Human computer interfaces for autism: assessing the influence of task assignment and output modalities

Several experimental studies have shown the usefulness of computers for autism, but software design remains poorly documented. Our multidisciplinary research focuses on educational HCI for autism. We compared two domains of learning: social dialogue understanding and spatial planning, our hypothesis being that people with autism will be less skilful in the first than in the second domain. Two sets of exercises were designed for each domain: one for training purposes and the other for performance assessment before and after training. We also tested the influence of the following output modalities: text, images, speech synthesis, visual and auditory feedback. Each exercise produced log files informing on duration, number of trials and successes. So far, eight teenagers with autism have completed a 13 week training program with one session a week. First analysis of log files suggests a significant progression in dialogic understanding but not in spatial planning; nor was significant influence of output modalities found.