09,310
So there's a little news
clipping about subway noise.

23
00:01:09,310 --> 00:01:12,880
So I don't ride the
Red Line very often.

24
00:01:12,880 --> 00:01:16,010
I ride a commuter train
from Lincoln, where I live,

25
00:01:16,010 --> 00:01:17,570
to North Station.

26
00:01:17,570 --> 00:01:20,580
And when that train comes
in and puts on its brakes,

27
00:01:20,580 --> 00:01:21,915
it's incredibly shrill.

28
00:01:21,915 --> 00:01:23,700
And I always plug my years.

29
00:01:23,700 --> 00:01:27,420
And about half the people
do and the other half

30
00:01:27,420 --> 00:01:30,574
have hearing loss, I
guess, or whatever.

31
00:01:30,574 --> 00:01:31,990
It's not really a
laughing matter.

32
00:01:36,220 --> 00:01:38,630
So any questions from last time?

33
00:01:38,630 --> 00:01:42,710
Of course, the brainstem
reflexes we were talking about

34
00:01:42,710 --> 00:01:45,980
are also good for
other functions,

35
00:01:45,980 --> 00:01:50,300
like reducing the
effects of noise masking,

36
00:01:50,300 --> 00:01:55,410
allowing selective attention,
those types of functions.

37
00:01:55,410 --> 00:01:58,930
I also have experimental
evidence in support of them.

38
00:02:01,480 --> 00:02:04,340
So if there aren't
questions from last time,

39
00:02:04,340 --> 00:02:07,620
we are going to shift
gears a little bit

40
00:02:07,620 --> 00:02:11,110
and talk about something
completely different, which

41
00:02:11,110 --> 00:02:12,593
is sound localization.

42
00:02:14,260 --> 00:02:20,400
And so today's
roadmap is titled--

43
00:02:20,400 --> 00:02:22,240
headlined-- "Sounds
Localization."

44
00:02:22,240 --> 00:02:25,620
And we're going to be talking
about the kind of localization

45
00:02:25,620 --> 00:02:29,275
using binaural cues,
where you have two ears.

46
00:02:30,610 --> 00:02:35,100
And because you have two ears
and because sound sources

47
00:02:35,100 --> 00:02:39,250
are located off your
midline for the most part,

48
00:02:39,250 --> 00:02:45,120
we have some very prominent
cues called Interaural Time

49
00:02:45,120 --> 00:02:50,330
Differences, ITDs, and
Interaural Level Differences,

50
00:02:50,330 --> 00:02:50,830
ILDs.

51
00:02:54,950 --> 00:02:58,450
OK, so we'll talk about what
those are, how big they are.

52
00:02:59,670 --> 00:03:05,260
We'll talk about performance
for localizing sounds in humans.

53
00:03:05,260 --> 00:03:08,025
How good are we at
doing that task?

54
00:03:09,785 --> 00:03:11,076
We'll have some demonstrations.

55
00:03:12,160 --> 00:03:16,260
One of them I'm going to
give you in the room here

56
00:03:16,260 --> 00:03:22,120
and the other three demos we'll
have to listen to in headphones

57
00:03:22,120 --> 00:03:27,170
because we want to use just one
of these cues, interaural time

58
00:03:27,170 --> 00:03:28,900
or interaural level differences.

59
00:03:29,930 --> 00:03:33,100
And for the most part in a room,
in an ordinary environment,

60
00:03:33,100 --> 00:03:35,690
they're mixed up together.

61
00:03:35,690 --> 00:03:36,550
They come together.

62
00:03:38,000 --> 00:03:41,050
But using headphones
we can isolate and play

63
00:03:41,050 --> 00:03:42,020
just one or the other.

64
00:03:43,820 --> 00:03:46,380
Then, we'll launch into
the neural processing

65
00:03:46,380 --> 00:03:52,530
of one of these cues, ITDs, in
a part of the superior olivary

66
00:03:52,530 --> 00:03:56,325
complex called the Medial
Superior Olive, or the MSO.

67
00:03:57,750 --> 00:04:00,350
And so just a
heads up here, last

68
00:04:00,350 --> 00:04:05,460
lecture we were talking about
the neurons in the olive called

69
00:04:05,460 --> 00:04:08,250
the medial
olivocochlear neurons.

70
00:04:08,250 --> 00:04:09,710
Those are completely different.

71
00:04:09,710 --> 00:04:12,480
They have nothing to do
with sound localization.

72
00:04:14,050 --> 00:04:16,149
So those were the MOC neurons.

73
00:04:16,149 --> 00:04:19,439
And today, we're talking
about MSO neurons.

74
00:04:19,439 --> 00:04:20,730
These are completely different.

75
00:04:20,730 --> 00:04:24,690
They're neurons in a major part
of the superior olive called

76
00:04:24,690 --> 00:04:26,490
the medial superior
olive because it's

77
00:04:26,490 --> 00:04:27,350
in the medial part.

78
00:04:28,950 --> 00:04:31,540
And then finally, we'll
end up with a discussion

79
00:04:31,540 --> 00:04:32,750
of the assignment.

80
00:04:32,750 --> 00:04:34,720
We have a written
assignment for audition

81
00:04:34,720 --> 00:04:36,980
that's due in a few weeks.

82
00:04:36,980 --> 00:04:40,610
But it is based, in large
part, on today's lecture.

83
00:04:41,980 --> 00:04:46,500
So it talks about the model of
neural processing in the MSO.

84
00:04:47,510 --> 00:04:48,790
And another heads up.

85
00:04:48,790 --> 00:04:52,530
I've added a tiny little bit
to the end of the assignment.

86
00:04:52,530 --> 00:04:54,360
And we'll talk about
what I've added.

87
00:04:56,780 --> 00:04:59,460
And that revised
assignment is now

88
00:04:59,460 --> 00:05:01,090
posted on the course website.

89
00:05:01,090 --> 00:05:03,250
We'll talk about that at
the end of today's class.

90
00:05:06,370 --> 00:05:10,110
So sound localization
using binaural cues.

91
00:05:10,110 --> 00:05:11,250
What are the two cues?

92
00:05:11,250 --> 00:05:13,920
Well, they're the
interaural time differences

93
00:05:13,920 --> 00:05:16,170
and interaural
level differences.

94
00:05:17,176 --> 00:05:21,890
And this subject,
cat, is listening

95
00:05:21,890 --> 00:05:23,840
to this sound source
at the black dot.

96
00:05:25,250 --> 00:05:27,280
The sound source
is emitting sound.

97
00:05:28,760 --> 00:05:35,020
And because sound doesn't travel
instantaneously through air,

98
00:05:35,020 --> 00:05:37,670
the sound located off to
the right of the subject

99
00:05:37,670 --> 00:05:40,590
strikes the subject's
right ear first,

100
00:05:40,590 --> 00:05:42,810
and then there's a
little bit of time

101
00:05:42,810 --> 00:05:46,040
before it gets around
to the left ear, which

102
00:05:46,040 --> 00:05:47,445
is pointed away from the source.

103
00:05:48,800 --> 00:05:53,040
OK, so if this sound source is
emitting a click as diagrammed

104
00:05:53,040 --> 00:05:57,170
here, so this y-axis could
be the sound pressure

105
00:05:57,170 --> 00:06:04,270
level at the subject's
right and left eardrums

106
00:06:04,270 --> 00:06:05,700
as a function of time.

107
00:06:05,700 --> 00:06:07,510
This is the time axis.

108
00:06:07,510 --> 00:06:10,240
So obviously, the sound
source is off to the right.

109
00:06:10,240 --> 00:06:12,591
The right ear is going to
receive the sound pressure

110
00:06:12,591 --> 00:06:13,090
first.

111
00:06:14,660 --> 00:06:17,730
And then a little bit
later, the left ear

112
00:06:17,730 --> 00:06:19,430
will receive the sound source.

113
00:06:21,230 --> 00:06:23,030
I mean, will receive
the click sound.

114
00:06:24,090 --> 00:06:28,930
If the sound emitted is, instead
of a click, a continuous wave,

115
00:06:28,930 --> 00:06:29,950
like a sinusoidal wave.

116
00:06:31,580 --> 00:06:36,140
It'll be delayed the same
amount right versus left.

117
00:06:36,140 --> 00:06:39,710
So here's the right ear
sound pressure level,

118
00:06:39,710 --> 00:06:41,480
and then here is
the delayed version

119
00:06:41,480 --> 00:06:43,440
that appears at
the left eardrum.

120
00:06:45,010 --> 00:06:46,230
Just a delayed version.

121
00:06:48,060 --> 00:06:49,550
So how much is the delay?

122
00:06:49,550 --> 00:06:55,110
Well, it depends on things like
the velocity of sound in air.

123
00:06:56,740 --> 00:06:57,890
Sound in air.

124
00:06:57,890 --> 00:07:00,875
I think we had this in
the very first lecture.

125
00:07:02,830 --> 00:07:09,355
340 meters per second in air.

126
00:07:18,250 --> 00:07:21,740
Now, it depends a little bit
on the temperature of the air.

127
00:07:21,740 --> 00:07:24,820
And it depends a little bit
on the barometric pressure,

128
00:07:24,820 --> 00:07:27,295
but these factors
change at less than 1%.

129
00:07:28,750 --> 00:07:31,670
So about 340 meters
per second in air.

130
00:07:32,920 --> 00:07:39,480
So knowing how many meters
between the right and left ear,

131
00:07:39,480 --> 00:07:44,185
you can calculate for various
positions of the sound source

132
00:07:44,185 --> 00:07:46,800
the interaural time difference.

133
00:07:46,800 --> 00:07:50,640
Now, if you have a big head,
so that the left and right ears

134
00:07:50,640 --> 00:07:53,470
are separated very far
apart, obviously you're

135
00:07:53,470 --> 00:07:56,430
going to get a bigger
interaural time difference

136
00:07:56,430 --> 00:07:58,100
than if you have a
tiny, little head.

137
00:07:58,100 --> 00:08:00,990
Like a mouse sort or a bat.

138
00:08:00,990 --> 00:08:05,240
The smallest animals
have very close eardrums,

139
00:08:05,240 --> 00:08:08,365
so they have very small
interaural time differences.

140
00:08:10,970 --> 00:08:12,920
OK, now a couple
other things I should

141
00:08:12,920 --> 00:08:15,540
say about both of these
cues, interaural time

142
00:08:15,540 --> 00:08:18,550
differences and interaural
level differences,

143
00:08:18,550 --> 00:08:22,120
is they help us detect
where the sound is coming

144
00:08:22,120 --> 00:08:27,520
from in the azimuthal plane,
which is the horizontal plane.

145
00:08:27,520 --> 00:08:30,710
So here's a schematic
of a person's head

146
00:08:30,710 --> 00:08:32,000
with the left ear.

147
00:08:32,000 --> 00:08:33,250
This is the front of the head.

148
00:08:33,250 --> 00:08:36,830
And the right ear is
behind, so you can't see it.

149
00:08:36,830 --> 00:08:41,020
And so azimuth is in
the horizontal plane.

150
00:08:41,020 --> 00:08:43,580
And that's where
interaural time difference

151
00:08:43,580 --> 00:08:45,580
changes as a
function of position.

152
00:08:47,140 --> 00:08:51,190
The other plane perpendicular to
that is the plane of elevation.

153
00:08:51,190 --> 00:08:55,060
So for me going straight
up and straight down

154
00:08:55,060 --> 00:08:57,970
is the elevation of
the sound source.

155
00:08:57,970 --> 00:09:01,740
And you can imagine if there's
a sound source straight ahead,

156
00:09:01,740 --> 00:09:05,000
it's going to strike my two
eardrums at the same time

157
00:09:05,000 --> 00:09:07,900
because the path length from
the source to the eardrums

158
00:09:07,900 --> 00:09:09,610
is the same.

159
00:09:09,610 --> 00:09:11,450
And if that sound
source moves up

160
00:09:11,450 --> 00:09:13,770
from being straight
ahead to, say,

161
00:09:13,770 --> 00:09:15,640
elevated from straight ahead.

162
00:09:15,640 --> 00:09:21,200
But still, it's in the same
place relative to the eardrums.

163
00:09:21,200 --> 00:09:24,630
The path from the sound
source to the two eardrums

164
00:09:24,630 --> 00:09:26,210
is going to be the same.

165
00:09:26,210 --> 00:09:31,310
So the ITD does not change
as a function of sound source

166
00:09:31,310 --> 00:09:31,810
elevation.

167
00:09:33,720 --> 00:09:38,860
So these binaural cues that
we're talking about here

168
00:09:38,860 --> 00:09:41,460
do not change as a
function of elevation.

169
00:09:41,460 --> 00:09:47,110
So how do we detect the change
in elevation of a sound source?

170
00:09:47,110 --> 00:09:49,170
Well, we talked
earlier in the course.

171
00:09:49,170 --> 00:09:51,370
I think in the
very first lecture

172
00:09:51,370 --> 00:09:57,280
about the so-called
"pinna" cues.

173
00:10:02,770 --> 00:10:05,970
And so we have these
external ears, the pinnae.

174
00:10:07,640 --> 00:10:11,160
And they help us
greatly in detecting

175
00:10:11,160 --> 00:10:16,250
sounds that differ in
elevation because they

176
00:10:16,250 --> 00:10:21,830
put onto the sound spectrum some
very prominent peaks and nulls.

177
00:10:21,830 --> 00:10:24,350
And those peaks and nulls
change as a function

178
00:10:24,350 --> 00:10:25,235
of sound elevation.

179
00:10:26,630 --> 00:10:29,930
So go back to the very first
lecture that I gave you

180
00:10:29,930 --> 00:10:31,310
and review that.

181
00:10:31,310 --> 00:10:37,100
And we read a paper where if you
distort the pinnae by putting

182
00:10:37,100 --> 00:10:40,230
little plastic or clay
ear molds in your pinnae,

183
00:10:40,230 --> 00:10:42,790
your detectability
of sound sources that

184
00:10:42,790 --> 00:10:45,640
change in elevation goes to pot.

185
00:10:45,640 --> 00:10:47,110
You can't do it anymore.

