Assessing Executive Function Using a Computer Game: Computational Modeling of Cognitive Processes
S. Hagler, H. B. Jimison, and M. Pavel
Early detection of decline in cognitive performance is important to providing proper medical care for older adults. Presently, cognitive decline is detected using clinical evaluations of cognitive performance administered by health-care professionals as part of a regular check up in a clinic. Visits to a clinic are infrequent and of short duration. Several visits may be required before cognitive decline can be detected, delaying the beginning of proper treatment. Earlier detection of cognitive decline can be obtained through regular and frequent measurement of cognitive performance in the older adult’s home. We make regular and frequent measurements of cognitive performance using Scavenger Hunt, a simple computer game. Scavenger Hunt is designed to mimic the Trail-Making Test, a standard, clinical neuropsychological test that resembles a child’s connect-the-dots puzzle. To make measurements of cognitive performance using play of Scavenger Hunt, we have constructed a mathematical model that describes the way a person plays the game. The model decomposes the process used to make each move in Scavenger Hunt into three additive stages: (1) recall, (2) search, and (3) motor. We measure the performance of the older adult separately for each of the three stages, providing a more detailed description of the older adult’s cognitive performance than given by the Trail-Making Test score. As Scavenger Hunt is designed to resemble the Trail-Making Test, the model also describes the way a person completes the test, allowing us to also use the performance measures made from the game to estimate expected performance on the test.