For many years, the focus of research in the area of human error was the prevention of erroneous actions and assessments through training and design. However, errors can never be eliminated completely. Therefore, the goal of more recent efforts is to minimize their negative consequences through support for error management, i.e., the detection, explanation, and recovery from erroneous actions. For the most part, these efforts have examined the first step in this sequence - error detection. In contrast, little is known about how operators explain and recover from errors. This is true especially for dynamic collaborative environments such as aviation. In this paper, we present findings from a survey and an incident report analysis that suggest the need for adapting the current model of error recovery. Specifically, we emphasize the importance of considering constraints imposed by specific domains in order to predict and explain the predominance and success of certain recovery strategies.
