Combining recognition, conflict-monitoring and feedback-related ERPs to detect concealed autobiographical information
Running Head: Detecting concealed information with fCIT
Xiaohong Allison Lin123a, Hong Li2,4,a, Tingwen Sheng2, Genyue Fu123, Liyang Sai123*
1. Zhejiang Philosophy and Social Science Laboratory for Research in Early Development and Childcare, Hangzhou Normal University, Hangzhou, China
2. Department of Psychology, Hangzhou Normal University, Hangzhou, China
3. Zhejiang Key Laboratory for Research in Assessment of Cognitive Impairments, Hangzhou Normal University, Hangzhou, China
4. State Key Laboratory of Cognitive Neuroscience and Learning & IDG/McGovern Institute for Brain Research, Beijing Normal University, Beijing, China
*Corresponding authors:
E-mail address: liyangsai@foxmail.com (L.S.)
a. These authors contributed equally to this work.
Abstract
This study examined neural signatures associated conflict-monitoring, recognition and feedback processing in a novel feedback Concealed Information Test (fCIT), and also examined whether all the ERPs can be used to detect concealed autobiographical information. Participants were randomly assigned to one of two groups (guilty or innocent) and then tested in the fCIT while undergoing electroencephalograms (EEGs). Results show that the probe (participants’ name) elicited more negative N200, more positive recognition P300 than irrelevants among guilty participants, and feedback following the probe elicited a larger feedback P300 than feedback following irrelevants. Further, we found that all indicators, including conflict-monitoring N200, recognition P300, and feedback P300, could significantly discriminate between guilty and innocent participants. Combining them is highly effective in discriminating between guilty and innocent participants (AUC = 0.96). These findings not only shed light on the neural processing of fCIT, but also suggest the potential of using fCIT to detect concealed autobiographical information.
Keywords: N200; P300; FRN; Feedback P300; Concealed Information Test; Principal Component Analysis
INTRODUCTION
For centuries, researchers have devoted much effort towards detecting concealed information. The CIT has been established as reliable and valid for detecting the status of information recalled by examinees. An Event-Related Potential (ERP) P300, which reflects the process of recognition and memory, has been widely used with the CIT (Allen, Iacono, & Danielson, 1992; Ellwanger, Rosenfeld, Sweet, & Bhatt, 1996; Meijer, Smulders, Merckelbach, & Wolf, 2007; for a review, see Rosenfeld, 2019). Recently, we presented evidence that ERPs associated with feedback processing during a novel fCIT can accurately identify concealed, crime-related information (Sai, Lin, Hu, & Fu, 2014; Sai et al., 2016; Sai et al., 2020; Zheng et al., 2022). This study concurrently examines ERPs associated with conflict-monitoring, recognition and feedback processing in an autobiographical information-based fCIT, and also whether those ERPs can be used to detect concealed autobiographical information.
ERPs (more specifically the P300) have been used in CIT for over thirty years (Ben-Shakhar, 2012; Kubo & Nittono, 2009; Lu et al., 2018; Rosenfeld, 2011; Rosenfeld, Biroschak, & Furedy, 2006; Rosenfeld, Sitar, Wasserman, & Ward, 2018). The P300 is an endogenous ERP component that occurs between 300 and 800 ms after stimulus onset, and is often elicited by rare, recognized, and meaningful items (Johnson, 1986). Since crime-related information (the probe) is highly meaningful to guilty suspects, the probe will elicit a larger P300 than irrelevant stimuli within that group; for innocent participants (who know nothing about the crime), the probe stimulus is indistinguishable from the irrelevants, in which case there is no significant difference in P300 amplitude between the two. A number of studies have demonstrated that P300 can effectively discriminate between guilty and innocent participants (Ben-Shakhar & Nahari, 2018; Farwell & Donchin, 1991; Lukács et al., 2016; Meixner & Rosenfeld, 2011, 2014; Rosenfeld, 2005; Rosenfeld et al., 1988; Rosenfeld, Labkovsky, Davydova, Ward, & Rosenfeld, 2017; Winograd & Rosenfeld, 2011).
However, limitations to the P300-based CIT exist. For example, CIT is susceptible to countermeasures and information leakage, leading to a significant reduction in lie detection efficiency (Rosenfeld et al., 2004; Zheng et al., 2022; Kim et al., 2022; Rosenfeld et al., 2008; also see a review by Rosenfeld, 2019). In consequence, many researchers have sought ways of enhancing the utility of the P300-CIT (Hu, Pornpattananangkul, & Rosenfeld, 2013; Hu & Rosenfeld, 2012; Rosenfeld et al., 2013; Rosenfeld, Hu, & Pederson, 2012).
