11Coding of frequency & pitch
The auditory system has two independent ways of telling what frequency a sound is: which fibres are firing, and when they fire. The first — the place code — is tonotopy read off the array of nerve fibres. The second — the temporal code — is hidden in the precise timing of the spikes, which lock to the cycles of low-frequency sounds. Normal hearing uses both; a cochlear implant has good access to only the first, and that single fact explains much of why electric hearing handles speech well but music poorly.
TCTwo codes for frequency
Frequency is represented twice over. The place code says which neurons are active: because of tonotopy, a given frequency excites fibres from a particular cochlear place, so the identity of the active fibres signals the frequency. The temporal code says when the neurons fire: at low frequencies the spikes are locked to the cycles of the waveform, so the timing of the spikes signals the frequency. The two are complementary, and the brain uses both.[2012]
TThe place code
The place code is the direct read-out of everything in Module 4: the travelling wave peaks at a frequency-dependent place, the inner hair cells there drive their fibres, and so the position of the active fibres along the tonotopic array encodes the frequency. It works across the entire range of hearing — it is the only frequency code available at high frequencies — and its resolution depends on how sharply tuned the cochlea is (which is why it degrades when the amplifier is lost).[2009]
TCThe temporal code — phase-locking
Now the timing. For low-frequency sounds, an auditory-nerve fibre does not fire randomly through the cycle: it fires preferentially at one phase of the waveform — classically near a particular point in each cycle — so the intervals between spikes reflect the stimulus period. This phase-locking, first characterised in detail by Rose and colleagues, gives a timing-based measure of frequency that is independent of place. Crucially, it has a ceiling: the precision falls off above about 4–5 kHz, beyond which neurons can no longer track individual cycles and only the place code remains. Sweep the frequency below.[1967]
CThe volley principle
A single fibre cannot fire on every cycle of even a 1 kHz tone — it needs recovery time between spikes. Wever and Bray resolved the apparent paradox with the volley principle: although no one fibre fires every cycle, different fibres fire on different cycles, so the population collectively marks every period. The timing information is preserved across the group even when each member skips cycles — a population solution that recurs throughout neuroscience.[1930]
CFrom frequency to pitch
Pitch is the perception, and it is not simply the place of peak excitation. The clearest evidence is the missing fundamental: a complex tone with harmonics at 200, 300, and 400 Hz is heard at the pitch of 100 Hz even though there is no energy at 100 Hz. The pitch comes from the periodicity — the 100 Hz repetition rate of the combined waveform — which is a temporal cue. Low-frequency pitch, where we are most sensitive and which carries melody and voice pitch, leans heavily on this temporal information.[2013]
FTFrequency coding in the cochlear implant
A cochlear implant has good access to the place code and only crude access to the temporal one. Frequency is conveyed mainly by which electrodeis stimulated — but there are only a couple of dozen electrodes against the cochlea's continuous map, the current from each spreads to overlapping neural populations, and the array's position rarely matches the natural place for each input frequency (the frequency-to-place mismatch of Module 4). Pulse rate on a single electrode gives some temporal pitch, but only up to a few hundred hertz and far more coarsely than phase-locking. Techniques like current steering / virtual channels try to create intermediate places between physical electrodes.[2009]
Speech recognition relies mostly on spectral shape — the pattern of formants — which a coarse place code conveys adequately. Music and voice pitch rely on fine frequency and periodicity, which need the temporal precision the implant largely lacks. This is why a typical recipient understands speech well yet finds melody and pitch hard — the single most common limitation recipients report, and a major target of current research.
The nerve's two great dimensions — intensity and frequency — are now covered. The remaining modules follow the coded signal beyond the nerve: up the central pathway, and across the two ears.
Which difference between normal and electric frequency coding best explains good speech but poor music and pitch?
What are the two codes the auditory system uses to represent frequency?
Above roughly what frequency does phase-locking fail, leaving only the place code?
Why do cochlear-implant recipients typically understand speech well but find music and pitch difficult?