[
This
thread has been abusively deleted. The Philpapers Team offered me the
opportunity to restore it.
"How
many threads do you need to restore? Combining multiple posts into
one would be a way to get around the limitation on 2 posts, and would
also be less work for you. Since they were previously accepted, we'll
make sure to accept them if you notify us ahead of time with the
subject heading." The PhilPapers Team
]
1
Hearing
George:
I hate ears!
me:
You sound just like Hate-Smurf. You did not mind the semi-circular
canals, so why hate ears?
George:
I don't mind the canals, in fact I love them. They're really fun. Any
time you move your head I get to whoosh from one pool to the
other!
me:
And riding the basilar membrane, that's not fun? That's just like a
trampoline, isn't it?
George:
Yeah, I guess it is. Or at least it would be, if I could play the
drums instead of having them bellowing in my head!
me:
I don't understand.
George:
Of course you don't. You're not supposed to. It's a homunculus thing,
and you're a scientist.
me:
Hey! I'm not a scientist!
George:
yeah, well, don't say it too loud. Not if you want them to take you
seriously!
Some References:
-
Mersenne: " L'Harmonie Universelle", 1636;
- von
Helmholz: "On the Sensations of Tone as a Physiological Basis
for the Theory of Music", 1862/1954;
- von Bekesy:
"Experiments in Hearing", 1960;
- Bauert: "Spatial
hearing", 1997;
- Beament: "How We Hear Music : The
Relationship Between Music and the Hearing Mechanism", 2003;
-
Gelfland; "Hearing: An Introduction To Psychological And
Physiological Acoustics", 2004;
- Schnupp et al: "
Auditory Neuroscience: Making Sense of Sound", 2011;
-
Heller: "Why You Hear What You Hear", 2013;
- Moore: "An
Introduction to the Psychology of Hearing", 2013;
- Langner:
"The neural Code of Pitch and Harmony", 2015. The
Physics of Hearing
Textbooks
on Hearing traditionally start with the physics of sound, which is as
a matter of fact quite different from the physics of hearing. Bauert
(1997) seems to be one of the rare exceptions. He makes the
distinction between a sound source and an auditory event, and refuses
the easy amalgam of the two. The first one is a physical phenomenon
that can be studied scientifically and can be the object of
(dis)agreement between different individuals; the second is a
personal sensory experience that can not be shared directly: what you
hear and what I hear are not necessarily the same thing.
There
must be of course a link between both, and the question then is
which. The distinction Bauert makes does not stop him from following
in the footsteps of his predecessors. He bases his analyses on the
same facts as they do, even if the nuance he brings in his approach
allows him to sometimes discern some discrepancies in the way certain
experiments have been set. Some authors who do not make this
distinction show results that are ambiguous because it is not clear
whether they apply to one, the sound source, or the other, the
auditory event. Ears and hearing can easily be compared with the way
a __telephone __works.
What I find particularly instructive
is the fundamental
difference in approach between acousticians and hearing aids
engineers. While
the first group concentrates on the mathematical description of
waves, the second made it their main task to understand sound as a
serial phenomenon. Cochlear implants are, just like the telephone,
based on the translation/transduction of waves into electrical
events. Those events cannot happen in a parallel way. After all,
there is only one membrane to receive those impulses at any time.
The
smallest bones in the body, the ones on which George would love to
sit and pound on the tympanic membrane using hands and feet at the
same time, seem to be relegated into obscurity while they are in fact
the most important factor in the whole process. They
are the first relevant phase in the sound production process.
Not
that the__ Ear and Head Transfer Functions__ are not important. The
way sound propagates from it source to humans is indispensable
knowledge for sound engineers, be they designers or sonic pollution
fighters.
Nonetheless,_ it is knowledge that is inaccessible
to the hearing brain_, and as such, irrelevant to the physics of
hearing, this in contrast with the physics of sound.
