9Pitch from Place: Tonotopy, Mismatch, and the Map

FThe cochlea as a piano keyboard

The normal cochlea is laid out tonotopically: high frequencies are coded at the base, low frequencies at the apex, in a smooth, orderly map. An implant exploits this directly. Stimulating a basal electrode produces a high-pitched percept and stimulating an apical electrode a low-pitched one, so moving along the array sweeps pitch up and down like running a finger along a keyboard. This is place pitch, and it is the second great pitch cue of electric hearing, independent of the timing code.

Because each electrode addresses a different cochlear location, the number and spread of electrodes set how many distinct place-pitch steps a listener can perceive. The orderly relationship between cochlear position and frequency is captured by the classic frequency-position function, which the implant is, in effect, trying to re-impose with a handful of contacts where nature used thousands of hair cells.

Place pitch and rate pitch are perceived as largely separate dimensions: a listener can tell that one stimulus differs from another in place, in rate, or in both, which is why early multidimensional studies recovered a two-dimensional perceptual space with place on one axis and rate on the other.[1990][1987]

TElectrode discrimination and place-pitch resolution

How finely a listener can resolve place is measured by electrode (place) discrimination and pitch ranking: can the listener tell two adjacent electrodes apart, and order them by pitch? Performance is usually good but imperfect and uneven along the array, and the different tasks, discrimination, pitch ranking and pitch scaling, do not always agree, reflecting both the spread of electrical current and the underlying pattern of surviving neurons.

Current from one electrode spreads to neighbouring neural populations, so two contacts can excite overlapping regions and become hard to tell apart. Where neural survival is patchy, some electrodes contribute little distinct pitch, producing dead or redundant places on the map. This is why simply adding electrodes does not linearly add resolvable pitches: the effective number of independent place channels is smaller than the physical electrode count.

Place-pitch resolution also interacts with the temporal code. The two cues can reinforce each other when they move a percept in the same direction, but they can also conflict, and the brain must combine a place cue and a rate cue that, in an implant, are not yoked together the way they are in the healthy ear.[1997][2008]

CFrequency-to-place mismatch

Here the clinic meets the cochlea. Electrode arrays are rarely inserted deeply enough to reach the apical, low-frequency region, and the speech processor assigns frequency bands to electrodes using a standard allocation table that does not know each patient’s true insertion depth. The result is a frequency-to-place mismatch: a band of, say, 200-400 Hz may be delivered to a cochlear location whose natural characteristic frequency is far higher. The patient hears that band shifted upward in pitch relative to a normal ear.

The mismatch is largest with shallow insertions and standard maps and varies with array length, cochlear size and surgical technique. It distorts the internal pitch scale, makes voices and music sound unnaturally high or thin at first, and can degrade speech, particularly when the two ears (in bilateral or bimodal users) are mismatched against each other and binaural cues no longer line up.

Encouragingly, the brain is not entirely fixed. Studies tracking pitch over months to years show that some users acclimatise, their pitch percepts drifting toward the frequencies the processor assigns rather than the raw cochlear place. This plasticity is partial and variable, which has driven interest in anatomy-based or image-guided frequency maps that place each band closer to its tonotopically correct electrode from the start.[2007][2008]

CImplications for music and frequency mapping

Combine the three facts of this chapter, an envelope-dominated signal, a temporal pitch ceiling near 300 Hz, and a place code that is coarse and often mismatched, and you have a clear explanation for why music is so hard through an implant. Melody depends on fine, accurate pitch; the implant offers few resolvable place steps, a shifted scale, and a timing channel that saturates exactly where melody lives. Rhythm, which rides on the envelope, survives well; pitch and timbre do not.

This shapes practical programming. Choosing frequency allocations that reduce mismatch, preserving and using residual low-frequency acoustic hearing, and matching maps across the two ears in bilateral and bimodal users all aim to give the place code a fair chance. Where temporal and place cues conflict, aligning them improves the coherence of the pitch percept.

For counselling, the honest summary is that an implant restores access to sound and speech but delivers pitch through a coarse, shifted, and capped system. Realistic expectations about music, alongside training and hearing preservation, let patients make the most of the pitch information the device can give.[2008][1990]