INFORMATION QUALITY IN MULTIVARIATE PROCESS CONTROL.

Human control of complex industrial processes poses unique man/machine interaction problems, with long training times and less-than-perfect productivity. Studies of the human operator in these multivariate processes have given useful data on factors affecting human performance but little guidance to designers who need to allocate tasks between machines and human operators. An experiment was performed to measure performance and strategy in a process control task and compare humans directly with an algorithm. The differences in results between human and algorithm have important implications for modeling human performance and for the allocation of function decision.