Strictly
speaking, even the role of the ossicles could be considered as
irrelevant: all we need is the effect of their actions on the
membrane, and its role in activating the auditory neurons. From
serial to sensory:
Vision,
as far as we know, is definitely a parallel process. We do not have
the time to "look at" each element of the visual field one
after the other. Hearing, even though it seems to us like we hear
many sounds simultaneously is just the opposite. Sounds that reach
the tympanic membrane make it move, and it makes in turn the ossicles
pound on the oval window one hit a time. The number of hits per
second, their frequency, will determine the reaction of the basilar
membrane, the last stage before neurons take over.
How
many hits does the basilar membrane need to know how to react? More
than two? The frequency is set by the first two hits, all the others
are only needed to indicate the duration of the stimulus, not its
quality. But are two hits physically sufficient to make the membrane
vibrate at the right frequency, or do we need many more than that?
This is quite a serious question. After all, if the stapes are
supposed to react fast enough to incoming sounds as to give us the
illusion of simultaneity, the less the number of hits needed for each
sound, the better. Puzzling phenomena like the missing
fundamental,
or the suppression
of the second sound,
might be more easily explained when taking into consideration the
seriality of sound sensations. There seems to be only__ two
fundamental physical properties of sound: frequency and amplitude__.
Everything else follows from those two. Likewise, those two
properties seem to be the only properties that can be relayed to the
brain. Which begs the question not only of pitch and timbre, after
all, those are only sensations, and we are used to the idiosyncrasies
of our brain. No, the mystery resides in the fact that these physical
properties seem to hide a third one, as tangible as can be: the
spatial location of sound.
We very often know where a sound comes from. Even assuming imperfect
knowledge based on experience, it remains a mystery how we
could know that. Where are the spatial codes hiding? The last puzzle
(for which I have no solution) until now is the convergence of so
many neurons (around 20) on the same inner hair cell.
2 Sound
Source Localization
The
most striking aspect of all the experiments concerning localization
of sound sources is their artificial character.
Not only are subjects very often restrained in their searching (head
and eye) movements, they are also generally expected to localize a
single one-time sound of limited duration. The results after more
than hundred years are very eloquent: nobody agrees with anybody, to
put it somewhat dramatically.
There are of course some trends
recognized by all, especially the fact that humans are much better at
localizing sounds that happen in front of them. As soon as sounds are
produced from above the horizon, above their head or behind them,
results become quite erratic, depending on the way each experiment
has been set up.
Also, the distinction between left and right
seems to be made quite accurately within a reasonably wide
range.
Last but not least, results seem to differ from one
individual to the other, making general conclusion even more
hazardous. A very interesting article is that of the long winded
title by Rankin et al "An Assessment of the Accuracy and
Precision of Localization of a Stationary Sound Source Using a
Two-Element Towed Hydrophone Array" (2013) in which they report
of their attempts... Well, the title says it all. Let me just add
that it happens on the open sea and that the sounds are produced by
underwater transducers while the towers are towed by a ship.
I
found the summary of their findings in the excellent abstract
[something that happens alas very rarely] very insightful:
"Bearings
to the sound source were determined based on the time-delay of the
signal arrival at two of the hydrophone elements. Localization was
estimated by visually inspecting the convergence of bearing angles to
the source. The accuracy of bearings and localization improved as the
ship approached the sound source." Here are thekey
words as
I see them.
- different arrival times at the towers are the main
acoustic source.
- The sounds are localized after a whole series
of recordings and computations.
- visual analysis is used to
implement acoustic findings.
- precise localization is
quasi-impossible: only a closing approximation of the source is
feasible. Except for the first one, all those points show how
unrealistic most experiments in this field are.
It is no wonder that it seems impossible to get any conclusive
results. If hearing, for humans, is a secondary sense as far as
spatial localization is concerned, if it is meant therefore to be
supplemented by vision and experience, than the researchers'
expectations may be just too high. The question is, would the same
image remain plausible once we take into account the first point?
