Departments of Psychology and Neuroscience, Northwestern University, Evanston, IL, USA.
Appl Psychophysiol Biofeedback. 2014 Mar;39(1):67-73. doi: 10.1007/s10484-014-9244-y.
In the first of two experiments, we compared the accuracy of the P300 concealed information test protocol as a function of numbers of trials experienced by subjects and ERP averages analyzed by investigators. Contrary to Farwell et al. (Cogn Neurodyn 6(2):115-154, 2012), we found no evidence that 100 trial based averages are more accurate than 66 or 33 trial based averages (all numbers led to accuracies of 84-94 %). There was actually a trend favoring the lowest trial numbers. The second study compared numbers of irrelevant stimuli recalled and recognized in the 3-stimulus protocol versus the complex trial protocol (Rosenfeld in Memory detection: theory and application of the concealed information test, Cambridge University Press, New York, pp 63-89, 2011). Again, in contrast to expectations from Farwell et al. (Cogn Neurodyn 6(2):115-154, 2012), there were no differences between protocols, although there were more irrelevant stimuli recognized than recalled, and irrelevant 4-digit number group stimuli were neither recalled nor recognized as well as irrelevant city name stimuli. We therefore conclude that stimulus processing in the P300-based complex trial protocol-with no more than 33 sweep averages-is adequate to allow accurate detection of concealed information.
在两项实验中的第一项实验中，我们比较了 P300 隐藏信息测试协议的准确性，该协议的准确性取决于被试经历的试验次数和调查人员分析的 ERP 平均值。与 Farwell 等人（Cogn Neurodyn 6(2):115-154, 2012）的研究结果相反，我们没有发现 100 次试验平均比 66 次或 33 次试验平均更准确的证据（所有数字的准确率都在 84-94%之间）。实际上，有一个趋势是倾向于使用最少的试验次数。第二项研究比较了 3 刺激协议与复杂试验协议中回忆和识别的无关刺激数量（Rosenfeld in Memory detection: theory and application of the concealed information test, Cambridge University Press, New York, pp 63-89, 2011）。再次与 Farwell 等人（Cogn Neurodyn 6(2):115-154, 2012）的预期相反，两种协议之间没有差异，尽管识别的无关刺激数量多于回忆的数量，而且无关的 4 位数字组刺激既未被回忆也未被识别，与无关的城市名称刺激一样。因此，我们得出结论，基于 P300 的复杂试验协议中的刺激处理——最多 33 次扫频平均值——足以允许准确检测隐藏信息。