186
00:10:48,530 --> 00:10:53,200
But if you go out and
re-experience sound

187
00:10:53,200 --> 00:10:55,810
with those little
distortions of your pinnae

188
00:10:55,810 --> 00:10:58,030
and come back in
a few weeks, you

189
00:10:58,030 --> 00:11:01,620
can re-learn how
to detect changes

190
00:11:01,620 --> 00:11:02,760
in sound source elevation.

191
00:11:05,220 --> 00:11:07,870
And in those kinds of
things, two comments.

192
00:11:07,870 --> 00:11:10,080
Number one, you
don't need two ears.

193
00:11:10,080 --> 00:11:13,000
You can detect change in
sound source elevation

194
00:11:13,000 --> 00:11:16,085
with just one ear, because
one ear has a pinna

195
00:11:16,085 --> 00:11:18,520
and it works just
fine in changing

196
00:11:18,520 --> 00:11:21,310
the spectrum from one ear.

197
00:11:21,310 --> 00:11:25,740
And number two, your
sensitivity to small changes

198
00:11:25,740 --> 00:11:29,620
in sound source elevation
is not very good.

199
00:11:29,620 --> 00:11:33,970
So if you change the sound
source elevation of 10 degrees,

200
00:11:33,970 --> 00:11:37,010
most people cannot
detect that change.

201
00:11:37,010 --> 00:11:40,975
The minimum changes in sound
elevation that are detectable

202
00:11:40,975 --> 00:11:45,360
are more like 30 degrees,
which is a pretty big change.

203
00:11:47,250 --> 00:11:51,810
We'll see using the binaural
cues to detect changes

204
00:11:51,810 --> 00:11:56,890
in sound source azimuth,
we're down to about one degree

205
00:11:56,890 --> 00:11:58,232
using these binaural cues.

206
00:11:58,232 --> 00:11:59,190
So they're much better.

207
00:11:59,190 --> 00:12:00,980
You're much more
accurate in terms

208
00:12:00,980 --> 00:12:05,595
of localizing sound in azimuth
than you are in elevation.

209
00:12:11,150 --> 00:12:13,610
Secondly, I should
have mentioned here

210
00:12:13,610 --> 00:12:18,220
when we had this time delay
for the ongoing signal,

211
00:12:18,220 --> 00:12:24,130
like a sinusoid wave, that
is equivalent to a phase

212
00:12:24,130 --> 00:12:24,630
difference.

213
00:12:25,760 --> 00:12:30,320
The phase is what
engineers use to define

214
00:12:30,320 --> 00:12:35,045
where and when a sinusoidal
source starts and ends.

215
00:12:36,090 --> 00:12:38,770
So engineers talk
about a sinusoid

216
00:12:38,770 --> 00:12:43,710
going through a 360 degree
phase for one cycle.

217
00:12:46,420 --> 00:12:54,060
And so this sinusoid is
delayed about a quarter cycle

218
00:12:54,060 --> 00:12:56,540
relative to the right ear one.

219
00:12:56,540 --> 00:13:00,550
And so it has a phase
lag or a phase difference

220
00:13:00,550 --> 00:13:02,305
of about 90 degrees.

221
00:13:03,530 --> 00:13:08,560
So you can talk about
interaural phase differences

222
00:13:08,560 --> 00:13:10,950
for continuous waveforms,
like sinusoids.

223
00:13:10,950 --> 00:13:14,170
And you talk about them
in terms of the number

224
00:13:14,170 --> 00:13:17,130
of degrees of phase difference.

225
00:13:18,380 --> 00:13:21,010
And of course, to convert
from one to the other you

226
00:13:21,010 --> 00:13:24,060
need to know the frequency
that you're talking about.

227
00:13:24,060 --> 00:13:27,940
Because if you're dealing with
a high-frequency sinusoid that

228
00:13:27,940 --> 00:13:31,360
goes back and forth
a lot of times,

229
00:13:31,360 --> 00:13:35,050
you have to know that to
calculate the interaural time

230
00:13:35,050 --> 00:13:36,210
difference.

231
00:13:36,210 --> 00:13:38,190
But some of our demos
that we listened to

232
00:13:38,190 --> 00:13:41,980
will quote the interaural
delay in terms of phase.

233
00:13:41,980 --> 00:13:43,315
An interaural phase difference.

234
00:13:47,880 --> 00:13:50,879
Now, the second cue is
interaural level difference.

235
00:13:50,879 --> 00:13:52,045
And here's the same subject.

236
00:13:52,045 --> 00:13:53,595
Here's the same sound source.

237
00:13:55,060 --> 00:13:57,390
The sound is coming to
the subject's right ear

238
00:13:57,390 --> 00:13:59,400
and it's a high level
there because there's

239
00:13:59,400 --> 00:14:00,255
a direct pathway.

240
00:14:02,340 --> 00:14:05,470
But the sound to get to
the left ear of the subject

241
00:14:05,470 --> 00:14:08,910
has to go around
the subject's head.

242
00:14:08,910 --> 00:14:12,380
And sound does
bend around a wall.

243
00:14:12,380 --> 00:14:16,684
So I can go out in the
hall and I can say, class,

244
00:14:16,684 --> 00:14:17,600
can you still hear me?

245
00:14:17,600 --> 00:14:19,390
Yes, you can still
hear me, right?

246
00:14:19,390 --> 00:14:20,692
The sound is bending around.

247
00:14:20,692 --> 00:14:22,150
Of course, some of
it's reflecting.

248
00:14:23,290 --> 00:14:25,430
Sound bends around,
but some sound

249
00:14:25,430 --> 00:14:29,930
bends around more easily and
more effectively than others.

250
00:14:29,930 --> 00:14:32,680
And obviously, this
sound bent around.

251
00:14:32,680 --> 00:14:35,450
There's still sound at
the subject's left ear.

252
00:14:35,450 --> 00:14:38,780
This should actually be
delayed because it took longer.

253
00:14:38,780 --> 00:14:42,840
But the purpose here is
that it's lower in amplitude

254
00:14:42,840 --> 00:14:44,770
and it's not vanishingly small.

255
00:14:44,770 --> 00:14:47,755
So there's less sound over
the subject's left ear.

256
00:14:48,990 --> 00:14:51,070
And this is the interaural
level difference.

257
00:14:53,120 --> 00:14:55,080
This interaural
level difference,

258
00:14:55,080 --> 00:14:57,900
as we'll see in just a
minute, depends greatly

259
00:14:57,900 --> 00:14:58,770
on sound frequency.

260
00:15:00,460 --> 00:15:04,000
Such that very
low-frequency sounds,

261
00:15:04,000 --> 00:15:06,260
because they have
a long wavelength

262
00:15:06,260 --> 00:15:09,230
relative to the object
they're bending around.

263
00:15:09,230 --> 00:15:11,650
Because of their
physical characteristics,

264
00:15:11,650 --> 00:15:13,310
they bend around very well.

265
00:15:14,690 --> 00:15:17,300
And so the interaural level
difference, as we'll see,

266
00:15:17,300 --> 00:15:19,835
is almost 0 for low frequencies.

267
00:15:21,410 --> 00:15:22,980
For very high
frequencies, there's

268
00:15:22,980 --> 00:15:25,240
a big, so-called sound shadow.

269
00:15:26,530 --> 00:15:29,440
They don't bend around
an object the size of the

270
00:15:29,440 --> 00:15:31,170
had very easily.

271
00:15:31,170 --> 00:15:34,470
And so there's a big
interaural level difference

272
00:15:34,470 --> 00:15:36,100
for high frequencies.

273
00:15:36,100 --> 00:15:39,950
So we think of these cues
then as being very important

274
00:15:39,950 --> 00:15:41,000
for high frequencies.

275
00:15:44,350 --> 00:15:46,750
So let me just note
that down here.

276
00:15:46,750 --> 00:15:52,830
So ILDs-- I'll use
a different color.

277
00:15:55,286 --> 00:15:55,786
Important.

278
00:16:05,530 --> 00:16:08,810
And let me show you some
data to support that then.

279
00:16:08,810 --> 00:16:15,870
These are some data taken for
the two cues for a human head.

280
00:16:17,010 --> 00:16:18,840
And how do we get these data?

281
00:16:18,840 --> 00:16:22,730
Well, we can take
two small microphones

282
00:16:22,730 --> 00:16:25,230
and put them down in our
ear canals very close

283
00:16:25,230 --> 00:16:26,040
to our eardrums.

284
00:16:26,040 --> 00:16:30,410
And we can measure the time
difference and the level

285
00:16:30,410 --> 00:16:30,910
difference.

286
00:16:30,910 --> 00:16:33,230
And so if you were
to do this, you'd

287
00:16:33,230 --> 00:16:36,220
seat a subject in
an anechoic room.

288
00:16:36,220 --> 00:16:39,270
An anechoic room is one that
doesn't have any echoes coming

289
00:16:39,270 --> 00:16:40,610
off the walls and the ceilings.

290
00:16:41,940 --> 00:16:45,430
So that the subject just
experiences the direct sound

291
00:16:45,430 --> 00:16:47,020
from whatever sound source.

292
00:16:47,020 --> 00:16:49,020
And the sound source we're
going to move around.

293
00:16:50,310 --> 00:16:54,290
So this x-axis is the
sound source position

294
00:16:54,290 --> 00:16:56,790
in angles from directly ahead.

295
00:16:56,790 --> 00:16:58,100
So directly ahead is 0.

296
00:17:00,170 --> 00:17:03,930
And higher angles is off,
let's say, to one side.

297
00:17:05,310 --> 00:17:07,664
And directly off to
the side is 90 degrees.

298
00:17:09,099 --> 00:17:13,060
And then behind the subject
is greater than 90 degrees.

299
00:17:13,060 --> 00:17:16,760
And directly behind the
subject will be 180 degrees.

300
00:17:16,760 --> 00:17:20,819
So this is degrees of
azimuth from directly ahead.

301
00:17:20,819 --> 00:17:25,560
This top graph shows you the
interaural time differences

302
00:17:25,560 --> 00:17:27,790
and the bottom graph shows
you the interaural level

303
00:17:27,790 --> 00:17:28,290
differences.

304
00:17:30,410 --> 00:17:34,365
So how do we compute time
from the microphone signals?

305
00:17:37,940 --> 00:17:44,310
Well, we simply
take our microphones

306
00:17:44,310 --> 00:17:48,130
and run them to an oscilloscope.

307
00:17:48,130 --> 00:17:49,200
So here's the head.

308
00:17:54,490 --> 00:17:59,590
Here's microphone 1 in the
ear canal on the left side.

309
00:17:59,590 --> 00:18:02,105
Here's microphone 2 in the
ear canal on the right side.

310
00:18:04,040 --> 00:18:10,825
You send the wires from those
two signals to an oscilloscope.

311
00:18:12,570 --> 00:18:15,740
The top channel
will be the signal

312
00:18:15,740 --> 00:18:17,055
coming from the left side.

313
00:18:18,240 --> 00:18:21,590
The lower channel is the signal
coming from the right side.

314
00:18:21,590 --> 00:18:23,710
And this device,
the oscilloscope,

315
00:18:23,710 --> 00:18:26,890
plots the voltage, which is
equivalent to the pressure

316
00:18:26,890 --> 00:18:28,120
as a function of time.

317
00:18:29,680 --> 00:18:34,160
And we can measure the
time when this sound starts

318
00:18:34,160 --> 00:18:40,670
and the delay, or
interaural time difference,

319
00:18:40,670 --> 00:18:42,860
when the second ear starts.

320
00:18:42,860 --> 00:18:44,170
So it's very simple to measure.

321
00:18:45,820 --> 00:18:50,070
Now that said, it's
also very simple

322
00:18:50,070 --> 00:18:52,460
to assume that the human
head is just a sphere.

323
00:18:53,740 --> 00:18:58,520
And it's simple
to compute knowing

324
00:18:58,520 --> 00:19:00,230
the distance
between the two ears

325
00:19:00,230 --> 00:19:01,900
and the angle of
the sound source.

326
00:19:01,900 --> 00:19:07,270
It's very simple to assume that
this human head is a sphere.

327
00:19:07,270 --> 00:19:10,700
And I think this solid line
is assuming it's a sphere.

328
00:19:11,710 --> 00:19:17,040
And the dashed line with x's
are the experimental data.

329
00:19:17,040 --> 00:19:20,050
The data are pretty
close to the assumption

330
00:19:20,050 --> 00:19:24,320
that the human head is a
sphere for interaural time

331
00:19:24,320 --> 00:19:25,190
differences.

332
00:19:25,190 --> 00:19:30,330
And here's the ITD plotted
as a function of the angle.

333
00:19:30,330 --> 00:19:34,230
As you would expect if the
sound source is straight ahead--

334
00:19:34,230 --> 00:19:37,690
that is 0 degrees
angle-- the ITD is 0.

335
00:19:38,970 --> 00:19:42,320
The sound takes the same
time to get to the two ears

336
00:19:42,320 --> 00:19:44,000
because it's straight ahead.

337
00:19:44,000 --> 00:19:48,060
Now as you move the
sound over to one side,

338
00:19:48,060 --> 00:19:50,475
it's going to get to one
ear first and the other ear

339
00:19:50,475 --> 00:19:51,840
a little bit later.

340
00:19:51,840 --> 00:19:54,610
And so the ITD
becomes bigger than 0.

341
00:19:55,620 --> 00:19:57,830
And it goes up.

342
00:19:57,830 --> 00:19:59,660
And the maximal
ITD, as you would

343
00:19:59,660 --> 00:20:01,580
expect when the
sound is directly off

344
00:20:01,580 --> 00:20:03,800
to the side, which
is at 90 degrees.

345
00:20:05,430 --> 00:20:07,720
And what are the units here?

346
00:20:07,720 --> 00:20:14,220
You probably can't read them,
but they go from 0 to 0.6.

347
00:20:14,220 --> 00:20:16,026
And those units
are milliseconds.

348
00:20:20,560 --> 00:20:33,540
So the ITDs go from
0 to 0.6 milliseconds

349
00:20:33,540 --> 00:20:34,740
for the human head.

350
00:20:36,270 --> 00:20:38,760
So millisecond is a
thousandth of a second.

351
00:20:38,760 --> 00:20:39,710
It's pretty small.

352
00:20:39,710 --> 00:20:44,210
It's less than 1 millisecond
for the maximum ITD.

353
00:20:44,210 --> 00:20:48,590
And so you can quote this in
terms of microseconds as well.

