ABSTRACT
Transcription typing is one of the basic visual-motor control tasks with practical importance. Brain imaging studies (fMRI and PET) have discovered 2 phenomena related to transcription typing, involving different activation patterns of brain areas. In behavioral studies, Salthouse (1986) summarized 29 quantitative behavioral phenomena of transcription typing, 19 of which were modeled by TYPIST (John, 1996)—the most comprehensive computational model of transcription typing. This paper first proposes a queuing network model that integrates queuing networks with reinforcement learning algorithms, and then describes how the model simulated the two brain imaging phenomena and three behavioral phenomena in learning process. Overall, the queuing network model successfully modeled not only all the phenomena modeled by TYPIST, but also 11 additional behavioral phenomena and the 2 brain imaging phenomena. All the phenomena emerged as outcomes of the natural operations of the human information processing queuing network, with no need to draw situation-specific scheduling charts that are required by TYPIST.
