Tactile Working Memory Capacity of Users Who Are Blind in an Electronic Travel Aid Application with a Vibration Belt
Jan B. F. van Erp, Katja I. Paul, Tina Mioch · 2020 · ACM Transactions on Accessible Computing · doi:10.1145/3372273
Summary
This paper investigates the tactile working memory capacity of users who are blind when using a vibrotactile belt for electronic travel aid (ETA) applications. ETAs that display obstacle information through vibration belts require users to integrate multiple tactile cues over time and space, building a mental picture of their surroundings. While previous research established that tactile working memory operates similarly to visual and auditory working memory, its capacity under real-world conditions—particularly while walking and in noisy environments—had not been studied in blind users. The study employed a vibrotactile belt with five tactors (vibration motors) arranged linearly around the participant's waist, spaced 8cm apart. Each tactor could produce two distinct patterns: a single burst (representing far obstacles) or double burst (near obstacles). This encoding scheme maps naturally to ETA use cases where vibration location indicates obstacle direction and pattern indicates distance. Fourteen adolescent participants who are blind (ages 15-20, 8 completely blind and 6 with remaining vision <8%) from a specialized school in the Netherlands completed the experiment. Participants received sequences of 1-5 vibration items and immediately recalled them using a whole report paradigm. Four conditions were tested: baseline (standing still, no noise), standing with distracting train station sounds (75-80 dB), walking with a white cane (no noise), and walking with both cane and noise. These conditions simulate the dual-task nature of real-world ETA use, where pedestrians must simultaneously process tactile navigation cues while attending to their physical environment and ambient sounds. Three performance measures were calculated: numerosity (how many items), location correct (where), and identity correct (both where and what pattern).
Key findings
Under ideal laboratory conditions (standing, no noise), tactile working memory capacity was surprisingly high—approximately 5 items for numerosity and 4.7 items for full identity recall (location plus pattern). These values exceed previous estimates from studies with sighted participants (typically 2-3 items), likely because blind individuals have greater experience with tactile information processing and the current study used a sequential spatiotemporal paradigm rather than simultaneous spatial presentation. However, performance dropped substantially under realistic conditions. When walking with a cane while ignoring train station sounds, capacity fell to 3.2 items for numerosity and just 1.6 items for full identity recall—a reduction of 40-65% from baseline. The effects of walking and noise were additive: walking alone reduced performance by 3-7% depending on the measure, while noise reduced it by 5-11%. Walking was perceived as significantly more mentally demanding than standing (11.2 vs 9.2 on a 21-point scale). Participants demonstrated strategic adaptations: experimenters observed that participants tended to slow down or stop walking during stimulus presentation, even when instructed to continue. This suggests users naturally trade off walking speed for better tactile processing. In post-experiment feedback, 10 of 13 participants agreed the belt should present no more than three items at once, aligning with measured capacity limits. All participants found the concept intuitive and expressed positive attitudes toward using such a belt for navigation.
Relevance
This research provides critical design guidance for vibrotactile ETAs and similar wearable navigation systems. The central recommendation—limit information to 2 simultaneous items in applied settings—has direct implications for how obstacle detection systems should filter and present environmental information to users. Presenting more obstacles than working memory can handle risks information overload, potentially degrading rather than enhancing safety. For practitioners developing tactile navigation aids, several design principles emerge. First, the dramatic difference between laboratory (5 items) and realistic conditions (1.6 items for identity) underscores that cognitive load testing must occur under dual-task conditions representative of actual use. Second, the additive effects of noise and walking suggest that systems may need to adapt their information density based on environmental context—providing fewer cues in busy, noisy environments. Third, the observed speed-accuracy tradeoff implies users may benefit from explicit training in strategy selection or interface features that support pacing. The study has limitations: participants were young (mean age 16.6) and healthy, potentially representing optimal performance; older adults or those in more demanding situations may have lower capacity. Additionally, with training (note: white cane proficiency requires ~100 hours), tactile processing could become more automatic and require fewer cognitive resources.
Tags: electronic travel aids · vibrotactile feedback · working memory · tactile displays · navigation · blind users · wearable technology · cognitive load
Standards referenced: EN ISO 8596