How different visualizations affect human reasoning about uncertainty: An analysis of visual behaviour. (March 2019)

• Record Type: Journal Article
• Authors: Reani, Manuele; Peek, Niels; Jay, Caroline
• Is Part Of: Computers in human behavior. Volume 92(2019)
• Page Start: 55
• Page End: 64
• Publication Date: 2019-03
• Subjects: Probabilistic reasoning -- Eye tracking -- Information visualization -- Decision making -- Visual analytics -- Human-computer interaction; Interactive computer systems -- Periodicals; Man-machine systems -- Periodicals; 004.019
• DOI: 10.1016/j.chb.2018.10.033
• Languages: English
• ISSNs: 0747-5632

Abstract

Abstract: Humans find reasoning about uncertainty difficult. In decision support systems and software for intelligence analysis, graphical representations are commonly used to display uncertainty. Nevertheless, our understanding of how people use the information presented in graphs displaying uncertainty to make decisions is limited. As many artificial intelligent systems require a human-in-the-loop who is able to actively take part in the analysis process, the understanding of high-level cognition involved in human-graph interaction is essential in the design of better tools for analysis. In this research, we investigate the visual behaviour that is associated with participants responses to problems testing probabilistic reasoning represented through two different visualizations (tree and Venn diagrams). Using the data from visual fixations and transitions, we present a description of different reasoning strategies covering both accurate and inaccurate reasoning for different visualization formats. The results show that gaze behaviour is related to reasoning accuracy. Moreover, this study shows that different graphs representing the same problem evoke different reasoning strategies, suggesting that higher level cognition is influenced by the graphical representation in which uncertainty is encoded. Highlights: We studied probabilistic reasoning in the context of human-graph interaction. We proposed an eye tracking method to study human reasoning. Probabilistic reasoning is affected by how uncertainty is graphically encoded. Gaze behaviour is linked to people performance in Bayesian reasoning tasks.