Dual process models have described addiction as a combination of relatively strong bottom-up cue-related neurocognitive processes and relatively weak top-down cognitive control processes (e.g. Bechara, 2005; Wiers & Stacy, 2006). I will describe some of the tests used to assess these bottom-up processes. In line with this perspective, we found across several studies a larger impact of memory associations and approach tendencies on behavior in adolescents with relatively weak cognitive control. Dual-process models have recently come under fire (e.g. Keren & Schul, 2009), but we think they can still be useful at a descriptive psychological level, while more work should be done to illuminate the underlying neurocognitive mechanisms (Gladwin et al., 2011). Moreover, dual process models inspired new interventions aimed at changing relatively automatic processes in addiction, varieties of Cognitive Bias Modification (CBM) paradigms (see Wiers et al., 2013). I will present work on attentional re-training in alcoholism (Schoenmakers et al., 2010) and on approach-bias re-training (Wiers et al., 2011; Eberl et al., 2013), which have yielded clinically relevant results. I will also argue that task requirements for assessment and modification are hard to reconcile: tasks that are optimal for assessment are not very suitable for modification and vice versa.