The influence of cognitive load, physical load and system response time on user productivity and satisfaction with computer systems

Forty years of research into computer system response time have produced many results but overly-general conclusions. The general consensus is that for various tasks with a computer, faster response times are generally better than slow ones, but the computer can respond too fast. When computer response is too slow, users tend to become annoyed and frustrated with the computer system and may perform their tasks more slowly. When computer response is too fast, users may adapt to this by working faster at the expense of committing more errors and generally feeling more stress. However, only the most general guidelines exist for which tasks require rapid response and those for which the user will tolerate some delay. The present study was undertaken to further determine how specific combinations of mental and physical load contribute to an optimal system response time.;A generic data-entry task was systematically manipulated to vary the amount of mental and physical effort a subject experienced when engaged in that task under a variety of delay levels in the hopes of identifying how changes in task difficulty interacted with delay to cause users to perform optimally and be the most satisfied with the system. While changes in task difficulty had a significant impact on several measures of user performance and satisfaction, no interaction was found between various levels of task difficulty and levels of delay with respect to how users performed on the task or how they felt about the system. This calls into question the conventional wisdom in the field that for more difficult tasks, users will tolerate or even prefer longer delays whereas for simpler tasks users will be more sensitive to delays and those delays will have a more significant impact on their performance. In light of these results, a new cost-based model of human-computer interaction is proposed in which delay is viewed as an integral part of the human-computer interaction rather than being viewed as somehow separate from the entire process.