Memoire Online - Contrainte Psycho-Physiques et Electrophysiologiques sur le codage de la stimulation électrique chez les sujets porteurs d'un implant cochléaire

L'objectif de cette étude a été d'évaluer sur une population de 5 sujets implantés cochléaires Digisonic, leur capacité de discrimination deux enveloppes temporellement asymétriques. Les signaux utilisés dérivent de l'étude de Patterson et al, 1994 effectuée chez les normo-entendants et sont directement transmis au récepteur interne via l'antenne sans passer par le processeur vocal.

Les résultats montrent que les sujets implantés discriminent mieux les asymétries temporelles du signal que les sujets normo-entendants. Cela pourrait être expliqué par l'absence de lissage due à la biomécanique cochléaire lors d'une stimulation électrique.

MRC, Applied Psychology Unit, 15 Chaucer Road, Cambridge CB2 2EF, United Kingdom and PCH Département de Psychologie Cognitive, Institut de Psychologie, Université Lumière Lyon II,

Laboratoire de Physiologie Sensorielle, "Perception & Mécanismes Auditifs," UPRESA CNRS 5020, Pav. U. Hôpital E. Herriot, 69437 Lyon Cedex 03, France

MRC, Applied Psychology Unit, 15 Chaucer Road, Cambridge CB2 2EF, United Kingdom

Several studies have recently demonstrated that normal-hearing listeners are sensitive to short-term temporal asymmetry in the envelopes of sinusoidal or noise carriers. This paper presents a study in which cochlear implantees were presented trains of current pulses with temporally asymmetric envelopes through one channel of an implant that stimulates the auditory nerve directly, thereby bypassing cochlear processes. When the level of the stimuli was adjusted to fit their audibility range, the implantees were able to discriminate temporal asymmetry over a much wider range than normal-hearing listeners. The results suggest that the perception of temporal asymmetry is limited by compression in the normal cochlea. (c) 1997 Acoustical Society of America. [S0001-4966(97)06707-6]

When a sound with an asymmetric temporal envelope is reversed in time, it often produces a marked change in the timbre of the sound. The effect is important because the power spectrum of a sound does not vary with the time reversai. Although the perceptual effect of the time reversai is well-known, it was largely ignored until recently, when Patterson (1994a, b) showed how the effects of the time reversai could be studied systematically using "damped" and "ramped" sinusoids. The term "damped sinusoid" refers to a segment of a sinusoid with a damped exponential envelope that is repeated cyclically to produce a sustained sound. The "ramped sinusoid" is simply the damped sinusoid reversed in time. The envelope "damps" down in the former case and "ramps" up in the latter case. Normal-hearing listeners can discriminate damped sinusoids from ramped sinusoids when the half-life of the envelope is in the range 1-50 ms. Compression in the cochlea of the normal-hearing listener reduces the dynamic range of damped and ramped sinusoids in the neural representations flowing from the cochlea. This led to the intriguing suggestion that cochlear implantees might exhiba even better asymmetry discrimination than normalhearing listeners because the implant bypasses the cochlea and any compression it might impose. We assume, as argued by Shannon (1986), that sensitivity to level differences in the central auditory system of cochlear implantees is about the same as that for normals when presented at the same sensation level. It is also the case that with implantees, one can restrict the stimulation to a single electrode and thereby preclude spectral cues to a degree that is not possible with normal listeners. Thus cochlear implantees would appear to pro- vide us with a unique opportunity to investigate the limits of retrocochlear processing of temporal asymmetry.

There have been several demonstrations that cochlear implantees can process temporal envelope information. Hochmair and Hochmair-Desoyer (1984) reported that implantees could distinguish among square, triangle, and sine waves for repetition frequencies up to 400 Hz. Shannon (1986) demonstrated near normal temporal integration, gap detection, and forward masking in two implantees, and more recently, Shannon (1992) reported that implantees could detect amplitude modulation at least as well as normal-hearing listeners. In these experiments, however, there is no manipulation of temporal asymmetry.

Pairs of ramped and damped envelopes with the same half-life were presented to cochlear implantees to assess their discrimination of temporal asymmetry. The electrical signal was delivered directly to a single electrode of a Digisonic DX10 implant without going through the preprocessor to avoid the compression normally applied by the device.

Five postlingually deaf listeners (SP, BM, LR, FJ, BR) participated in the experiment. Clinical information about these patients is presented in Table I. They were ail implanted with a Digisonic DX10 device (MXM), which is a transcutaneous 15-channel cochlear implant with an intracochlear electrode array (Beliaeff et al., 1994). Activation was limited to the most basal electrode which delivered biphasic current pulses; the remaining 14 electrodes in the array were connected together to serve as the return path for the current

Age Patient (years)		Cause of deafness	Duration of implant use (months)	Pulse amplitude (V)	Min Max (us/phase) (,us/phase)		2 a C
BM	64	head trauma	7	1.5	12	55
SP	44	head trauma	6	1.3	14.5	45
LR	60	unknown	3	1.3	10	45
FJ	45	head trauma	48	1.6	25	95
BR	36	progressive deafness	1	1.3	10	40	ri a
							C o o

("common ground" mode of stimulation). X-ray photographs revealed that electrode positioning was roughly the same for all implantees. All stimuli were presented directly to the implanted electrode without going through the speech processor. For each implantee, the pulse amplitude was fixed and the pulse duration was adjusted from threshold (min) to comfort level (max); the pulse amplitude and the min and max values for each implantee are presented in Table I. All implantees were experienced in two-interval, two-alternative forced choice tasks (21, 2AFC).

