From: malcolm@interval.com (Malcolm Slaney)
Date: Mon, 24 Apr 1995 12:06:26 -0700
Subject: Alternatives to axonal-delays in localization models
Message-Id: <v0211010babc19eded4dc@[199.170.106.94]>

We continue with the binaural talks at the CCRMA Hearing Seminar.

Just what does the neural machinery look like that figures out relative
timing differences between our two ears?  Axonal delays are easy,
electrical pulses can only travel so fast down an axon but are they long
enough?  Suga has talked about cortical oscillators, but their mechanism is
a bit unclear (to me at least.)  Ben Bonham has just finished his PhD
thesis at Berkeley and he'll be talking about cochlear propogation delays.

We've heard a similar proposal at the Hearing Seminar by Shihab Shamma and
their have been chips built to implement Shihab's ideas.  Come to CCRMA to
hear more about Ben's ideas and see if the model fits the neurophysilogical
data.

        Who:    Ben Bonham (UCB EE)
        What:   Alternatives to axonal-delays in localization models
        When:   Thursday April 27 at 11AM
        Where:  CCRMA Library (Top Floor of the Knoll at Stanford)

See you at CCRMA!

-- Malcolm


Alternatives to axonal-delays in localization models
Ben Bonham (UCB EE)

In this meeting we will discuss two biologically plausible schemes for
localization of low frequency sounds. Each of these schemes exploits cochlear
propagation delays, rather than axonal propagation delays (as proposed by
Jeffress in 1948), to compensate for interaural time differences (ITD's) and
create binaural neurons that respond best to sounds with particular ITD's. In
the first of these two schemes, each model binaural neuron compares neuronal
signals from a single eighth nerve afferent channel from each ear; and those
two channels have different characteristic frequencies (CF's). In the second
scheme, each model binaural neuron computes a weighted sum of the neuronal
signals from several eighth nerve channels from each ear, and compares the
resulting sums. The weighting factors in the summation arose from simulations
of activity-dependent development, and are such that different binaural neurons
respond best to sounds with different ITD's, but the two monaural CF's of each
binaural neuron are identical. Experiments with both schemes suggest that the
relationship of best-interaural phase to stimulus frequency for these model
binaural neurons is consistent with some (but not all) auditory brainstem
neurons.