Precedence-Effect with cochlear implant simulation

Bernhard U. Seeber and Ervin Hafter
University of California, Berkeley

The Precedence-Effect describes the perceptual suppression of a
delayed sound copy in the presence of a leading sound. From the view
of auditory scene analysis precedence can be seen as the unability to
segregate lead and lag into two separate events. One important
segregation cue for harmonic sounds is fundamental frequency which is
largely missing in wispered speech or for patients with cochlear
implants (CIs). So far we could not find CI-patients who show
precedence.  

The experiments study for ongoing stimuli processed with a noise-band
vocoder (CI-simulation): 
1) if precedence can be evoked,
2) the impact of interaural correlation (IAC) in the synthesizing
   noise of the vocoder as future CIs will transmit some interaural
   time differences (ITD),   
3) binaural cues responsible for precedence,
4) different signal processing schemes for extraction and
   synthetization of ITDs.  

In experiments 1 and 2 subjects localize precedence stimuli in the
free-field and with virtual acoustics. For short delays subjects
localize a single sound at the lead position whereas for longer delays
two sounds are heard at the lead and lag positions.  

In experiment 3 subjects lateralize sounds processed with a noise-band
vocoder for varying IACs of the synthesizing noise. The results
exhibit some precedence for all correlations, but the auditory scene
partly breaks up into two images at the lead and lag side. The IAC
affects image width and blurs the image to the middle, especially for
later parts of a CVC-word.  

Experiment 4 studies the contribution of envelope-ITDs to precedence
by suppressing ITD-cues through temporal envelope quantization. As the
results show little change to experiment 3 it is concluded that
precedence with noise-band stimuli can be evoked solely by ILD cues.
Support provided by NIH and NOHR.