The stimuli were trains of biphasic current pulses of fixed amplitude. The pulse duration as a function of time was shaped by a ramped or damped function. Equation (1) shows the general foret of a damped envelope:

hl is the half-life of the damped envelope, and c is a constant (-0.693 147), that brings the envelope to [max --Inin]/2 in hl ms. T is the repetition period which is 50 ms. The pulse rate was 800 Hz, which was the maximum pulse rate provided by the device used to perform direct stimulation. The ramped envelopes were produced by reversing the damped envelopes in time. The envelopes were digitally generated on a PC by a 16-bit D/A converter at a sampling frequency of 44.1 kHz. The electrical stimuli were generated using a Digistim system (MXM) connected to the PC via a serial port. The duration of the stimuli was 500 ms; the silent interval between stimuli was 500 ms. The comfort level, max, was adjusted as the half-life was increased so that the stimuli were of approximately equal loudness. The loudness of the stimuli was equated by asking each implantee to adjust the level of each damped stimulus to that of a damped stimulus with a 8-ms half-life. Adjustment was performed using an initial step size

FIG. 1. Ramped (left column) and damped (right column) current pulse trains with a half-life of 1 ms (top section), 8 ms (middle section), and 64 ms (bottom section). The carrier frequency is 800 Hz. The repetition period is 50 ms.

of 5us/phase and a final step size of 0.5 ,us/phase. Three estimates were collected and averaged. The average max values measured for each half-life and for each implantee are presented in Table II.

Segments of the current pulse trains are presented in Fig. 1; the left and right panels show ramped and damped fonctions, respectively. The half-life of the exponential is 1 ms in the top section, 8 ms in the middle section, and 64 ms in the bottom section of Fig. 1. The trains of current pulses show that the Digistim system preserves the temporal envelope.

Implantees sat in front of a keyboard connected to the computer controlling the experiment. The task was 21, 2AFC. In a given trial, implantees were presented a ramped stimulus in one interval and a damped stimulus with the same half-life in the other interval, and asked to choose the interval with the "more tonal quality." In each trial, inter-

TABLE II. Max values (in ms/phase) producing stimuli of equal loudness as the half-life is increased from 0.25 to 1024 ms. The data are presented for the five cochlear implantees of the study.

hl (ms)	0.25	0.5	1	2	4	16	32	64	128	256	512	1024
BM	65	60	60	60	55	55	50	45	40	40	37	40
SP	53	52.5	50	48	46.5	44	43	42	41	40	38	37.5
LR	38	40	41	43	45	45	45	45	40	37	35	35
FI	80	86.5	90	90	95	95	85	77	71	65.5	67	65
BR	35	35	35	37	42.5	36	31.5	30	27.5	23	23	23

FIG. 2. Average psychometric function for the five cochlear implantees, showing performance as a function of hl. Error bars show one standard deviation about the mean performance of the five implantees. For comparison, the data are piotted along with the mean of the data obtained with normal-hearing listeners.

vals were presented in random order. A correct response corresponded to a ramped response. The half-life was fixed within a block and was varied from 0.25 to 1024 ms from block to block. Each block contained 30 trials. Implantees received visual feedback conceming the accuracy of their response after each trial. They received 15 min of preliminary training before participating in the experiment.

The average psychometric function for the five implantees is presented by the solid line with filled circles in Fig. 2. The dotted line without symbols show the mean of the data obtained by Patterson (1994b) and Irino and Patterson (1996) with normal-hearing listeners, using a carrier frequency of 800 Hz and a repetition rate of 50 ms. The results show that implantees can discriminate ramped and damped envelopes better than normal-hearing listeners. Performance is above chance for half-lives as short as 0.5 ms and as long as 500 ms; the range for normal-hearing listeners is about 1-50 ms. The individual psychometric functions were very similar in shape to that of the average psychometric function. Table III

TABLE III. Short threshold half-life and long threshold half-life for the five cochlear implantees. The number of doublings between the short and long thresholds half-lives is presented in the last column.

Patient	Shortest hl (ms)	Longest hl (ms)	Number of doublings
BM	0.57	610	10.06
SP	0.9	780	9.75
LR	0.2	300	10.55
FJ	0.69	860	10.28
BR	0.4	460	10.16

presents estimates of the shortest and the longest justdiscriminable half-lives for the individual psychometric functions, and it shows that the short and long thresholds covary. The number of doublings between the short and long threshold is presented in the right-hand column; it is close to ten for all of the implantees indicating that the size of the range where discrimination is possible is highly consistent across implantees.

