The statistical properties of the Survivor Interaction Contrast

The Survivor Interaction Contrast (SIC) is a powerful tool for assessing the architecture and stopping rule of a model of mental processes (Townsend & Nozawa, 1995). Despite its demonstrated utility, the methodology has lacked a method for statistical testing until now. In this paper we briefly describe the SIC then develop some basic statistical properties of the measure. These developments lead to a statistical test for rejecting certain classes of models based on the SIC. We verify these tests using simulated data, then demonstrate their use on data from a simple cognitive task.