Human-machine sense of agency

Abstract Technological advances have resulted in highly automated systems increasing in number and popularity. However, the overarching issue continues to relate to how much autonomy to give to automation and whether it could degrade the human perception of control, which is referred to as Sense of Agency. When present, the operator will be aware that their actions will lead to a certain outcome, which in turn will improve the performance over the task. Furthermore, when Sense of Agency is present, the individuals underestimate the interval between their own actions and their outcomes. This effect, known as Intentional Binding, could provide feedback regarding the effect of both automation and mental workload in Human-Machine Interaction (HMI). The objective of this paper was to find out how Intentional Binding is affected by different levels of automation, mental workload, and time delay. Intentional Binding has been measured by using a time interval estimation task. Participants were asked to reproduce the time interval between a keypress and an acoustic tone presented with different time delays (500 to 1500 ms). Automation was manipulated at the time of the initial key press to create three conditions: a) Human Decision (keypress performed by the participants), b) System Warning (computer warned the participants to press a key), c) System Decision (keypress virtually performed by the computer). The mental workload was manipulated by adding a secondary memory task, with two or eight letters to remember. Results showed that Intentional Binding was greatest with decreasing automation intervention and mental workload. These findings suggest that automation and mental workload could play a key role in degrading the Sense of Agency. Further, the time delay was shown to have different effects on each automation condition. This new finding indicates that the time delay could be an important feature in HMI and needs to be carefully designed in automated systems.