A Framework for Cognitive Bias Detection and Feedback in a Visual Analytics Environment

This paper presents a framework that supports the detection and mitigation of cognitive biases in visual analytics environments for criminal analysis. Criminal analysts often use visual analytics environments for their analysis of large data sets, for gaining insights on criminal events and patterns of criminal events, and for drawing conclusions and making decisions. However, due to the nature of human cognition, these cognitive processes may lead to systematic errors, so-called cognitive biases. The most prominent and relevant cognitive bias in the intelligence field is the confirmation bias, in which an analyst disproportionally considers and selects information that supports the initial expectation and hypothesis. The framework presented in this paper describes a model, how the possible occurence of the confirmation bias can be detected automatically, while the analyst makes use of the visual environment. Moreover, based on this information, different feedback methods are employed that support and encourage the mitigation of the confirmation bias. This framework is in a work-in-progress state and contains research objectives and directions, the framework design, initial implementations, plans for further development and integration, as well as user-centric evaluation.