Auditory cortical images of cochlear-implant stimuli: coding of stimulus channel and current level

This study quantified the accuracy with which populations of neurons in the auditory cortex can represent aspects of electrical cochlear stimuli presented through a cochlear implant. We tested the accuracy of coding of the place of stimulation (i.e., identification of the active stimulation channel) and of the stimulus current level. Physiological data came from the companion study, which recorded spike activity of neurons simultaneously from 16 sites along the tonotopic axis of the guinea pig's auditory cortex. In that study, cochlear electrical stimuli were presented to acutely deafened animals through a 6-electrode animal version of the 22-electrode Nucleus banded electrode array (Cochlear). Cochlear electrode configurations consisted of monopolar (MP), bipolar (BP + N) with N inactive electrodes between the active and return electrodes (0 < or = N < or = 3), tripolar (TP) with one active electrode and two flanking return electrodes, and common ground (CG) with one active electrode and as many as five return electrodes. In the present analysis, an artificial neural network was trained to recognize spatiotemporal patterns of cortical activity in response to single presentations of particular stimuli and, thereby, to identify those stimuli. The accuracy of pair-wise discrimination of stimulation channels or of current levels was represented by the discrimination index, d', where d' = 1 was taken as threshold. In many cases, the threshold for discrimination of place of cochlear stimulation was < 0.75 mm, and the threshold for discrimination of current levels was < 1 dB. Cochlear electrode configurations varied in the accuracy with which they signaled to the auditory cortex the place of cochlear stimulation. The BP + N and TP configurations provided considerably greater sensitivity to place of stimulation than did the MP configuration. The TP configuration maintained accurate signaling of place of stimulation up to the highest current levels, whereas sensitivity was degraded at high current levels in BP + N configurations. Electrode configurations also varied in the dynamic range over which they signaled stimulus current level. Dynamic ranges were widest for the BP + 0 configuration and narrowest for the TP configuration. That is, the configuration that showed the most accurate signaling of cochlear place of stimulation (TP) showed the most restricted dynamic range for signaling of current level. These results suggest that the choice of the optimal electrode configuration for use by human cochlear-prosthesis users would depend on the particular demands of the speech-processing strategy that is to be employed.

Adaptation in the compound action potential response of the guinea pig VIIIth nerve to electric stimulation

An experimental study, carried out in guinea pigs, was designed to investigate whether forward masking measured psychophysically in 3M-House cochlear implant users might have a correlate in VIIIth nerve activity. The study was based on electrically evoked VIIIth nerve compound action potentials (ECAPs), using a masking paradigm comparable to the one used in the psychophysical study. Trains of 50 maskers with inter-masker-intervals of 509 ms appeared to induce a long-term fatigue effect that could influence the recovery from adaptation measurements. Fatigue stabilized within about 1 to 3 min when masker trains were repeated with intervening silent intervals of 10.5 s. The change in amplitude of probe-evoked ECAPs with increasing masker-probe delays was determined within the steady fatigue state. The recovery-from-adaptation functions obtained from these measurements resembled the forward masking functions found in 3M-House cochlear implant users. No correlate of psychophysical backward masking was found at the VIIIth nerve level. To examine whether hair cells were involved in fatigue and recovery from adaptation, the measurements described above were carried out in intact cochleas and in cochleas without hair cells. Results were essentially the same in the different preparations. The results suggest that processes at the level of the VIIIth nerve could, at least partly, account for forward masking found in 3M-House cochlear implant users. Backward masking must be attributed to mechanisms located centrally to the VIIIth nerve.

Effects of stimulus level on nonspectral frequency discrimination by human subjects

Frequency difference limens were determined as a function of reference-stimulus level for pulsatile electrical stimuli in 5 postlingually deaf human subjects with Nucleus-22 cochlear implants and for sinusoidally amplitude-modulated acoustic white noise stimuli in 4 normal-hearing humans. Subjects were tested at levels throughout the dynamic range and extending to the lowest detectable levels. Response stability was measured over the course of 10 sessions. For electrical stimulation in the deaf ears, difference limens decreased as a function of level throughout much or all of the dynamic range of hearing. This result contrasts with the case for nonspectral acoustic stimulation of normal-hearing subjects, where nonspectral frequency difference limens were strongly affected by level only near the detection threshold. These data suggest differences in the acoustic and electrical response spaces that must be considered in the design of auditory prosthesis processors.

Cochlear wall titanium implants for auditory nerve stimulation

Genetically deaf dalmatian dogs and ototoxically deafened macaque monkeys were implanted with electrodes housed in cochlear wall titanium implants to assess long-term stability, tolerance, and performance. Short-term human implantation, followed by trials of stimulation, was performed in 4 unilaterally deaf patients. In the dog experiments, cochlear wall electrode stimulation produced consistent electrophysiologic thresholds that were higher, by approximately 6 dB, than those obtained with bipolar scala tympani stimulation. Clinical testing revealed electrically evoked middle latency response, auditory brain stem response, and/or behavioral detection responses in 3 of 4 patients, at levels below those for facial nerve activation and pain sensation. Electrode place discrimination studies, with controls for loudness cues, revealed near-perfect discrimination in a monkey subject. Eleven of the 12 animal implants were found to be rigidly fixed in the cochlear bone, with direct contract between bone and implant over 8% to 23% of the implant surface for the 6 implants examined in detail. These results suggest that long-term fixation of titanium cochlear wall implants occurs by virtue of intimate implant-bone contact in restricted areas. This approach to prosthetic stimulation demonstrates encouraging performance characteristics in achieving auditory activation.