Concepts like Interaural
Time Difference (ITD
or arrival time of the sound in respectively the left or right ear),
and Interaural
Level Difference (__ILD__
or difference in sound intensity as experienced by each ear, also
called IID for Intensity. See Popper et al "Sound
Source Localization", 2005) are fundamental concepts in this
context. They
are also typical homunculus concepts.
They assume that the brain can compare two input streams and then
take the appropriate action. Which would mean that the brain is able
to "look at" both inputs in a neutral, analyzing way, just
like we are able to in everyday life or in a scientific
experiment. That
is the difficulty with homunculus illusions. They seem so plausible.
After all, we are capable of comparing two sounds, so our brain must
somehow be able to do that also. Hearing
sounds is a sensory process, analyzing them is an intellectual one.
Some theories, starting with Helmholz, tend to consider the boundary
between both stages as quasi inexistent. But
it really becomes a problem when the intellectual process seems to
precede the sensory one.
That is if we somehow seem able to determine, even before we hear the
sounds, what the difference is between them. Which seems the only way
we could assign to each sound its place in our experience.
If the
experience of the difference between two sounds does precede its
analysis, then there is no problem with considering the intellectual
process as fundamental in processing these inputs. It becomes a
normal theoretical debate.
It remains however very problematic to
assign such intellectual processes to all animals or even humans
(babies, children, illiterates, mentally
challenged...). ILD and ITD concepts are very
similar to that of binocular disparity. As I have tried to show, it
is very unlikely that parts of the brain are able to compare what
each hemisphere is receiving. This is something only the whole brain
or organism can do. Still, whatever intellectual processes we believe
are involved in the process, they have to work with whatever our
sensory organs provide us with, and the corresponding sensations.
_Once we accept the idea that we are somehow "aware" of a
difference in arrival time or intensity in one ear relative to the
other, we can ask ourselves whether the brain really needs anything
else to determine the direction of the sound source. _Would the
sensation that a sound has reached our left ear before our right ear
not make all further computations redundant?__ ILD and ITD seem
to me as the logical consequences of a mathematical/computational
approach that denies the existence of sensations and must therefore
seek refuge in complex calculations to explain what the brain can do
so naturally. What
does Tastee Wheat taste like?
The
people in "The Matrix" had no way of knowing that, just
like congenially deaf people have no way of knowing how voices or
sounds in general are supposed to sound. Still, they are able to
learn with the help of cochlear implants. A Youtube clip of a young
child, under seven, shows him as having a conversation with the
interviewer and other people present while continuing to play with
what looks to be Lego-like blocks. Quite a technological and human
achievement that would have been unthinkable a few decades ago.
There
is no doubt that these implants will keep being improved with each
year passing, a very positive development for people all over the
world. A
Fundamental Flaw?
I
hope to convince the reader that I wish in no way to belittle the
progress made in this area, but I am myself convinced that the
approach to cochlear implants, however successful it might seem, is
fundamentally wrong.
The following quote from Clark "Cochlear
Implants: Fundamentals and Application ", 2003, is
representative of the current approach:
"Cochlear
implants should aim to reproduce
the coding
of sound
in the auditory system as
closely as possible, for best sound perception."
(ch.5). (my emphasis) There was once a documentary on TV concerning a
tribe that honored their elders by chewing their food for them when
they lacked the tooth to do it themselves. This is the impression I
get when I read this quote. The implants are doing the work that the
patients auditory system is supposed to be able to do unaided .
The
fact that with only a very limited number of electrodes (about 22
according to the 2005 edition) relative to the 20.000 inner hair
cells patients are able to learn so much is truly astounding. I will
come back to the theoretical significance of this ratio in a little
while. First I want to stand still by the question of pre-coding and
its disadvantages.
The makers of cochlear implants have to remove
the basilar membrane in its entirety if it is still present.
Unluckily, instead of trying to emulate its effects, the technicians
have followed the erroneous views of theoreticians of brain processes
in general, and those of the auditory system in
particular.
