Method for detecting auditory steady-state potentials recorded from humans.

Auditory steady-state potentials were recorded from the scalp of adult humans. The stimuli were 100-microseconds clicks presented at a rate of 39.1/s. Four stimulus levels were used (-20, 0, 10, and 20 dB SL). The presence or absence of a response was determined by three frequency-domain methods and examiners. The frequency-domain methods were: magnitude-squared coherence (MSC), phase coherence (PC), and magnitude only (MO). The MSC method generally had the highest d' values, indicating that it was the most sensitive method for detecting responses. The hearing threshold predicted by the MSC method was the lowest, and it was within 4 dB of the behaviorally measured threshold for the click stimuli. Further, the amplitude of the response was significantly more variable than its phase, and no relation was found between the variability of the amplitude of the response and the amplitude of the noise. In summary, response detection methods that incorporate phase information (such as MSC and PC) should be chosen over methods which incorporate only amplitude information.