2

I am looking for the advice regarding the modelling of the results of a working (short-term) memory test. In a typical case, performance in such tests can be characterized using the hit rate and false alarm rate, which can be nicely combined into d-prime, and the reaction time. My question is whether there are smart concepts how all these parameters may be fused together to generate some sort of the "working memory performance index"? (Toy example: d-prime that is weighted by the log-transformed reaction time.) Any references will be very much appreciated.

Upd: I am interesting in the specific application and possible specific pitfalls for the working memory research.

1 Answers1

1

This is a tough question, one that I have had to face in my own research. Your question has already been answered/addressed in How to analyze reaction times and accuracy together?

In short, there is no universal method. If you don't find the references or methods listed in that question useful, then I suggest you will require a computational model such as the Linear Ballistic Accumulator. The LBA is a (structurally and computationally) framework that informs process underling both RT and error data (drift rate vs response criterion).