Researchers concentrate their effort in ameliorating
the speech-processing abilities of their implants while they should
be directing all their effort towards a better delivery of sound in
the form of electrical impulses. In plain words,they
should be trying to replace the effects of the basilar membrane
instead of creating an auxiliary auditory system outside of the
brain. Number
of electrodes
The
ratio of electrodes and inner hair cells necessary for a working
hearing system would seem to indicate that just like with vision,
each auditory fiber is capable of relaying all, or at least many
sound sensations. Also, technical results show that applying serial
instead of simultaneous sounds give better hearing results. This is
in accordance with the serial nature of the stapes and therefore of
the acoustic data that impinges on the basilar membrane, and from
there on the auditory fibers.
Put very simply, implant technicians
need "merely" to transduce the sounds received by the
microphone into electrical impulses and relay them through as many
electrodes as they can to as many locations in the cochlea as
possible.
I am not even sure whether they should bother with
the same spatial distribution frequency-wise as seen in the natural
basilar membrane. It could well be that the distinction between high
(at the base) and high frequencies (at the apex) is related to the
mechanical properties of the membrane, more than to the processing
abilities of the corresponding neurons.
In
summary: get
the sound (impulses) to the auditory system and let it do its work.
3
The homunculus explained
me:
George, would you mind?
George:
You know you are asking me to reveal secrets that have been kept,
well, secret, since the dawn of Man? Gosh! I have always wanted to
say that. Just for allowing me this opportunity, I think I will help
you out a little bit.
me:
Much obliged.
George:
Now you sound just like Elvis.
me:
would you get on with it!
George:
Okay! Okay! Imagine you are listening to a sound that is reaching
both your ears, but not at the same time.
me: ILD or ITD?
George:
Whatever. What I do is take what you hear, not me, since they are
your ears, if you catch my drift. [George giggles!] Then I decide
which sound arrived first, and what they look, I mean, what they
sound like, and then I whisper these differences in your ear. I mean
in your mind.
me:
How can you do that?
George:
Easy. Wait! Do what?
me:
How can you know the difference if you don't use your own
ears?
George:
Helloo! Because I don't need to! Homunculus? Moi? Remember?
me:
George, you don't think you have a little George in you, do
you?
George:
This is really sick, you know! That'll teach me to... Teach humans!
4
Calculations
and Sensations
I
have repeatedly hammered on the necessity not to lose sensations out
of sight because all our calculations not only are based on them but,
eventually, point back to them. Ignoring them compels the researcher
to look for artificial, biologically implausible artifacts that would
explain those calculations. Sound is no exception, or rather, it is a
beautiful illustration of the intimate relationship between
calculations and sensations. Why
the Basilar Membrane?
After
all, the percussions of the stapes on the oval window are what it is
all about.
The problem of course is for the brain to interpret
serial, telegraph- or Morse-like streams of data, and turn them into
a meaningful whole. Each series has to be distinguished from the
next, and for that the brain would need a neural decoder which it
does not have. This is where the basilar membrane comes into
play. Sounds
are distinguished through their respective frequency without any
computation or decoding process. Physical and mechanical laws take
care of that. The
membrane reacts differently to different rhythms. Just like the
jumping rope of little girls reacts differently to each wrist and arm
movement. The rope does not need to compute all those variables
before it can react accordingly. And neither does the basilar
membrane when it reacts to a seemingly meaningless stream of
percussions. The basilar membrane is seen by all as a biological
frequency analyzer,
or even a Fourier
analyzer by
those who need the mathematical security blanket.
One percussion
moves the membrane in an indeterminate way, but the second one
defines the final reaction, the one following either continuing the
same pattern of breaking it.
The movements of the basilar membrane
activate different neurons according to their specific patterns. The
details are, for now, irrelevant. What counts is the following
paradigm: series of percussions --> membrane movement -->
neuronal activation Still, membrane movements can be considered as
continuous, and even when they are not, the brain need some means of
distinguishing one series of movements from the other.
