| Last Updated:
Creating DOI. Please wait...
We present a comprehensive empirical evaluation of the ACT-R based model of sentence processing developed by Lewis and Vasishth (2005) (LV05). The predictions of the model are compared with the results of a recent meta-analysis of published reading studies on retrieval interference in reflexive-/reciprocal-antecedent and subject-verb dependencies (Jäger, Engelmann, & Vasishth, 2017). We show that the model has only partial success in explaining the data, and that two independently motivated theoretical constructs are necessary: memory accessibility (prominence), and a theory of multi-associative cues. We implement these two constructs within the LV05 model and quantitatively compare the predictions of the original and the extended model with the results of the meta-analysis. Our simulations show that the extended model furnishes a superior fit. These results show that cue-based retrieval models need to take into account differences in the accessibility of items in memory, and the effect of context-based feature-selectivity. The simulations thus shed new light on the cognitive mechanisms underlying interference effects, and should be considered in the interpretation of existing empirical results and in the design of future experiments.
CC-By Attribution 4.0 International