A Human Motor Throughput Model in One-Dimensional Pointing Tasks under Temporal Constraints

Abstract

In Fitts’ law, motor throughputis defined as the index of difficulty divided by corresponding movement time and can be experimentally obtainedfrom the reciprocal of the slope relating movement time and the index of difficulty, which applies only to spatially-constrained pointing tasks

throughput models provides frameworks for analyzing the performance of a given pointing device and can be used to compare performances on pointing tasks in diverse settings. However, for temporally-constrained pointing tasks, there is no such model or definition that can represent the motor throughput for a given pointing task. In this study, a definition of human motor throughput in pointing tasks under temporal constraints is proposed. Basedon the proposed definition, a novel motor throughput model is mathematically derived. According to the proposed model, motor throughput is a function of interface size, inherent noise in human movements, and sensitivity to the interference induced by the movement for a given movement time. Interestingly, in temporally-constrained experiments with three different pointing devices (mouse, stylus, touch screen) for four levels of movement time (556, 682, 870, and 1200 ms), for each pointing devices, the throughput values calculated by the proposed throughput modelwere similar to one another over a wide range of interfacesizes, regardless of time allowed for movement which means that the throughput value can be a unique value for a given pointing task. Overall, mouse (low noise and high interference) showed the highest throughput and touch screen (high noise and middle interference) showed the lowest throughput. However, due to a smaller interference, stylus (middle noise and low interference) seemed to have higher throughput than the mouse when the information size became large. In addition, further experiment was conducted to test whether the proposed model can also be applied to spatially-constrained pointing tasks whose throughput has been modeled byFitts’ law. Twenty participants performed a spatially-constrained pointing task using a stylus pen for four groups of ‘target size-movement amplitude’ pairs, each of which consisted of six different pairs with the same index of difficulty. The experimental results showed that throughput remained relatively constant regardless of the index of difficulty which means that the throughput value can be a representative value for a given pointing task under spatial constraints.