354
00:20:48,590 --> 00:20:51,330
So this would be a
total, a maximal ITD

355
00:20:51,330 --> 00:20:52,416
of 600 microseconds.

356
00:20:55,255 --> 00:20:55,755
OK.

357
00:21:02,270 --> 00:21:04,675
Now, how about interaural
level differences?

358
00:21:07,040 --> 00:21:09,950
They are in the lower panel.

359
00:21:09,950 --> 00:21:12,550
And these are all
measured values.

360
00:21:12,550 --> 00:21:13,425
They're not computed.

361
00:21:14,840 --> 00:21:16,840
I don't know why, you
could compute them easily.

362
00:21:18,470 --> 00:21:20,050
Maybe it's because of the pinna.

363
00:21:20,050 --> 00:21:24,030
The pinna introduces
another piece of complexity

364
00:21:24,030 --> 00:21:25,980
that's a little bit
different from the head

365
00:21:25,980 --> 00:21:27,455
being an absolute sphere.

366
00:21:28,660 --> 00:21:30,390
So these are measured.

367
00:21:30,390 --> 00:21:34,090
And these are, again, plotted
as a function of the azimuth.

368
00:21:34,090 --> 00:21:38,480
0 degrees is straight ahead,
90 degrees is off to one side,

369
00:21:38,480 --> 00:21:42,090
and 180 degrees of azimuth
is behind the subject.

370
00:21:43,230 --> 00:21:46,915
And these, I said the
ILDs depend on frequency.

371
00:21:47,990 --> 00:21:50,009
So these are separate
plots for a bunch

372
00:21:50,009 --> 00:21:51,050
of different frequencies.

373
00:21:52,240 --> 00:21:55,910
Down at the bottom is a very
low frequency, 200 Hertz.

374
00:21:58,050 --> 00:22:01,300
This is our standard
center of the human hearing

375
00:22:01,300 --> 00:22:03,730
range, 1,000 Hertz.

376
00:22:03,730 --> 00:22:05,370
And this is a very
high frequency

377
00:22:05,370 --> 00:22:09,540
at the top, 6,000 Hertz.

378
00:22:09,540 --> 00:22:12,250
And as you can see quite
clearly, a 200 Hertz,

379
00:22:12,250 --> 00:22:15,930
like I said before,
this frequency of sound

380
00:22:15,930 --> 00:22:19,830
bends around very
nicely for objects

381
00:22:19,830 --> 00:22:21,720
the size of the human head.

382
00:22:21,720 --> 00:22:24,350
And so even if you have the
sound source all the way

383
00:22:24,350 --> 00:22:29,227
to the right, The ILD
for 200 Hertz is 0.

384
00:22:30,720 --> 00:22:32,830
You don't have an
ILD, essentially,

385
00:22:32,830 --> 00:22:34,065
for such low frequencies.

386
00:22:35,420 --> 00:22:39,080
For our mid-human
hearing frequency,

387
00:22:39,080 --> 00:22:42,770
the ILD starts out
at 0, straight ahead,

388
00:22:42,770 --> 00:22:48,336
and climbs to perhaps 6 or 8 dB.

389
00:22:48,336 --> 00:22:50,660
As the sound source
moves off to the side,

390
00:22:50,660 --> 00:22:53,610
it's not behaving
like a perfect sphere.

391
00:22:53,610 --> 00:22:55,130
So a perfect sphere,
the ILD would

392
00:22:55,130 --> 00:22:58,050
go up and be maximal
at 90 degrees.

393
00:22:58,050 --> 00:23:00,670
So maybe this is the
effect of the pinna.

394
00:23:00,670 --> 00:23:04,020
This is not doing that, but
it's certainly climbing.

395
00:23:06,850 --> 00:23:10,590
The biggest ILD is found at
these highest frequencies.

396
00:23:10,590 --> 00:23:14,422
So for 6,000 Hertz,
the ILD climbs from 0

397
00:23:14,422 --> 00:23:18,120
up to an ILD of
20 dB as the sound

398
00:23:18,120 --> 00:23:19,835
is located off to one side.

399
00:23:21,140 --> 00:23:26,730
And that's a huge cue for
localization of sounds.

400
00:23:26,730 --> 00:23:30,760
So maybe you can remember
back to the very first

401
00:23:30,760 --> 00:23:35,200
of my lectures, where we had
the decibel demonstration.

402
00:23:35,200 --> 00:23:36,675
We changed decibel levels.

403
00:23:37,710 --> 00:23:40,010
I think this was a
noise that we presented

404
00:23:40,010 --> 00:23:42,720
in changing steps of 6 dB.

405
00:23:42,720 --> 00:23:44,770
And it was clearly
obvious when you

406
00:23:44,770 --> 00:23:50,180
change-- this is now
binaural level at 6 dB.

407
00:23:50,180 --> 00:23:53,760
3 dB changes in noise
level were obvious.

408
00:23:53,760 --> 00:23:59,480
A 1 dB change in noise
level as we stepped through

409
00:23:59,480 --> 00:24:01,980
was obvious after you'd
gone through a few steps.

410
00:24:01,980 --> 00:24:06,185
But maybe from one to the
next step was not obvious.

411
00:24:07,310 --> 00:24:10,770
Clearly, a 20 dB difference
between the two ears

412
00:24:10,770 --> 00:24:11,770
is a huge cue.

413
00:24:14,840 --> 00:24:18,160
How can we confirm
that that's a huge cue?

414
00:24:18,160 --> 00:24:22,704
Well, we can take a these
cues and present them

415
00:24:22,704 --> 00:24:23,495
through headphones.

416
00:24:24,880 --> 00:24:28,950
And it's very easy
to set up a circuit

417
00:24:28,950 --> 00:24:33,160
so that you have a signal
coming into the two headphones.

418
00:24:33,160 --> 00:24:35,410
And in one channel,
you put a device

419
00:24:35,410 --> 00:24:39,050
called an attenuator that
attenuates the signal.

420
00:24:39,050 --> 00:24:42,540
So why not use an amplifier
that amplifies the signal?

421
00:24:42,540 --> 00:24:45,320
Well, amplifiers
are active devices.

422
00:24:45,320 --> 00:24:48,629
And if you amplify
sounds, inevitably you

423
00:24:48,629 --> 00:24:49,920
add a little bit of distortion.

424
00:24:51,640 --> 00:24:54,340
So it's easier to just
attenuate the sound.

425
00:24:54,340 --> 00:24:56,120
Attenuators are
passive devices and you

426
00:24:56,120 --> 00:24:59,840
can cut the sound
to one channel 20 dB

427
00:24:59,840 --> 00:25:01,780
and it sounds like
a huge effect.

428
00:25:01,780 --> 00:25:02,810
And you say, oh, wow.

429
00:25:02,810 --> 00:25:04,310
That sound disappeared
in one ear.

430
00:25:04,310 --> 00:25:06,060
It sounds like it's
coming from the other.

431
00:25:06,840 --> 00:25:09,790
You can cut the sound
by 6 dB, a huge effect.

432
00:25:09,790 --> 00:25:12,610
Cut it by 3 dB, a big effect.

433
00:25:12,610 --> 00:25:14,550
And you cut it by
1 dB and the person

434
00:25:14,550 --> 00:25:17,175
says, well, before when the
sound was the same in the two

435
00:25:17,175 --> 00:25:19,030
ears, the sound
was straight ahead.

436
00:25:19,030 --> 00:25:21,370
When you changed it 1 dB,
it sounded like it just

437
00:25:21,370 --> 00:25:23,320
moved a little bit
from straight ahead.

438
00:25:24,430 --> 00:25:30,546
So it turns out that 1 dB is not
only our just detectable level

439
00:25:30,546 --> 00:25:34,010
in sounds when we present
them to the two ears,

440
00:25:34,010 --> 00:25:37,340
but when we vary the
interaural level difference.

441
00:25:39,350 --> 00:25:49,890
So 1 dB is our just
noticeable difference for ILD.

442
00:25:52,600 --> 00:25:54,920
OK, we can play that
trick with headphones.

443
00:25:54,920 --> 00:25:56,828
Do good tests on that.

444
00:26:01,810 --> 00:26:05,420
So clearly, these cues are very
important at high frequencies

445
00:26:05,420 --> 00:26:07,470
because they're
the salient cues.

446
00:26:07,470 --> 00:26:08,980
What about at low frequency?

447
00:26:08,980 --> 00:26:12,680
Since we don't have an
interaural level difference,

448
00:26:12,680 --> 00:26:15,800
you would assume that we're
going to use interaural time

449
00:26:15,800 --> 00:26:17,350
differences at low frequencies.

450
00:26:17,350 --> 00:26:18,990
And you'd be correct.

451
00:26:18,990 --> 00:26:21,160
Much of the evidence
suggests that ITDs

452
00:26:21,160 --> 00:26:25,510
are used at low
frequencies and ILDs,

453
00:26:25,510 --> 00:26:27,690
because they're big at
high frequencies-- those

454
00:26:27,690 --> 00:26:29,450
are the cues used there.

455
00:26:29,450 --> 00:26:32,290
And why don't we use
ITDs at high frequencies?

456
00:26:34,330 --> 00:26:40,310
OK, there is nothing in the
physical characteristics

457
00:26:40,310 --> 00:26:40,810
of sound.

458
00:26:41,830 --> 00:26:46,040
For example, the sound velocity
doesn't depend on frequency.

459
00:26:46,040 --> 00:26:49,545
It's constant no matter what
the frequency of the sound.

460
00:26:51,060 --> 00:26:54,754
So you're going to get the
same ITD for low frequencies

461
00:26:54,754 --> 00:26:55,795
and for high frequencies.

462
00:26:57,440 --> 00:26:59,190
Well, if you think
about it a little bit,

463
00:26:59,190 --> 00:27:03,580
there is a reason where
high-frequency ITDs

464
00:27:03,580 --> 00:27:04,515
are less useful.

465
00:27:07,540 --> 00:27:10,730
At some frequencies,
this sound waveform

466
00:27:10,730 --> 00:27:15,150
is going to go back
and forth so quickly

467
00:27:15,150 --> 00:27:16,910
that it might hit the right ear.

468
00:27:18,110 --> 00:27:21,220
And then by the time it
leaks around to the left ear,

469
00:27:21,220 --> 00:27:22,990
it goes through
a complete cycle.

470
00:27:25,540 --> 00:27:27,550
Then, what are we left with?

471
00:27:27,550 --> 00:27:30,785
We're left with right and
left looking exactly the same.

472
00:27:32,050 --> 00:27:37,170
There's a big time difference,
an interaural phase difference

473
00:27:37,170 --> 00:27:39,490
of 360 degrees.

474
00:27:39,490 --> 00:27:41,190
But we'll never be
able to perceive

475
00:27:41,190 --> 00:27:43,970
that because the
sound is, again,

476
00:27:43,970 --> 00:27:45,335
the same at the two ears.

477
00:27:46,850 --> 00:27:49,790
And it turns out,
of course, that

478
00:27:49,790 --> 00:27:52,790
depends on how widely
spaced your ears are.

479
00:27:53,980 --> 00:27:58,750
For the size of a
human head, the phase

480
00:27:58,750 --> 00:28:00,990
goes completely
through one cycle

481
00:28:00,990 --> 00:28:06,415
at a frequency of 1.6 kilohertz.

482
00:28:07,550 --> 00:28:11,690
And that's pretty centered
in your hearing range.

483
00:28:11,690 --> 00:28:14,410
1 kilohertz is about the
middle of your hearing range.

484
00:28:14,410 --> 00:28:15,790
1.6 is pretty close.

485
00:28:17,520 --> 00:28:24,423
So above 1.6 kilohertz,
ITDs become ambiguous.

486
00:28:26,320 --> 00:28:33,590
And so we think of that as
the time where ITDs fall out

487
00:28:33,590 --> 00:28:36,280
and ILDs become important.

488
00:28:36,280 --> 00:28:46,940
So ITDs are ambiguous for
humans above 1 kilohertz.

489
00:28:46,940 --> 00:28:53,165
Now, that argument, of course,
is for ongoing interaural time

490
00:28:53,165 --> 00:28:53,665
differences.

491
00:28:54,750 --> 00:28:57,190
If you start the sound,
no matter what frequency,

492
00:28:57,190 --> 00:28:59,135
you're going to have
an onset difference.

493
00:29:00,420 --> 00:29:03,635
You can start a 2
kilohertz sound.

494
00:29:05,016 --> 00:29:07,640
The left ear will get it first,
and then the right ear a little

495
00:29:07,640 --> 00:29:09,350
bit later, depending
on where it's

496
00:29:09,350 --> 00:29:12,870
located and the degree of
separation of your two ears.

497
00:29:12,870 --> 00:29:16,010
But that's just one cue.

498
00:29:16,010 --> 00:29:19,070
And if these things are repeated
back and forth thousands

499
00:29:19,070 --> 00:29:21,180
of times, depending
on how long they're

500
00:29:21,180 --> 00:29:24,160
on for, for low-frequency
sounds you'll

501
00:29:24,160 --> 00:29:27,560
get those thousands of cues.

502
00:29:29,360 --> 00:29:31,560
But for high frequencies,
you'll just get the onset.

503
00:29:33,910 --> 00:29:39,430
So for high frequencies, we
think of these frequencies

504
00:29:39,430 --> 00:29:43,570
as not being useful for
ITDs for these ongoing cues,

505
00:29:43,570 --> 00:29:45,125
like interaural
phase differences.

506
00:29:47,470 --> 00:29:48,705
OK, is that clear?

507
00:29:49,890 --> 00:29:50,590
Any questions?

508
00:29:54,030 --> 00:29:54,565
All right.

509
00:29:58,030 --> 00:30:00,690
Let's see how good we are
in terms of performance.

510
00:30:03,290 --> 00:30:07,010
Sound localization is
accurate and humans

511
00:30:07,010 --> 00:30:08,615
have excellent performance.

512
00:30:10,662 --> 00:30:12,120
How do we measure
such performance?

513
00:30:12,120 --> 00:30:13,203
Well, we take an observer.

514
00:30:14,580 --> 00:30:18,270
We sit them down in a
room or in free field

515
00:30:18,270 --> 00:30:21,110
where there's only the
direct sound coming

516
00:30:21,110 --> 00:30:22,290
from the sound source.