This is
where we realize that we
cannot do without the sensations.
A percussion or movement pattern evokes a definite sensation
depending on its frequency. The
brain does not need first to determine a frequency to feel the
corresponding sensation.
In fact, once the sensation, and its intensity, has been felt, there
is no need anymore for the brain to record any of those two. And even
if it wanted to, that is if we decided that brain must somehow code
for frequency and amplitude, we know that it cannot. There is nowhere
in the neuron where such codes could be hidden. A second series of
percussions will evoke another sensation, and that is the way the
brain can differentiate between different sounds. It
does not need to know when a series ends and another one starts. It
just feels it. The
same way that the little girls' rope just knows when one of them is
ready to jump. Or is it the other way around? Can little girls
compute the frequency of the rope before deciding when to get under
it and jump? Do our researchers have some equations hidden at the
bottom of a drawer somewhere that would explain such wondrous
abilities of our children?
5
Calculations
and Sensations (2)
The
cocktail party effect has created many insoluble problems for the
researchers.. Bregman's "Auditory Scene Analysis : The
Perceptual Organization of Sound", 1994, is considered as a
classic in the field. This is how the same author describes the
problem of hearing in his foreword to a more recent work by
Divenyi, editor of "Speech Separation by Humans and
Machines", 2004: "There
is a serious problem in the recognition of sounds. It derives from
the fact that they do not usually occur in isolation but in an
environment in which a number of sound sources (voices, traffic,
footsteps, music on the
radio,
and so on) are active at the same time. When these sounds arrive at
the ear of the listener, the complex pressure waves coming from the
separate sources add together to produce a single, more complex
pressure wave that is the sum of the individual waves. The problem is
how
to form separate mental descriptions of the component sounds,
despite the fact that the “mixture wave” does not directly reveal
the waves that have been summed to form it."
(my
italics) Understood as a non-serial series the numerous sounds that
reach us seem to be all part of a sonic soup we are supposed to drink
with forks.
[I agree with George. It really sounds
terrible!] This of course does not mean that sounds cannot be
scrambled beyond recognition. But to think that the brain has a way
of filtering different frequencies beforehand is a homunculus
illusion.
Attention and concentration are processes which
still await a plausible explanation as to how they are able to
enhance certain sound streams above others, it remains
nonetheless the question whether they do that with the help of
frequency filters, therefore at the source, or at the end of the
acoustic process: after different sounds have been perceived.
6
Calculations
and Sensations (3)
The
conundrum faced by Bregman and the different participants to
Divenyi's (2004) is I think insoluble mathematically without assuming
previous knowledge of the different sound sources contributing to
form a specific spectrogram. In other words,__ The mathematical
approach either relies implicitly on a homunculus, or it admits it is
powerless to analyze sound events in real time.__
That
certainly does not mean that a scientific analysis is therefore
impossible.
Taking example of the way biological hearing systems have solved the
problem we must acknowledge the fact that a technical, or
technological, solution is necessary instead of brute mathematical
computations. What mathematics cannot do is quite elementary for an
artificial membrane and corresponding stapes.
It would certainly
not be the first time that physical manipulation took the place of
computation. Just think of the way transistors are built. Computers
would not exist if not the peculiar reaction of two physical surfaces
put in close relationship to each other (Levinshtein&Simin
"Transistors: From Crystals to Integrated Circuits", 1998).
Man does not create electricity, he makes it possible by combining
matter. We much too easily forget this fundamental phase in which
electrical currents suddenly seem to surge from nowhere just because
two different molecular configurations were allowed to interact with
each other. Electricity is not a mathematical concept, even if all
its manifestations can be expressed mathematically. It is a real
physical phenomenon.
The same way, maybe frequency,
and amplitude, are more than mathematical parameters, but also real
physical attributes of physical events. If that is the case, physical
manipulation might be the only way to distinguish them in a complex
environment. Maybe
we need the physical vibrations to further analyze sounds
mathematically, and not only at the start of the analysis.
