Decomposing retrieval and integration in memory for actions: a multinomial modeling approach.

Typically, action phrases are recalled better if participants are asked to enact the phrases than if they are just asked to remember them. When investigating which processes constitute this enactment effect a difficulty is that observable effects in standard memory tests are ambiguous because such tests require several processes. In the present article, we introduce a multinomial model that decomposes observable memory performance into a retrieval parameter and a parameter concerning the item-specific processing and integration of an action phrase. These parameters are estimated from free recall and cued recall performance. The model fitted the data of two experiments designed to test it. Experiment 1 demonstrated the basic usefulness of the model by showing expected differences in the integration parameter in the absence of unexpected differences in the retrieval parameter. Experiment 2 extended the conditions under which the model is useful by showing expected differences in the retrieval parameter even in the presence of unexpected differences in the integration parameter. Together, these findings support our theoretical framework according to which enactment generally boosts integration of action phrases, but increases retrieval only for phrases with context cues.