Plasticity in the auditory system

Question 1

Distinguish the causes of hearing loss

Answer 1

conductive and sensorineural

Question 2

conductive hearing loss

Answer 2

tends to be uniform across frequency spectrum

Question 3

Sensorineural hearing loss

Answer 3

• frequency selective
• most commonly due to OHC damage

Question 4

Requirements for cochlear implant

Answer 4

• The electrode array must be small and selectively spaced close to the spiral ganglion afferents along the length of the cochlea (wouldn’t help to stimulate the whole AN!)
• Sensorineural hearing loss: AN afferents must be intact (following prolonged IHC loss, AN afferents may degenerate)
• Sensorineural hearing loss is often due to OHC damage, in which case IHCs and connected afferents remain healthy and ready to receive input
• Need computational strategy for converting sound signals into pattern of electrical impulses for stimulation of AN axons

Question 5

Cochlear Implants

Answer 5

Sound picked up by the microphone is converted by the speech processor into a pattern of stimulating pulses for each electrode (we’ll see how this is done in a minute). This signal is transmitted to the receiver and passed on to an array of 8-12 stimulating electrodes.

Question 6

Transmitter coil:

Answer 6

-speech processor with microphone
-Receiver with stimulating electrode array (fitted under scalp

Question 7

The stimulating electrodes

Answer 7

• The basic idea is to thread the electrode array through so that each electrode sits at a different frequency location along the cochlea
• The idea is to thread the electrode through the round window, hugging the centre of the cochlear turn (keeping to the bony modiolus).
• This positions the electrode contacts closer to the spiral ganglion afferents and reduces the risk of damaging the organ of Corti.

Question 8

Basic strategy for basic strategy

Answer 8

• The cochleogram shows the vibration energy of the BM according to frequency location along its length. You can see the formants in the vowel part of the word reflected in the BM vibration.
• This is what would set up spiking patterns across a range of AN afferents according to their preferred frequencies. You would want to mimic this by direct stimulation of the afferents.
• For example, to get the formants you would want to stimulate low frequency areas to capture the harmonics around 0.2, 0.4 … 0.8 kHz, less stimulation for the weaker harmonics around 1-1.6 kHz and stronger again for the harmonics of the next formant around 2kHz. You would also want to reproduce the amplitude modulation over time.
• So stimulation of the low frequency locations would start 100ms into the word, but you would have some stimulation of high frequency locations from the beginning to capture the aperiodic ‘h’.
• At the end you would have a brief burst of stimulation over a range of frequency locations to capture the transient glottal stop, ‘d’.

Question 9

Processing strategies

Answer 9

• In C the waveform of A is split into 6 different frequency channels. The red line shows the temporal modulation (amplitude modulation) of different frequency components of the overall acoustic waveform.
• You can also see the amplitude modulation of the different frequency bands in B, but you have to infer it from the colour-coded energy levels as you move forward in time.
• The amplitude modulation for each frequency channel in C will be used to control the intensity of stimulating pulses delivered independently to each electrode, corresponding to frequency locations shown.

Question 10

Processing strategies 2

Answer 10

• A train of stimulating pulses is delivered to each electrode. The amplitude modulation in each frequency channel is used to modulate the intensity of these pulses, to stimulate a greater or smaller proportion of AN afferents at or near that frequency/place location.
• In this case we have just 6 frequency channels and each would be delivered to a separate electrode. With good placement each electrode should stimulate a different population of AN afferents.
• Increasing the stimulating current will increase the number of afferents depolarized above threshold within that population, hopefully without overlap. This can also be minimized by staggering (interleaving) the pulse timing between electrodes.
• Ultimately what we are doing is a crude simulation of cochlear place coding (but with just 6 electrodes as opposed to 3000 IHCs).

Question 11

Pitch perception is poor – why?

Answer 11

• Frequency resolution is good enough for major formants of speech, but not to resolve full stack of harmonics
• Temporal resolution is good enough for formant changes that mark phonetic contrasts important for speech perception, but not the full temporal structure of sound (importance of periodicity for pitch)
• Obviously, this rules out enjoyment of music. Also, pitch can be in important cue in auditory object recognition – keeping track of a speaker against a noisy background of other conversation, a form of scene segmentation. Implant users are much less effective at this (ie worse at understanding speech against background).

Question 12

Discuss the limitations of current implant technology

Answer 12

• The electrode array does not reach the most apical turn of the cochlea – unable to stimulate AN afferents tuned to lowest frequencies
• Although the technology is available to make very small electrodes, in practice the number is limited because of short-circuiting of electrodes by the perilymph and cross talk
• Effective number of frequency channels may be limited to 8-9
• Although electrodes are available with up to 25 contacts. The main point is that the device aims to replace the tonotopically-organized inputs from 3000 IHCs, but it has vastly fewer than 3000 channels to do this with.