Miner L A, McFarland D J, Wolpaw J R
Wadsworth Center, New York State Department of Health and State University of New York, Albany 12201-0509, USA.
Arch Phys Med Rehabil. 1998 Sep;79(9):1029-33. doi: 10.1016/s0003-9993(98)90165-4.
To demonstrate that humans can learn to control selected electroencephalographic components and use that control to answer simple questions.
Four adults (one with amyotrophic lateral sclerosis) learned to use electroencephalogram (EEG) mu rhythm (8 to 12Hz) or beta rhythm (18 to 25Hz) activity over sensorimotor cortex to control vertical cursor movement to targets at the top or bottom edge of a video screen. In subsequent sessions, the targets were replaced with the words YES and NO, and individuals used the cursor to answer spoken YES/NO questions from single- or multiple-topic question sets. They confirmed their answers through the response verification (RV) procedure, in which the word positions were switched and the question was answered again.
For 5 consecutive sessions after initial question training, individuals were asked an average of 4.0 to 4.6 questions per minute; 64% to 87% of their answers were confirmed by the RV procedure and 93% to 99% of these answers were correct. Performances for single- and multiple-topic question sets did not differ significantly.
The results indicate that (1) EEG-based cursor control can be used to answer simple questions with a high degree of accuracy, (2) attention to auditory queries and formulation of answers does not interfere with EEG-based cursor control, (3) question complexity (at least as represented by single versus multiple-topic question sets) does not noticeably affect performance, and (4) the RV procedure improves accuracy as expected. Several options for increasing the speed of communication appear promising. An EEG-based brain-computer interface could provide a new communication and control modality for people with severe motor disabilities.
证明人类能够学会控制特定的脑电图成分，并利用这种控制来回答简单问题。
四名成年人（其中一名患有肌萎缩侧索硬化症）学会使用感觉运动皮层上的脑电图（EEG）μ节律（8至12赫兹）或β节律（18至25赫兹）活动来控制视频屏幕顶部或底部边缘目标的垂直光标移动。在随后的实验环节中，目标被替换为“是”和“否”两个词，参与者使用光标回答来自单主题或多主题问题集的口头“是/否”问题。他们通过响应验证（RV）程序确认答案，即在词的位置切换后再次回答问题。
在最初的问题训练后的连续5个实验环节中，参与者平均每分钟被问到4.0至4.6个问题；他们64%至87%的答案通过RV程序得到确认，其中93%至99%的答案是正确的。单主题和多主题问题集的表现没有显著差异。
结果表明：（1）基于脑电图的光标控制可用于高精度地回答简单问题；（2）对听觉问题的关注和答案的形成不会干扰基于脑电图的光标控制；（3）问题的复杂性（至少以单主题与多主题问题集来表示）不会明显影响表现；（4）RV程序如预期那样提高了准确性。几种提高交流速度的选择似乎很有前景。基于脑电图的脑机接口可为严重运动障碍患者提供一种新的交流和控制方式。
