Authors: Pilar Castillo¹; Rafael Delgado²; Ross Maddox³; Joaquin Valderrama⁴
¹University of Granada; ²Intelligent Hearing Systems Corp. / University of Miami; ³University of Michigan; ⁴University of Granada / Macquarie University
Background: The parallel auditory brainstem response (pABR) enables simultaneous acquisition of frequency-specific ABRs across ears, substantially reducing recording time compared to serial paradigms. In pABR, responses are estimated by averaging EEG segments associated with Poisson-distributed stimuli. However, simultaneous stimulation introduces overlapping responses that may degrade estimation quality.
Method: A simulation framework was developed to characterise the effect of increasing the number of simultaneous stimuli on pABR estimation quality. Noise-free EEG signals were synthesised as convolutions of Poisson-distributed event sequences with up to 10 different ABR templates. Responses were estimated using conventional pABR averaging and multi-response deconvolution. Monte Carlo simulations evaluated response quality in terms of signal-to-noise ratio (SNR).
Results: With conventional pABR averaging, SNR progressively decreases as the number of simultaneous stimuli increases, whereas deconvolution maintains stable SNR across conditions. Despite the SNR degradation observed with conventional averaging, the resulting effect on ABR morphology was found to be negligible due to the large temporal jitter of Poisson-distributed sequences. Deconvolution preserved both SNR and waveform morphology independently of the number of overlapping responses.
Conclusion: Increasing the number of simultaneous stimuli in the pABR paradigm reduces estimation SNR when using conventional averaging; however, this degradation has minimal practical impact on ABR morphology. Multi-response deconvolution minimises the SNR decrease, providing a robust strategy for pABR analysis in paradigms involving large numbers of simultaneous stimuli and preserving response quality even in highly overlapping conditions.
