Behavioral dishonesty in multiscenes: Associations with trait honesty and neural patterns during (dis)honesty video-watching

Abstract

Cross-situational inconsistency is common in the expression of honesty traits; yet, there is insufficient emphasis on behavioral dishonesty across multiple contexts. The current study aimed to investigate behavioral dishonesty in various contexts and reveal the associations between trait honesty, behavioral dishonesty, and neural patterns of observing others behave honestly or dishonestly in videos (abbr.: (dis)honesty video-watching). First, the results revealed limitations in using trait honesty to reflect variations in dishonest behaviors and predict behavioral dishonesty. The finding highlights the importance of considering neural patterns in understanding and predicting dishonest behaviors. Second, by comparing the predictive performance of seven types of data across three neural networks, the results showed that functional connectivity in the hypothesis-driven network during (dis)honesty video-watching provided the highest predictive power in predicting multitask behavioral dishonesty. Last, by applying the feature elimination method, the midline self-referential regions (medial prefrontal cortex, posterior cingulate cortex, and anterior cingulate cortex), anterior insula, and striatum were identified as the most informative brain regions in predicting behavioral dishonesty. In summary, the study offered insights into individual differences in deception and the intricate connections among trait honesty, behavioral dishonesty, and neural patterns during (dis)honesty video-watching.

Keywords: anterior insula; behavioral dishonesty; medial prefrontal cortex; striatum; trait honesty.

FIGURE 1

Experimental procedure. Participants first completed an online trait and demographic questionnaire 1–7 days before the fMRI measurement. In the fMRI experiment, participants went through 8 min of resting‐state functional scanning before the movie‐watching session. In the movie‐watching session, participants watched six videos with self‐regarding lying plots, six videos with other‐regarding lying plots, four videos with truth‐telling plots, and four documentaries without any lying plots. Three to six days after the fMRI experiment, participants completed Task 1 (single‐player task including matrix puzzles, difference spotting puzzles, Chinese idioms puzzles, and logical reasoning puzzles) and Task 2 (two‐player task including revised sender‐receiver games and dictator games) subsequently in the lab. Seven days after that, participants reported their lying frequencies in the past 24 h on six random days for two consecutive weeks (Task 3).

FIGURE 2

Participants' dishonest extent in eight subtasks, multitask behavioral dishonesty, and honesty trait (N = 97). (a) Individuals' adjusted deceptive extent in each subtask (y‐axis on the left) was shown, ranked by multitask behavioral dishonesty (y‐axis on the right; upper panel) and trait honesty (y‐axis on the right; lower panel). (b) Intrasubject correlations of adjusted deceptive extent in eight subtasks, ranked by multitask behavioral dishonesty (upper panel) and trait honesty (lower panel). (c) Honest–humility scores negatively correlated to multitask behavioral dishonesty.

FIGURE 3

Performance prediction (R ²) of demography, personality, and FCs extracted from seven data types in three brain networks (Honesty trait‐driven networks, non‐hypothesis‐driven network, and hypothesis‐driven network) in predicting multitask behavioral dishonesty. SES: socioeconomic status; HEXACO: Honesty‐Humility, Emotionality, Extraversion, Agreeableness, Conscientiousness, and Openness to Experience; RS: resting‐state fMRI data; Neutral: fMRI data during watching video clips without lying plots; Self‐regarding: fMRI data during watching videos with self‐regarding lying plots; Other‐regarding: fMRI data during watching videos with other‐regarding lying plots; Honest: fMRI data during watching videos with truth‐telling plots; SOH: combined data from watching self‐regarding, other‐regarding, and honest videos; SOHN: combined data from watching self‐regarding, other‐regarding, honest, and neutral videos; Frontal_med: superior medial gyrus; Frontal_sup: superior frontal gyrus; Frontal_mid: middle frontal gyrus; Frontal_inf: inferior frontal gyrus; MCC: middle cingulate cortex; MPFC: medial prefrontal cortex; ACC: anterior cingulate cortex; DLPFC: dorsolateral prefrontal cortex; TPJ: temporoparietal junction; PCC: posterior cingulate cortex.

FIGURE 4

Performance (R ²) in each iteration by using FCs of the hypothesis‐driven network from seven data types to predict honesty trait (i.e., honesty‐humility) (a) and multitask behavioral dishonesty (b). Features in the highest predictive models were visualized by chord diagrams on the right sides. (c) Results of feature eliminations in using FCs of the hypothesis‐driven network to predict multitask behavioral dishonesty. The removal of ROI features (in purple) that led to the largest reduction (differences between the solid line and dashed line) in the highest R ²s (dash line) were displayed on the left. R ² change (differences between the solid line and dashed line) of using SOH data after eliminating features from different brain regions were shown on the right. RS: resting‐state fMRI data; Neutral: fMRI data during watching non‐honesty related video clips; Self‐regarding: fMRI data during watching videos with self‐regarding lying plots; Other‐regarding: fMRI data during watching videos with other‐regarding lying plots; Honest: fMRI data during watching videos with truth‐telling plots; SOH: combined data from watching self‐regarding, other‐regarding, and honest videos; SOHN: combined data from watching self‐regarding, other‐regarding, honest, and neutral videos; MPFC: medial prefrontal cortex; ACC: anterior cingulate cortex; DLPFC: dorsolateral prefrontal cortex; TPJ: temporoparietal junction; PCC: posterior cingulate cortex.