Methods for Discovering Cognitive Biases in a Visual Analytics Environment

Abstract

This chapter discusses three clusters of approaches to measure and detect cognitive biases in the context of a Visual Analytics Environment (VAE). The first approach is called theory-driven since it refers to purely expert-driven methods. One of these methods is based on the definition of design recommendations for interactive data visualizations to reduce the potential effects of cognitive biases. Another expert-driven method is the in-depth analysis of a VAE and the cognitive biases it may induce. The second, empirical approach, encompasses behavioral observations, as well as experimental methods to operationalize cognitive biases. We describe methods to measure the Confirmation Bias and the Clustering Illusion. Finally, the third approach refers to data-driven methods that enable a non-invasive measurement of cognitive biases. This method is based on data-mining techniques to analyze and interpret user interaction patterns in terms of cognitive biases.