Investigação do rastreamento ocular durante a observação de sinais da mentira em adultos

Abstract

The lie as a subject is of great relevance, because it deals with an apparently recurrent behavior that is present in different intensities and degrees in mankind. Although, there is still a scientific gap, perhaps because of that, researchers have been looking to develop and to improve techniques to detect lie from verbal and non verbal signs. The main objective of this study was to utilize the eye tracking of videos with and without lie content, interpreted by male and female targets, in order to investigate the eye movement pattern in the detection of the lying signs, in young adults. A experimental delimitation was used with repeated measureswith two conditions: control condition (videos with targets telling the truth) and experimental condition (videos with targets telling lies), in a counterbalanced order. The videos that were used as stimuli had people’s faces (targets) that were telling the truth or lying. The tracking of the participants ocular movements was performed with an Eye Tracker Tobii TX300 model. In this study, 16 students (eight men and eight women) with mean age of 22 years (SD = 6.81). Firstly, the participants answered in the following order the sociodemographic questionnaire, the visual acuity test (optotypes E of Rasquin) and Beck’s anxiety and depression scales. The lie detection task initiated with the instructions and the calibration of the eye tracker. After this, each participant watched four videos, with 30 seconds intervals in between, where they judged if the video was true or false in a answer sheet. The data of the ocular tracking for each video were grouped in a spreadsheet accordingly to a condition and analyzed with the SPSS software version 21. Accordingly with the normality criteria (Shapiro-Wilk test) the data was analyzed with a ANOVA mixed factorial, in order to evaluate the main effects and the interaction of the variables among and between the participants. The results of the ocular tracking showed significant differences in the number of fixations in the videos of true and lie, because it was verified an interaction effect [F (1. 14) = 6.02, p = 0.028, η2 = 0.30)] and a main effect of the video type [F (1. 14) = 72.78, p = 0.001, η2 = 0.84)]. This main effect showed a bigger number of fixations when the video was a video with truths (M = 129.53; SD = 37.08) than when the video was a video with lies (M = 78.37; SD = 19.21) [t (15) = 8.78; p = 0.001]. The total time of the fixation duration of the participants was bigger in the videos of truth with female targets (M = 73,182.50, SD = 857.08) than when the target was male (M = 67,868.75, SD = 979.94) [t (15) = -8.60; p = 0.001] Considering the differences in the area of interest in the face (nose, mouth and eyes) for each type of video (truth and lie), the one with the bigger number of fixations from the participants was the nose area in the videos of truth and lie (only in the male targets). The categorical analysis of the signs of lie (DePaulo et al., 2003) revealed that 83% of the participants’ detected signs were related to non verbal facial expressions. In a decreasing order, the most detected signs of lie identified by the participants were: facial expressions contrary to the speach, such as, disgusted, joy, microexpressions: wiggle the corner of the mouth, tigthen the eyes (33.5%); false smiles and nervousness (29%); the direction of the look, like looking to the side and looking away (19%); seriousness (9.5%); cognitive effort such as prolongated responses, thinking too much to answer and pauses in the speech (4.5%); and lack of enthusiasm (4.5%). Therefore, the results suggests that the ocular tracking pattern in the lie detection differs significantly according to the areas of interest of the face and the type of video. Also, that the face can demonstrate signs of lie through the targets’ expressions, mainly through non verbal signs, which were more perceived by the observers, possibly being clues of the true signs of lies.