Naturalistic decision-making studies of intelligence analysis have generally focused on information search, collection, and synthesis processes, deemphasizing the initial "problem formulation" phase, in which analysts interpret and contextualize the information request to determine which information to collect. We present the results of two studies focusing on this phase. In the first study, we performed a cognitive task analysis via semistructured interviews with 22 active-duty U.S. Army intelligence analysts to uncover factors that arise in operational environments that complicate problem formulation. The factors discovered (e.g., vague and/or overly narrow intelligence requests) led to a second study probing 6 active-duty U.S. Army intelligence analysts' cognitive strategies with a "think-aloud" protocol as they interpreted and evaluated representative information requests. The study revealed that analysts actively interpret and contextualize an information request. The analysts reframed and broadened the request so that they could respond meaningfully to the underlying intent, then used contextual cues and metainformation to determine the most useful collectors and how effectively the request could be answered in the time allotted. We discuss these results and their implications for both the cognitive modeling of intelligence analysis and the development of training and decision aids for more effective framing and contextualization of information requests.

Information, as well as its qualifiers, or metainformation, forms the basis of human decisionmaking. Modeling human reasoning therefore requires the development of representations of both information and meta-information. However, while existing models and modeling approaches may include computational technologies that support meta-information analysis, they generally neglect its role in human reasoning. Herein, we describe the application of Bayesian Belief Networks to model how humans calculate, aggregate, and reason about metainformation when making decisions.

No abstract

No abstract

No abstract