517
00:30:22,290 --> 00:30:25,690
And no reflections off
the walls or echoes

518
00:30:25,690 --> 00:30:26,880
off the ceiling or floor.

519
00:30:28,490 --> 00:30:31,250
And we say to this
observer, OK, here's

520
00:30:31,250 --> 00:30:32,990
the sound directly ahead.

521
00:30:34,260 --> 00:30:35,810
And then we move
the sound source.

522
00:30:35,810 --> 00:30:38,370
And this can be a
physical speaker.

523
00:30:38,370 --> 00:30:40,470
We move the speaker
over a little bit

524
00:30:40,470 --> 00:30:41,760
and so we change azimuth.

525
00:30:43,100 --> 00:30:50,155
And we ask the person, was
that azimuth detectable?

526
00:30:51,470 --> 00:30:56,690
We say, is that the same or
different from straight ahead?

527
00:30:56,690 --> 00:30:59,790
So if straight ahead is here
and we move the speaker over

528
00:30:59,790 --> 00:31:04,000
to here and we say to the
person, same or different?

529
00:31:04,000 --> 00:31:05,830
Same position or
different position?

530
00:31:07,170 --> 00:31:08,830
And the person says, yeah, sure.

531
00:31:08,830 --> 00:31:09,570
That's different.

532
00:31:09,570 --> 00:31:12,770
And then we instead say,
OK, straight ahead is here

533
00:31:12,770 --> 00:31:14,140
and we're going to move it less.

534
00:31:14,140 --> 00:31:16,840
The person says, sure,
I can detect that.

535
00:31:16,840 --> 00:31:17,370
No problem.

536
00:31:17,370 --> 00:31:18,665
It moved off to the right.

537
00:31:18,665 --> 00:31:21,250
And this is, again,
without any visual cues.

538
00:31:21,250 --> 00:31:23,205
So the speaker has to
be behind a screen.

539
00:31:24,270 --> 00:31:26,510
So it's only auditory
cues that allow

540
00:31:26,510 --> 00:31:29,010
you detect this
change in movement.

541
00:31:29,010 --> 00:31:31,130
So we do it a third
time, straight ahead,

542
00:31:31,130 --> 00:31:35,580
and we move it such a small
degree that the person says,

543
00:31:35,580 --> 00:31:37,580
I don't hear that change at all.

544
00:31:37,580 --> 00:31:40,440
And we titrate the
level of movement

545
00:31:40,440 --> 00:31:46,930
until we get what's called
the minimum audible angle

546
00:31:46,930 --> 00:31:47,575
in degrees.

547
00:31:50,030 --> 00:31:52,670
And we can do that for a whole
bunch of different frequencies.

548
00:31:56,500 --> 00:31:59,550
And when we do that
with sound sources,

549
00:31:59,550 --> 00:32:03,400
such that the initial position
is 0 degrees straight ahead,

550
00:32:03,400 --> 00:32:07,350
we get this curve linked
by the black dots.

551
00:32:08,950 --> 00:32:13,630
Such that at low frequencies,
the minimum audible angles

552
00:32:13,630 --> 00:32:15,385
approach, in the
best performance,

553
00:32:15,385 --> 00:32:17,245
1 degree of azimuth.

554
00:32:18,570 --> 00:32:21,160
An observer can
tell the difference

555
00:32:21,160 --> 00:32:25,920
between 0 degrees straight ahead
and 1 degree to the right or 1

556
00:32:25,920 --> 00:32:27,200
degree to the left.

557
00:32:27,200 --> 00:32:28,575
It's very impressive
performance.

558
00:32:31,170 --> 00:32:35,580
In the middle frequencies,
where the ITDs

559
00:32:35,580 --> 00:32:41,180
are starting to break
down and become ambiguous,

560
00:32:41,180 --> 00:32:43,910
our performance is not so good.

561
00:32:43,910 --> 00:32:46,825
Minimum audible angles go
up to perhaps 3 degrees.

562
00:32:49,590 --> 00:32:51,520
And then they get
a little bit better

563
00:32:51,520 --> 00:32:52,790
at the very high frequencies.

564
00:32:52,790 --> 00:32:55,880
And remember here, we said
that the interaural level

565
00:32:55,880 --> 00:32:58,225
differences were very salient.

566
00:32:59,820 --> 00:33:03,470
So at 5 kilohertz, there
was a 20 dB difference.

567
00:33:03,470 --> 00:33:06,700
And we come back down,
maybe not to 1 degree

568
00:33:06,700 --> 00:33:08,820
minimum audible
angle, but maybe to 2.

569
00:33:08,820 --> 00:33:10,380
So almost as good.

570
00:33:11,490 --> 00:33:13,600
And then at very
high frequencies,

571
00:33:13,600 --> 00:33:14,660
we get worse again.

572
00:33:19,040 --> 00:33:23,600
Now, we can take
these same subjects

573
00:33:23,600 --> 00:33:27,375
and we can go back
to the physical cues.

574
00:33:29,860 --> 00:33:36,210
Let's say at a 1 degree
minimum audible angle

575
00:33:36,210 --> 00:33:40,700
at low frequencies where
ITDs are important cues, what

576
00:33:40,700 --> 00:33:44,070
is the interaural time
difference that a subject can

577
00:33:44,070 --> 00:33:45,140
just barely detect?

578
00:33:46,420 --> 00:33:48,040
Well, if you had a
magnifying glass,

579
00:33:48,040 --> 00:33:49,460
you could read it
off this curve.

580
00:33:49,460 --> 00:33:53,395
Because 1 degree is a very
small angle on this scale.

581
00:33:55,680 --> 00:33:59,520
But you can calculate it
from the sphere, right?

582
00:33:59,520 --> 00:34:01,830
And it turns out
that the prediction

583
00:34:01,830 --> 00:34:07,880
is that for the minimum
audible angle of 1 degree,

584
00:34:07,880 --> 00:34:10,217
the interaural time
difference is 10.

585
00:34:10,217 --> 00:34:11,300
Now, this is microseconds.

586
00:34:14,800 --> 00:34:16,825
It's calculated to
be 10 microseconds.

587
00:34:18,150 --> 00:34:24,370
Where we perform the best,
where we're using ITDs,

588
00:34:24,370 --> 00:34:26,770
this minimum audible
angle, change

589
00:34:26,770 --> 00:34:29,960
in sound source of
1 degree corresponds

590
00:34:29,960 --> 00:34:34,300
to a change of ITD
of 10 microseconds.

591
00:34:34,300 --> 00:34:38,790
That seems like an unbelievably
small instant in time.

592
00:34:39,840 --> 00:34:44,015
But once again, if you
take the subject and now

593
00:34:44,015 --> 00:34:46,650
instead of an attenuator
in one channel,

594
00:34:46,650 --> 00:34:51,560
you have a circuit that delays
one ear relative to the other.

595
00:34:52,989 --> 00:34:56,100
And you say to the
subject, OK, with no delay,

596
00:34:56,100 --> 00:34:57,740
what did it sound like?

597
00:34:57,740 --> 00:34:59,260
Straight ahead.

598
00:34:59,260 --> 00:35:00,030
OK?

599
00:35:00,030 --> 00:35:01,447
Then you put your
delay in and you

600
00:35:01,447 --> 00:35:02,821
say, what does it
sound like now?

601
00:35:02,821 --> 00:35:04,620
And the subject says,
it just barely sounds

602
00:35:04,620 --> 00:35:07,250
like it moved a little
bit off to one side.

603
00:35:07,250 --> 00:35:12,160
And in fact, 10 microseconds is
the just noticeable difference

604
00:35:12,160 --> 00:35:24,320
for ITD in headphones,
which is evidence then

605
00:35:24,320 --> 00:35:27,762
that we are using these
ITDs at the low frequencies

606
00:35:27,762 --> 00:35:28,720
for sound localization.

607
00:35:32,180 --> 00:35:36,200
Now, you can, as we said,
play the same trick here.

608
00:35:36,200 --> 00:35:39,850
Minimum audible
angle at 2 degrees.

609
00:35:39,850 --> 00:35:41,665
Go to the ILD curves.

610
00:35:43,350 --> 00:35:46,100
And I'll try to get your
magnifying glass out here

611
00:35:46,100 --> 00:35:47,085
at 5 kilohertz.

612
00:35:48,160 --> 00:35:50,270
Look at where 2 degrees is.

613
00:35:50,270 --> 00:35:54,120
And it turns out that the
interaural level difference

614
00:35:54,120 --> 00:35:56,595
there is 1 dB.

615
00:35:58,880 --> 00:36:00,050
Play it in headphones.

616
00:36:00,050 --> 00:36:01,485
The subject says,
yeah, it sounds

617
00:36:01,485 --> 00:36:03,110
like it's just a
little bit off center.

618
00:36:04,425 --> 00:36:09,220
So it all fits with these
presentations in free fields

619
00:36:09,220 --> 00:36:12,937
and these interaural level
cues measured in free fields

620
00:36:12,937 --> 00:36:14,270
with what you get in headphones.

621
00:36:16,680 --> 00:36:19,965
Now, we have some
demos of these things.

622
00:36:21,910 --> 00:36:24,190
And I have several
kinds of demos.

623
00:36:24,190 --> 00:36:26,175
So let's do the
real fun one first.

624
00:36:27,630 --> 00:36:30,310
The problem with fun demos
is sometimes they don't work.

625
00:36:31,400 --> 00:36:33,970
So this has worked
in some classes

626
00:36:33,970 --> 00:36:36,190
and has not worked
in other classes.

627
00:36:36,190 --> 00:36:37,290
I have a sound source.

628
00:36:39,000 --> 00:36:41,360
It has a lot of
high frequencies.

629
00:36:41,360 --> 00:36:43,940
You can probably appreciate
that, jangling keys.

630
00:36:45,120 --> 00:36:47,850
And I would like to
be able to just say,

631
00:36:47,850 --> 00:36:51,150
OK, I'm going to just give you
interaural level differences.

632
00:36:51,150 --> 00:36:53,440
Probably, that's
what you'll get here

633
00:36:53,440 --> 00:36:55,270
with high frequency sounds.

634
00:36:55,270 --> 00:36:57,610
But I can't assume
that it's also not

635
00:36:57,610 --> 00:37:00,200
going to have some ITDs
because in this room

636
00:37:00,200 --> 00:37:02,550
there's going to be
differences in time

637
00:37:02,550 --> 00:37:04,870
from the sound source
to your two ears.

638
00:37:04,870 --> 00:37:06,650
So I'm really going
to give you both cues,

639
00:37:06,650 --> 00:37:10,440
but let's start with this demo
before we go on to headphones.

640
00:37:12,430 --> 00:37:14,890
I'm going to jangle these keys.

641
00:37:14,890 --> 00:37:16,560
And I'm going to
ask you, because you

642
00:37:16,560 --> 00:37:20,950
are such visual people,
to cover your eyes while I

643
00:37:20,950 --> 00:37:21,830
jangle the keys.

644
00:37:21,830 --> 00:37:24,170
Don't cover yet, because
what I want you to do

645
00:37:24,170 --> 00:37:26,460
is to tell me where
the keys are coming

646
00:37:26,460 --> 00:37:28,280
from with your other hand.

647
00:37:28,280 --> 00:37:32,620
I want you to point where you
think the keys are coming from.

648
00:37:32,620 --> 00:37:36,710
And at the end, after you
hear the sound, keep pointing,

649
00:37:36,710 --> 00:37:39,750
and look and see
how good you were.

650
00:37:39,750 --> 00:37:40,802
OK?

651
00:37:40,802 --> 00:37:43,020
So everybody, cover your eyes.

652
00:37:45,762 --> 00:37:47,350
[KEYS JANGLING]

653
00:37:47,350 --> 00:37:47,850
OK?

654
00:37:49,042 --> 00:37:50,750
Now, uncover your eyes
and keep pointing.

655
00:37:52,360 --> 00:37:54,810
OK, so we're doing
pretty well here.

656
00:37:54,810 --> 00:37:59,230
Maybe 20 degree angle, but
most of you guys are on.

657
00:37:59,230 --> 00:38:01,257
All of you guys are on.

658
00:38:02,420 --> 00:38:04,810
All right, let's
do it once more.

659
00:38:04,810 --> 00:38:06,485
So blindfold yourself.

660
00:38:06,485 --> 00:38:08,655
Going to move the
position of the keys.

661
00:38:08,655 --> 00:38:12,830
[KEYS JANGLING]

662
00:38:12,830 --> 00:38:15,030
OK, open your eyes
and keep pointing.

663
00:38:15,030 --> 00:38:16,910
OK, you guys are good.

664
00:38:16,910 --> 00:38:19,150
Maybe like 20 degrees.

665
00:38:19,150 --> 00:38:21,130
I changed the elevation
a little bit too,

666
00:38:21,130 --> 00:38:24,030
so you guys are making
mistakes in elevation.

667
00:38:24,030 --> 00:38:24,890
You guys are good.

668
00:38:25,961 --> 00:38:26,460
OK.

669
00:38:26,460 --> 00:38:30,750
Now, since I claimed to you
that we're using two ears,

670
00:38:30,750 --> 00:38:35,420
I would like you to plug
one ear, blindfold yourself.

671
00:38:35,420 --> 00:38:38,460
Now, it's going to be a little
bit of a trick to point, right?

672
00:38:38,460 --> 00:38:40,490
So I don't know how
you're going to point.

673
00:38:40,490 --> 00:38:42,630
You can point with
your elbow or you

674
00:38:42,630 --> 00:38:44,480
can do whatever
you want to point.

675
00:38:44,480 --> 00:38:45,992
AUDIENCE: Close your eyes.

676
00:38:45,992 --> 00:38:47,450
PROFESSOR: You can
close your eyes.

677
00:38:47,450 --> 00:38:48,750
OK, if you're truthful.

678
00:38:50,900 --> 00:38:52,429
You can close your eyes, OK.

679
00:38:53,866 --> 00:38:58,840
[KEYS JANGLING]

680
00:38:58,840 --> 00:39:01,590
OK, open up and we'll
see how you did.

681
00:39:01,590 --> 00:39:02,465
You guys are good.

682
00:39:03,800 --> 00:39:05,360
There's an error.

683
00:39:05,360 --> 00:39:07,145
That's the biggest
error I've seen so far.

