Cochlea -- quality of filtering

Tony Woolf tony at howl.demon.co.uk
Fri Sep 22 04:12:59 EST 1995


In article <43tb8l$jt8 at geraldo.cc.utexas.edu>
           sirianni at uts.cc.utexas.edu "Jeffrey Sirianni" writes:

> In article <43oll3$gt2 at hollywood.cinenet.net>, erc at cinenet.net (Eric Smith)
>  says:
> 
>   when
> a sound of this magnitude is presented to the cochlea, many auditory
> nerve fibers on both sides of the tonal frequency are activated.  So why
> do we not perceive this as a complex sound?  The brain's 2nd and 3rd order
> neurons suppress the activity on the sidebands.  This is an oversimplied
> answer to this question, but I would suggest reading Kim and Molnar (1979)
> J. Neurophysiol. 42:16-30.  

I'm not a specialist in this field - I'm an acoustician.  However I'm
interested and have done some reading, and it's obvious that there's
been a great deal of progress since the 1970s.  My information comes
from a text book "An introduction to the physiology of hearing" (James
O. Pickles, Academic Press 1988 ISBN 0-12-554754-4).  (This is a very
readable general introduction.)  Also from various conference sessions
and a few discussions with Prof. E.F.Evans of Keele University UK who
did some of the experimental work.  A lot of this work was done in the
early 1980s.
 
Using some very sophisticated measuring techniques, it's been shown,
not just that individual hair cell tuning is as sharp as as that of the
cochlear nerve fibres, but that the basilar membrane vibration also
shows the identical sharp tuning!  However this is true only if the
subject is in very good physiological shape, which is why the earlier
experiments done on cadavers or heavily anaesthetised animals didn't
show it.

Quote from the UK Institute of Acoustics Bulletin July/August 1994,
article by E.F.Evans:
"These cochlear nerve filters are quite remarkable: they have
half-power bandwidths in the one-third to one-sixth octave range and
cut-off slopes of the order of 100-200dB/octave for characteristic
frequencies above 2kHz or so, approaching 1000dB/octave on the high
frequency cut-offs."  (Generally they cut off faster on the high
frequency side than the low frequency side.)

As I said, the modern belief is that this frequency selectivity (in
mammals) arises from the basilar membrane vibration itself.  There is
a good hypothesis for this which accounts for the known facts although
AFAIK the exact mechanism remains unknown.

The hypothesis is that there is a positive feedback mechanism which
sharpens the tuning.  This is what happens when you get a howl-round
in a hearing aid or PA system.  When the gain is just short of
howl-round the sensitivity increases, and frequency selective
effects are increased.  (Early radio receivers used this under the
name of "reaction" or "regeneration" to increase sensitivity and
sharpen tuning, but it fell out of use because it was difficult to
control.)

In the basilar membrane, the feedback mechanism has to be partly
mechanical.  It is known that the hairs on the outer hair cells change
length in step with an applied voltage, and this seems to be the way
that a reinforcing vibration is fed back into the basilar membrane.

Damage to the outer hair cells stops this effect.  This seems to
be the main mechanism of loss of hearing caused by excessive noise, at
least in the early stages.  This explains why the critical bandwidth
of the ear usually increases when there is hearing loss related to the
cochlea nerves, as stopping the positive feedback both reduces
sensitivity and stops the tuning being sharp.

-- 
Tony Woolf  (tony at howl.demon.co.uk)



More information about the Audiolog mailing list