Precedence Effect with Cochlear Implants - Simulation and Results

Bernhard Seeber and Ervin Hafter
University of California, Berkeley


In recent years cochlear implants (CIs) were successful in restoring 
the ability in many patients to understand speech in quiet and in 
acoustically dry environments. However, patients still encounter
great difficulties in understanding in situations of
speech-in-noise. The present study investigates if CI-patients show
the precedence effect, i.e. the suppression of the directional
influence of the lagging sound on the localization of a leading
sound. Previous studies showed that CI-patients rely on the
evaluation of interaural level cues (ILDs) for localization, and
interaural temporal cues (ITDs) are ignored. Since precedence is
predominantly based on the evaluation of ITDs at low frequencies, it
is interesting to see if the altered cues with CIs carry enough
information for the precedence effect.  

Experiments are done in the simulated open-field environment, a
calibrated loudspeaker setup in our anechoic chamber. A burst of
wide-band-noise or the speech utterance "shape" are played with
equal level, but varying interstimulus-delay, from speakers at ±30°.
Subjects judge the apparent location of the sound using a movable
visual pointer. 

Four subjects with bilateral CIs were tested so far, of which 2
subjects showed some localization ability in the localization
baseline test. None of those subjects showed the precedence effect.
If sounds were played concurrently, the sound was localized in the
middle between both loudspeakers, equivalent to summing localization
in normal hearing. If a time delay was introduced between lead and
lag, sounds were localized at both loudspeakers separately, even for
short delay times. This is surprising because we assumed summing
localization to take place over an extended range of delay times.
Studies with vocoder-simulations of CIs are underway to clarify if
the information present in ILDs and envelope-ITDs is sufficient to
evoke the precedence effect in normal hearing subjects. Support
provided by NIH and NOHR.