Understanding game actions: The development of a post-processing method for audio-visual scene analysis

摘要

This paper introduces a video processing method designed specifically to give researchers access to players' behaviors within any commercially available off-the-shelf PC game. Insights into players' experiences with videogame violence are achieved from automatically processing significant elements of the audio-visual communication/feedback (i.e. moving image and sound) over long periods of uninhibited game play. The method exploits a wealth of significant and (behaviorally) influential information that is conveyed to the player as they progress through a game. Such information can relate to virtual progress (e.g. maps, mission logs), player reserves (e.g. acquisitions) or vitality (e.g. health or stamina levels), to name but a few. By analyzing what is transmitted to the player during play, feedback-based game metrics are produced that utilize the same content that the player perceives and processes during their gameplay, thus, tying the method directly to individual `player experiences.' This method was developed for a project that addressed the persistent concerns over the impact of game violence and its translation into societal violence. The method provided the project with an alternative approach to understanding the use and function of violence within game systems, taking research in a new direction from the approach and research-designs employed by experimental psychological over the last few decades.