P29. Differential linear and nonlinear brain activation for probabilistic aversive outcomes in pathological gamblers and controls

Objective Functional imaging studies on decision making under uncertain risk of negative (aversive) outcomes are scarce. To get a better understanding of the underlying processes, we tested models of decision making including participants that were gambling regularly despite negative long-term consequences. Primarily, we sought to detect group differences in cortical and subcortical regions involved in decision processes. Also, we wanted to investigate, if current economic decision making models account for observable physiological data. Do we process aversive contingencies of decisions in a linear or a nonlinear manner, or both? Methods We studied 22 pathological and 18 habitual gamblers, as well as 23 control subjects, evaluated by psychologists with a semi-structured interview and self-reported Southern Oaks Gambling Scale. Participants were scanned with blood oxygenation level-dependent (BOLD) functional magnetic resonance imaging (fMRI) and engaged in a card game in which un-ambiguous situations of various risk levels for punishment (unpleasant, but not painful electric shocks) were created. Apart from neural responses to punishment, our paradigm allowed us to measure how neural activity during expectation of a punishment is modulated by the probability to be punished (risk) or the certainty of the outcome (certainty). Results Extending previous work and testing our hypothesis, based on economic studies, we modelled expected outcomes according to the prospect theory. Activations differed spatially when situations were modelled according to risk or certainty. With linearly increasing chances of avoiding punishment we found that the BOLD signal increased in the orbitofrontal cortex, caudate nucleus, precuneus, and the posterior cingulate cortex ( Fig. 1 a). Conversely, with increasing certainty of an outcome (U-shaped from most certain punishment to most certain avoidance), we could demonstrate nonlinear activations in medial superior frontal cortex, anterior cingulate cortex, anterior insula and angular gyrus ( Fig. 1 b). Interestingly, we found no group differences in cerebral activation during the task, which dovetailed with ‘offline’ behavioural analyses of decision making in the same groups. Conclusion Our results support the notion that economic decision making contains both linear and nonlinear cognitive processes largely following contingencies for risk and certainty and provide evidence that these cognitive processes are represented in different sets of brain regions. Correlates of the risk for punishment were found in areas known to be involved in dopaminergic reinforcement learning, while regions known to sub serve decision making, conflict monitoring and punishment show correlation with certainty of the outcome. While differences in reward perception in pathological gamblers are frequently described in the literature, we found no evidence for changes in punishment perception.