[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Hair cell selectivity



Hi Rick,

Yes - that is a good one. The problem in simply working out the hair 
cell's frequency selectivity is that it depends on the preceding 
signals - nonlinearity. Further as you mention there is some sort of 
feedback in-vivo. In your opinion, where is that feedback ?

This is precisely why linear models have not succeeded to capture the 
frequency selectivity of the hair cell in the past.

Let us assume that we can extract a hair cell and keep it alive in the 
necessary fluids and in a system which resembles the Scala tubes and the 
Organ of Corti. In this case, we can probe the nature of the hair cell 
with no feedback. We expect to see signals which change the response 
nature of the hair cell due to the nonlinearity of the cell. This must 
be accounted for - if your hair cell is behaving linearly, then it is 
dead !

We must - as you mention - conduct a frequency selectivity test with the 
hair cell in its original environment. This is difficult as well. How do 
you stimulate a single Hair cell and record it without altering the 
state of the animal you are operating on ?
Perhaps the closest we can get accurately is to measure anesthetised
animal's nerve responses. Alas, these responses are also altered by 
drugs and exposure...

Quite obviously, in order to conduct a frequency selectivity test you 
need a good physiological model. This physiological model should include 
active protein processes, and model passive ionic protein processes.

Apart from that you need a complete physiological model of the 
environment of the Hair Cell - if there is feedback, including the 
structures which feedback. You now have your very own 'Cochlea 
amplifier'

The system readily becomes very complex. Whilst it is specified 
analytically, they almost always end up being numerical computations.
It is possible to do all of this analytically using one ion type, 
however as soon as you add a few ion types, they become coupled... this 
is difficult to solve analytically.

The key point in such hair cell and hearing models is - what do I 
include and what do I leave out ???
Once you have decided on those major issues, the next step is to 
collect the necessary hardware and software to construct the model. You 
are now almost there, you have your model and your physical modelling 
environment. Whats left ? Months and months of computation... thousands 
of lines of code to extract and plot your results... and so on...

Matt

On Mon, Oct 01, 2007 at 01:29:03PM -0400, Rick wrote:
> I have a rather naive question, but please, if you have a second, indulge me?
> 
> When you model hair cell sensitivity to frequency, how do you take
> into account active feedback via the outer hair cells' ability to
> change the response characteristics of the membrane? Wouldn't that
> upset any kind of predictability?
> 
> R
> 
> 
> 
> eOn 10/1/07, AUDITORY automatic digest system <LISTSERV@xxxxxxxxxxxxxxx> wrote:
> > There is 1 message totalling 108 lines in this issue.
> >
> > Topics of the day:
> >
> >  1. HC selectivity ... was Re: Physiological models of cochlea activity -
> >     alternatives to the travelling wave
> >
> > ----------------------------------------------------------------------
> >
> > Date:    Mon, 1 Oct 2007 13:25:39 +1000
> > From:    Matt Flax <flatmax@xxxxxxxx>
> > Subject: Re: HC selectivity ... was Re: Physiological models of cochlea activity - alternatives to the travelling wave
> >
> > Hi Reggie, Bob and the group,
> >
> > In the linear case I agree with you Reggie.
> >
> > But hair cell physiology is not linear. The closest they get is with the
> > Goldman-Hodgkin-Katz [1] equation. The furthest they get is with kinetic
> > algorithms [references not listed].
> >
> > For this reason I would write .. actually BEGIN to write the membrane
> > frequency dependence with respect to other signals like this
> >
> >                              1
> > wc(t) =  =========================================
> >           ___          _
> >          |__ \        (_)
> >             ) |  _ __  _
> >            / /  | '_ \| |
> >           / /_  | |_) | |   r(t)c(t)
> >          |____| | .__/|_|
> >                 | |
> >                 |_|
> >
> >
> > A nonlinear impedance. The time factor being the important change ...
> >
> > I don't believe that it is valid to give hair cell selectivity a first
> > order electrical model.... Sorry ... as simple as that...
> >
> > For this reason, previous assumptions on the frequency selectivity of
> > the cochlea hair cell is - in terms of frequency distributed power -
> > incorrect.
> >
> > A lot of mind or mathematical models may have incorrect assumptions to
> > date.
> >
> > The time dependent model should be able to explain frequency dispersion
> > and a certain amount of hair cell adaptation.
> >
> > sincerely
> > Matt
> >
> > [1] @BOOK{Hille:2001,
> >  title = {Ion Channels of Excitable Membranes (3rd Edition)},
> >  publisher = {Sinauer Associates},
> >  year = {2001},
> >  author = {Hille, B.},
> >  month = {July},
> >  url =
> > {http://www.amazon.fr/exec/obidos/ASIN/0878933212/citeulike04-21}
> > }
> >
> >
> >
> > On Sat, Sep 29, 2007 at 11:23:03AM -0500, Reggie Weece wrote:
> > > Bob and Matt -
> > >
> > > The difference between your two values for the cut off frequency
> > > appears to be the 2pi factor.  From
> > > http://en.wikipedia.org/wiki/RC_circuit,
> > > wc = 1/(RC)  rad/s
> > > fc = 1/(2piRC)  Hz
> > >
> > > Thus wc = 25000 rad/s (not 25 kHz, as Matt suggested!)
> > > Bob's calculation of fc = ~4kHz is correct.
> > >
> > > - Reggie
> > >
> > > > Date:    Fri, 28 Sep 2007 08:24:35 -0400
> > > > From:    Bob Masta <audio@xxxxxxxxxxx>
> > > > Subject: Re: HC selectivity ... was Re: Physiological models of cochlea activity - alternatives to the travelling wave
> > > >
> > > > On 28 Sep 2007 at 12:55, Matt Flax wrote:
> > > >
> > > > <snip>
> > > >
> > > > > With respect to a hair cell at rest ... take the following general
> > > > > values for membrane resistance and capacitance :
> > > > > R= 40 Meg Ohms  ...  taken from [4] for example
> > > > > C= 1 pF         ...  taken from [5] for example
> > > > > The first order cut-off frequency would be about 25 kHz.
> > > > > fc=25 kHz.
> > > >
> > > > How do you obtain this 25 kHz value?  Using
> > > >    f = 1 / (2 * pi * R * C)
> > > > I come up with about 4 kHz.
> > > >
> > > > Best regards,
> > > >
> > > >
> > > >
> > > > Bob Masta
> >
> > --
> > http://www.flatmaxstudios.com/
> > http://www.flatmax.org
> >
> > Public Projects :
> > http://sourceforge.net/search/?type_of_search=soft&words=mffm
> > http://www.psysound.org
> >
> > ------------------------------
> >
> > End of AUDITORY Digest - 29 Sep 2007 to 30 Sep 2007 (#2007-223)
> > ***************************************************************
> >
> 
> 
> -- 
> ======================
> Dr. Rick Nance
> Composer at Large
> PlasticMusic.Net

-- 
http://www.flatmaxstudios.com/
http://www.flatmax.org

Public Projects :
http://sourceforge.net/search/?type_of_search=soft&words=mffm
http://www.psysound.org