Divided attention is a vital cognitive ability used in important daily activities (e.g., driving), which tends to deteriorate with age. As with Alzheimer's and other neural degenerative conditions, treatment for divided attention problems is likely to be more effective the earlier it is detected. Thus, it is important that a method be found to detect changes in divided attention early on in the process, for both safety and health care reasons. We present here a new method for detecting divided attention unobtrusively, using performance on a computer game designed to force players to attend to different dimensions simultaneously in order to succeed. Should this model prove to predict scores on a standard test for divided attention, it could help to detect cognitive decline earlier in our increasingly computer-involved aging population, providing treatment efficacy benefits to those who will experience cognitive decline.