Developing a naturalistic categorization task for testing intuitive decision making.

Intuitive Decision Making (IDM) depends on knowledge that cannot be easily articulated. It does not reflect explicitly learned rules and guidelines. Rather, it is hypothesized to rely on implicit learning (IL). Basic science research on the phenomenon of IL provides a theoretical framework for understanding the acquisition of knowledge outside of conscious awareness from practical experience. This framework has the potential to accelerate the development of IDM during training and speed the acquisition of expertise. Here we describe a procedure and present a program of research based on adapting a more operationally relevant task to controlled laboratory conditions to bridge basic science and enhanced simulation-based training. The underlying task is one in which a complex decision is made based on environmental terrain characteristics, such as the formation in which to proceed with a patrolling infantry squad. This decision process is analogous to laboratory tasks in which participants learn to discriminate among a set of visual categories, but requires a new kind of task in which the visual stimuli are constructed from complex terrain dimensions. We defined a stimulus space based on four environmental dimensions: vegetation density, topography (hilliness), time of day and weather conditions. An artificial category structure was then defined within this stimulus space around three hidden prototypes. Participants learned these categories through trial-and-error with feedback about their decisions. Across three experiments, participants exhibited learning, increasing their decision accuracy across a range of task parameters selected to promote reliance on IL and use IDM. The resulting protocol will serve as a testbed for quantification of IDM effects and allow future work to examine training and educational interventions aimed at improving effective use of IDM. In addition, the task development process can serve as a model for bridging basic science research and operationally relevant domains.