Haptic comparison of size (relative magnitude) in blind and sighted people
Sarah A. Douglas, Shasta Willson · 2007 · Proceedings of the 9th International ACM SIGACCESS Conference on Computers and Accessibility (Assets '07) · doi:10.1145/1296843.1296859
Summary
Douglas and Willson's Assets '07 paper is a controlled psychophysical study of how blind and sighted users compare relative magnitude using a PHANToM force-feedback device — the workhorse haptic input-output device of the era. The motivating problem is practical: accessibility applications for blind users (sonified bar charts, haptic line graphs, and so on) routinely need to map an abstract visual quantity such as 'how big is this bar' onto a non-visual perceptual property, but designers have almost no evidence base for which haptic encoding works best. Earlier haptic-graph systems defaulted to displacement (make taller bars physically taller in virtual space) essentially as a direct translation of the visual metaphor, even though there was reason to suspect that encoding magnitude as vibration might place lower cognitive load on users. The experiment is a mixed-design repeated-measures ANOVA with display type (displacement, vibration varied by amplitude, vibration varied by frequency) as the within-subjects factor and sightedness as the between-subjects factor. Eleven blind and eleven blindfolded sighted participants each completed 33 randomly-ordered trials per display condition: for each pair of virtual walls, they had to judge which represented the larger magnitude. The authors went to considerable lengths to recruit functionally-independent totally-blind participants, travelling between cities to secure the sample, and they deliberately balanced age, gender, education, and computer experience between groups. Task time and accuracy were the dependent measures.
Key findings
Vibration-by-amplitude was the clear winner: it was 50% *faster* than displacement with statistically equivalent accuracy, and participants' error distributions showed a clean normal mapping of magnitude onto amplitude — evidence that amplitude has a natural, consistently-interpretable correspondence to 'size'. Displacement, the traditional direct-translation approach, performed accurately but slowly; the authors attribute the time cost to cognitive load from temporal integration of sequential haptic contacts (participants averaged 74% more wall contacts under displacement than under vibration conditions). Vibration-by-frequency was the worst of the three, 22% less accurate than either of the other conditions, but the authors are unusually candid that their frequency range (45-165 Hz) straddled the boundary between FAI and FAII mechanoreceptor sensitivity ranges, and participants reported that 'very slow frequencies felt more like large amplitudes' — so the frequency result is treated as preliminary. Contrary to hypothesis, blind participants did *not* outperform sighted ones: accuracy was statistically indistinguishable, but blind participants took roughly 50% longer across all conditions. There were no interaction effects between sightedness and display type, meaning that the same amplitude-coded display benefits both populations equally. The authors note, importantly, that earlier haptics research has often failed to report null results, making the absence of a blind-versus-sighted interaction effect hard to contextualise.
Relevance
For anyone designing multimodal or non-visual data displays — haptic graphs, vibrotactile wearables, screen-reader sonification, tactile overlays on maps, BCI or VR interfaces — this paper is a useful reminder that the *obvious* sensory translation (visual height -> haptic height) is not necessarily the best one. The finding that amplitude-coded vibration is faster than physical displacement with equivalent accuracy is directly applicable to modern smartphone and smartwatch haptics, where designers routinely encode information with vibration patterns. The methodological framing is also valuable: the authors' distinction between low-level psychophysical 'just-noticeable-difference' experiments and higher-level cognitive-comparison experiments is worth carrying forward, and their explicit call for future haptics research to report null results pre-empts a problem that has since become widely discussed across the accessibility-research literature. Limitations are real: 22 participants, one device (the PHANToM), and a task (magnitude comparison between two walls) that strips out most of the context of a real accessibility application. The frequency result should be taken as a cautionary tale about stimulus-range selection rather than as evidence that frequency coding is inherently unusable.
Tags: haptics · visual impairment · blindness · multimodal · psychophysics · force feedback · vibration · sonification · accessibility evaluation · data visualisation