684
00:39:08,820 --> 00:39:11,587
This is an error, like
at least 45 degrees.

685
00:39:11,587 --> 00:39:12,420
This is an era here.

686
00:39:12,420 --> 00:39:14,470
So we're doing more poorly.

687
00:39:14,470 --> 00:39:17,540
Let's plug the other ear
and I'll repeat the sound.

688
00:39:18,790 --> 00:39:20,750
OK, close your eyes.

689
00:39:22,714 --> 00:39:29,110
[KEYS JANGLING]

690
00:39:29,110 --> 00:39:31,300
OK, open your eyes
and keep pointing.

691
00:39:32,830 --> 00:39:34,178
These are some errors.

692
00:39:36,050 --> 00:39:37,387
Oh, this is a big error.

693
00:39:38,670 --> 00:39:40,250
This is an error.

694
00:39:40,250 --> 00:39:44,151
OK, so we're clearly,
anecdotally, getting worse.

695
00:39:44,151 --> 00:39:44,650
OK.

696
00:39:44,650 --> 00:39:48,070
Now, the ultimate
test, plug one ear.

697
00:39:48,070 --> 00:39:50,860
And in your remaining
ear, distort your pinna

698
00:39:50,860 --> 00:39:54,222
because I claimed that your
pinna cues were helping you.

699
00:39:54,222 --> 00:39:56,908
And I really don't know how
you're going to point here.

700
00:40:00,324 --> 00:40:05,180
[KEYS JANGLING]

701
00:40:05,180 --> 00:40:05,680
OK.

702
00:40:05,680 --> 00:40:06,950
Now, open.

703
00:40:06,950 --> 00:40:08,710
Just continue to
point if you can.

704
00:40:09,946 --> 00:40:11,880
OK, so that's a 90 degree error.

705
00:40:13,420 --> 00:40:14,680
You guys are pretty good.

706
00:40:14,680 --> 00:40:16,370
You have an error here.

707
00:40:16,370 --> 00:40:17,420
You guys are good.

708
00:40:17,420 --> 00:40:18,185
This is an error.

709
00:40:19,730 --> 00:40:20,980
And where were you pointing?

710
00:40:20,980 --> 00:40:22,880
You were pointing
correctly, yeah.

711
00:40:22,880 --> 00:40:23,880
So good.

712
00:40:23,880 --> 00:40:24,640
OK.

713
00:40:24,640 --> 00:40:28,100
Let me try it one more time with
distorting pinna and plugging

714
00:40:28,100 --> 00:40:28,620
one ear.

715
00:40:31,536 --> 00:40:37,100
[KEYS JANGLING]

716
00:40:37,100 --> 00:40:39,676
OK, show me where
you were pointing.

717
00:40:39,676 --> 00:40:40,820
You guys are good.

718
00:40:40,820 --> 00:40:41,550
AUDIENCE: Nice.

719
00:40:41,550 --> 00:40:43,707
PROFESSOR: This is a
little-- was it harder--

720
00:40:43,707 --> 00:40:44,790
AUDIENCE: It's impossible.

721
00:40:44,790 --> 00:40:45,680
PROFESSOR: It's impossible.

722
00:40:45,680 --> 00:40:46,554
So you were guessing?

723
00:40:47,778 --> 00:40:49,189
That's why it doesn't work.

724
00:40:49,189 --> 00:40:50,564
AUDIENCE: It all
sounds the same.

725
00:40:50,564 --> 00:40:51,330
It all literally
sounds the same.

726
00:40:51,330 --> 00:40:52,326
PROFESSOR: What's that?

727
00:40:52,326 --> 00:40:53,700
AUDIENCE: It all
sounds the same.

728
00:40:53,700 --> 00:40:56,158
PROFESSOR: It all sounds the
same with the pinna distorted.

729
00:40:56,158 --> 00:40:57,010
Right.

730
00:40:57,010 --> 00:40:59,874
OK, this is a huge error.

731
00:40:59,874 --> 00:41:00,557
AUDIENCE: Yeah.

732
00:41:00,557 --> 00:41:01,140
PROFESSOR: OK.

733
00:41:01,140 --> 00:41:02,880
I think it worked.

734
00:41:02,880 --> 00:41:07,160
Pinna is very important
in distinguishing front

735
00:41:07,160 --> 00:41:07,875
versus back.

736
00:41:09,310 --> 00:41:12,340
Interaural time and
level differences

737
00:41:12,340 --> 00:41:14,440
are equivalent for
front and back.

738
00:41:14,440 --> 00:41:16,140
So how do we know
what's front and back?

739
00:41:16,140 --> 00:41:18,290
Well, the pinna cues are
very important for that.

740
00:41:19,390 --> 00:41:22,260
Otherwise, you have
subjects reporting

741
00:41:22,260 --> 00:41:23,980
confusion between
front and back.

742
00:41:25,640 --> 00:41:30,540
A lot of the times to eliminate
front-back confusions,

743
00:41:30,540 --> 00:41:32,332
experimenters will
require subjects

744
00:41:32,332 --> 00:41:34,040
to point in the frontal
[? hemisphere. ?]

745
00:41:34,040 --> 00:41:36,140
And say, it's not in the back.

746
00:41:36,140 --> 00:41:38,082
I give you that
it's in the front.

747
00:41:39,450 --> 00:41:43,270
OK, so we have more
control demonstrations

748
00:41:43,270 --> 00:41:46,450
that can be presented
during headphones.

749
00:41:46,450 --> 00:41:48,920
And we have demonstration
of these two cues.

750
00:41:50,047 --> 00:41:51,505
And there are three
demonstrations.

751
00:41:52,980 --> 00:41:59,075
So the first two
demonstrations are ITDs.

752
00:42:01,680 --> 00:42:03,695
And the first one uses tones.

753
00:42:04,750 --> 00:42:06,370
So a tone is a sine wave.

754
00:42:07,880 --> 00:42:15,090
And it's going to give you tones
of 500 Hertz and 2,000 Hertz.

755
00:42:15,090 --> 00:42:18,390
So the low frequency
and higher frequency

756
00:42:18,390 --> 00:42:22,230
are heard with alternating
interaural phases

757
00:42:22,230 --> 00:42:25,555
of plus and minus 45 degrees.

758
00:42:26,640 --> 00:42:32,860
So let me just remind you, we
saw here an interaural phase

759
00:42:32,860 --> 00:42:34,440
difference of about 90 degrees.

760
00:42:34,440 --> 00:42:36,660
So this is going
to be half of that.

761
00:42:38,420 --> 00:42:41,080
Interaural phase
differences of 45 degrees

762
00:42:41,080 --> 00:42:43,290
for these 2 pure tones.

763
00:42:44,880 --> 00:42:47,930
The second one for ITDs
is going to use clicks.

764
00:42:49,185 --> 00:42:51,685
And they're going to be repeated
just like my hand clapping.

765
00:42:52,710 --> 00:42:56,420
Next, the interaural arrival
time of a click is varied.

766
00:42:56,420 --> 00:43:00,110
And it doesn't tell
you how big the ITD is.

767
00:43:00,110 --> 00:43:03,750
The apparent location of
the click appears to move.

768
00:43:03,750 --> 00:43:06,520
And to me, it sounds like
the clicks are going around.

769
00:43:07,970 --> 00:43:10,740
Somebody clapping their hands
moving around in a circle.

770
00:43:10,740 --> 00:43:14,125
Now, when you listen to these
demos with earphones or ear

771
00:43:14,125 --> 00:43:16,910
buds, sometimes you
get the impression

772
00:43:16,910 --> 00:43:18,690
that wherever it's
coming from, it's

773
00:43:18,690 --> 00:43:20,520
sort of like on the
surface of your head.

774
00:43:20,520 --> 00:43:22,980
It's not way out there.

775
00:43:22,980 --> 00:43:25,360
That may be because
you have lost

776
00:43:25,360 --> 00:43:27,740
some of the room acoustics.

777
00:43:27,740 --> 00:43:29,790
Certainly, you're not
listening to reflections.

778
00:43:29,790 --> 00:43:33,710
But you've also lost the
effects of the pinna, which

779
00:43:33,710 --> 00:43:35,500
filter things in
such a way that they

780
00:43:35,500 --> 00:43:37,960
sound more roomy or out there.

781
00:43:37,960 --> 00:43:39,650
These things sort
of sound internal,

782
00:43:39,650 --> 00:43:42,460
but they still appear
to change in azimuth.

783
00:43:42,460 --> 00:43:45,790
And I think this click
demonstration is the most vivid

784
00:43:45,790 --> 00:43:47,700
of the three.

785
00:43:47,700 --> 00:43:50,355
The final demonstration is
interaural level differences.

786
00:43:51,360 --> 00:43:53,510
The overall interaural
level difference

787
00:43:53,510 --> 00:43:58,020
of a 100 and 4,000
tone is varied.

788
00:43:58,020 --> 00:43:59,260
Now, heads up here.

789
00:43:59,260 --> 00:44:02,070
This low-frequency
tone is never going

790
00:44:02,070 --> 00:44:06,496
to have an ILD in normal
free-field acoustics.

791
00:44:06,496 --> 00:44:08,120
It's just never going
to happen because

792
00:44:08,120 --> 00:44:09,300
of the size of your head.

793
00:44:09,300 --> 00:44:12,710
You have to have a huge
head to cause a sound shadow

794
00:44:12,710 --> 00:44:14,490
for that low of a frequency.

795
00:44:14,490 --> 00:44:18,820
But you can certainly just-- as
easily as for high frequencies,

796
00:44:18,820 --> 00:44:20,800
present that in headphones.

797
00:44:20,800 --> 00:44:24,860
And it is perceptually
obvious, OK?

798
00:44:24,860 --> 00:44:26,220
So there's these three demos.

799
00:44:26,220 --> 00:44:28,519
So how many people have
them on your machines

800
00:44:28,519 --> 00:44:29,310
and have earphones?

801
00:44:31,210 --> 00:44:33,715
OK, so you guys can
just go right ahead.

802
00:44:35,470 --> 00:44:36,410
How does it appear?

803
00:44:40,020 --> 00:44:40,795
Are they in order?

804
00:44:42,820 --> 00:44:43,685
Do they have names?

805
00:44:43,685 --> 00:44:45,350
AUDIENCE: You can just
download them from Stellar.

806
00:44:45,350 --> 00:44:46,016
PROFESSOR: Yeah.

807
00:44:46,016 --> 00:44:49,400
You can download them from
Stellar if you haven't already.

808
00:44:49,400 --> 00:44:51,200
They should be on
the course website.

809
00:44:51,200 --> 00:44:52,475
So just do them in sequence.

810
00:44:54,180 --> 00:44:56,810
People who don't have
them on their laptops

811
00:44:56,810 --> 00:44:58,400
can listen up here.

812
00:44:58,400 --> 00:44:59,745
I have these listening stations.

813
00:45:01,800 --> 00:45:04,160
So whoever wants to come
up and listen can come up.

814
00:45:14,300 --> 00:45:16,510
Could you hear them
going around your head?

815
00:45:16,510 --> 00:45:18,650
That's, to me, the most vivid.

816
00:45:18,650 --> 00:45:19,850
Comments about the others?

817
00:45:22,670 --> 00:45:26,180
This last one is
kind of piercing

818
00:45:26,180 --> 00:45:28,265
for the high frequency, right?

819
00:45:28,265 --> 00:45:30,580
And it still sounds
like it's moving around.

820
00:45:30,580 --> 00:45:33,150
What about for this
lower frequency?

821
00:45:33,150 --> 00:45:35,910
Did it sound like it
was moving around?

822
00:45:35,910 --> 00:45:38,630
OK, even though
that stimulus is not

823
00:45:38,630 --> 00:45:39,830
present in free-field sound.

824
00:45:39,830 --> 00:45:42,720
You will never hear
that without headphones.

825
00:45:42,720 --> 00:45:44,690
It's perceptually obviously.

826
00:45:44,690 --> 00:45:47,250
Now, how about this ITD?

827
00:45:47,250 --> 00:45:51,810
To me, this 500 Hertz interaural
phase difference is obvious,

828
00:45:51,810 --> 00:45:53,564
but the 2,000 Hertz wasn't.

829
00:45:53,564 --> 00:45:54,480
AUDIENCE: [INAUDIBLE].

830
00:45:54,480 --> 00:45:55,690
PROFESSOR: Clear for you.

831
00:45:55,690 --> 00:45:58,170
Any speculation on why
that would be true?

832
00:46:02,204 --> 00:46:03,120
AUDIENCE: [INAUDIBLE].

833
00:46:12,974 --> 00:46:14,640
PROFESSOR: OK, what
we're saying is then

834
00:46:14,640 --> 00:46:27,570
that this is the left
ear for the low frequency

835
00:46:27,570 --> 00:46:30,355
and the right ear is, what?

836
00:46:30,355 --> 00:46:31,390
45 degrees?

837
00:46:31,390 --> 00:46:33,616
So it's going to be
sort of like this.

838
00:46:43,460 --> 00:46:45,515
OK, that's for maybe
the lower frequency.

839
00:46:46,990 --> 00:46:52,270
And the higher frequency is
quite a bit higher, right?

840
00:46:54,930 --> 00:46:59,000
It's more than
twice the frequency.

841
00:46:59,000 --> 00:47:00,850
So it's more than twice as much.

842
00:47:00,850 --> 00:47:03,600
I'm a terrible
artist, but it's going

843
00:47:03,600 --> 00:47:05,765
to go back and forth faster.

844
00:47:10,340 --> 00:47:15,790
And this is going to
be delayed 45 degrees.

845
00:47:15,790 --> 00:47:19,080
So because this is
going faster, it's

846
00:47:19,080 --> 00:47:26,865
a smaller time difference
for the high frequency.

847
00:47:27,920 --> 00:47:31,630
So you might say, OK, it's
just a smaller time difference.

848
00:47:31,630 --> 00:47:34,350
And that's why it's harder
for us to distinguish it.

849
00:47:34,350 --> 00:47:37,070
But remember the concept
that we had of phase

850
00:47:37,070 --> 00:47:39,420
locking of the auditory nerve.

