Divided Attention Amplifies the Importance of Expectation-Aligned Visualization Design

Visualization design research often assumes a single-task environment, but real-world users frequently interpret data while managing concurrent tasks like monitoring alerts or responding to messages. This study investigates whether divided attention—the need to allocate cognitive resources across multiple tasks—exacerbates the performance penalties associated with visualization designs that violate user expectations (e.g., mapping light colors to higher values when users expect dark colors to represent higher values).

The researchers conducted two experiments using a colormap interpretation task where participants judged data values based on color scales. In the dual-task condition, participants simultaneously monitored and responded to incoming phone messages. Experiment 1 allowed unlimited viewing time, while Experiment 2 imposed strict time constraints. To understand the underlying cognitive mechanisms, the authors applied a Linear Ballistic Accumulator (LBA) framework, which models the decision-making process as the accumulation of evidence toward a threshold.

The study found that divided attention consistently amplifies the cost of expectation-violating designs. In the unlimited-time condition, this manifested as increased response times and lower accuracy. Under time pressure, the effect was even more pronounced, leading to a higher rate of failed responses (misses). The LBA modeling revealed that divided attention reduces the separation between evidence for correct and incorrect interpretations, effectively making it harder for users to distinguish the correct answer when the design conflicts with their mental models.

These findings suggest that design principles for "ideal" conditions may be insufficient for high-stakes, real-world environments where attention is fragmented. Designers should prioritize expectation-aligned mappings—such as dark-is-more or high-is-more—when creating interfaces for dashboards, monitoring systems, or mobile displays where users are likely to be multitasking. The study also demonstrates the value of process-oriented modeling in predicting how specific design choices impact user performance under cognitive load.