A number of recent studies have introduced a new variant of CIT that adds feedback regarding participants’ concealment performance. The advantage of this fCIT is to combine both recognition P300 and feedback-related ERPs to detect concealed information. For example, Sai et al. (2016) asked participants to conduct a mock crime (theft of a ring), and then undergo the fCIT to detect concealed information. The authors focused on two well-known feedback-related ERPs: a negativity around 200-300ms (called the feedback-related negativity, FRN, or a reward-related positivity after principal component analysis method) (Luo, Sun, Mai, Gu, & Zhang, 2011; Sato et al., 2005; Yeung and Sanfey, 2004; see a review, Proudfit, 2015); and a positivity around 300-500ms (called the feedback P300) after feedback stimuli. According to their findings, for guilty participants, feedback following a probe elicited greater feedback-related negativity (FRN) and feedback P300 than feedback following irrelevants. On the other hand, for innocents, no significant differences were found for FRN and feedback P300 between feedback following a probe and feedback following irrelevants. Individual analyses showed that both FRN and feedback P300 proved highly effective in discriminating guilty from innocent participants. Furthermore, combining feedback-related ERPs and recognition P300 yielded a near-perfect detection efficiency (AUC = 99%). Further, Sai et al. (2020) repeated the fCIT mock crime (Sai et al., 2016) and investigated whether the feedback stage would influence the detection efficiency in the recognition CIT stage. This study successfully replicated the high detection efficiency and found that the detection efficiency of recognition P300 was basically the same in the classic CIT or in the fCIT (CIT with feedback). The most recent study (Zheng et al., 2022) also adopted the fCIT in mock crime to explore whether the mental countermeasures could compromise the deception efficiency of feedback-related ERPs in fCIT, just like the recognition P300 in the classic CIT paradigm. They found that, unlike the decreased recognition P300 efficiency, the FRN and feedback P300 remain the same efficiency with or without the countermeasures in the fCIT. These findings suggest great promise for the use of feedback-related ERPs in detecting concealed information in mock crime.
Although the high detection efficiency was consistently found in fCIT, most of the above studies examined the utility of using fCIT to detect crime-related information. Specifically, participants in those studies were asked to perpetrate a mock crime (stealing a ring) and then attempt to conceal their knowledge of crime-related information. What remains unclear is whether the fCIT might be used to detect concealed autobiographical information (e.g., a name) that has obvious significance in terms of societal goals. For example, terrorists conceal their identities to commit attacks, while spies conceal theirs to more easily engage in illegal activities (Meixner & Rosenfeld, 2011; Verschuere, Crombez, Koster, & De Clercq, 2007). Thus, another purpose of the present study is to examine whether the fCIT can be utilized to detect this sort of concealed autobiographical information.
In addition, N200 has also been used to detect concealed information in several previous studies. N200 is a negative ERP component that appears around 200-300ms and is found to be associated with cognitive control such as conflict monitoring (Donkers & Van Boxtel, 2004; Enriquez-Geppert, Konrad, Pantev, & Huster, 2010). Since concealing a probe requires more cognitive control than telling the truth to the irrelevant, researchers have hypothesized that probe elicit a more negative N200 than irrelevants. However, the findings are mixed. Some studies have found that probe elicit a more negative N200 than irrelevants, while others have not (for a meta-analysis, see Sai, Cheng, Shang, Fu, & Verschuere, 2023). It should be noted that most previous studies used the baseline to peak method to analyze the N200. It is uncertain if the N200 effect observed in prior research is accurate, as the N200 and P300 are close together and often overlap. To address this issue, the present study used PCA (Principal Component Analysis) to untangle the N200 from the P300 to examine the N200 effect in the CIT.
Taken together, the present study examined whether fCIT can be used to detect autobiographical information. Besides feedback-related ERPs, we also examined the N200 and recognition P300 in fCIT to see whether combining recognition P300, conflict monitoring N200 and feedback-related ERPs could further improve the detection of autobiographical information. We anticipated that (1) probe stimuli would elicit a bigger N200 than irrelevant stimuli in guilty participants, but not in innocent participants; (2) probe stimuli would elicit a greater recognition P300 than irrelevant stimuli in guilty participants, but not in innocent participants; (3) feedbacks following a probe would elicit a larger FRN and feedback P300 than feedbacks following irrelevants in guilty participants, but not in innocent ones; and (4) combining N200, recognition P300, FRN, and feedback P300 might potentially yield very high deception-detection classification efficiencies.