- About Us
- Apply to CoMPLEX
- For Students
- Students & Alumni
<Graduate Open Day>
- CoMPLEX will be at the Graduate Open Day of the Faculty of Mathematics and Physical Sciences on 24 January 2014
A new publication by PhD student Nicolas Jaccard
Dr. Nicol Harper
My work deals with information processing in the auditory system. Natural selection suggests that the auditory system will be optimised to improve an animal’s chances of survival and reproduction. One function that the auditory neural population code may be optimised for is the accurate representation of sounds in the world, given constraints such as intrinsic neuronal noise. I present evidence for such optimisation for two different auditory features.
First I consider the optimal coding of the interaural time difference (ITD) of sound arrival, a major sound localisation cue. Results from small mammals suggest it is represented by the relative activation of two distinct neural populations. This contradicts the classic model, consistent with barn owl data, of neurons tuned to every ITD. Here I present a simple generic model which qualitatively reproduces the different codes of mammals and barn owls by optimising the accuracy of a noisy population code for ITD given the animal’s head size and ITD-sensitive frequency range. The model further predicts that humans should use different codes depending on frequency. I examine the model’s robustness to different parameters, and also compare in detail the predictions of the model with published data from many different species, including analysing new data from the macaque. A number of qualitative predictions hold.
Second I examine, this time for sound intensity, whether optimisation, or at least coding improvements, can occur on a behavioural timescale. Responses were recorded to guinea pig inferior colliculus neurons to continuous broadband noise with randomly-fluctuating intensity. When some intensities were made more common than others, the neurons altered their intensity-response curves, typically in seconds, such that the neural population’s highest coding accuracy is over those common intensities. This suggests that the neural population adapts to accurately represent the more common sound levels in the environment. I conclude that the optimality approach to the auditory system has some value.
I am now a Sir Henry Wellcome Postdoctoral Fellow exploring the neural representation and processing of sound at the Redwood Center for Theoretical Neuroscience, UC Berkeley.
Page last modified on 06 sep 10 15:54