There were developments in two strands of our work with patients with GSP and GAD over the past 12 months: The first concerns the specific nature of the functional impairment seen in GSP. In previous work, we have shown that GSP does not simply represent a heightened amygdala response to social threats (angry expressions). Instead, there appears additionally to be atypical self-referential processing of social information. Following this work, we considered that it might be possible to reduce patients'heightened amygdala response to social threats by altering their self-referential processing. Specifically, we investigated whether the presentation of simple statements that either praised the self, criticized the self or were neutral had an impact on the amygdala response to social threats. Notably, patients with GSP showed exaggerated amygdala responses to social threats (angry expressions) relative to comparison individuals that were particularly increased if they followed a self-critical statement. In contrast, group differences in amygdala response to angry expressions were significantly reduced following self-praising statements. These data indicate the important modulatory role of self-referential processing over the anxiety responses of patients with GSP and stress the importance of developing treatments focused on positive self referential processing. The second concerns the specific nature of the functional impairment seen in GAD. In particular, we have been examining whether some of the problems in emotional responding in GAD that we observed in our preliminary work with patients with this disorder might manifest in difficulties on decision-making tasks. In previous work, we had demonstrated that patients with GAD, relative to patients with GSP and healthy adults, show impairment on very simple reinforcement-based decision-making tasks. Such a task is the passive avoidance task. On this task, the participant must learn to respond to some stimuli as doing so engenders reward (monetary gain). However, they must also passively avoid others (do nothing) as responding to these bad stimuli will engender punishment (monetary loss). Participants with GAD respond less to the good stimuli and more to the bad stimuli than patients with GSP and healthy adults. Our goal this year was to determine the computational basis of this impairment at the neural level. In other work, we have investigated the neural representation of prediction errors (the difference between the reward expected by the person and that received) and expected value (the expected reward from committing the action). The accurate representation of prediction errors is critical for learning;the greater the error, the greater the learning that should occur. The accurate representation of expected value is critical for successful decision-making;it is important to select the response associated with the greatest reward. Using a modified version of the passive avoidance task specifically designed for fMRI, we determined that patients with GAD show significant impairment in the representation of prediction error within caudate relative to patients with GSP and healthy participants. It is possible that uncertainty in decision choice, consequent on this impairment, contributes to the worry shown by patients with this disorder.
