Simulating HCI for Special Needs
Pradipta Biswas · 2007 · SIGACCESS Accessibility and Computing · doi:10.1145/1328567.1328569
Summary
This paper describes the development of a simulator designed to model human-computer interaction for users with physical disabilities, addressing a significant gap in HCI research. The author argues that existing HCI models such as the GOMS family (KLM, CMN-GOMS, CPM-GOMS) and cognitive architectures (Soar, ACT-R, CORE) were designed primarily for able-bodied expert users and do not account for the diverse range of abilities and interaction patterns of disabled users. Meanwhile, assistive interface research has focused on building specific applications or novel input devices rather than developing systematic modeling tools. Practical user trials with disabled participants are difficult to arrange and time-consuming, creating a need for simulation-based evaluation. The simulator takes a task definition and the locations of interface objects as input, then predicts cursor traces and task completion times for different input device configurations (mouse, single switch scanning) across different disability profiles and skill levels. Its architecture comprises three components: an Application Model that breaks tasks into atomic subtasks, an Interface Model that selects and configures the appropriate interface for a given user, and a User Model that simulates interaction patterns using the Model Human Processor framework. The User Model itself contains perception, cognitive, and motor behaviour submodels — the perception model processes keyboard/mouse events and bitmap images to produce eye-movement sequences, the cognitive model uses interacting Markov processes based on dual space theory, and the motor behaviour model predicts completion times and interaction patterns from statistical analysis of disabled users' cursor traces.
Key findings
The paper maps existing HCI models onto a two-dimensional space of user skill level versus physical ability, revealing that most models cluster in the able-bodied expert quadrant, leaving the disabled-user and novice-user regions largely unaddressed. The AVANTI project is noted as one of the few efforts to model assistive interfaces, but it was limited to web browsers and did not address underlying perceptual, cognitive, and motor behavior. Research by Keates et al. found that motor-impaired users have significantly greater motor action times than able-bodied users when measured against the Model Human Processor, confirming that standard HCI timing parameters do not generalize to disabled populations. The simulator had been validated for novice able-bodied users at the time of writing, with the next step being to populate the remaining model components and validate with physically disabled participants. The approach is notable for aiming to work across both able-bodied and disabled users, across different skill levels, and to remain accessible to interface designers without requiring deep expertise in cognitive architectures.
Relevance
This work addresses a persistent challenge in accessibility practice: the difficulty of evaluating assistive interfaces without recruiting participants with specific disabilities for every design iteration. A reliable simulator could dramatically accelerate the design cycle for accessible interfaces by allowing designers to predict usability issues before conducting user trials, reducing the burden on disabled participants who are frequently over-recruited for studies. For accessibility practitioners, the three-component architecture (application, interface, user model) provides a useful conceptual framework for thinking about how disability affects interaction at multiple levels — not just motor output, but also perception and cognition. The paper also highlights an important limitation of mainstream HCI methods: models built for able-bodied expert users systematically exclude the very populations who stand to benefit most from well-designed interfaces. Although this is an early-stage proposal from 2007, the approach anticipated the growing interest in inclusive user modeling and simulation-based accessibility evaluation.
Tags: human-computer interaction · motor impairment · user modeling · simulation · assistive interface · adaptive interface · input device