15Measuring Electric Hearing, and Using What You Find

FWhy we measure perception, not just impedance

Two implant users with identical surgery, identical devices and identical impedances can perform worlds apart, and the difference usually lives in perception rather than in any electrical reading. To understand and improve an individual outcome we have to measure what the listener actually hears: how sharply each electrode is resolved, how finely the spectrum can be carved up, how well rapid changes in time are followed, and how loudness grows with current. These are the questions the psychophysical toolbox answers.

The value of these measures is that they are diagnostic and prescriptive at once. A poor result on a spectral test and a poor result on a temporal test point to different problems and therefore to different fixes. Treating the implant as a black box tuned only by comfort levels leaves these levers untouched; measuring perception turns programming from guesswork into something closer to a targeted intervention.[2007][2010]

TThe spectral tools: spread, tuning and ripple

The first family of measures concerns spectral, or place, resolution. Forward masking estimates the spread of excitation from a single electrode: a brief probe is set just audible, then maskers on neighbouring electrodes are raised until they just mask it, tracing how far current and excitation spread along the array. Plotting masker level against electrode position yields the forward-masked spatial tuning curve, the electric analogue of the sharply tuned filter a healthy cochlea provides. Broad tuning means heavy channel interaction.

Spectral-ripple discrimination asks the listener to tell a spectrum with peaks and valleys from its phase-inverted version as the ripples are packed more densely; the densest ripple still discriminated is a single number summarising usable spectral resolution. Spectral-modulation detection is a close cousin that varies the depth rather than the density of spectral peaks. Both are quick, language-free and correlate with speech and music outcomes, which is why they have become workhorse bench tests. Together this family tells you whether a user’s problem is a blurred spectrum, and how much room there is to sharpen it.[2008][2007][2009]

CThe temporal and loudness tools, and how measures become decisions

The second family probes time. Modulation-detection thresholds measure the smallest amplitude fluctuation a listener can detect on an electrode, and across-electrode patterns of this sensitivity reveal good and poor regions of the array; better modulation sensitivity tracks with better speech recognition. Gap detection and rate discrimination round out the temporal picture. Loudness scaling, in which the listener rates the loudness of increasing current, maps the steep electric growth of loudness and sets the comfortable and threshold levels that anchor every map.

The point of all this measuring is action. Broad tuning curves and poor ripple scores argue for current focusing such as tripolar stimulation, for deactivating electrodes that interact badly, and for frequency-allocation choices that avoid wasting bands on indistinguishable places. Poor modulation sensitivity in a region argues for site-specific level adjustments or steering processing toward better channels. Loudness data set the dynamic range that compression and mapping must respect. Where psychophysics reveals limits that no electrode can overcome, especially for low-frequency pitch, the answer is to recruit acoustic hearing through electric-acoustic stimulation. In this way the bench measures feed directly into the coding strategies, the fitting workflow and the device decisions covered in the coding, speech-coding and programming chapters.[2011][2008][2009]

CPutting the toolbox to work, and where it leads

In practice a focused battery is achievable even in a busy clinic: a spectral-ripple or modulation-detection test takes minutes, requires no language, and can be repeated to compare strategies or settings for the same patient. Using the patient as their own control sidesteps the huge between-person variability that bedevils group studies and lets the clinician pick the map that genuinely sounds better to this ear.

These measures also define the frontier. Each future improvement, sharper focusing, fine-structure coding, optical stimulation, drug-eluting arrays, will have to prove itself by moving the very numbers this toolbox produces, because those numbers are the closest bench proxy we have for what the patient will hear. Mastering the psychophysics of electric hearing is therefore both a clinical skill for today and the yardstick by which tomorrow’s devices will be judged.[2010][2007]