False Beliefs and Intentions: Can the Brain Tell the Difference?

Background: The results from neuroimaging studies of Theory of Mind (ToM) are heterogeneous, implicating numerous brain regions, including the medial prefrontal cortex, temporoparietal junction, and superior temporal regions. A recent review (Carrington & Bailey, 2009) concluded that although a clear confound, paradigmatic variation could not adequately account for the heterogeneity of these findings. Nevertheless, there was some evidence suggesting that false belief and deception might be associated with dissociable neural substrates. Reduced or spatially altered activity has been reported in several brain regions typically associated with ToM in individuals with ASD (e.g. Gilbert et al., 2009; Kana et al., 2009). To the best of our knowledge, however, within-subject comparison of activity evoked by different mental states has not been investigated in ASD. 

Objectives: To determine whether the attribution of false beliefs (FB) and intentions (INT) was associated with dissociable patterns of brain activity in both typically developing (TD) individuals and individuals with an ASD.

Methods: Behavioural and fMRI data were acquired for 22 TD individuals and 20 individuals with ASD. ToM was explored using a comic strip paradigm based on the task developed by Sarfati et al. (1997). The original paradigm included three conditions, one involving mental states (INT) and two physical causality control conditions, one including characters (CP) and one with objects only (CO). Novel stimuli were developed for each of these conditions and an additional ToM condition: false belief (FB).

Results: Both groups were slowest and least accurate in the FB condition. The ASD group were consistently slower and less accurate than the TD group. In both groups, both INT and FB evoked significantly more activity than either CP or CO in a region of the posterior cingulate cortex and the ventral precuneus (PCC/precuneus). Moreover, CP did not evoke significantly more activity than CO in this region, suggesting that ToM-related activity was not attributable simply to the presence of people. FB evoked significantly more activity than any of the other conditions in the dorsomedial prefrontal cortex (dmPFC). Although similar patterns of activity were observed in both groups, there was a small cluster in the PCC/precuneus where the increase in activity evoked by INT compared with CP was significantly greater in the ASD group compared with TD (p<0.005, uncorrected for multiple comparisons). Furthermore, post hoc ROI analyses and comparison of each condition with baseline indicated less selective engagement of the dmPFC for FB in the ASD group.

Conclusions: Consistent with previous neuroimaging studies, ToM was associated with activity in the dmPFC and PCC/precuneus. The data extended previous knowledge by demonstrating that the PCC/precuneus appeared to be engaged during ToM reasoning regardless of the mental state, while the dmPFC was more selectively engaged for the attribution of FB. Group differences were subtle, but suggested less specialisation of the dmPFC for FB in ASD. Furthermore, activity in the PCC/precuneus differentiated less between INT and CP in the TD group compared with ASD, perhaps suggesting more spontaneous mental state attribution evoked by the presence of people.