We most
probably cannot, armed solely with the mathematical concepts,
distinguish one frequency from the other in a noisy environment. But
a physical device might do just that.
Too often have scientists
thought that whatever a device can do could be reduced to a set of
mathematical equations or algorithms. This is certainly the deep
conviction of Rodney Brooks ("The relationship between matter
and life", 2001) in his quest of creation of intelligent,
embodied robots. He rejects categorically the idea that the way to
better AI is through the discovery of "new stuff". I
wholeheartedly agree with him. But not the way he would hope. What I
am proposing is in fact very "old stuff": matter. I will
leave the mind equation out for the moment. Suffice to say that not
all brain or even computation processes can be solved mathematically
of computationally. Sometimes the only way to understand them is to
manipulate matter and take note of the changes.
In other words,
maybe auditory scene analysis (ASA) or computational auditory scene
analysis (CASA) are not mathematical or computational problems, but
merely technical issues. [Bregman's approach is less mathematical
than it is based on Gestalt principles. ]
It is ironic that the
trend of inventing all kinds of mathematical devices for brain
processes should have blinded most of the researchers to this
elementary truth: the brain can solve complex problems with very
simple means: before being mathematically reduced, Archimedes' lever
had been used for thousands of years by non-scientists. No
mathematical analysis a priori could have discovered it.
What is
even more staggering is that the idea of an artificial membrane is
far from new. Already Nobel laureate Bekesy in the 1930's had built
models of a basilar membrane to study its reactions to sound
stimulations. You would have thought that researchers after almost a
century would have grown accustomed to the idea that a rudimentary
physical device could do what mathematics could not (microphones for
instance still need a membrane). Apparently, the ideological hold of
mathematics and the computational view is just to strong to be
ignored. Strangely enough, Brooks' approach is based on the idea that
intelligence cannot be disembodied. The next step would be to
acknowledge that manipulations of the physical world (beyond mere
bodily actions in the world) are a fundamental way of gaining new
knowledge. Just like the ancient Greek, we call that physics. [Brooks
would of course not object to such a view.]
The question "Why
a basilar membrane?" sounds hopefully even less theoretical than
before: such a device was a physical necessity, and not only an
evolutionary shortcut. [I apologize for speaking on Evolution's
behalf. Nobody should be allowed to do that. The only thing that
Evolution teaches unambiguously is that things change. The reasons
how and why are of our own making.] This also shows the necessity
for any scientific analysis to go beyond a mere mathematical
description of brain processes and take the issue of their physical
realization as even more fundamental than abstract equations than can
be interpreted in so many ways.
7
An
empirical test
The
question whether the stapes and therefore the basilar membrane, react
serially to many different sounds, some simultaneous, some not, and
how this affects the frequency and amplitude analysis of each
particular sound; or whether they react to the acoustic garbage
formed by all the sounds received within a certain period, can be
considered as an empirical matter. It should be possible to answer
the question unambiguously with physical experiments. And if it turns
out that there no easy answers, that would be a very important result
as well.
In fact, we already have such empirical tests.
Telephones and microphones do exactly that: they convert different
sounds, of which many are produced simultaneously, in a serial series
of sounds that we can distinguish.
Why
would then the brain re-scramble that which has been so dutifully
separated by the basilar membrane? Spatial
codes of hearing
It
sounds so strange when we think that sound knows only two fundamental
properties: frequency and amplitude. The Ear and Head Related
Transfer Functions are supposed to explain the metamorphosis of
ubiquitous sound into directional sound. The changes that the
external ear and the head bing about must explain how the brain is
thought to be able to compute equations like the Minimum
Audible Angle (Mills
"On the minimum audible angle", 1958; see also Hartmann "
On the minimum audible angle - A decision theory approach",
1989) and other equations that help localizing a sound source
in space.
But directional sound is something that we Humans
cannot perceive [unless we step into the path of an ultra-sound wave.
