Tensions between Preference and Performance: Designing for Visual Exploration of Multi-frequency Medical Network Data

TL;DR

The paper investigates tensions between task-driven visualization design and aesthetics in visualizing multi-frequency EEG brain networks. It reports a three-phase design study that yields two high-fidelity prototypes (A: aesthetically constrained layered doughnut; B: bar+heatmap) evaluated with domain experts and lay participants. Results show higher objective performance for Prototype B but stronger aesthetic preference for Prototype A, illustrating the trade-off between effectiveness and engagement. The work argues for parallel high-fidelity prototyping and consideration of affective goals in bespoke visual analysis tooling, with implications for production deployment and future guidelines.

Abstract

The analysis of complex high-dimensional data is a common task in many domains, resulting in bespoke visual exploration tools. Expectations and practices of domain experts as users do not always align with visualization theory. In this paper, we report on a design study in the medical domain where we developed two high-fidelity prototypes encoding EEG-derived brain network data with different types of visualizations. We evaluate these prototypes regarding effectiveness, efficiency, and preference with two groups: participants with domain knowledge (domain experts in medical research) and those without domain knowledge, both groups having little or no visualization experience. A requirement analysis and study of low-fidelity prototypes revealed a strong preference for a novel and aesthetically pleasing visualization design, as opposed to a design that is considered more optimal based on visualization theory. Our study highlights the pros and cons of both approaches, discussing trade-offs between task-specific measurements and subjective preference. While the aesthetically pleasing and novel low-fidelity prototype was favored, the results of our evaluation show that, in most cases, this was not reflected in participants' performance or subjective preference for the high-fidelity prototypes.

Figures (4)

• Figure 1: Schematic representation of the EEG data. The grey matter (outer layer of the brain) was parceled into 72 ROIs (red dots), and each ROI's electrical activity was reconstructed from the EEG signal. The frequency content of the ROI signal is displayed for each frequency band in the box. Once the connectivity between each ROI is estimated, a brain network (ROI x ROI) is built for each frequency band. Then, nodal network metrics – such as the clustering coefficient – are computed for each ROI.
• Figure 2: The three low-fidelity prototypes: (a) using bar charts (metric) and a dot plot (connectivity), (b) using parallel coordinates (metric) and doughnut charts (connectivity), and (c) using a layered doughnut chart (metric and connectivity). Remark: user interface elements (e.g., menu for selecting ROIs) have been excluded to focus on the visual encoding of the data.
• Figure 3: High-fidelity prototypes with all 72 ROIs. Prototype A (based on low-fidelity prototype 3) shows a layered doughnut chart (metric and connectivity), histograms (metric and connectivity), and a brain plot with 72 ROIs. Prototype B (based on low-fidelity prototype 1) shows bar charts (metric), a heatmap (connectivity), and a brain plot with 72 ROIs.
• Figure 4: High-fidelity prototype A with three subnetworks containing five, three, and five ROIs, respectively.