Unintentional cues in psychophysical spectral ripple tests: Insights from models of acoustic and electric hearing

Authors: Savine Martens1, Johan Frijns1, Jeroen Briaire1

1Leiden University Medical Center

Background: The spectral ripple test (Won, 2007 JARO) and the spectrally modulated ripple test (SMRT) (Aronoff & Landsberger, 2013 JASA) assess spectral resolution with CIs. However, the validity of these tests has been questioned. The upper limit of discrimination in the former test in CI users is assumed to be caused by multiple peaks falling within a single filter. For the SMRT, an inadequate carrier density caused aliasing. As a result, the psychometric curve for normal hearing participants was non-monotonous, as at 16 RPO the stimuli were more discriminable than at 10 RPO. Narne et al. (2016, JASA) hypothesized that spectral smearing explains why normal hearing subjects reach higher RPOs in the SMRT than in the other test. Here, we propose a pipeline to assess the presence of these unintentional cues in silico.

Method: Through computational models of the neural responses of the cochlear nerve for acoustic and electric hearing (with CI), which include the cochlear tonotopy, we could portray the neural response comparable to a spectrum and a spectrogram. The model of Bruce et al. (2018, Hearing Research) was used to assess normal hearing. The electric hearing model consists of a speech processor, a volume conduction model, an active nerve model and a stochastic adaptation model (Kalkman, 2022 Hearing Research; Van Gendt, 2020 Hearing Research).

Results: For the spectral ripple test, we could show the limitation in the speech processor’s output and the cochlear nerve’s output, and the slight benefit of current steering. With SMRT, we could show aliasing in the lower carrier density. In line with psychophysics, the ripple resolution was reduced in our electric hearing results. Lastly, spectral smearing appears to add an unintentional cue in the SMRT.

Conclusion: These initial results demonstrate that our framework can serve as a first-step assessment for future psychophysical tests. Moreover, the pipeline may be used to test new speech coding strategies.