A hierarchical Bayesian approach to distinguishing serial and parallel processing

Research in cognitive psychology often focuses on how people deal with multiple sources of information. One important aspect of this research is whether people use the information in parallel (at the same time) or in series (one at a time). Various approaches to distinguishing parallel and serial processing have been proposed, but many do not satisfactorily address the mimicking dilemma between serial and parallel classes of models. The mean interaction contrast (MIC) is one measure designed to improve discriminability of serial-parallel model properties. The MIC has been applied in limited settings because the measure required a large number of trials and lacked a mechanism for group level inferences. We address these shortcomings by using hierarchical Bayesian analyses. The combination of the MIC with hierarchical Bayesian modeling gives a powerful method for distinguishing serial and parallel processing at both individual and group levels, even with a limited number of participants and trials.