What cochlear implants can tell us about pitch perception in normal hearing

Temporal models of pitch perception postulate that pitch depends on auditory nerve inter-spike intervals. Such models were assessed with cochlear implant recipients by stimulating via a single electrode with two distinct pulse timing patterns. Pitch-ranking results using on-off modulated pulse trains were not consistent with a published autocorrelation model, but were consistent with an alternative model that analyses first-order inter-spike intervals. Place of stimulation cues were investigated by choosing a fixed pulse rate, high enough to avoid temporal cues (1800 pps), and then varying the electrodes that were activated. Pitch-ranking results suggested that when a group of neighbouring electrodes was activated, the percept depended on the centroid of the stimulation pattern. This is similar to a model of brightness (timbre) perception in normal hearing. Cochlear implant recipients performed a melody perception test with melodies presented by (i) varying the pulse rate on a single electrode (temporal cues only), and (ii) varying the centroid of the stimulation pattern (place cues only). Scores were worse than those of normal-hearing subjects listening to tones containing only resolved harmonics, but similar to those obtained with unresolved harmonics. Such tones contain only temporal cues to pitch and evoke a weak pitch sensation. Given the dissimilarities in the neural firing patterns, it is surprising that place cues alone and temporal cues alone can each provide useful information about melodic pitch.