Radvansky, Spieler, and Zacks, (1993, see also Radvansky & Zacks, 1991) report an experiment in which the fan of either the object or the location was manipulated from 1 to 3 while the fan of the other was held constant at 1. In their Experiments 1 and 2, in which inanimate objects were placed in locations, they found large effects of object fan and weak effects of location fan. In their Experiments 3 and 4 in which people were placed in locations, these effects were reversed (i.e., large effects of location fan and weak effects of person fan). In their experiments, 5 and 6, in which small locations were used, the effect of location fan was even larger while the effect of person fan was even weaker. Radvansky et al. interpret their results in terms of whether the participants organize their memory into location models (where all the objects are in the same location), for Experiments 1 and 2, or whether they organize them according to people models (where the history of a person is represented going from location to location) for Experiments 3 and 4 and particularly for Experiments 5 and 6, where the locations typically hold a single person. If all the objects are in one location or all of the locations are associated with a single person, there is only one model to be activated and participants do not have to search through multiple models. They ascribe the fan effect to the need to search through multiple models. As they acknowledge, it is a bit mysterious why there would not be an effect of the number of things in a model (objects in a room or places a person goes to). It should take longer to sort through more things. In fact, there are weak effects of fan for the other dimension. However, they suggest that such size-of-model effects are much less than the effects of having to retrieve multiple room models in Experiments 1 and 2 or retrieve multiple person models in Experiments 3-6.
ACT-R Model (experiment 1 and 2)
ACT-R Model (experiment 3 and 4)