A task-based taxonomy of erroneous human behavior

Unexpected, erroneous human interactions often contribute to failures in complex systems. Human factors engineers and researchers have developed taxonomies that allow engineers, designers, and practitioners to think about and model erroneous behavior to improve the safety of human-interactive systems. However, the two leading erroneous behavior taxonomies are based on incompatible phenomenological and genotypical perspectives. Further, neither of these are formulated in terms of task analytic methods, analysis and modeling techniques human factors engineers use for documenting how humans normatively achieve goals when interacting with a system. In this work, we introduce a new erroneous human behavior taxonomy based on where and how human behavior diverges from task analytic models of human behavior. By describing where a human diverges from a normative task, and by identifying what information the human failed to properly attend to that produced the divergence, this taxonomy seeks to unify the phenomenological and genotypical perspectives. We describe the theory behind this taxonomy and the different erroneous behavior classifications that result from it. We then show how it is compatible with the leading phenomenological and genotypical taxonomies. Finally, we discuss the implications of this new taxonomy and avenues of future research.