论文部分内容阅读
认知心理学发现,视觉、听觉接收到信息有冲突时,大脑皮层电位会发生扰动,由此可探索认知冲突控制的“刺激-反应”机制。视觉认知冲突实验较多,成果丰硕,而相应的听觉实验很少,并且得到不一样的结论。本研究利用冲突和非冲突的语音信号刺激,分析研究脑电信号,提出基于三阶段听觉认知控制的时域特征模型。研究人脑听觉通道在出现语音认知冲突时的认知控制的规律下的单次试验脑电数据特征提取方法。根据得到的认知规律,单次试验脑电样本被分成3个部分。被分割的每个阶段使用时域上的平均幅值和Lempel-Ziv复杂度(LZC)进行计算,从而联合3个阶段的特征作为听觉认知脑电样本的特征。结果表明:(1)先发现的认知冲突相关的混合脑电成分“N1-P2&N2&Late-SW”分别体现了听觉认知控制的3个阶段;(2)一个更完整的听觉认知控制过程应包括3个阶段的时域特征:感知阶段:110~140 ms,识别阶段:260~320 ms,解决阶段:500~700 ms;(3)提出针对单次听觉认知控制脑电样本的特征提取方法,联合使用平均幅度和LZC可以获得最好的识别率(99.33%)。实验结果证明了提出的方法能够有效地检测听觉认知控制脑电数据,进而提供人脑认知控制能力评价的声学方法。
Cognitive psychology found that when the information is received in the visual or auditory sense, the potential of the cerebral cortex fluctuates, and thus the “stimulus-response” mechanism of cognitive conflict control can be explored. Visual cognitive conflict experiment more fruitful, and the corresponding auditory experiments are few, and get different conclusions. In this study, we used stimulus of conflict and non-conflict speech signals to analyze and study EEG signals and proposed a time-domain feature model based on three-stage auditory control. To study the method of single-feature extraction of EEG data in the auditory control of human auditory pathway in the presence of phonetic cognitive conflicts. According to the rules of cognition obtained, a single trial of EEG samples is divided into three parts. Each stage of segmentation is calculated using the average amplitude in the time domain and the Lempel-Ziv Complexity (LZC), thus combining the characteristics of the three stages as a feature of the auditory cognitive EEG sample. The results show that: (1) The mixed EEG component “N1-P2 & N2 & Late-SW ” related to cognitive conflict first revealed three stages of auditory control; (2) a more complete auditory cognitive control The process should include three stages of time-domain features: Perception stage: 110-140 ms, Identification stage: 260-320 ms, Resolution stage: 500-700 ms; (3) Proposed for a single auditory cognitive control EEG samples The feature extraction method, combined with the average amplitude and LZC can get the best recognition rate (99.33%). The experimental results show that the proposed method can effectively detect auditory control of EEG data, and then provide an acoustic method to evaluate the cognitive control ability of human brain.