| In legal contexts,children are routinely questioned about transgressions that allegedly perpetrated by adults.A crucial matter concerns the extent which interviewers can safely press children for information without risking false accusation.The dilemma is that children who have witnessed nothing are susceptible to suggestions of wrongdoing,whereas children who have witnessed wrongdoing may have been forced into silence.In order to solve this situation,a large amount of research have focused on the interview processes,attempt to find a way to elicit true report without increasing false allegations.However,these methods can only raise children’s willingness to speak the truth,can not differentiate between truth and lie.The current study used computer version technology combined with proven effective interview method,examine nonverbal facial expressions of children(4-9 years old)telling truth and lie,as well as which facial movement can be used as marked indicators of children’s lying.In study 1,we exam whether there is a significant difference in the accuracy that automatic facial expression analysis techniques to distinguish between children’s lying and telling the truth under different instructions.By comparing the facial expressions under free recall and putative confession conditions,using support vector machine and cross validation algorithm to classify facial expression data and calculate the average accuracy.In study 2,we exam whether maltreated and nonmaltreated children’s facial expressions in rapport building can predict disclosure in free recall and close-ended question.Present study concluded that:1.There was no significant difference between the accuracy of using facial expression to distinguish children’s lying and telling the truth.In putative confession condition,first non-broken toy elaboration part had the highest correct rate was 90.72%.2.In putative confession condition,children who telling the truth were more likely to show surprise expression,children who lied were more likely to experience conflict and struggle.In free recall condition,children who telling the truth were more likely to show facial expressions associate with happy’ and positive emotions,and children who lied were more likely to show facial movements associate with disgust and anger.3.Children’s facial expression in rapport building can effectively predict whether children disclose in free recall and close-ended question,and the accuracy rate of disclosure prediction based on AU in nonmaltreated children group was significant higher than maltreated children group.In close-ended question,nonmaltreated children group has the highest disclosure prediction correct rate was 91.36%.4.Maltreated children and nonmaltreated children who telling the truth in free recall and close-ended question were more likely to display fear expression during rapport building,children who lied in free recall and close-ended question were more likely to display disgust and contempt expression during rapport building.Furthermore,AU5 is a notable indicator to differentiate between lying and telling the truth for nonmaltreated children group.5.The accuracy based on AU was significant higher than based on emotion when differentiating and predicting lying and telling the truth. |
