A Galvanic Skin Response Interface for People with Severe Motor Disabilities
Melody M. Moore, Umang Dua · 2004 · Proceedings of the 6th International ACM SIGACCESS Conference on Computers and Accessibility (Assets 04) · doi:10.1145/1028630.1028640
Summary
This paper explores the use of Galvanic Skin Response (GSR) — changes in the electrical conductivity of the skin caused by sweat gland activity linked to the sympathetic nervous system — as a non-muscular input channel for people with locked-in syndrome. The research was conducted with a 42-year-old male subject diagnosed with ALS in 1997 who had progressively lost all motor function, becoming completely locked-in approximately one year before the study. Initially he could move his eyes slightly, but lost even that ability during the study period; his ophthalmologist crafted special glasses to brace his eyelids open. The subject had 14 months of experience controlling his GSR before data collection began. GSR was measured using a commercial polygraph system with two steel electrodes attached to the index and middle fingers, with sensitivity raised from the default 20 to 250-300. The experimental system superimposed a threshold target on the polygraph display, projected onto a 72-inch screen at the foot of his bed. The subject could generate a "yes" by raising his GSR above the threshold and a "no" by keeping it stable below threshold. Auditory versions of all interfaces were implemented so the subject could operate the system without having his eyes propped open. Applications included a binary yes/no answering system, a binary speller, a number program, a phrase chart, and a chess game. Data was collected over 30 independent sessions across six months.
Key findings
The subject achieved an overall unbiased accuracy of 61.78% in generating yes/no responses via GSR control, compared to baseline accuracy of essentially random (10% yes correct, 94.1% no correct — indicating no intentional GSR activation during baseline). Accuracy varied considerably across sessions, ranging from 30% to 90%, with the fourth and final day of the initial intensive week showing consistently strong performance (90% on the first run). Performance was affected by fatigue, with accuracy declining within sessions, and by environmental factors — room temperature correlated with GSR control (73°F versus the subject's usual 69°F). Artifact detection proved critical: the subject's automated hospital bed caused electrode movement producing large false peaks, and periodic patterns approximately 45 seconds apart were traced to bed position shifts. Pilot studies explored methods to improve accuracy: sudden changes in visual input (watching golf on TV) caused sharp GSR spikes, while auditory input (listening to a book on tape) produced lower, more stable GSR, suggesting that selective attention to visual versus auditory stimuli could be leveraged for control. Choice anticipation — when the subject could predict the next requested response — dramatically improved accuracy to near 100% for both yes and no responses, though it also occasionally caused anticipatory false activations.
Relevance
This research pushes the boundaries of assistive technology into territory where no other input channel is available. For people with complete locked-in syndrome who have lost even eye movement, GSR represents one of the few remaining voluntary communication channels, since the sympathetic nervous system that drives skin conductance changes can be influenced by conscious emotional and cognitive effort. The 61.78% accuracy, while insufficient for practical text entry via spelling, is significant for someone with literally no other way to communicate — it enables rudimentary yes/no interaction. For accessibility practitioners, the study highlights several important principles: the critical importance of artifact detection in physiological computing systems, the influence of environmental factors (temperature, bed movement, ambient stimuli) on biometric interfaces, and the potential to leverage selective attention to different sensory modalities as a control mechanism. Key limitations include the single-subject design, the modest accuracy rate, the fatigue effects that limit session duration, and the long GSR stabilization time between responses that severely limits information transfer rate. The dedication of the paper to the subject's memory underscores the urgency of this work for people whose communication window may be closing.
Tags: locked-in syndrome · galvanic skin response · brain-computer interface · biometric control · ALS · assistive technology · physiological computing · motor disability · biofeedback