Created by Jeff KnowlesThe Model
Sensory perception of auditory stimuli involves complex mechanical and biological computations in order to transduce a pressure waveform into perceivable sounds and auditory objects. This is a model of the basilar membrane, the organ in the inner ear responsible for the transduction of sound into neural signals. I have available for download a set of matlab scripts that run a numerical simulation the basilar membrane, hair cells and the auditory nerve as the system analyzes real sound data.
The model is powered by an original algorithm that decomposes the basilar membrane into a single dimensional array of interconnected oscillators each powered by the traveling waveform in the cochlea. By adjusting its parameters, the model can be adjusted to deconstruct sound at frequencies ranging from 20 hertz, the lower threshold of hearing for the human ear, up to hundreds of kilohertz, the upper threshold in bats.
Just like the biological organ, the oscillators at one end of the array have a low resonance frequency which increases as one travels away from the source of the sound wavefront. The result is that sound data is spatially segregated by frequency, a transformation analogous to Fourier Analysis. The output from this model could likely be mapped onto typical Fourier transforms, but the transformation would be non-invertible. Rather, the model adheres to the detailed physiological characteristics of the basilar membrane and it can be integrated with detector cells (Inner Hair Cells) and amplifier cells (Outer Hair Cells) in order to achieve a coherent model of the entire transduction process.
Limited only by computation time, one can increase the temporal, spatial and frequency resolutions of the simulation. This is important because the perception of sound by the brain utilizes both the disturibution of sound energy through the frequency spectrum, as encoded by the distrubution of ossilation on the basilar membrane, and the phase of ossilation of the membrane in order to get around biological limitations. This model can be used to capture both the spacial (tonotropic) and temporal (phase locked or volley encoded) components of the transduction process.
The programs available on this page include batsho.m, a computationally useful program that inputs already recorded data to run at any resolution necessary, and rtsho.m, a lower resolution version that runs in real time using matlab's audio capture functions. A third program, bmsim.m, is an earlier verson of the model by only considers energy rather than kinimatic data concearning the membrane's displacment and which inputs sound data via a fast fourier transform. This script also includes transduction by the inner hair cells, amplification and modulation by the outer hair cells, afferent signals from the auditory nerve, and a modulatory feedback mechanism controlled by a hypothetical "brainstem dampener nucleus," a transduction and modulation circuit is inspired by this diagram from Sheinberg 2007. The programs I have posted here can stand alone, but they are somewhat out of date. Upon contact, I can also send you the latest verson of a more complete verson of the model which utilizes the numerical technique discussed above to better simulate the complex elements of the cochlear transduction process.
If you have any comments, questions or suggestions, please let me know. If you have any interesting modifications to the code, send them along.
If you do not have Matlab, you can check out some sample videos on the right column, although the model is much more useful if you can import sound clips of your own.
For more information on the basilar membrane, check out the wikipedia page, this short animation, or this quick drawing of mine. The similarity between the cochlear physiology and formal Fourier Anlaysis is an amazing convergence between mathimatics and biology that can only be truley appreciated by this kind of a model.
The methods utilized in this model are original, but they are inspired by the sources listed below. Check back soon for updates to the code, and a more formal text that integrates this model into prior literature.
Last updated May 2009.
Sources and Background information on the Basilar Membrane:
Bear, Mark, Connors, Barry, Paradiso, Michael. Neuroscience: Exploring the Brain. Philadelphia: Lippincott, Williams and Wilkins, 2006.
Brugge, John F and Poon, Paul WF Eds. Central Auditory Processing and Neural Modeling. New YorkL Plenum Press, 1998.
Geisler, C Daniel. From sound to synapse: physiology of the mammilian ear. New York: Oxford Univerity Press, 1998.
Sheinberg, David. Neural Systems: Course Text for NEUR1030 at Brown University. Published and distributed as course material by Sheinberg, 2007. Contact
Copyright Jeff Knowles, 2009.