851
00:47:39,420 --> 00:47:43,280
In the auditory nerve,
the left auditory nerve

852
00:47:43,280 --> 00:47:45,961
is going to fire some spikes
at some point in the stimulus

853
00:47:45,961 --> 00:47:46,460
waveform.

854
00:47:47,600 --> 00:47:49,280
The right auditory
nerve is going

855
00:47:49,280 --> 00:47:52,430
to fire some spikes at
a corresponding point

856
00:47:52,430 --> 00:47:55,470
in the stimulus waveform
for the first cycle.

857
00:47:55,470 --> 00:47:57,810
OK, let's repeat it
for the second cycle.

858
00:47:57,810 --> 00:47:59,362
Going to fire in
there somewhere.

859
00:47:59,362 --> 00:48:01,195
This one's going to
fire in there somewhere.

860
00:48:02,310 --> 00:48:04,940
And remember, cycles are
going by pretty quickly.

861
00:48:04,940 --> 00:48:06,530
There's lots of time.

862
00:48:06,530 --> 00:48:09,380
So you can build up
a response pattern

863
00:48:09,380 --> 00:48:12,010
where you have
thousands of spikes

864
00:48:12,010 --> 00:48:14,850
there, thousands of spikes here.

865
00:48:17,390 --> 00:48:20,450
And these two auditory nerves
are coming into the brain.

866
00:48:20,450 --> 00:48:23,650
They're synapsing in the left
and right cochlear nucleus.

867
00:48:23,650 --> 00:48:25,180
Cochlear nuclei.

868
00:48:25,180 --> 00:48:29,110
The cochlear nuclei are
then projecting centrally.

869
00:48:29,110 --> 00:48:32,911
And at some places, the left
and the right-side pathways

870
00:48:32,911 --> 00:48:33,410
converge.

871
00:48:35,090 --> 00:48:38,540
There's a comparator then that's
comparing the arrival time

872
00:48:38,540 --> 00:48:41,070
and it's saying the left ear
signal is getting to me first

873
00:48:41,070 --> 00:48:42,820
and the right ear is
a little bit delayed.

874
00:48:44,680 --> 00:48:46,760
What happens with
the higher frequency?

875
00:48:46,760 --> 00:48:48,700
So this is an interesting
high frequency.

876
00:48:48,700 --> 00:48:51,570
This is the 2,000 Hertz.

877
00:48:51,570 --> 00:48:53,500
What happens to
the phase locking

878
00:48:53,500 --> 00:48:57,120
for frequencies
above 1,000 Hertz?

879
00:48:57,120 --> 00:48:59,330
It declines, right?

880
00:48:59,330 --> 00:49:02,660
So instead of a nice
synchronized pattern,

881
00:49:02,660 --> 00:49:05,525
this left auditory nerve
is going to respond.

882
00:49:05,525 --> 00:49:06,800
The right is going to respond.

883
00:49:08,280 --> 00:49:12,060
But from successive
cycle to cycle,

884
00:49:12,060 --> 00:49:14,660
the pattern is not
going to be synced

885
00:49:14,660 --> 00:49:17,515
to a particular point in
the stimulus waveform.

886
00:49:18,640 --> 00:49:24,627
So instead of getting a
nice synchronized pattern,

887
00:49:24,627 --> 00:49:25,960
it's going to be unsynchronized.

888
00:49:27,300 --> 00:49:29,530
And maybe for some
stimulus presentations,

889
00:49:29,530 --> 00:49:31,450
the right-side spike
is going to come

890
00:49:31,450 --> 00:49:33,480
in earlier than the left side.

891
00:49:33,480 --> 00:49:37,260
And the comparator
is going to say, WTF.

892
00:49:37,260 --> 00:49:39,965
I don't know where the
sound is coming from, right?

893
00:49:41,170 --> 00:49:44,560
So the fact that phase
locking breaks down

894
00:49:44,560 --> 00:49:48,100
means that the timing at
wherever the central comparator

895
00:49:48,100 --> 00:49:50,662
is, it's not synchronized.

896
00:49:53,020 --> 00:49:54,320
Timing here is synchronized.

897
00:49:54,320 --> 00:49:56,430
The timing here is
not synchronized.

898
00:49:56,430 --> 00:50:02,510
And we're claiming with
this psychophysical metrics

899
00:50:02,510 --> 00:50:06,160
that we can detect a
difference between the left

900
00:50:06,160 --> 00:50:09,133
and the right ear minimally
at 10 microseconds.

901
00:50:12,790 --> 00:50:14,980
With synchronized
patterns at least.

902
00:50:16,000 --> 00:50:19,930
Now, let me just draw--
it's not too surprised.

903
00:50:19,930 --> 00:50:23,490
Everybody understands why when
you break down phase locking,

904
00:50:23,490 --> 00:50:25,855
you don't have a
temporal code anymore?

905
00:50:27,940 --> 00:50:35,540
Let me show you the spike
waveforms for one spike coming

906
00:50:35,540 --> 00:50:36,920
in in the left
side and one spike

907
00:50:36,920 --> 00:50:40,700
coming in the right
side that are delayed

908
00:50:40,700 --> 00:50:42,930
by this minimal time
of 10 microseconds.

909
00:50:45,310 --> 00:50:50,890
So here is spike, let's
say the left side.

910
00:50:54,990 --> 00:50:58,300
So to draw a spike coming in
from the right side delayed

911
00:50:58,300 --> 00:51:02,330
10 microseconds, I need to
know the time base here.

912
00:51:02,330 --> 00:51:08,170
How long does it
take a spike to fire

913
00:51:08,170 --> 00:51:09,500
in the central nervous system?

914
00:51:09,500 --> 00:51:10,700
What is the duration here?

915
00:51:10,700 --> 00:51:12,280
What's the time scale here?

916
00:51:12,280 --> 00:51:13,570
AUDIENCE: 1 millisecond.

917
00:51:13,570 --> 00:51:15,190
PROFESSOR: 1
millisecond, very good.

918
00:51:16,840 --> 00:51:20,840
OK, so now I'm going to draw
a right-side spike coming

919
00:51:20,840 --> 00:51:22,765
in that's delayed
by 10 microseconds.

920
00:51:22,765 --> 00:51:24,170
And it's pretty easy to do.

921
00:51:24,170 --> 00:51:26,500
I'm going to do it
in a different color,

922
00:51:26,500 --> 00:51:29,940
but it overlays
here very clearly.

923
00:51:29,940 --> 00:51:32,675
I didn't draw anything because
on this time [? base, ?]

924
00:51:32,675 --> 00:51:35,036
they're almost perceptually
indistinguishable.

925
00:51:39,180 --> 00:51:49,730
OK, so the right side here
delayed 10 microseconds.

926
00:51:49,730 --> 00:51:53,490
That's the wonderful
property of this system

927
00:51:53,490 --> 00:51:59,100
that we have relatively
large, long inputs coming

928
00:51:59,100 --> 00:51:59,865
into the CNS.

929
00:52:01,200 --> 00:52:04,570
And we can, at the
limits of our perception,

930
00:52:04,570 --> 00:52:08,570
distinguish delays of
a very short time scale

931
00:52:08,570 --> 00:52:10,420
on the order of 10 microseconds.

932
00:52:10,420 --> 00:52:14,460
That's the impressive nature
of this central comparator,

933
00:52:14,460 --> 00:52:15,530
wherever it is.

934
00:52:15,530 --> 00:52:19,880
And we're about to talk
about where it is now.

935
00:52:19,880 --> 00:52:27,280
So let's look at where
it is in the brain.

936
00:52:31,250 --> 00:52:34,730
We had our block diagram of
the central auditory pathway

937
00:52:34,730 --> 00:52:36,010
before.

938
00:52:36,010 --> 00:52:38,165
And here's kind of a
simplified diagram.

939
00:52:39,270 --> 00:52:41,270
This is the left side
and the right side

940
00:52:41,270 --> 00:52:44,650
of the brainstem with the
two auditory nerves coming

941
00:52:44,650 --> 00:52:46,410
into the two cochlear nuclei.

942
00:52:47,570 --> 00:52:49,510
And in the cochlear
nuclei, we discussed

943
00:52:49,510 --> 00:52:52,060
that the auditory nerve
synapses and cochlear

944
00:52:52,060 --> 00:52:55,530
nucleus neurons then
pick up the message.

945
00:52:55,530 --> 00:52:58,060
Now, in one sense the
cochlear nucleus neurons

946
00:52:58,060 --> 00:53:02,645
know only about what's going
on on that side of the brain,

947
00:53:02,645 --> 00:53:04,820
or that auditory nerve.

948
00:53:04,820 --> 00:53:07,500
They're not getting inputs
from the other side.

949
00:53:07,500 --> 00:53:10,580
So they're not
binaural, if you will.

950
00:53:10,580 --> 00:53:15,040
The first places where
you have binaural input

951
00:53:15,040 --> 00:53:17,190
in the auditory
pathway are centers

952
00:53:17,190 --> 00:53:20,380
like the superior
olivary complex.

953
00:53:20,380 --> 00:53:32,110
And we talked about that before,
Superior Olivary Complex,

954
00:53:32,110 --> 00:53:37,180
or SOC, having a bunch of
not only binaural inputs

955
00:53:37,180 --> 00:53:39,700
but a bunch of
different sub-nuclei.

956
00:53:39,700 --> 00:53:41,790
That's why it's
called a complex.

957
00:53:41,790 --> 00:53:44,110
One of the most important
of those sub-nuclei

958
00:53:44,110 --> 00:53:48,120
is the Medial Superior
Olive indicated here by MSO.

959
00:53:49,190 --> 00:53:53,680
And the MSO gets input from
the left cochlear nucleus

960
00:53:53,680 --> 00:53:55,080
if it's the left MSO.

961
00:53:55,080 --> 00:53:58,135
And it also gets input from
the right cochlear nucleus.

962
00:53:59,260 --> 00:54:04,130
And here is a very
good guess, at least,

963
00:54:04,130 --> 00:54:07,680
for where the
central comparator is

964
00:54:07,680 --> 00:54:10,260
on the basis of interaural
time differences.

965
00:54:11,690 --> 00:54:15,360
And this was appreciated
from very early time point.

966
00:54:15,360 --> 00:54:18,870
If you draw MSO neurons here,
they have two dendrites.

967
00:54:20,270 --> 00:54:22,770
One's going to the left side,
one's going to the right side.

968
00:54:22,770 --> 00:54:26,760
And they get numerous synaptic
inputs onto each dendrite.

969
00:54:27,810 --> 00:54:31,900
If you make a lesion, so you
interrupt the inputs coming

970
00:54:31,900 --> 00:54:35,960
from, for example, the
right side, all the inputs

971
00:54:35,960 --> 00:54:38,440
onto the right
dendrites drop off.

972
00:54:38,440 --> 00:54:40,000
So it looks like
all these inputs

973
00:54:40,000 --> 00:54:41,370
are coming from the right side.

974
00:54:42,660 --> 00:54:45,470
They've dropped off when
their pathway has been cut.

975
00:54:45,470 --> 00:54:48,335
And the left ones remain
because their pathway is intact.

976
00:54:49,750 --> 00:54:53,550
So clearly, the MSO neurons
get input from the two sides.

977
00:54:54,702 --> 00:55:00,840
Now, way back in the
1940s, a psychologist

978
00:55:00,840 --> 00:55:05,090
whose name was Lloyd
Jeffress proposed the model

979
00:55:05,090 --> 00:55:07,020
for detection of the ITDs.

980
00:55:08,510 --> 00:55:13,490
And he guessed at several
locations in the brain

981
00:55:13,490 --> 00:55:16,330
where this model could
actually be present.

982
00:55:16,330 --> 00:55:19,250
And one of his guesses
was in the MSO.

983
00:55:19,250 --> 00:55:23,030
And it turns out the MSO was the
correct of his several guesses.

984
00:55:24,510 --> 00:55:28,490
And this is a very
interesting model

985
00:55:28,490 --> 00:55:30,930
for neural processing of ITDs.

986
00:55:32,240 --> 00:55:35,080
And it has several
important assumptions.

987
00:55:37,160 --> 00:55:41,310
First, the MSO receives
input from the left

988
00:55:41,310 --> 00:55:43,450
and the right side, as
we have just gone over.

989
00:55:45,510 --> 00:55:48,480
So these are axons coming
in from the left side.

990
00:55:48,480 --> 00:55:50,900
And these are axons coming
in from the right side.

991
00:55:50,900 --> 00:55:54,270
And the dots there are the
MSO neurons themselves.

992
00:55:56,850 --> 00:56:01,350
The dots are very
fussy kind of neurons.

993
00:56:01,350 --> 00:56:03,335
They don't just
respond to any input.

994
00:56:04,600 --> 00:56:06,150
They're very discerning.

995
00:56:06,150 --> 00:56:08,920
They say, OK, I got some
input from the left side.

996
00:56:10,100 --> 00:56:11,640
Not a big deal.

997
00:56:11,640 --> 00:56:14,960
We get some input
from the right side,

998
00:56:14,960 --> 00:56:17,020
I'm not going to get
excited about that.

999
00:56:17,020 --> 00:56:20,330
But I'm going to get very
excited if I get input

1000
00:56:20,330 --> 00:56:24,625
from the left and the right
side at the same time.

1001
00:56:26,107 --> 00:56:27,106
That is, coincidentally.

1002
00:56:29,120 --> 00:56:31,870
And so the MSO
neurons are sometimes

1003
00:56:31,870 --> 00:56:34,420
called coincidence detectors.

1004
00:56:34,420 --> 00:56:37,570
That is, they detect
and they respond only

1005
00:56:37,570 --> 00:56:41,080
when they get coincident
input from the left

1006
00:56:41,080 --> 00:56:41,880
and the right side.

1007
00:56:43,011 --> 00:56:44,450
Well, how's that
going to help us

1008
00:56:44,450 --> 00:56:48,560
if we're delaying one
side versus the other?

1009
00:56:48,560 --> 00:56:52,140
Well, the second major
component of the Jeffress model

1010
00:56:52,140 --> 00:56:55,730
is that the axons providing
input, which we now know

1011
00:56:55,730 --> 00:56:58,550
are coming from the
cochlear nucleus,

1012
00:56:58,550 --> 00:57:01,730
they run down the
length of the MSO.

1013
00:57:01,730 --> 00:57:04,585
And as they run down the
length, they give off branches

1014
00:57:04,585 --> 00:57:09,260
to each MSO neuron in this long
chain going from left to right.

