Joe Thompson

March 04, 2014

Digital records of performance become more common as our natural behavior becomes increasingly computer mediated. Such data provide a new purview from which to view real-world performance. One such opportunity comes from competitive real time strategy video games. Detailed records of second by second performance can be automatically recorded and collected online, allowing for large samples of what is, for many of our participants, a part of their natural cognitive functioning. The present work considers age-related changes in the cognitive-motor speed, as measured by latencies between view-screen shifts and action, of 3305 StarCraft 2 players from the ages of 16-44. Linear regression of looking doing latency on age and skill revealed the existence of age-related change. We estimated when decline began using piecewise linear regression of age on looking doing latency, and concluded that aging probably begins around 24 years of age. These analyses revealed no evidence that such declines were ameliorated in more skilled players. However, in order to consider whether players could indirectly compensate for these declines by changing other aspects of their gameplay, we performed an exploratory analysis on other performance variables known to be related to StarCraft 2 skill, identifying a few aspects of the game interface that older players may be using to reduce cognitive load. Our results suggest that the appearance of skill preservation throughout adolescence and early adulthood may hide complex developmental processes where the quality of performance is maintained by the cognitive systems adaptations to age-related change.

Joe Thompson is a PhD student working in the Cognitive Science Lab at Simon Fraser University, where is part of a research team under Mark Blair, that uses large samples of video game telemetry data to study skill development, maintenance, and decline. His theoretical interests lie in the general complexity of the extended learning process, and so he is interested in any methodology that allows for a better characterization and understanding of this complexity.