A search on the web will provide a wealth of technical information on
the subject, and a plethora of commercial products that are supposed
to enhance our experience. Fish are supposed to have directional
hearing, but that is something I really know nothing about. See Fay
and Popper "Comparative Hearing: Fish and Amphibians, 1999; also
the ch.3 of Popper et al mentioned below).
Let us not dwell on the
homunculus aspect of these computations and accept them at face
value. Let us even accept the idea that those equations do not allow
a perfect navigation instrument and must be supplemented by vision,
head and eye movement, and experience in general.
[I would
like to remind the reader though that whatever effects the external
ear and the head can have on how the sounds reach our brains, this
knowledge is only available to an external observer. As listeners we
have no way of distinguishing between a sound that has impinged
directly on our inner ear, and one that had to literally pass over
our head.] The fact remains that we are apparently able to recognize
a sound after it has changed its aspect by moving to another
location. We can worry later about how we do that exactly.
So
there must be something in the sound itself that not only keeps it
recognizable, which does not "sound" extremely improbable,
but also that indicates its possible location. And that is exactly
the problem, isn't it?
How
could previous experience explain the fact that we recognize the same
sound as having now another location?
1)
We must recognize the sound as such, even if it has now somehow
changed. No problem.
2) We must have previous knowledge of the
location of the same sound relative to our body. For instance, we
have previously identified a sound source in front of us, and had
experienced the sound from different angles by moving around it. The
identification of the different sounds as being of the same nature or
origin does not necessarily mean that those sounds could be linked to
the same source objectively by a sound instrument. Maybe their
spectrum, Fourier shapes or whatever it is that sound experts like to
play with, would seem completely unrelated to each other. The
association of all the sounds to relative locations and sources is a
psychological one, not a scientific endeavor.
This is how Fay
expresses it in ch.3 of Popper et al (2005): "it
is also quite clear that binaural processing is subject to
several essential ambiguities that must be solved using other
strategies. These include the fact that all sources on the
median sagittal plane result in equal (zero) interaural differences,
and that, in general, “cones of confusion” exist bilaterally on
the surfaces of which interaural cues are equal for many possible
source locations. These ambiguities seem to be solved adequately
among humans and many terrestrial species that use head or
pinnae movements, information from other senses, judgments of
the plausibility (...) of potential locations, and a processing
of the head-related transfer function (HRTF)
through
frequency analysis..."
(my
italics) Another interesting article is " Infants' Localization
of Sounds within Hemifields: Estimates of Minimum Audible Angle",
by Morrongiello and Rocca of 1990 where they show that human babies
and children learn with time how to localize sound more and more
accurately. Already in 1940 Wallach had pointed at the role of extra
auditory cues in the localization process of sound ("The role of
Head Movements and Vestibular and Visual Cues in Sound
Localization"). All
in all, sound localization seems to me be very similar to how we know
how hard or how high to throw the ball to score in a basket ball game
even though our brain has no way of computing the different
trajectories. Still, experience
cannot explain the sensations we have that a sound seems to come from
a certain direction,
however vague those sensations might be.
Experience
does not create corresponding sensations, only corresponding behavior
or conceptions. Where
do therefore our auditory spatial sensations come from? If
we want to avoid turning around in circles, we will have to point
either at the sound itself, or at our own body as the origin of those
sensations. Our brain, once again, cannot produce spatial sensations
out of nowhere. Or so I will assume.
8
Are
our ears symmetrically located on our head?
Not
according to scientific measurements (Abaza and Ross"Towards
Understanding the Symmetry of Human Ears: A Biometric Perspective",
2010). The difference is nigh undetectable, and it is really doubtful
whether this minimal asymmetry can account by itself for our spatial
sensations.
