11th Speech in Noise Workshop, 10-11 January 2019, Ghent, BE

Listening through a cochlear implant: the hidden costs

Helen F Willis(a)
University College London, UK

(a) Presenting

Astounding speech perception outcomes can be achieved with the cochlear implant (CI), even when noise is present in the background. However, apparent efficacy is not necessarily synonymous with efficiency. The underlying auditory processing requires listening effort (LE). This is the amount of cognitive resources deliberately allocated to the listening task. It is possible that there is a hidden cost of excessive LE in order to enable successful CI performance.

To assess this, a dual-task paradigm design was utilised, where the participant is required to perform two tasks simultaneously. The dual-task paradigm operates on the cardinal principle that cognitive capacity is finite. Accordingly, it was hypothesised that, if a primary listening task (i.e. recalling British English Lexicon sentences) is performed together with a secondary visual task (i.e. counting the number of targets seen within a stream of flashing digits), performance in the visual task will decrease as the listening task becomes harder. This deterioration in secondary visual task performance can then provide an index of LE.

Three participant groups were tested: 30 normal hearing (NH) participants listening to normal speech, 30 NH participants listening to CI simulations, and 25 CI users listening through their speech processors. Performance in quiet conditions was compared to that in different levels of background noise. Adaptive tracking procedures were used in an attempt to ensure that the challenge of noise was equal for all participants. This principle was also applied to equalise difficulty in terms of the number of channels used in the spectral resolution of the CI simulations. As expected, NH participants only exhibited significant deterioration in visual accuracy when noise was present (p<.001) suggesting increased LE. Strikingly, CI users showed a decrease in visual accuracy when listening in quiet (p<.001), thus implicating higher levels of LE despite optimal listening conditions. Interestingly, this trend of increased LE in quiet was also found in NH controls when CI simulations were applied (p<.001). Therefore, it appears that the degraded auditory input provided by CI induces LE, even in ideal listening conditions.

There is evidence that excessive cognitive load can cause stress-related physical and mental health problems. If LE begins in quiet, efficiency will be compromised further in noise, despite excellent speech perception. There is no clinical test to measure the cost of LE, which means any harmful increases in LE are hidden in the CI population. Active monitoring is needed, as hearing well should mean hearing healthily.

Last modified 2018-12-08 00:23:30