Addicted individuals pursue substances of abuse even in the clear presence of positive outcomes that may be foregone and negative outcomes that may occur. Computational models of addiction depict the addicted state as a feature of a valuation disease, where drug-induced reward prediction error signals steer decisions toward continued drug use. Related models admit the possibility that valuation and choice are also directed by 'fictive' outcomes (outcomes that have not been experienced) that possess their own detectable error signals. We hypothesize that, in addiction, anomalies in these fictive error signals contribute to the diminished influence of potential consequences. Using a simple investment game and functional magnetic resonance imaging in chronic cigarette smokers, we measured neural and behavioral responses to error signals derived from actual experience and from fictive outcomes. In nonsmokers, both fictive and experiential error signals predicted subjects' choices and possessed distinct neural correlates. In chronic smokers, choices were not guided by error signals derived from what might have happened, despite ongoing and robust neural correlates of these fictive errors. These data provide human neuroimaging support for computational models of addiction and suggest the addition of fictive learning signals to reinforcement learning accounts of drug dependence.
We have not found any resources mentioned in this publication.
SciCrunch is a data sharing and display platform. Anyone can create a custom portal where they can select searchable subsets of hundreds of data sources, brand their web pages and create their community. SciCrunch will push data updates automatically to all portals on a weekly basis. User communities can also add their own data to SciCrunch, however this is not currently a free service.