MMN Analysis for Chinese Lexical Tones Based on Enhanced ICA

This paper describes the physiological coding mechanism of Chinese lexical tones with the single-trial mismatch negativity (MMN) study using independent component analysis (ICA). The enhanced ICA algorithm combined with subspace filtering performs MMN preprocessing, the evoked components are extracted again to improve the performance of removing artifact. Experiment results show that MMN responses are elicited distinctly by the different stimulation blocks at the prefrontal electrodes, and especially better contrast responses are obtained with the falling-rising and falling tone deviants among the flat tone standards. The peak latency and amplitude have large fluctuation in the blocks with the falling tone deviants, P3 is also distinct in them.