Examining Training Comprehension and External Cognition in Evaluations of Uncertainty Visualizations to Support Decision Making

Abstract

Recent advances in uncertainty visualization research have focused not only on design features to support decision making, but also on challenges of evaluating the effectiveness of uncertainty visualizations, such as the degree to which individuals' baseline task comprehension may alter their performance on experimental tasks regardless of a visualization's effectiveness. Building on recent work, we investigated the effect of training comprehension on performance across varying representations of uncertainty and varying degrees of visualization interactivity using a simulated course of action selection task. Additionally, we explored how extended cognition theory can be applied to visualization evaluations by incorporating interface features that afford externalization of knowledge within the task environment. Our findings suggest that regardless of how uncertainty is represented, training comprehension leads to superior transfer, reduced workload, more accurate metacognitive judgments, and higher cognitive efficiency. Our findings also suggest that external cognition during decision making leads to improved accuracy and cognitive efficiency. The present study contributes to research on the design and evaluation of uncertainty visualizations. In addition, this study extends previous work by demonstrating how extended cognition theory can inform the design of human-machine interfaces to support decision making.