Testing the Impact of Instrumentation Location and Reliability on Ecological Interface Design: Control Performance

Recent research has shown that an ecological interface for a simulated pasteurization plant leads to most effective failure diagnosis performance only if the interface is supported by an adequate set of sensors. In this investigation we examine the effect of interface on control performance, focusing especially on between-subjects variability in control actions taken and state variable values. Results provide some support for the idea that when the interface is supported by an adequate set of sensors, an ecological interface produces less between-subjects variability in the values of higher-order state variables, but more between-subjects variability in the number of control actions taken. We interpret these findings to indicate that when goals and higher-order properties of a system are revealed through effective interface design, the degrees of freedom for action at lower levels are more fully exploited, which is all the better to achieve the higher-order purposes of control.