Digital Games as Experiment Stimulus

Digital games offer researchers a rich but demanding experimental stimulus, and this work argues that their careful use can advance experimental psychology and human-computer interaction. Games are familiar and engaging to most participants and can safely evoke emotions or simulate ethically sensitive scenarios that would be difficult to create in a laboratory. The authors support this case with survey evidence that 72 percent of U.S. households and 54 percent of Finnish respondents are active gamers, underscoring the broad population reach of the medium. At the same time, they caution that interactive content varies from player to player because of differences in skill, artificial intelligence behavior, randomness, and prior familiarity, which makes controlling the stimulus far harder than in conventional experimental materials.

The methodological guidance centers on four considerations for researchers: task type, data segmentation, participant compatibility, and data collection. Commercial titles offer realism and engagement but often lack built-in logging and resist modification, while custom-built games give experimenters control at the cost of reduced engagement and a risk of experimenter bias. Practical recommendations include matching games on quality, using critic ratings as a control, and selecting tasks with frequently repeating, isolated events. Recruiting participants with comparable gaming experience is advised, though the authors acknowledge that perfectly identical stimuli across players are rarely achievable. For event-based analysis, they recommend game logs or external recording with calibrated timestamps, with linear mixed models suited to the hierarchical, repeated-measures structure of such data.

A case study illustrates these principles. The racing game GTR 2, paired with Motec i2 Pro telemetry, was used to test whether a performance health drink affected concentration, performance, emotion, alertness, and stress. Racing games suit such work because their laps provide repetitive, predictable events and built-in performance metrics, while fixed difficulty, a fixed car, automatic gears, randomized circuits, a within-subject design, and a training session limit learning effects and skill differences. From this experience the authors propose a four-step framework covering task matching, event coding and segmentation, participant compatibility, and data collection planning. They conclude by calling for more standardized and thoroughly documented procedures so that findings across game-based research can be meaningfully compared.