1015
00:57:10,910 --> 00:57:14,250
And if you know anything about
spikes that are traveling down

1016
00:57:14,250 --> 00:57:18,260
axons or nerve
fibers, you'll know

1017
00:57:18,260 --> 00:57:20,550
that the spikes
don't get instantly

1018
00:57:20,550 --> 00:57:23,400
to the very tip of the axon.

1019
00:57:23,400 --> 00:57:26,840
But it takes them time
to travel down the axon.

1020
00:57:26,840 --> 00:57:31,100
And so for example in this axon,
the impulse is coming down here

1021
00:57:31,100 --> 00:57:33,790
and it gets to this
leftmost branch first.

1022
00:57:35,130 --> 00:57:36,980
And then a little
bit later in time,

1023
00:57:36,980 --> 00:57:38,230
it gets to the next branch.

1024
00:57:39,330 --> 00:57:41,290
And so on and so
forth until it gets

1025
00:57:41,290 --> 00:57:44,870
to the rightmost branch
at the longest delay.

1026
00:57:46,560 --> 00:57:50,570
So Jeffress said, the inputs
to the MSO are, if you will,

1027
00:57:50,570 --> 00:57:51,636
delay lines.

1028
00:57:53,310 --> 00:57:58,340
That is, axonal impulse
propagation takes time.

1029
00:57:58,340 --> 00:58:01,400
You can set these lines up
so that they are delay lines.

1030
00:58:02,950 --> 00:58:05,825
The inputs on the right
have corresponding delays.

1031
00:58:08,980 --> 00:58:10,676
Now, how big are the delays?

1032
00:58:13,390 --> 00:58:15,210
And the flip side
of that question

1033
00:58:15,210 --> 00:58:21,894
is, how long does it take
impulses to travel down axons?

1034
00:58:24,535 --> 00:58:36,930
So another name for that is the
conduction velocity in axons.

1035
00:58:36,930 --> 00:58:39,130
Well, these are,
let's say, myelinated

1036
00:58:39,130 --> 00:58:43,430
axons of pretty big
size, like 5 micrometers.

1037
00:58:44,600 --> 00:58:51,110
So let's say they're
myelinated, a large diameter.

1038
00:58:55,800 --> 00:58:57,910
5 micrometers, let's say.

1039
00:58:59,470 --> 00:59:02,560
It turns out that such
a conduction velocity

1040
00:59:02,560 --> 00:59:05,700
for those kinds of axons is
about 10 meters per second.

1041
00:59:08,850 --> 00:59:11,430
And Jeffress was
sharp enough to know

1042
00:59:11,430 --> 00:59:15,430
that in the dimensions
of the brain,

1043
00:59:15,430 --> 00:59:17,640
those conduction
velocities work out

1044
00:59:17,640 --> 00:59:20,460
to predict about the
right delay for the kinds

1045
00:59:20,460 --> 00:59:22,660
of interaural time
differences that we're

1046
00:59:22,660 --> 00:59:26,920
talking about for sounds
that differ in azimuth.

1047
00:59:27,990 --> 00:59:32,350
So Jeffress, at the time he
was postulating his model

1048
00:59:32,350 --> 00:59:35,150
in the 1940s, there
were good measurements

1049
00:59:35,150 --> 00:59:36,935
of axonal conduction velocity.

1050
00:59:37,990 --> 00:59:40,700
And he realized that
these delay lines

1051
00:59:40,700 --> 00:59:45,210
were pretty good for
predicting or compensating

1052
00:59:45,210 --> 00:59:46,715
for the interaural
time differences.

1053
00:59:47,950 --> 00:59:49,300
Now, how does this model work?

1054
00:59:49,300 --> 00:59:53,340
Well, I have a little demo of
the model, which is a movie.

1055
00:59:56,025 --> 00:59:58,490
Which I have in a
different PowerPoint.

1056
00:59:58,490 --> 01:00:03,815
And I'm going to show
this coincidence model.

1057
01:00:09,620 --> 01:00:12,450
And I guess I didn't
credit the person who

1058
01:00:12,450 --> 01:00:15,749
made this movie,
which is Tom Yin.

1059
01:00:20,137 --> 01:00:21,720
And he works on the
auditory brainstem

1060
01:00:21,720 --> 01:00:24,094
and he's based at the University
of Wisconsin in Madison.

1061
01:00:24,880 --> 01:00:28,400
So what this demo will
show you is you'll

1062
01:00:28,400 --> 01:00:30,280
be looking down
onto the brainstem.

1063
01:00:31,320 --> 01:00:35,550
And the MSO on the left side and
the right side will be present.

1064
01:00:35,550 --> 01:00:39,710
The model is set up to demo
the MSO on the right side.

1065
01:00:39,710 --> 01:00:42,440
There will be a
cochlea on the left

1066
01:00:42,440 --> 01:00:44,620
and a cochlea on the right.

1067
01:00:44,620 --> 01:00:46,950
The auditory nerve coming
into the cochlear nucleus

1068
01:00:46,950 --> 01:00:48,890
on the left and
the auditory nerve

1069
01:00:48,890 --> 01:00:51,260
coming into the cochlear
nucleus on the right.

1070
01:00:51,260 --> 01:00:55,100
And those two nuclei will be
providing inputs to the MSO.

1071
01:00:55,100 --> 01:00:58,030
And action potentials,
or impulses, along

1072
01:00:58,030 --> 01:01:01,170
these nerve fibers
will be indicated

1073
01:01:01,170 --> 01:01:03,530
by little highlights--
little yellow lights.

1074
01:01:04,800 --> 01:01:08,410
And the demonstration
will show you

1075
01:01:08,410 --> 01:01:11,980
what happens to these
incoming impulses that

1076
01:01:11,980 --> 01:01:16,879
converge at the MSO for a
sound that's straight ahead.

1077
01:01:16,879 --> 01:01:18,420
So a sound that's
straight ahead will

1078
01:01:18,420 --> 01:01:21,040
strike the two
sides at the same--

1079
01:01:21,040 --> 01:01:23,220
will strike the two
pathways at the same.

1080
01:01:23,220 --> 01:01:26,055
And you'll see what
happens to the MSO

1081
01:01:26,055 --> 01:01:29,090
and which neuron in the
coincidence detector

1082
01:01:29,090 --> 01:01:31,050
array lights up.

1083
01:01:31,050 --> 01:01:33,775
Then, I think it'll play that
same demo in slow motion.

1084
01:01:35,240 --> 01:01:38,110
Then, the second part
of the demo I think

1085
01:01:38,110 --> 01:01:42,440
has the sound source displaced
off to the left side.

1086
01:01:42,440 --> 01:01:46,160
So the sound wave front will
come and strike the left side

1087
01:01:46,160 --> 01:01:49,520
first and the right
side after a delay--

1088
01:01:49,520 --> 01:01:51,550
the interaural time difference.

1089
01:01:51,550 --> 01:01:55,260
And you'll see what happens
to the impulses and the MSO

1090
01:01:55,260 --> 01:01:59,645
neurons with that second
sound source position.

1091
01:02:00,830 --> 01:02:06,670
So here are the two
cochleas, left and right.

1092
01:02:06,670 --> 01:02:07,580
Here is the MSO.

1093
01:02:11,350 --> 01:02:14,305
This is the right
cochlear nucleus.

1094
01:02:14,305 --> 01:02:16,070
This is the left
cochlear nucleus.

1095
01:02:17,280 --> 01:02:19,270
This is the MSO we're
not talking about.

1096
01:02:19,270 --> 01:02:21,120
This is the right MSO.

1097
01:02:21,120 --> 01:02:24,240
There was a sound wavefront
that hit the two sides equally.

1098
01:02:24,240 --> 01:02:26,250
And it was a little
hard to appreciate,

1099
01:02:26,250 --> 01:02:32,140
but I think this neuron up here
in this part of the MSO lit up.

1100
01:02:32,140 --> 01:02:34,140
This is going to be
the same wavefront

1101
01:02:34,140 --> 01:02:38,640
in slow motion activating the
two cochleas at the same time,

1102
01:02:38,640 --> 01:02:41,910
the two cochlear nuclei
at the same time,

1103
01:02:41,910 --> 01:02:44,576
and coming in to the MSO.

1104
01:02:44,576 --> 01:02:46,950
This one gets there first
because it's on the right side.

1105
01:02:48,550 --> 01:02:51,600
And the two impulses
arrive coincidentally

1106
01:02:51,600 --> 01:02:53,380
at MSO neuron number 2.

1107
01:02:53,380 --> 01:02:58,060
And that's the one, because it
gets coincident input-- that's

1108
01:02:58,060 --> 01:02:59,430
the one that fires off.

1109
01:03:01,090 --> 01:03:03,750
Here's a second part of the
demo where the sound is now

1110
01:03:03,750 --> 01:03:06,080
located off to the left side.

1111
01:03:06,080 --> 01:03:09,740
First, it's going to
show you in fast motion

1112
01:03:09,740 --> 01:03:11,010
and then in slow motion.

1113
01:03:14,410 --> 01:03:17,190
Left cochlea activates
first, right second.

1114
01:03:18,510 --> 01:03:21,570
And now, the MSO neuron
that got coincident input

1115
01:03:21,570 --> 01:03:25,616
is located down here,
neuron number 6 I believe.

1116
01:03:25,616 --> 01:03:27,930
Now, it's going to
show you that offset

1117
01:03:27,930 --> 01:03:29,360
sound source in slow motion.

1118
01:03:30,650 --> 01:03:34,420
Left cochlea gets activated
first, right second.

1119
01:03:35,490 --> 01:03:38,880
Left cochlear nucleus first,
right cochlear nucleus second.

1120
01:03:40,390 --> 01:03:44,120
Now, the delay
lines are set up so

1121
01:03:44,120 --> 01:03:47,440
that neuron number
six in the MSO

1122
01:03:47,440 --> 01:03:50,600
is the one that
responds because it now

1123
01:03:50,600 --> 01:03:52,190
is the one that gets
coincident input.

1124
01:03:54,410 --> 01:03:55,190
Is that clear?

1125
01:03:56,590 --> 01:04:00,880
So what you've set
up then in the MSO

1126
01:04:00,880 --> 01:04:08,460
is an array of neurons where
0 ITD is mapped up here

1127
01:04:08,460 --> 01:04:13,280
and left leading ITDs
are mapped down here.

1128
01:04:13,280 --> 01:04:16,930
You've mapped interaural
time difference

1129
01:04:16,930 --> 01:04:19,880
to position along
the MSO in the brain.

1130
01:04:21,922 --> 01:04:23,172
And that's the Jeffress model.

1131
01:04:32,028 --> 01:04:41,610
So the Jeffress model has
been tested experimentally

1132
01:04:41,610 --> 01:04:45,630
by going in and recording
from single MSO neurons.

1133
01:04:47,380 --> 01:04:53,010
So easy to say and extremely
difficult to do in practice.

1134
01:04:54,520 --> 01:04:57,580
It's not absolutely clear why.

1135
01:04:57,580 --> 01:05:00,640
It may be that the
MSO neurons are small

1136
01:05:00,640 --> 01:05:04,620
and there are thousands of
big inputs coming to them,

1137
01:05:04,620 --> 01:05:07,350
so that you get what
are called big field

1138
01:05:07,350 --> 01:05:11,075
potentials in your recordings
and very small spikes.

1139
01:05:12,590 --> 01:05:17,400
So the number of studies
of actual MSO recordings

1140
01:05:17,400 --> 01:05:20,840
can probably be listed on
the fingers of one hand.

1141
01:05:20,840 --> 01:05:22,700
So we don't have very much data.

1142
01:05:23,940 --> 01:05:27,950
What data we have
from MSO neurons

1143
01:05:27,950 --> 01:05:32,050
shows clearly that
the firing rate

1144
01:05:32,050 --> 01:05:34,940
is dependent on the
interaural time difference.

1145
01:05:34,940 --> 01:05:37,580
And that's what this
graph shows here.

1146
01:05:37,580 --> 01:05:41,990
I'm sorry it's not very
clear, but 0 interaural time

1147
01:05:41,990 --> 01:05:44,980
difference is right here
with the dashed line.

1148
01:05:46,620 --> 01:05:49,010
The firing rate is
plotted on the y-axis.

1149
01:05:51,750 --> 01:05:54,960
These dots over here
indicate the firing rate

1150
01:05:54,960 --> 01:05:57,020
for just left ear sound.

1151
01:05:58,320 --> 01:06:00,890
Or in the other one,
just right ear sound.

1152
01:06:00,890 --> 01:06:04,620
So there's not a very big
firing for presentation

1153
01:06:04,620 --> 01:06:06,810
of sound in just one
ear or the other.

1154
01:06:06,810 --> 01:06:09,710
That's consistent with
the Jeffress model.

1155
01:06:09,710 --> 01:06:12,150
Also, consistent
with the Jeffress

1156
01:06:12,150 --> 01:06:17,810
model is that if you
get a particular ITD

1157
01:06:17,810 --> 01:06:20,550
from the particular neuron
that you're recording from,

1158
01:06:20,550 --> 01:06:22,375
that neuron fires a great deal.

1159
01:06:23,940 --> 01:06:29,650
And other ITDs elicit
much less firing.

1160
01:06:29,650 --> 01:06:33,400
That is, the delay lines
didn't allow coincident input

1161
01:06:33,400 --> 01:06:36,930
to come and excite that neuron.

1162
01:06:36,930 --> 01:06:40,780
So firing rate that changes a
great deal as a function of ITD

1163
01:06:40,780 --> 01:06:42,450
is consistent with
the Jeffress model.

1164
01:06:44,360 --> 01:06:48,120
There, probably because
there's so few data,

1165
01:06:48,120 --> 01:06:52,300
the idea of this
mapping along the MSO

1166
01:06:52,300 --> 01:06:55,810
is not borne out by the
scanty experimental distance.

1167
01:06:55,810 --> 01:06:57,450
So we don't really
know that there's

1168
01:06:57,450 --> 01:07:01,680
a map as a function of a
particular brain distance.

1169
01:07:01,680 --> 01:07:04,700
This is the
anterior-posterior distance.

1170
01:07:04,700 --> 01:07:06,530
And they put this line here.

