Hierarchical Bayesian mixture models of processing architectures and stopping rules

Systems Factorial Technology is a methodology that allows researchers to identify properties of cognitive processing systems, such as the system's architecture and the decisional stopping rule. It assumes that the cognitive system will use the same architecture and decisional stopping rule on every trial of an experiment. Through simulation, we aim to explore the predictions of models that allow for a mixture of architectures and decisional stopping rules across trials. Our simulation reveals a pattern of interaction contrasts across different mixtures, some of which mimic each other. Given the mimicry, we developed a hierarchical Bayesian inference method that can identify mixture probabilities for mixtures of architectures and stopping rules. The method is also a useful inference tool that can determine what architecture and stopping rule a participant is using and has many promising future developments, such as simultaneous group and subject level inference.