期刊
JOURNAL OF NEUROSCIENCE METHODS
卷 313, 期 -, 页码 77-94出版社
ELSEVIER SCIENCE BV
DOI: 10.1016/j.jneumeth.2018.12.010
关键词
Eye tracking; Eye movements; EEG; FRP; Regression; rFRP; Free viewing
资金
- DFG [VO 1683/2-1]
- ERC [617891]
- European Research Council (ERC) [617891] Funding Source: European Research Council (ERC)
Background: In the analysis of combined ET-EEG data, there are several issues with estimating FRPs by averaging. Neural responses associated with fixations will likely overlap with one another in the EEG recording and neural responses change as a function of eye movement characteristics. Especially in tasks that do not constrain eye movements in any way, these issues can become confounds. New method: Here, we propose the use of regression based estimates as an alternative to averaging. Multiple regression can disentangle different influences on the EEG and correct for overlap. It thereby accounts for potential confounds in a way that averaging cannot. Specifically, we test the applicability of the rERP framework, as proposed by Smith and Kutas (2015b), (2017), or Sassenhagen (2018) to combined eye tracking and EEG data from a visual search and a scene memorization task. Results: Results show that the method successfully estimates eye movement related confounds in real experimental data, so that these potential confounds can be accounted for when estimating experimental effects. Comparison with existing methods: The rERP method successfully corrects for overlapping neural responses in instances where averaging does not. As a consequence, baselining can be applied without risking distortions. By estimating a known experimental effect, we show that rERPs provide an estimate with less variance and more accuracy than averaged FRPs. The method therefore provides a practically feasible and favorable alternative to averaging. Conclusions: We conclude that regression based ERPs provide novel opportunities for estimating fixation related EEG in free-viewing experiments.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据