Sound localization at high frequencies and across the frequency range
Introduction
The direction of sound is computed in two branches of the auditory pathway. The first branch has an important relay station in the medial superior olive (MSO) and process lower frequencies. Higher frequencies are processed in the lateral superior olive (LSO). Both the MSO and the LSO are the first nuclei in the sound localization pathway which collect inputs from both sides. To achieve sufficient time precision, in both systems, coincidence detection (CD) is the spike generation mechanism. In an old result by Srinivasan and Bernard [17], CD is shown to give a multiplication of firing rates. They assume asynchronous spike trains as input. We have adapted their observations for synchronous spike trains. We propose two variants of spike generation mechanisms, which are consistent with real spike trains statistics. To distinguish them, we call them excitatory coincidence detection (ECD) and inhibitory coincidence detection (ICD). As in [17], the mechanisms include timing jitter (random delay in spike arrival, denoted ΔJ) and probabilistic spike delivery (denoted pD). Experimentally observed synchronization precision in MSO and LSO is the highest spike timing precision ever described in neural systems. We study how the system copes with the jittery spike propagation.
In the MSO the sound direction clue is the disparity of timing (i.e. phase) between the two ears, interaural time delay. A classical theory of Jeffres [5] postulates that the right timing is picked up along the delay line, where individual neurons are specialized as CDs. The classical theory for the LSO states that the disparity of sound level between the two ears, interaural level difference, is calculated via subtraction of firing rates (SFR) of two inputs. From the excitatory input at ipsilateral side the inhibitory input at contralateral side is subtracted [6].
We have shown previously that neurons perform coincidence detection thanks to the specific dynamics of ion channels [10]. Our simulation and theory, and also similar theory by other authors [1], were based on experiments in chicken brainstem slices [15], [16]. We also studied limits, which are imposed upon the CD mechanism basically by the presence of refractory period and firing stochasticity [11]. The ECD described here takes place in MSO (this agrees with previous theories, we only show the spike statistics differently) and further we describe the ICD. We assume that the ICD is implemented in LSO (this is a novel theory, described here and in [12] for the first time).
Section snippets
Coincidence detection mechanisms
We will describe spike generation mechanisms as a calculation with spikes. For the biological context we refer to the discussion section. Spike trains in both sides derive from the sound input. For the purpose of following calculations, all neural events will be written as functions of time. Let us write the sound input as a periodic train SZ=∑i=1nδ(t−iTZ), of the Dirac delta pulses δ, where TZ=fZ−1 is the period of the sound (inverse of its frequency). Train of spikes at times T1,T2,…,Tn will
Discussion
What is the neural site of the two CD mechanisms? We attribute the windows W following spikes to the low pass-filtering nature of axons, synapses and dendrites. The window in the ICD with negative phase preceding the positive phase is just regular spike followed by the hyperpolarizing currents, but inverted, because it is passed through the inhibitory synapse. The ICD is an alternative mechanism to the SFR in the LSO. The SFR is postulated to take place in the LSO according to current opinion.
Acknowledgements
Supported by Physiome.cz, Ministry of Education no. 111100008, Charles University no. 201352. Thanks to N. Dorrell.
Petr Marsalek received his MD in 1990, his BS in Computer Science in 1992 and his Ph.D. in Biophysics in 1999, all degrees from the Charles University, Prague, The Czech Republic. Since 1993 he has been with the Department of Pathological Physiology, First Medical Faculty, Charles University. In the past, he spent several years as a postdoctoral fellow. He worked at the Caltech in the CNS Program and at the Johns Hopkins University in the Mind Brain Institute. He currently prepares his
References (17)
- et al.
How the threshold of a neuron determines its capacity for coincidence detection
Biosystems
(1998) Coincidence detection in the Hodgkin-Huxley equations
Biosystems
(2000)Neural code for sound localization at low frequencies
Neurocomputing
(2001)- et al.
The role of dendrites in auditory coincidence detection
Nature
(1998) Behaviors of spike output jitter in the integrate-and-fire model
Phys. Rev. Lett.
(1997)- et al.
A neuronal learning rule for sub-millisecond temporal coding
Nature
(1996) - et al.
Response of binaural neurons of dog superior olivary complex to dichotic tonal stimulisome physiological mechanisms of sound localization
J. Neurophysiol.
(1969) A place theory of sound localization
J. Comp. Physiol. Psychol.
(1948)
Cited by (0)
Petr Marsalek received his MD in 1990, his BS in Computer Science in 1992 and his Ph.D. in Biophysics in 1999, all degrees from the Charles University, Prague, The Czech Republic. Since 1993 he has been with the Department of Pathological Physiology, First Medical Faculty, Charles University. In the past, he spent several years as a postdoctoral fellow. He worked at the Caltech in the CNS Program and at the Johns Hopkins University in the Mind Brain Institute. He currently prepares his “habilitation” thesis at Charles University, to become an Associate Professor.
Jiri Kofranek received his MD in General Medicine in 1974 and his Ph.D. in Normal and Pathological Physiology in 1982, both from Charles University, Prague. Since 1975 he has been with the Department of Pathological Physiology, First Medical Faculty, Charles University, now as an Assistant Professor. His main interests are in modeling homeostasis and the development of simulation and educational software. He has been the principal investigator of numerous grants in his field. In recent years he became a coordinator of the research initiative “physiome.cz”. He is currently the Head of the Biocybernetics Laboratory.