Is it really a mismatch negativity? An assessment of methods for determining response validity in individual subjects.

Mismatch negativity (MMN) responses were collected from 86 normal school-age children in response to synthesized speech syllables, /wa/and two variants of /ba/. Waveform characteristics and statistical properties of the responses were analyzed across stimulus conditions in order to assess methods for determining response validity in individuals. Methods were compared using signal detection theory techniques. Criteria based on measurements of response area, onset latency, and duration were the best indicators of response validity. Also a promising indicator of validity was the interval of significance based on Z transformations determined by considering the variance of the underlying noise distribution. Correlations of individual responses with the grand average and integral calculations of the response negativity showed somewhat lower d' values. Statistical methods which utilized response subaverages were the poorest indicators of response validity. Likely the methods are limited primarily by the signal to noise ratio of the MMN compared to the underlying physiologic noise. Improvement of the signal to noise ratio remains a significant factor in the interpretation of MMN for individual subjects.