Toward the trichotomy method of reaction times: Laying the foundation of stochastic mental networks (Q1812709)

Cognitive psychologists try to find out how subjects solve problems. One approach assumes that a complex task is divided into several partial tasks which are solved sequentially, and that subjects cannot begin any partial task until they have completed the preceding one. More complex models allow for both sequential and concurrent mental processes. For more than one hundred years researchers have been trying to identify the underlying mental architectures by designing factorial experiments, the factors of which are assumed to affect the mental processes. By analyzing the mean reaction times resulting from the different combinations of factor levels, they hoped to obtain information on the number and nature of partial tasks, whether the processing is ``serial'' or ``parallel'', and to find answers to other questions concerning the influencing factors. In the present article, stochastic discrete PERT networks are used as a model of mental architecture. Mental processes are identified as arcs in a directed acyclic network. When an experimental factor has a direct effect on the processing time distribution of exactly one process, this is called a selective influence. Two theorems concerning types of selective influence are proved.