Neuroimaging studies of addiction seek to understand the brain mechanisms that predispose to and support the maintenance of addictive behaviors. Traditional studies are case-control cross-sectional studies, i.e. they conceptualized individuals suffering from addiction as a homogenous group, and report lab-based experiments conducted at one particular point in time. In this talk, I will argue that a refined understanding of addictive behaviors requires the use of dimensional longitudinal studies. Using dimensions will reveal the existence of heterogenous profiles within diagnostic groups, and allow researchers to incorporate individual variability in their models. In turn, using longitudinal follow-up measures should allow researchers to determine whether brain-related abnormalities are predictive of symptoms in real-life. I will illustrate these points using a few example studies from the literature.

None Declared