Quantitative comparison of electrically and acoustically evoked auditory perception: implications for the location of perceptual mechanisms

Electrical stimulation of the human auditory system produces different patterns of spatial and temporal neural activity than those that occur in the normal, acoustically stimulated system. Quantitative comparison of psychophysical performance measured with acoustic and electrical stimulation may allow us to infer the physiological locus of perceptual mechanisms. In this paper we compare psychophysical data on temporal resolution from normal-hearing listeners, cochlear implant listeners, and patients electrically stimulated on the cochlear nucleus. Measures of gap detection, forward masking, and modulation detection will be compared. These comparisons demonstrate that temporal processing is relatively similar across these three groups once the obvious differences in dynamic range are taken into consideration. In addition, preliminary results with speech processors indicate that implant patients can utilize all temporal information in speech. Thus, implant patients have relatively normal temporal resolution and can integrate temporal cues normally for the recognition of complex acoustic patterns such as speech. These results imply that the central auditory systems of implant patients are able to fully utilize the non-natural patterns of temporal neural information produced by electrical stimulation. The differences in the microstructure of the neural pattern (phase locking, stochastic independence of fibers, spatial distribution of activity, etc.) between electrical and acoustic stimulation are apparently not necessary for temporal processing. Thus, the physiological locus of temporal processing mechanisms must be more central in the auditory system than the cochlea and cochlear nucleus.

Electrical stimulation of the auditory nerve: effects of pulse width on frequency discrimination

Effects of pulse width on discrimination of simultaneous changes in frequency and level of electrical pulse trains were studied in a monkey subject with a cochlear implant. At test-stimulus levels where performance was minimum, frequency difference limens were larger for longer-duration pulses than that for shorter-duration pulses. Several factors may have contributed to these differences.

Discrimination of complex electrical stimulation through a multichannel intracochlear implant

A model has been developed to describe the electric fields generated in the inner ear when electrical stimuli are presented through a multichannel implant in the scala tympani of the cochlea. The model relies on the hypothesis that stimuli which excite the largest number of neural elements provide the greatest probability of successful discrimination by the implanted subject. It suggests that the effective stimulus is determined by the linear combination of electrical fields produced by the individual channels, and that excitation takes place in a spatially restricted area of the auditory nerve in the vicinity of the stimulating electrodes. The model was tested by biophysical measurements of the potential developed in the stimulated cochlea, and by a psychophysical study of the ability of a monkey to discriminate complex electrical signals using dual channel stimulation. The experimental findings are in agreement with the computer simulations.

Effects of level on nonspectral frequency difference limens for electrical and acoustic stimuli

The purpose of this experiment was to study the effects of stimulus level on discrimination of frequency as represented in the temporal waveforms of acoustic and electrical signals. The subjects were four nonhuman primates in which one ear had been deafened and implanted with an electrode array and the other ear was untreated. Frequency difference limens for 100 Hz electrical sinusoidal stimulation via a cochlear implant in the deafened ear were compared to those for 100 Hz sinusoidally amplitude-modulated white noise (SAM noise) acoustic stimuli to the normal-hearing contralateral ear. To correct for loudness cues, levels of the test stimuli were varied relative to the reference-stimulus level. The test-stimulus levels at which the percent responses were minimum were determined. These levels were used to measure the frequency difference limens. Frequency difference limens for the electrical stimuli decreased as a function of reference-stimulus level through most of the dynamic range, while those for the acoustic stimuli reached a minimum at 20 dB to 40 dB above threshold. For the electrical stimuli the slopes and relative positions of the frequency difference limen vs. level functions varied from subject to subject, and with changes in electrode configuration within a subject. These differences were related to threshold level and dynamic range. At higher levels of stimulation, frequency difference limens for acoustic and electrical stimuli fell in the same range. The slopes and relative positions of the frequency difference limen vs. level functions for electrical stimuli did not parallel those of level difference limen vs. level functions collected simultaneously from the same ears. The data suggest that nonspectral frequency discrimination may depend on the number of nerve fibers stimulated. With prostheses in cochleas with less than a full complement of auditory nerve fibers, the data suggest that stimulation level is an important variable influencing discriminability.

Comparisons of psychophysical and neurophysiological studies of cochlear implants

This paper compares psychophysical and neural studies of electrical stimulation of the auditory nerve with the goal of evaluating the relevance of single-unit animal models for the development of cochlear prostheses for profoundly deaf humans. Comparative psychophysical studies with implanted deaf subjects indicate that animal models, at least nonhuman primates, provide a close match to humans, though this is not always true for acoustic stimulation of normal-hearing subjects. However, the human-animal comparisons, especially those involving electrical stimuli, need further study using more carefully matched conditions. Comparisons of psychophysical and neurophysiological thresholds for electrical stimulation in animals reveal consistently higher thresholds in the neural studies. A number of factors which may account for these differences are discussed. A partial resolution of the problem could result from conducting neurophysiological and behavioral studies in the same animal. Finally, comparison of psychophysical and neurophysiological studies of temporal encoding suggest that there may be more information encoded in the auditory nerve than is used by the system, at least for nonspectral frequency discrimination. This points to a need for further analysis of the processing of this information at higher levels in the auditory pathway.