Spectral loudness summation for electrical stimulation in cochlear implant users

Abstract

Objective: This study investigates the effect of spectral loudness summation (SLS) in the electrical domain as perceived by cochlear implant (CI) users. Analogous to SLS in the acoustical domain, SLS was defined as the effect of electrode separation at a fixed overall stimulation rate. Design: Categorical loudness scaling (CLS) was conducted at three overall stimulation rates using single-electrode stimuli and multi-electrode stimuli presented interleaved on two or four electrodes. The specific loudness of the pulses in the multi-electrode stimuli were equalized based on single-electrode measurements at the same overall stimulation rate. At a fixed overall stimulation rate and a fixed loudness perception, SLS was calculated as the difference in mean current between single-electrode and multi-electrode stimuli. Study sample: Ten postlingually deafened adult CI users. Results: The amount of SLS varied between subjects and between the number and location of the stimulated electrodes in the multi-electrode configuration. SLS was significantly higher than 0 for a subset of the subjects. Conclusions: For a subpopulation of CI users, loudness models should account for nonlinear interactions between electrodes (in the perceptual domain). Similarly, SLS should be accounted for when using CLS outcomes for fitting purposes, at least in a subpopulation of CI users.

Key Words::

• Cochlear implant
• categorical loudness scaling
• spectral loudness summation
• loudness perception
• loudness growth

Acknowledgements

We would like to thank the subjects for their time and effort. Also we would like to thank Cochlear for their technical support. Furthermore we would like to thank Nan van Geloven for her support with the statistical analysis. Finally, we would like to thank Rolph Houben, Bas van Dijk, and Filiep Vanpoucke for their comments on earlier versions of the article.

Notes

Declaration of interest: The authors report no conflicts of interest.

Notes

1. Brand and Hohmann (2001) referred to this procedure as adaptive categorical Loudness scaling (ACALOS).

2. The data at 900 pps were also used in Theelen et al (2014), to assess the reproducibility of the measurement tool.

3. For practical reasons, the fitting model used in the present study is different than that used by Theelen et al (2014) for a subset of the dataset and other expansive models that have been used in the literature for loudness judgments (of any kind) and current in μA (e.g. Fu & Shannon, 1998; Chatterjee, 1999; Cohen et al, 2009).

4. To assess the impact of using a different fitting model, we retrospectively repeated all analyses of SLS for LGFs obtained with the exponential fitting model used by Theelen et al (2014) rather than the piecewise linear fitting model. In doing so, we mathematically corrected for the actual stimulation levels used during the measurements. The results indicated that none of the conclusions about SLS would have been different if the exponential rather than the piecewise linear fitting model had been used.

5. For the purpose of equalizing the specific loudness of the pulses in the multi-electrode configuration according to the single-electrode LGFs at the same overall stimulation rate, the piecewise linear model was applied to all single-electrode measurement data instead of to the median values per loudness category. Fitting through all data instead of the median values leads to slightly different ‘predicted stimulation levels’ and thus slightly different SLS values. We performed a two-sided paired t-test between the SLS values based on both types of ‘predicted stimulation levels’. This test indicated that the effect of fitting through all data or through the median values per loudness category on SLS was not significant (F_1,240 = − 0.67, p = 0.50). The results presented here are based on the single-electrode LGFs fitted through the median values per loudness category.

6. Some pairwise comparisons were not based on the complete dataset. For example, for 40 of the 50 available multi-electrode LGFs, SLS data was available up to ‘Soft – Medium’ (20 LU) or ‘Medium’ (25 LU) loudness perceptions. This was the consequence of limiting the analysis of SLS to loudness categories within the test range.

7. This analysis was done up to a loudness perception of ‘Loud – Very loud’ (40 LU) because only a few data points were available above this loudness perception.

8. For a loudness perception of ‘Medium to Loud’ (30 LU) only two data points were available (both for the 2-electrode stimuli at the basal electrodes). Therefore, this loudness category was omitted in Figure 5. However, it should be noted that these two data points did not agree with the trend described in the text for loudness categories up ‘Medium’ (25 LU).