Kashyap Rajan, Ouyang Guang, Sommer Werner, Zhou Changsong
Department of Physics, Hong Kong Baptist University, Kowloon Tong, Hong Kong; Centre for Nonlinear Studies and the Beijing-Hong Kong-Singapore Joint Centre for Nonlinear and Complex Systems (Hong Kong), Institute of Computational and Theoretical Studies, Hong Kong Baptist University, Kowloon Tong, Hong Kong.
Department of Psychology, Humboldt-Universität zu Berlin, Berlin, Germany.
Brain Res. 2016 Feb 1;1632:58-72. doi: 10.1016/j.brainres.2015.12.001. Epub 2015 Dec 9.
The late components of event-related brain potentials (ERPs) pose a difficult problem in source localization. One of the reasons is the smearing of these components in conventional averaging because of trial-to-trial latency-variability. The smearing problem may be addressed by reconstructing the ERPs after latency synchronization with the Residue Iteration Decomposition (RIDE) method. Here we assessed whether the benefits of RIDE at the surface level also improve source localization of RIDE-reconstructed ERPs (RERPs) measured in a face priming paradigm. Separate source models for conventionally averaged ERPs and RERPs were derived and sources were localized for both early and late components. Jackknife averaging on the data was used to reduce the residual variance during source localization compared to conventional source model fitting on individual subject data. Distances between corresponding sources of both ERP and RERP models were measured to check consistency in both source models. Sources for activity around P100, N170, early repetition effect (ERE/N250r) and late repetition effect (LRE/N400) were reported and priming effects in these sources were evaluated for six time windows. Significant improvement in priming effect of the late sources was found from the RERP source model, especially in the Medio-Temporal Lobe, Prefrontal Cortex, and Anterior Temporal Lobe. Consistent with previous studies, we found early priming effects in the right hemisphere and late priming effects in the left hemisphere. Also, the priming effects in right hemisphere outnumbered the left hemisphere, signifying dominance of right hemisphere in face recognition. In conclusion, RIDE reconstructed ERPs promise a comprehensive understanding of the time-resolved dynamics the late sources play during face recognition.
事件相关脑电位（ERP）的晚期成分在源定位方面存在难题。原因之一是由于每次试验潜伏期的变异性，这些成分在传统平均过程中会被模糊。可以通过使用残差迭代分解（RIDE）方法在潜伏期同步后重建ERP来解决模糊问题。在此，我们评估了RIDE在表面水平的优势是否也能改善在面孔启动范式中测量的RIDE重建ERP（RERP）的源定位。我们推导了传统平均ERP和RERP的单独源模型，并对早期和晚期成分进行了源定位。与基于个体受试者数据的传统源模型拟合相比，对数据进行留一法平均以减少源定位期间的残余方差。测量ERP和RERP模型相应源之间的距离，以检查两个源模型的一致性。报告了P100、N170、早期重复效应（ERE/N250r）和晚期重复效应（LRE/N400）周围活动的源，并在六个时间窗口评估了这些源中的启动效应。从RERP源模型中发现晚期源的启动效应有显著改善，特别是在颞中叶、前额叶皮层和颞叶前部。与先前的研究一致，我们发现在右半球有早期启动效应，在左半球有晚期启动效应。此外，右半球的启动效应数量超过左半球，这表明右半球在人脸识别中占主导地位。总之，RIDE重建的ERP有望全面理解晚期源在人脸识别过程中所起的时间分辨动态作用。
