LEARNING TO DETECT LIES IN AUTISM

Background: A key feature of autism is the inability to mentalize, or attribute mental states to others. Mentalizing is necessary to understand the meanings and intentions of others. Frith states (1989), “the mentalizing mechanism enables the [typically developing] child to learn surprisingly fast about… deception.” Because of their impairments in mentalizing, individuals with autism may be more susceptible to being deceived.

Objectives: This fMRI study investigated the neural activity of high functioning individuals with autism and neurotypical individuals while they tried to detect lies in computer animated videos.

Methods: 15 high-functioning individuals with autism and 15 age and IQ matched controls viewed pairs of computer animated avatars uttering a sentence. The task was to choose the video out of each pair in which the avatar was lying. Lying was implemented through several auditory and visual clues previously found to be associated with lying (DePaulo et al., 2003). The experiment consisted of a pretest, a training session with feedback, and a posttest.

Results: Both before and after training, the autism group had greater activation than the control group during the lie discrimination task. With training, the autism group showed a significantly smaller increase in inter-region synchronization of activation than the control group. Throughout the experiment the behavioral performance was comparable between the two groups.

Conclusions: The greater activation in the autism group throughout the experiment suggests that they recruited more neural resources, while showing the same behavioral performance as the control group. The smaller increase in synchronization of activation with training for the autism group suggests that they less efficiently automatized the lie discrimination task than did the controls.