Attacking Cognitive Bias in Intelligence Analysis: A Framework for Investigating the Efficacy of Bias Mitigation Techniques

Donald Kretz, University of Texas at Dallas and Raytheon Co., USA

About the author:

Donald Kretz has more than 30 years’ experience in cognitive science research, intelligence collection and analysis, and information technology research and development. His service as both a military and civilian intelligence analyst included coordinating the development of threat assessments for the White House and the U.S. Congress, and he has worked as a lead engineer on projects for DARPA and U.S. military customers. Don is a Principal Investigator in Raytheon’s Analytical Sciences Group in areas such as cognitive “debiasing” techniques for intelligence analysis; automated tools for human profiling, matching, and exploitation; user-centric analytics to improve knowledge worker efficiency; and enhanced pattern classification for recognizing and labeling patterns of human activity for activity-based intelligence. Earlier this year, his research was featured in a BBC Horizon documentary titled “How You Really Make Decisions” together with Daniel Kahneman and Dan Ariely. Don holds a B.A in Psychology from the University of Texas at San Antonio and a M.S. in Applied Cognition and Neuroscience from the University of Texas at Dallas, where he is completing work on his Ph.D.

About the talk:

Visualizations are developed for a variety of purposes. Some are intended to help viewers understand fairly simple and straightforward concepts, while others are designed to enable users to interactively solve cognitively complex problems. Intelligence analysis is an example of a complex cognitive process. Analysts receive substantial training and develop deep expertise in their tradecraft; yet, analytic outcomes continue to be negatively impacted by cognitive factors. In a series of post-9/11 studies and reports, cognitive bias mitigation was explicitly identified as a focus of attention, and research shows that even experienced, trained personnel sometimes commit serious errors in judgment as a result of cognitive heuristics. Despite this knowledge, there is yet no “recipe book” for experimentally-validated tools and techniques to avoid judgment biases when solving complex intelligence problems. Bias mitigation can take many forms, including advanced visualizations designed to derail certain mental shortcuts. But how do we determine whether or not a bias mitigation is effective? In this talk, I will describe one approach that we believe will prove effective in investigating the efficacy of specific tools and techniques in overcoming cognitive biases in intelligence analysis. In doing so, I will offer some lessons learned and identify some challenges for future work.