18th May 1999 - Periodicity coding at the lower limit of pitch perception

The UK Section's May lecture was given by Katrin Krumbholz, a Research Associate at the Medical Research Council's Centre for the Neural Basis of Hearing. Katrin's lecture considered how we perceive pitch at low frequencies and how recent experimental techniques have shed light on the complex workings of the ear-brain system.

Katrin started by giving a brief overview of the hearing mechanism with particular emphasis on the complex workings of the inner ear.

Sound vibrations pass to the inner ear's oval window via the stapes. This is the final bone in the familiar chain of middle ear bones. The inner ear contains many chambers which perform the main functions of balance and orientation and, of course, hearing. The most intricate part of the inner ear is the cochlea which is encased in bone and is spiral in shape.

The cochlea spiral comprises two main fluid-filled cavities separated by a partition for most of the length. The partition is bounded by two membranes. One of these membranes, the basilar membrane, supports the main receiving organ known as the organ of Corti. The cochlea contains a multitude of inner hair cells which bend under the influence of the fluid pressure waves originating at the oval window and activate nerve impulses that get routed to the auditory cortex of the brain via the auditory nerve.

(The nerve responses appear to be polarised and trigger at a particular point in the pressure cycle thus phase locking nerve firing to the stimulus. A single cell does not 'recover' fast enough to respond to every cycle of the incoming signal at high frequencies but large numbers of cells tend to group together to provide continuous sampling. To complicate matters further there are also outer hair cells which appear to react to signals fed back from the brainstem by altering their length and, consequently, affecting the resonant characteristics of the system. The cochlea is not simply a passive receptor but is an active device capable of generating its own acoustic output in response to brainstem stimuli - a kind of super-regenerative receiver.)

Katrin's ongoing research is into the way these cochlea signals are coded to allow the fairly low resolution mechanics of the cochlea to provide the information that permits the ear-brain system to work with such remarkable definition. She described the work of early pioneers like Hermann von Helmholtz who described the basilar membrane as being like a series of mechanical resonators allowing each frequency component of a complex sound to be represented by signals from the appropriately tuned detector. Von Helmholtz suggested that high frequencies could be detected by cells near the base of the cochlea with lower frequencies being detected towards the apex. This became known as the Place theory of pitch perception and appeared to be supported, some years later, by direct observations made by Georg von Bekesy.

Katrin suggested that the Place theory alone could not explain our ability to resolve pitch at higher frequencies. Direct observation showed that the basilar membrane's resolution for higher harmonics was not that good so some other form of detection or coding (more akin to pattern recognition) must be taking place.

She then described and demonstrated a series of auditory pitch perception tests using notes composing a fundamental plus a series of harmonics. With harmonically-rich notes it was easy to pick out a difference between two notes or a melody when the fundamental was around 128Hz but very difficult with fundamentals below 50Hz. Clearly, the upper harmonic content was helping; but was the ear-brain's resolution failing as the harmonics became more closely spaced for the lower fundamentals?

Katrin went on to describe similar tests with notes synthesized so that phase alternated between successive harmonics. Waveform changes appeared to change the perception of pitch by altering the Repetition Frequency Discrimination Threshold (RFDT).

Katrin gave one example of current thinking. The basilar membrane can be thought of as two parallel filter sets whose signals get summed further on in the processing. One of the filter sets forms a series of allpass filters (delays) causing combing and modulation at the combined output. We tend to respond to the simple frequency component at low frequencies but to temporal structure at higher frequencies. Katrin suggested that we may be doing 'Wavelet analysis' rather than 'Fourier Analysis'.

The lecture concluded with a description of on-going research using binaural tests. Pitch perception tests were conducted using headphones where masking noises were 'positioned' in the apparent sound field using level and phase panning. Our ability to localise and separate sounds in a space appears to be provided by processes in the brain. If RFDTs can be altered by binaural conditions then it would appear that at least some measure of pitch discrimination is happening in the brain and not just the result of cochlea tuning.

Tests with a centrally positioned pitch sequence masked by noise showed that RFDTs reduced as the noise was panned away from the centre or separated by an interaural polarity reversal. This suggestedthat binaural processing and temporal structure can, indeed, play an important part in our ability to perceive pitch at low frequencies.

Research continues..... An in-depth question and answer session followed with some members forwarding personal observations and interesting anecdotes. The audience then thanked Katrin for an interesting and thought-provoking lecture.

Those interested in further reading may wish to visit the following web sites:

Roy Patterson's site for the Centre for the Neural Basis of Hearing - www.mrc-cbu.cam.ac.uk/personal/roy.patterson/cnbh/

Institute of Hearing Research - www.ihr.mrc.ac.uk/

Jim Cousins