Still, combined with head and body movements, this
small asymmetry might just do the trick. Spatial
sensation might be nothing else but the impact of a sound on one ear
before the other. This
is I think much more plausible than the topological arguments used by
Rauschecker and Tian in "Mechanisms and streams for processing
of ‘‘what’’ and ‘‘where’’ in auditory cortex"
(2000). I really have no idea how computations can create any
sensation at all, and certainly not the sensation of "where",
even if parts of brain (and not a human individual) could make use of
geometrical equations. A view that does not cease to amaze
me. Auditory
Field and External Space
What
would happen if the only factors allowing us to localize,
approximately, sounds were ILD or ITD?
All sounds reaching the right or left ear would seem like not only
coming from the same direction, but also from the same location. What
is strange is that sounds coming from a loudspeaker are in fact all
from the same location. Nonetheless, we do not have the sensation
that they all occupy the same space. I
think we make much too easily the amalgam between auditory space and
real external space. Vision has spoiled us so to speak. What we see
is also where we see it. All
Sounds at One Location?
Two
sounds of different frequency will be felt at different locations.
This the basis of the place theory as advocated by Bekesy. It seems
to work quite well for certain frequencies, but stops being of any
use at some fundamental frequencies like those used by speech and at
very high frequencies.
The problem really is the fact that we
still have on clue where the spatial codes can be hidden. How
does the fact that neurons activated at specific places from the
basilar membrane explain the spatial character of auditory
sensations?
That
the auditory field, like the visual field, moves with our own head is
understandable.
But while we expect a perfect correspondent
between the visual field and 3D external space, that relationship is
lost when it comes to sound.
Frequencies
seem to have a location in auditory space that turns out very often
to be completely unrelated to real sound sources.
We
localize sounds in the objective external space by ear (ILD
and ITD as sensations, not computations), but we hear
different sounds at different locations in auditory space instead of
throwing them all in a heap. Distinguishing
them "audio-spatially" help us interpret them better
[something indispensable for speech production and comprehension]
even if it creates some confusion as to where they are in real
space. Does
sound explain hearing?
The
basilar membrane definitely reacts to sound. But that is certainly
not the case of auditive fibers. They are activated by hair cells
which react, just like with touch, to movements of their cilia. So
yes, hearing certainly involves sound indirectly, but maybe
we should consider hearing more like a particular modus of
touch. That
would in a way explain why we feel sounds as belonging to an auditory
space.
Maybe what we are feeling is comparable to what we feel when
something touches our skin, or our body hair. There is also always a
spatial sensation accompanying the touch sensation. That
would also explain why those spatial sensations do not give us
precise knowledge concerning
the source of sounds in the external space. They convey the
spatiality of our cochlea, just like touch conveys the spatiality of
our body as a whole. We feel where we have been touched, or where our
skin itches. What makes such a view especially interesting is that it
can perhaps be empirically
falsified.
This could be done by showing that there are no correlations between
the objective position of the part of the cochlea admittedly involved
in the perception of a specific frequency and the spatial
sensation.
I do not think we can hope to get clear-cut
statistics that would show a definitive relationship between
sensations and spatial position of parts of the cochlea. I am not
sure the position of the different parts of the cochlea relative to
the body would allow such unambiguous conclusions. But it might be
possible to find some correlation between auditory sources and
cochlea positions. In other words, that when
we feel that a sound is coming from behind us, we are actually
feeling a part of our cochlea that could be said to lie caudally, to
the back, relative to where we are facing.
A
negative result would not infirm the view that auditory spatial
sensations are cochlea and not space related, but it would leave it a
little bit dangling in the air, like any other speculative idea.
[I
could not find any data concerning the localization ability of sounds
among patients with a cochlear implant. All the efforts are
concentrated around speech comprehension, music being a far second.
But If am right, this kind of patients should have more difficulties
in localizing sounds. The problem will be to to prove that it has
anything to do with the number of electrodes. After all, if they can
learn to recognize speech maybe the idea that each auditive fiber can
convey all or many acoustic sensations can be complemented with the
corresponding spatial sensations. In other words, it might be almost
impossible to find definitive answers to all the questions we have. A
pragmatic attitude, certainly concerning implants, might be our best
bet.]