Addressing very short stimulus encoding times in modeling schizophrenia cognitive deficit

It is well known that encoding times in persons with paranoid schizophrenia are longer than those of normal controls. Neufeld and others have argued that this is the consequence of additional subprocesses being executed during the encoding process in the case of schizophrenia. In general they expressed an encoding time as the sum of k′ independent exponentially-distributed subprocesses, each executed with rate v. A troubling consequence of their application of this model to real data was that some individuals appeared to encode instantaneously (i.e., k′=0 was observed). This was accommodated in Neufeld et al. (2010) by placing a Poisson distribution on k′. In this paper the view is taken that k′=0 is not realistic and an alternative model is developed in which k′ is restricted to positive integers. This is made compatible with very short encoding times by introducing a task parameter α into the model. The problem of estimating α is addressed at length.