Cognitive modeling of lure discriminability in the Mnemonic Similarity Task.

Abstract

Lure discrimination in the Mnemonic Similarity Task (MST) has been widely used to measure pattern separation. However, the classic index of lure discrimination in the MST has arbitrary assumptions with limited supporting evidence. The present study has thus developed several models with different assumptions on the process underlying MST as well as the different model-derived indices of lure discrimination. Furthermore, we have assessed and compared these models in a measurement-based approach. We found that the model for the classic lure discrimination index fails to accurately predict the responses in an MST from > 150 participants. Instead, a new index based on the unidimensional signal detection model provides the best fits of the empirical dataset. This work highlights the value of model-based approaches in measuring lure discriminability in the MST.