Dylan Cashman

Tactile graphics are often adapted from visual chart designs, yet many of these encodings do not translate effectively to non-visual exploration. Blind and low-vision (BLV) people employ a variety of physical strategies such as measuring lengths with fingers or scanning for texture differences to interpret tactile charts. These observations suggest an opportunity to move beyond direct visual translation and toward a tactile-first design approach. We outline a speculative tactile design framework that explores how data analysis tasks may align with tactile strategies and encoding choices. While this framework is not yet validated, it offers a lens for generating tactile-first chart designs and sets the stage for future empirical exploration. We present speculative mockups to illustrate how the Tactile Perceptual Grammar might guide the design of an accessible COVID-19 dashboard. This scenario illustrates how the grammar can guide encoding choices that better support comparison, trend detection, and proportion estimation in tactile formats. We conclude with design implications and a discussion of future validation through co-design and task-based evaluation.

New consumer devices with refreshable tactile displays promise to allow visually impaired people to analyze data through tactile representations of graphical visualizations. To understand whether results based on visual perception translate to tactile perception, we present a study replicating the formative study by Cleveland and McGill (1984) on graphical perception to tactile representations suitable for visually impaired users. To assess how tactile graphics can convey complex graphical information, we investigate the effectiveness of tactile data visualizations compared to reported results on visual graphical primitives, examining the accuracy and inference times of visually impaired versus sighted users. We find that visually impaired users interpret simpler tactile formats such as bar charts with significantly greater accuracy and speed than more complex formats like bubble charts.

This interactive article presented a game that helps the reader understand how machine learning models can fail in applied machine learning use cases. It uses visualization to illustrate a thought experiment presented in a famous learning theory paper from the 1980s. I presented this article at the 6th workshop on Visualization for AI Explainability at the 2023 IEEE VIS conference. It received a Best Submission award.