Slow-varying normalization explains auditory steady-state masking interactions in human EEG

The inherent ability of sensory neurons to entrain to modulations in the temporal structure of an auditory stimulus gives rise to the auditory steady-state response (ASSR). Simultaneous presentation of multiple acoustic stimuli by frequency tagging them to generate ASSRs at different frequencies is routinely employed for hearing threshold determination and cognitive studies. However, the nature of ASSR interactions as a function of competing modulation frequencies in the absence of overt behaviour and the underlying neural mechanisms have not been well studied. Such interactions have previously been studied with visual stimuli that generate steady-state visually evoked potentials (SSVEPs) and explained using a normalization model. Here, we tested whether similar interactions are observed in the auditory domain, and if so, can be explained using a similar model. We played sinusoidally amplitude-modulated stimuli simultaneously modulated at two frequencies while we recorded a 64-channel EEG from subjects who passively listened to these sounds with closed eyes. We used multiple modulation frequencies and depths to characterize the ASSR modulation masking response profile. We observed that modulation frequencies closer to each other suppress the ASSR strongly than frequencies that are farther apart, similar to the interaction observed in the visual domain using SSVEPs. The observed suppression was captured by a slow-varying normalization model, which was initially used to explain SSVEP interactions. We obtained a band-pass shaped suppression profile with a low-pass cutoff that matched to that observed for SSVEP interactions. Our findings highlight the universality of the normalization model in accounting for masking interactions across modalities.