1171
01:07:07,580 --> 01:07:10,110
Really, the data are
all over the map.

1172
01:07:10,110 --> 01:07:13,160
So it's not clear that
there's an organized mapping.

1173
01:07:13,160 --> 01:07:17,500
It is clear that there's
a function of firing

1174
01:07:17,500 --> 01:07:18,760
when you change the ITD.

1175
01:07:23,535 --> 01:07:27,565
So there are some updates
to the Jeffress model.

1176
01:07:29,020 --> 01:07:31,960
And I'm not going to go
through these in detail,

1177
01:07:31,960 --> 01:07:34,523
but I want to point
them out to you

1178
01:07:34,523 --> 01:07:36,773
because this is part of the
answer to your assignment.

1179
01:07:37,950 --> 01:07:42,870
The assignment says, sort of
here's the Jeffress model.

1180
01:07:42,870 --> 01:07:45,080
Give me a quick outline of it.

1181
01:07:45,080 --> 01:07:47,360
So that's just what I said.

1182
01:07:47,360 --> 01:07:48,540
I've given you that part.

1183
01:07:49,640 --> 01:07:52,680
The second part says,
well, the Jeffress model

1184
01:07:52,680 --> 01:07:55,040
is maybe currently
under discussion.

1185
01:07:58,280 --> 01:08:00,470
What are new
experimental evidence--

1186
01:08:00,470 --> 01:08:03,860
new I mean from the
last 15 years-- that's

1187
01:08:03,860 --> 01:08:06,810
not perfectly consistent
with the Jeffress model?

1188
01:08:07,840 --> 01:08:10,070
And you should go
to this paper, which

1189
01:08:10,070 --> 01:08:14,850
is the assigned paper for
today's lecture, in which they

1190
01:08:14,850 --> 01:08:18,689
discuss point number
1, point number 2,

1191
01:08:18,689 --> 01:08:19,880
and several other points.

1192
01:08:19,880 --> 01:08:25,319
At least one other point is
demonstrated in that paper

1193
01:08:25,319 --> 01:08:28,620
to show some
experimental evidence

1194
01:08:28,620 --> 01:08:31,859
from MSO recordings
which is not perfectly

1195
01:08:31,859 --> 01:08:33,510
consistent with
the Jeffress model.

1196
01:08:33,510 --> 01:08:35,010
Or makes you think,
well, maybe they

1197
01:08:35,010 --> 01:08:41,230
Jeffress model is not
complete, or is outright wrong.

1198
01:08:43,700 --> 01:08:46,840
And these are recordings
from Brandt et al.

1199
01:08:48,399 --> 01:08:51,939
The earlier slide I showed you
was from recordings in Cat.

1200
01:08:51,939 --> 01:08:54,645
Cat has become less and
less the experimental model.

1201
01:08:55,750 --> 01:08:58,920
And these are now recordings
from smaller animals which

1202
01:08:58,920 --> 01:09:01,332
are more in vogue to
use experimentally.

1203
01:09:01,332 --> 01:09:02,790
In this case, it's
from the gerbil,

1204
01:09:02,790 --> 01:09:04,145
which is a popular animal.

1205
01:09:04,145 --> 01:09:08,680
It has a big MSO, good
low-frequency hearing

1206
01:09:08,680 --> 01:09:11,359
where you use
prominent ITD cues.

1207
01:09:13,649 --> 01:09:16,939
And this paper
clearly is a challenge

1208
01:09:16,939 --> 01:09:18,319
to the Jeffress model.

1209
01:09:18,319 --> 01:09:21,279
I don't think it
completely rules it out,

1210
01:09:21,279 --> 01:09:25,300
but clearly there are
some data from this paper

1211
01:09:25,300 --> 01:09:27,670
to suggest that it
might not be everything.

1212
01:09:29,479 --> 01:09:31,210
Now, the second part
of the assignment.

1213
01:09:33,206 --> 01:09:34,580
Actually, that's
the second part.

1214
01:09:34,580 --> 01:09:38,880
The third part of
the assignment comes

1215
01:09:38,880 --> 01:09:41,770
from some other
experimental data

1216
01:09:41,770 --> 01:09:44,100
that I'm not going to
give you because you

1217
01:09:44,100 --> 01:09:45,500
don't need to know about them.

1218
01:09:46,649 --> 01:09:51,220
But some labeling studies have
asked the question, OK, I'm

1219
01:09:51,220 --> 01:09:54,360
going to inject a neural
tracer into my cochlear nucleus

1220
01:09:54,360 --> 01:09:55,460
neuron.

1221
01:09:55,460 --> 01:09:57,540
I'm going to trace
its axon and I'm

1222
01:09:57,540 --> 01:10:00,360
going to find this
nice, ladder-like delay

1223
01:10:00,360 --> 01:10:01,330
line in the MSO.

1224
01:10:02,760 --> 01:10:05,570
Those labeling studies
haven't been particularly

1225
01:10:05,570 --> 01:10:09,940
gratifying in that they
don't fit this model so well.

1226
01:10:12,450 --> 01:10:14,500
I say here, at
first it was thought

1227
01:10:14,500 --> 01:10:17,680
that there were delay
lines from both sides.

1228
01:10:17,680 --> 01:10:20,300
Labeling studies suggest
that there is only

1229
01:10:20,300 --> 01:10:23,730
a delay line for
contralateral input.

1230
01:10:23,730 --> 01:10:26,410
OK, so that's not exactly
consistent with the Jeffress

1231
01:10:26,410 --> 01:10:26,910
model.

1232
01:10:27,930 --> 01:10:30,360
And even more recent
studies since I

1233
01:10:30,360 --> 01:10:33,800
wrote this suggests maybe there
aren't even delay lines at all.

1234
01:10:33,800 --> 01:10:36,010
So experimentally,
someone doesn't

1235
01:10:36,010 --> 01:10:38,435
see a delay line, that
doesn't mean it's not there.

1236
01:10:38,435 --> 01:10:40,390
It just means maybe
it's not so obvious.

1237
01:10:41,470 --> 01:10:44,180
But people have started
thinking about maybe

1238
01:10:44,180 --> 01:10:49,820
there are other ways to
provide delays in inputs

1239
01:10:49,820 --> 01:10:53,630
to the MSO that might make
the Jeffress model work.

1240
01:10:54,880 --> 01:10:57,270
If you have coincident
detectors and they

1241
01:10:57,270 --> 01:11:03,270
respond only when the delay a
certain ITD-- matches the ITD.

1242
01:11:03,270 --> 01:11:05,710
So the third part
of the assignment

1243
01:11:05,710 --> 01:11:11,260
asks you for other ways
that you can think of

1244
01:11:11,260 --> 01:11:31,230
to create delays to MSO neurons.

1245
01:11:36,340 --> 01:11:39,890
And let me give you a
couple hints to the answers

1246
01:11:39,890 --> 01:11:40,780
that I'm looking for.

1247
01:11:41,920 --> 01:11:49,320
I think you should think about
the synapse between the input

1248
01:11:49,320 --> 01:11:51,530
from the cochlear nucleus
to the MSO neuron.

1249
01:11:52,950 --> 01:11:56,550
How could you create
different types

1250
01:11:56,550 --> 01:12:00,305
of delays using
synaptic properties?

1251
01:12:11,960 --> 01:12:15,090
So this is kind of
a thought question

1252
01:12:15,090 --> 01:12:17,950
because these haven't
been measured yet.

1253
01:12:17,950 --> 01:12:21,800
Secondly, there is another
way to create delays.

1254
01:12:21,800 --> 01:12:24,510
And it comes from
properties of the cochlea.

1255
01:12:27,550 --> 01:12:30,390
And that brings us to
our reading for today.

1256
01:12:30,390 --> 01:12:31,490
We always have a reading.

1257
01:12:32,720 --> 01:12:36,865
The reading is from this
obscure document, also called

1258
01:12:36,865 --> 01:12:38,500
our textbook.

1259
01:12:38,500 --> 01:12:43,610
OK, and the reading
is on page 61.

1260
01:12:51,130 --> 01:12:53,030
Page 61 is the early
part of the book.

1261
01:12:53,030 --> 01:12:56,400
It's talking about
the ear, the cochlea.

1262
01:12:56,400 --> 01:13:00,060
It says, "But you may also note
that the vibration of parts

1263
01:13:00,060 --> 01:13:03,000
of the basilar membrane
tuned to frequencies below 1

1264
01:13:03,000 --> 01:13:08,450
kilohertz-- very low-- appear
time shifted or delayed

1265
01:13:08,450 --> 01:13:11,355
relative to those tuned
above 1 kilohertz.

1266
01:13:12,370 --> 01:13:15,070
This comes about because
of the mechanical filters

1267
01:13:15,070 --> 01:13:17,270
that make up the
basilar membrane.

1268
01:13:17,270 --> 01:13:19,950
They're not all in
phase with each other.

1269
01:13:19,950 --> 01:13:22,670
If you look at Figure 2.4--"

1270
01:13:22,670 --> 01:13:26,140
So everybody should look
at Figure 2.4 in the text.

1271
01:13:26,140 --> 01:13:28,230
So this is my devious
way of getting

1272
01:13:28,230 --> 01:13:30,670
you to actually
open the textbook.

1273
01:13:30,670 --> 01:13:33,010
You will see that the impulse
responses of the lower

1274
01:13:33,010 --> 01:13:35,890
frequency filters rise
to the first peak later

1275
01:13:35,890 --> 01:13:37,710
than those of the
higher frequency ones.

1276
01:13:39,240 --> 01:13:42,390
OK, so don't just quote the
textbook in your answer.

1277
01:13:42,390 --> 01:13:48,360
Tell me how you could get
a delay that would make up

1278
01:13:48,360 --> 01:13:52,580
for the interaural time
difference using this cochlear

1279
01:13:52,580 --> 01:13:56,100
property that's mentioned
in the textbook page 61.

1280
01:13:56,100 --> 01:14:00,165
And it's illustrated in
Figure 2.4 of the textbook.

1281
01:14:01,680 --> 01:14:05,470
That's another way people
are thinking of an alternate

1282
01:14:05,470 --> 01:14:07,350
to Jeffress delay lines.

1283
01:14:08,650 --> 01:14:11,886
And then finally,
because last time I

1284
01:14:11,886 --> 01:14:13,510
thought that it was
too easy-- you guys

1285
01:14:13,510 --> 01:14:16,305
are too smart-- I added
something to the assignment.

1286
01:14:17,540 --> 01:14:22,820
This fits with my son's view--
my son is in high school

1287
01:14:22,820 --> 01:14:26,111
and he says, teachers
love to load on homework.

1288
01:14:26,111 --> 01:14:26,985
The more, the better.

1289
01:14:28,350 --> 01:14:31,365
So I added something to
make this assignment more

1290
01:14:31,365 --> 01:14:31,865
challenging.

1291
01:14:33,020 --> 01:14:34,840
And here's what I added.

1292
01:14:34,840 --> 01:14:37,215
And so I think this has now
been posted on the website.

1293
01:14:38,740 --> 01:14:41,400
It doesn't add a great
deal of difficulty,

1294
01:14:41,400 --> 01:14:45,790
but I think it makes it
more relevant to our course.

1295
01:14:48,040 --> 01:14:49,670
I haven't updated this.

1296
01:14:49,670 --> 01:14:52,920
OK, well, look on the
website, course website.

1297
01:14:52,920 --> 01:14:54,870
I haven't updated
on my computer yet.

1298
01:14:54,870 --> 01:14:56,550
Look on the course website.

1299
01:14:56,550 --> 01:14:59,700
The very last part of
the assignment, there's

1300
01:14:59,700 --> 01:15:08,410
one sentence that says, how
would a cochlear implant user

1301
01:15:08,410 --> 01:15:12,785
have trouble using the Jeffress
model to localize sounds?

1302
01:15:15,050 --> 01:15:17,360
Even if the cochlear
implant user

1303
01:15:17,360 --> 01:15:20,635
had a cochlear implant in the
left ear and a cochlear implant

1304
01:15:20,635 --> 01:15:22,040
in the right ear?

1305
01:15:22,040 --> 01:15:23,600
And I think this
is a fair question.

1306
01:15:23,600 --> 01:15:25,940
We spent a lot of
time on our course

1307
01:15:25,940 --> 01:15:27,445
talking about cochlear implants.

1308
01:15:28,900 --> 01:15:34,680
And cochlear implant processing
is clearly very different

1309
01:15:34,680 --> 01:15:37,230
than we, as normal hearers,
have the processing

1310
01:15:37,230 --> 01:15:38,590
on our auditory nerve.

1311
01:15:38,590 --> 01:15:40,420
So think about that.

1312
01:15:40,420 --> 01:15:43,650
How would the
Jeffress model not be

1313
01:15:43,650 --> 01:15:47,070
able to be used very well by
a cochlear implant user who

1314
01:15:47,070 --> 01:15:49,340
had implants in the
left and right side?

1315
01:15:52,930 --> 01:15:54,440
So this is a written assignment.

1316
01:15:54,440 --> 01:15:57,350
I think before we talked
about how long it should be.

1317
01:15:57,350 --> 01:16:02,135
And I can't remember how--
maybe five pages is plenty.

1318
01:16:05,860 --> 01:16:08,925
In the very beginning, it
talks about the Jeffress model.

1319
01:16:08,925 --> 01:16:10,285
So give me a quick sketch.

1320
01:16:12,360 --> 01:16:18,255
It's due on December 4, which
is the day of the lab tour.

1321
01:16:19,760 --> 01:16:22,630
And you could send
them to me by email

1322
01:16:22,630 --> 01:16:25,970
or you can bring a printed
copy to the lab your.

1323
01:16:25,970 --> 01:16:28,660
Or, you can bring a
printed copy to my office,

1324
01:16:28,660 --> 01:16:32,920
but that's at Mass Eye and
Ear where the lab tour is.

1325
01:16:32,920 --> 01:16:35,690
And the idea behind
having it due then is

1326
01:16:35,690 --> 01:16:37,850
I can look them
over and grade them,

1327
01:16:37,850 --> 01:16:41,000
and then we can talk
about what I thought

1328
01:16:41,000 --> 01:16:45,280
is the correct answer to this
at the review session, which

1329
01:16:45,280 --> 01:16:47,960
is the class after December 4.