Shared Mental Models in Human-Machine Systems

Imagine a future where humans and machines are able to share tasks, to monitor each other’s performance, and to interchange (control) authority whenever required or desired. In aviation, this vision was conceptualized almost twenty-five years ago by the late Charles Billings and is formally known as Human-Centered Automation. Although the aviation community has embraced this perspective, it proves to be difficult to realize this envisioned level of human-machine collaboration, especially for cognitive tasks. To achieve a breakthrough, I argue that we first have to consider what seems to be missing from current forms of automation that is fundamental to effective inter-human collaboration: the possibility to share mental models (or representations) of the problem(s) to solve. When looking at human-human interaction, productive team thinking and problem-solving efforts are accomplished when teammates have a “common ground” or shared understanding of the work to be done and the various ways to do it. Similarly, when work would be distributed over human and automated agents, the constraints introduced by the other agents are properties of the work domain and must therefore be shared. But how can more information be shared while not overloading the human’s capacity to learn and solve problems? In this paper I argue that the Ecological Interface Design paradigm can provide the means and guidelines to pursue shared human-automation mental models that will facilitate productive thinking. I will illustrate this by means of an example in the field of aircraft conflict detection and resolution.