P300和CNV在GKT的延时反应范式中测谎效果的分离

在杀人游戏情境中，设置无反馈和有反馈两个实验组，采用GKT（guilt knowledge test）的延时反应范式，考察P300和CNV两种ERP成分的测谎效果。结果发现：刺激呈现后大约450ms诱发了一个正性ERP成分（P300），大约800ms诱发了一个负性ERP成分（CNV）。在有、无反馈两种实验范式下，P300和CNV两种测谎指标的变化情况发生了分离——有无反馈不影响P300峰值大小；而在有反馈条件下，“杀手”对于犯罪细节词语（探测刺激）的CNV明显增大。使用Bootstrap波幅差异分析法，分别计算P300和CNV在有、无反馈情况下的检出率，结果发现：两个指标的检出率在有、无反馈两种实验范式下的变化情况也发生了分离。这些结果表明：P300和CNV两个指标具有不同的测谎功能，这可能和它们所反映的不同欺骗心理过程有关。

The Functionally Separation of P300 and CNV in Lie Detection

Some researches which have proved that the P300 and the CNV are effective indicators in lie detection, and they are affected by different factors. In the lie detection, the P300 is found to be closely related to stimulus attribute, while the CNV is only related to the response of a subject (deception or honesty). However, there is no research which induces the two ERPs synchronously in one experiment, and no any studies reveals the functional distinction of the two indicators in lie detection and further compares their consequences in lie-detection. Based on previous research, our study assumes that the P300 and the CNV are functionally different in lie-detection and they might reflect different cognitive phases in deception. In order to compare and explore the characteristics of the P300 and the CNV in lie-detection, this study induced “spontaneous deception” under the condition of a “murder game”, and combined GKT (guilt knowledge test) which was often used in the P300 lie-detection research with delayed-response paradigm which induced the CNV to design a new paradigm, so as to induce the two ERPs synchronously in lie-detection and separate them by controlling feedback. There were 33 subjects who participated in this research. They were divided into two groups: 17 of them got feedbacks concerning the outcome of lie-detection while other 16 didn’t.This experiment included two steps. Subjects played a “murder game” before lie detection, and their event-related brain potentials (ERPs) were measured during the detection. We mainly analyzed the ERPs evoked by three kinds of stimuli. Our result revealed a distinctive tendency between the P300 and the CVN in both groups. Whether there was feedback did not affect the amplitude of P300. In both of the two groups, the amplitude of P300 elicited by probe stimulus was significantly larger than that elicited by the irrelevant stimulus no matter whether feedback was acquirable. The P300 could detect lie effectively. In contrast, absence of feedback could affect the amplitude of the CNV. The amplitude of the CNV elicited by the probe stimulus was significantly larger when people got feedbacks than when they didn’t. The CNV could only detect lie effectively in the condition of having feedback. The data concerning lie-detection rates of the two indicators reflected that: the lie-detection rate of CNV elicited in feedback group was significantly larger than that in no-feedback group, which were 88.24% and 43.75% respectively; while the rates of the P300 elicited in the two groups were 70.6% and 75% , respectively. The results revealed that the P300 and the CNV were separate on the function of cognitive process. The former was related to the cognitive process of stimulus in deception, while the latter was related to the processes such as response to deception and generating deceptive intention etc. In addition, the two indicators were separate concerning detection rate in different conditions. P300 could detect lie effectively in both conditions while the CNV could only be used as an available indicator of lie detection in the condition of getting having feedback.