@inproceedings {Ashida1809_2016,
	year = {2016},
	author = {Ashida, Go and Kretzberg, Jutta and Tollin, Daniel},
	title = {Coding Amplitude-Modulated Sounds by Coincidence Detection in the Lateral Superior Olive},
	booktitle = {Assoc. Res. Otolaryng. MidWinter Meeting (ARO)},
	URL = {http://c.ymcdn.com/sites/www.aro.org/resource/resmgr/Abstract_Archives/UPDATED_2016_ARO_Abstract_Bo.pdf},
	abstract = {Background
Neurons in the mammalian lateral superior olive (LSO) detect
interaural level differences by comparing excitatory inputs from
the ipsilateral cochlear nucleus with inhibitory inputs driven
by contralateral sounds. LSO neurons also show sensitivity
to binaural phase-differences of amplitude-modulated
(AM) sounds. Although binaural coding by LSO has been
extensively studied both theoretically and experimentally,
monaural response characteristics of LSO to AM sounds are
only marginally understood. Previous in vivo recordings in
cat LSO showed that spiking rates of LSO neurons generally
decrease with increasing modulation frequency, but that
variations of modulation-frequency dependence across
neurons were considerably large (Joris and Yin, 1998, J.
Neurophysiol.). In this study, we aim to reveal the underlying
mechanisms for this monaural AM coding using a simple
computational model of LSO.
Methods
Phase-locked excitatory inputs to LSO were modeled as an
inhomogeneous Poisson process with a periodic intensity
function, while spontaneous inhibition was modeled as
homogeneous Poisson process. Similar to a previous
modeling study (Franken et al., 2014, Front. Neural Circuits),
the LSO neuron was modeled as a counter of coincident
inputs. Namely, if the number of excitatory inputs within a preset
coincidence window reached or exceeded the threshold,
an output spike was generated. Then the model is in the
refractory period, in which no more spikes are generated.
Effects of inhibition were modeled as a transient increase in
threshold. We systematically varied parameters of the model
and examined how they affect AM-tuning of the model neuron.
Results
By changing the model parameters, most variations of AMtuning
curves observed in vivo were reproduced. Frequencydependence
of input parameters (spike rates, degrees of
phase-locking, and spontaneous inhibition) had only minor
effects on LSO output spike rates. In contrast, increasing
coincidence threshold or shortening the coincidence window
resulted in lower output spike rates. Moreover, lower
coincidence thresholds led to higher half-peak positions of
AM-tuning curves. The duration of the refractory period
affected the AM-tuning curve only below 300 Hz.
Conclusion
Our modeling results suggest that coincidence detection is one
of the most fundamental operations in LSO and that variations
of coincidence parameters may explain the empirical neuronto-
neuron variations in AM coding. Investigating the relations
between the abstract parameters of our coincidence counting
model and underlying biophysical factors (such as membrane
and synaptic properties) would be an important subject of
future study.
Funding
Supported by the Cluster of Excellence “Hearing4all” (GA,
JK), by the NIH Grant DC011555 (DJT), and by a Hanse-Wissenschaftskolleg
(HWK) Fellowship (DJT).}
}