Independent concurrent processing in dual-task performance: Evidence for adaptive executive control of task scheduling.

During daily life, people must often attempt to perform two distinct perceptual-motor or cognitive tasks simultaneously (e.g., talking on a cellular telephone while driving an automobile). According to some prominent psychological theories, such attempts require using a mental information-processing mechanism that slows dual-task performance because it can only make decisions and select responses for one task at a time. However, the four experiments presented here raise strong doubts about the existence of this putative decision and response-selection bottleneck. The first two experiments use the psychological refractory-period (PRP) procedure to characterize how people perform multiple tasks concurrently. For both experiments, a primary choice-reaction task was paired with a secondary choice-reaction task that had two levels of response-selection difficulty. Secondary task response-selection difficulty was varied by manipulating stimulus-response (S-R) compatibility. The effect of this factor on secondary-task reaction times (RTs) decreased reliably as the stimulus onset asynchrony (SOA) decreased, suggesting that the responses for the two tasks were selected concurrently. Experiments 3 and 4 used a new dual-task procedure. On dual-task trials, participants performed concurrent auditory-vocal and visual-manual choice-reaction tasks with equally high priorities. On single-task trials, only one such task was performed. Although the tasks typified those in the PRP procedure, RTs and response accuracy were essentially the same for each trial type. These data reveal that after moderate amounts of practice, two basic perceptual-motor tasks may be performed simultaneously with essentially the same response latencies and accuracy as when each task is performed alone. They also reveal that when slowing of dual-task performance does occur, it may stem not from an immutable bottleneck mechanism but rather from flexible output schedules used by executive mental processes to control the serial order of physical movements. These results, contrary to structural-bottleneck hypotheses, support assumptions about human information-processing capacities made in EPIC computational models of Meyer and Kieras (1997a, 1997b, 1998).Ph.D.Cognitive psychologyPsychologyUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/131535/2/9909989.pd