To characterize some of the benefits available from using two cochlear implants compared with just one, sound-direction identification (ID) abilities, sensitivity to interaural time delays (ITDs) and speech intelligibility in noise were measured for a bilateral multi-channel cochlear implant user.
Sound-direction ID in the horizontal plane was tested with a bilateral cochlear implant user. The subject was tested both unilaterally and bilaterally using two independent behind-the-ear ESPRIT (Cochlear Ltd.) processors, as well as bilaterally using custom research processors. Pink noise bursts were presented using an 11-loudspeaker array spanning the subject’s frontal 180° arc in an anechoic room. After each burst, the subject was asked to identify which loudspeaker had produced the sound. No explicit training, and no feedback were given. Presentation levels were nominally at 70 dB SPL, except for a repeat experiment using the clinical devices where the presentation levels were reduced to 60 dB SPL to avoid activation of the devices’ automatic gain control (AGC) circuits. Overall presentation levels were randomly varied by ±3 dB. For the research processor, a “low-update-rate” and a “high-update-rate” strategy were tested.
Direct measurements of ITD just noticeable differences (JNDs) were made using a 3 AFC paradigm targeting 70% correct performance on the psychometric function. Stimuli included simple, low-rate electrical pulse trains as well as high-rate pulse trains modulated at 100 Hz.
Speech data comparing monaural and binaural performance in noise were also collected with both low, and high update-rate strategies on the research processors. Open-set sentences were presented from directly in front of the subject and competing multi-talker babble noise was presented from the same loudspeaker, or from a loudspeaker placed 90° to the left or right of the subject.
For the sound-direction ID task, monaural performance using the clinical devices showed large mean absolute errors of 81° and 73°, with standard deviations (averaged across all 11 loudspeakers) of 10° and 17°, for left and right ears, respectively. Fore bilateral device use at a presentation level of 70 dB SPL, the mean error improved to about 16° with an average standard deviation of 18°. When the presentation level was decreased to 60 dB SPL to avoid activation of the automatic gain control (AGC) circuits in the clinical processors, the mean response error improved further to 8° with a standard deviation of 13°. Further tests with the custom research processors, which had a higher stimulation rate and did not include AGCs, showed comparable response errors: around 8 or 9° and a standard deviation of about 11° for both update rates. The best ITD JNDs measured for this subject were between 350 to 400 μsec for simple low-rate pulse trains. Speech results showed a substantial headshadow advantage for bilateral device use when speech and noise were spatially separated, but little evidence of binaural unmasking. For spatially coincident speech and noise, listening with both ears showed similar results to listening with either side alone when loudness summation was compensated for. No significant differences were observed between binaural results for high and low update-rates in any test configuration. Only for monaural listening in one test configuration did the high rate show a small significant improvement over the low rate.
Results show that even if interaural time delay cues are not well coded or perceived, bilateral implants can offer important advantages, both for speech in noise as well as for sound-direction identification.