What to do and how to do it: sequence learning of action effects and transformation rules.

Many goal-oriented actions, for example in tool use, can be characterized by the involvement of two components: The effect that the acting person wants to achieve and the transformation rule that defines the relationship between a bodily movement and the associated action effect. Both in conjunction specify the concrete action that has to be executed. In our experiments, we utilized a sequence learning paradigm to investigate whether these components are represented separately or are bound together to form a more holistic representation of the action. Participants switched between different transformation rules to achieve certain action effects. In one group, there was a fixed sequence of action effects, with randomly varying transformation rules. In the other group, the sequence of transformation rules was fixed, whereas the action effects were randomly varying. Results indicate that both kinds of sequences can be learned and provide evidence for the notion that transformation rules and action effects are represented separately.