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

Re: Effect of duration on pitch perception

Hi Martin,

I assume you're referring to the Patterson et al paper, which only shows that complex tones evoke a pitch percept faster than pure tones. It doesn't really provide any evidence about mechanisms.

Anyway, I don't want to beat this subject to death, but periodicity detection based on place/patterns doesn't need three serial steps. The first step, frequency filtering, obviously already occurs in the cochlea (and its non-negotiable, as I heard someone once say). The only other step that is required afterwards is to have a neuron with excitatory projections from fibers with CFs on the appropriate harmonic grid, and inhibitory projections from fibers with all other CFs.

This an artificially simple implementation and there may not be actual evidence for its existence. Nonetheless, I think that pitch detection based on place/pattern vs temporal mechanisms is not necessarily more complex. At least it doesn't *need* to be more complex, or more time-consuming.



Erik Larsen
Speech and Hearing Bioscience and Technology
Harvard-MIT Division of Health Sciences and Technology
Cambridge MA

Martin Braun wrote:
Dear William, Erik, Roy, and others:


One would think that the ear architecture is ideal for parallel processing
in the frequency domain. In that case, one would not expect f0 pitch to be
discriminated more slowly than pure-tone pitch due to the additional
neural processing.


It doesn't take much processing to detect neural activity at a
set of harmonically related frequencies. It doesn't matter that more
neurons activated because the processing can be done in parallel.

Of course it would not matter if more neurons were involved. But the point
is that pitch extraction via detection of a spatial pattern could NOT be
done in any kind of parallel processing. It would have to be a SERIAL
First step: Frequency filtering.
Second step: Storage of the main spectral lines in a neural matrix.
Third step: Detection of a pattern in this matrix that fits a harmonic
reference pattern.

Not two of these three steps could be done in parallel.

Periodicity detection in a neural system, however, can easily be done as an
integration of parallel processes. It necessarily has the potential to be
much faster than place-pattern detection. This had been evident already 22
years ago.



Martin Braun
Neuroscience of Music
S-671 95 Klässbol
web site: http://w1.570.telia.com/~u57011259/index.htm