When damped and ramped envelopes with the same half-life are presented to cochlear implantées at the optimum intensity and without compression, they can discriminate the temporal asymmetry over a much broader range of half-lives than normal-hearing listeners. The stimulation was restricted to a single electrode, thereby precluding the use of spectral cues. As stimuli were equated in loudness, it is also unlikely that loudness cues played a role in the discrimination task. Discrimination at the shortest half-lives may have been limited by the pulse rate of the Digistim device, 800 Hz, so envelopes with half-lives less than 1.25 ms would not be properly represented. However, extension of the discrimination range to shorter half-lives would only strengthen the conclusion.

The obvions explanation for the superior discrimination performance of the hearing-impaired listeners would appear to be that compression in the normal cochlea reduces sensitivity to level differences, and temporal asymmetry is encoded as differences in level differences. Without compression, and assuming the same level processing mechanism in the central auditory system, hearing-impaired listeners are operating on larger level differences and so do better, but of course, only within their much reduced dynamic range. This is consistent with Shannon's (1992) finding that implantees often exhibit greater sensitivity to amplitude modulation than normal-hearing listeners when the modulation rate is less than 300 Hz.

Implantees cannot discriminate temporal pitch above 300 Hz (Tong et al., 1982). All the implantees who participated in the presented study labeled ramped stimuli as "smooth," "uniform," or "continuous," and damped stimuli as "interrupted" or "with a drum-like quality." This suggests that the "tonal quality" used to instruct listeners was interpreted by implantees as a "smooth quality."

Finally, the results indicate that it is important to ensure that hearing aids and cochlear implants do not restrict temporal asymmetry unduly when applying compression to control level. Asymmetry is a prominent property of speech sounds and hearing-impaired listeners are likely to make good use of it when it is available.

The first author was supported by a post-doctoral grant from the FYSSEN Foundation. The second author was supported by a CIFRE doctoral grant from the MXM Company. We wish to thank two anonymous reviewers for extensive comments on earlier versions of the paper.

Beliaeff, M., Dubus, P., Leveau, J. M., Repetto, J. C., and Vincent, P. (1994). "Sound processing and stimulation coding of DIGISONIC DX10 15-channel cochlear implant," in Advances in Cochlear Implant, edited by E. S. Hochmair (Vienna, Manz), pp. 198-203.

Hochmair, E. S., and Hochmair-Desoyer, I. J. (1984). "Aspects of sound signal processing using the Vienna intra- and extracochlear implants," in Cochlear Implants, edited by R. A. Schindler and M. M. Merzenich (Ras/en, New York), pp. 101-110.

Irino, T., and Patterson, R. D. (1996). "Temporal asymnetry in auditory perception and a "delta-gamma" theory of asymmetric intensity enhancement in the peripheral auditory system," J. Acoust. Soc. Am. 99, 23162331.

Patterson, R. D. (1994b). "The sound of a sinusoid: Time-interval models," J. Acoust. Soc. Am. 96, 1419-1428.

Shannon, R. (1986). "Temporal processing in cochlear implants," in Sensorineural Hearing Loss: Mechanisms, Diagnosis and Treatment, edited by M. J. Collins, T. J. Glattke, and L. A. Harker (University of Iowa, Iowa City), pp. 349-368.

Shannon, R. (1992). "Temporal modulation transfer functions in patients with cochlear implants," J. Acoust. Soc. Am. 91, 2156-2164.

Tong, Y. C., Clark, G. M., Blamey, P. J., Busby, P. A., and Dowell, R. C. (1982). "Psychophysical studies for 2 multiple-channel cochlear implant patients," J. Acoust. Soc. Am. 71, 153-160.

L'article précédent démontre la grande faculté des patients implantés à analyser l'enveloppe du signal lors que l'on stimule directement le récepteur. Nous avons voulu évaluer cette capacité de discrimination dans des conditions réelles, avec le traitement du processeur vocal.

Contrainte Psycho-Physiques et Electrophysiologiques sur le codage de la stimulation électrique chez les sujets porteurs d'un implant cochléaire

Article 11 :

DISCRIMINATION OF TEMPORAL ASYMETRY IN COCHLEAR IMPLANTEES
C. Lorenzi, S. Gallégo, R.D. Patterson

Contrainte Psycho-Physiques et Electrophysiologiques sur le codage de la stimulation électrique chez les sujets porteurs d'un implant cochléaire

Article 11 :

DISCRIMINATION OF TEMPORAL ASYMETRY IN COCHLEAR IMPLANTEESC. Lorenzi, S. Gallégo, R.D. Patterson

DISCRIMINATION OF TEMPORAL ASYMETRY IN COCHLEAR IMPLANTEES
C. Lorenzi, S. Gallégo, R.D. Patterson