Traditional associative learning research focused primarily on low-level associative processes that are assumed to operate in both human and non-human animals. In fact, many prominent learning theories (e.g., Rescorla & Wagner) were developed primarily on the basis of research in animals but are assumed to hold also for humans. As such, associative learning as a phenomenon (i.e., change in behavior as the result of relations between events in the world) was confounded with association formation as an underlying mechanism. In line with the overall aims of our project, in WP3 we explore the merits of dissociating these two aspects of learning research. Inspiration for this line of research comes from so-called propositional models of learning that focus on the role of high-level propositional processes in learning (e.g., Mitchell et al.). The impact of these processes can be examined in its most pure form in studies on learning via instructions (see Lovibond). For instance, after instructing participants about the fact that a light will be followed by an aversive shock, the light will evoke a conditioned response even when the light has never actually been followed by a shock. Although instructed conditioning effects are known to occur, they have not been investigated systematically, probably because they do not fit well with the low-level process models that dominated learning research for the past 100 years. Within WP3, we engage in such a systematic study of associative learning via instruction, both at the behavioral and neural level. Moreover, for the first time ever, this research is extended to non-associative forms of learning.