発表

3B-051

Neural Correlates of Action and Speech Memory Based Deception and Its Detection: A Near-infrared Spectroscopy (NIRS) Study of Inter-brain Coupling Using a Mutual Feedback Task

[責任発表者] 齋藤 洋典:1
[連名発表者] 齋藤 五大:2,4, [連名発表者] 行場 次朗:2, [連名発表者] 厳島 行雄:3
1:名古屋大学, 2:東北大学, 3:日本大学, 4:日本学術振興会

Introduction
As our cognitive functions emerge from interpersonal interactions, understanding these interactions enriches our comprehension of cognitive processes within a single individual's brain. We examined inter-brain coupling of paired Liar and Detector in a face-to-face communication condition in various feedback tasks using near-infrared spectroscopy (NIRS) (Saito G. et al., 2016; Saito H. et al., 2017).

Methods
Participants: Twenty-two male students (all right-handed, age: 18.7 ± 0.4 years) took part in the present study and were randomly assigned to pairs of Liar and Detector.
Materials: We used 40 simple (Object-Particle-Verb) kana-only sentences (e.g.: ペン で かく) in the Learning, Recognition, and Lying Sessions. The particles DE (で) and WO (を) were used in a 1:1 ratio. An additional five sentences were used for practice. Procedure: The Learning Sessions for Liars contained two tasks: Action and Speech consisted of twenty trials, respectively. In the Learning Session, Liars read sentences presented on a monitor and performed Action (enactment of) or Speech (reading aloud of) a presented sentence according to a given command word (動作: Action or 音読:Speech) simultaneously presented on the monitor. In the Recognition Session, Liars indicated their performance for each sentence presented in the Learning Session by a key press. In the Lying Session, Liars read the same sentences and then performed one of the following tasks by key press and immediate verbal report: Truth and Lying (honestly and dishonestly report performance conducted in the Learning Session according to a given command word (真実:Truth or 虚偽:Lying) ). The Detector, who sat diagonally across from the Liar, was also simultaneously presented the 40 sentences, but the command words for Liars to lie or tell the truth were not presented to them. Detectors were presented the sentences so that they were aware of what the Liar was considering to lie or tell the truth about. The Detector was instructed to silently read the sentences and watch the Liar, and to judge whether the Liar’s verbal report was a Lie or the Truth. After the L-T judgment by key press, the Detector received the T-L command words for the paired Liar on the Detector’s monitor, i.e., feedback for Detector’s judgment, and performed verbal report:.正答 Hit or 誤答Miss, i.e., feedback for Liar. We measured reaction times by Liars and Detectors in the Lying Session as well as the brain activities (the concentration of oxygenated hemoglobin (Coxy-Hb)) of paired participants during the Lying Session using NIRS.
Tasks: Detectors completed a sentence evaluation task in which they were asked to rate each presented sentence in terms of the five question items (comprehension, imagineability, Action frequency done, frequency seen, criminability) during the Liar’s Learning and Recognition sessions in a different room.
Behavior Data Analysis: Liars made errors in the Recognition and/or the Lying Session: they chose an incorrect option (e.g., choosing Action instead of Speech) and/or made extremely slow responses (3 SD above the mean reaction time (RT) for each participant). Trials with errors were excluded from further behavior and neural analyses. In addition, the Lying Session data was excluded if the Recognition Session data was an error. Furthermore, as this study used paired paradigms (i.e., Liar and Detector), errors were first analyzed in the behavioral data for the Liar, and the excluded error trials for the Liar were also excluded for the Detector.
NIRS Data Analysis: A 96-channel NIRS unit (LABNIRS; Shimadzu, Japan) was used to measure the temporal changes in the Coxy-Hb. The 98 channels were divided into two sets of 48 channels, and each set was assigned for a pair of Liar and Detector to cover their left and right hemisphere by 24 channels each.

Results & Discussion
The detection rate of a Detector was approximately 50%. NIRS data of paired Liar and Detector during the post Feedback period (3 seconds: after the beginning of feedback to the Detector) were analyzed to test whether the paired brain activations showed positive correlations according to the following three time lag periods: When we focus on the Liar’s activation in each channel for the Feedback period, the time lag zero (0) in Table 1 is synchronous activation period between Liar and Detector, minus two (-2) is the Detector’s activation preceding two seconds to the Liar’s one, and time lag plus two (+2) is the anteceding two seconds. The NIRS profiles in the three time lags in Feedback only for Detector’s judgment task (Saito H. et al., 2017) showed significant positive correlations (e.g., in their right IFG) for correct Detectors’ responses, but not for the error responses, showing inter-brain coupling. In contrast, the NIRS profiles in the present mutual feedback task, including feedback for Detector and Liar, showed significant positive correlations (e.g., right STS) regardless of Detectors’ correct or error responses. Taken together, the remaining question is why Detectors show a lie detection rate at a chance level regardless of implicit (Saito G. et al., 2016) or explicit (Saito H. et al., 2017) feedback types, even though their brain could show inter-brain coupling with the Liar's brain when Detectors’ response is correct.

詳細検索