April 22, 2025
Modeling the relationships among online Solitaire gameplay and measures of cognition
Authors:
Sam Ihlenfeldt, Gregory K. W. K. Chung, Susan Lyons, Jordan Lawson, and Elizabeth J. K. H. Redman
In this technical document, we investigate the potential of using Solitaired.com as a tool for cognitive assessment. Solitaire can sustain motivation and engage cognitive processes relevant to constructs like mild cognitive impairment (MCI). Solitaire gameplay also addresses many of the limitations of traditional neuropsychological evaluations, such as their unnatural format and lack of inclusivity for low-literacy individuals. In this document, we explore the research by Gielis and colleagues, which reinforces Solitaired as a feasible platform for measuring MCI and identifies game performance metrics that are sensitive to players’ cognitive differences.
An initial pilot study using Solitaired.com data examined the associations among gameplay metrics and MCI. Players self-reported their cognitive status, and metrics such as game completion time and average move time were analyzed. Significant correlations aligned with earlier findings by Gielis and colleagues, supporting the viability of Solitaired.com as a cognitive assessment tool. The study then extended this approach with a larger sample, employing a random group design and regression modeling to predict aspects of mental acuity, as measured by TestMyBrain (TMB) testlets from The Many Brains Project.
This study shows that Solitaired.com gameplay variables are statistically and strongly related to working memory, processing speed, and visual short-term memory, as measured by TMB tests. Based on players’ interactions with the game, we can predict players’ mental acuity scores on three validated assessments of cognition:
- Working memory score (Flicker Change Detection)
- Processing speed score (Choice Reaction Time)
- Visual short-term memory score (Digit Symbol Matching)
A comparison of each of those scales suggests a high degree of overlap, so presenting players’ scores from all three models is not advised. Ultimately, we recommend presenting players with an overall percentile score from the Digit Symbol Matching model and a percentile score that compares players to all other players within their age bracket. Because age is such a meaningful variable in the model, older players will, on average, have lower mental acuity scores. Consequently, allowing those players to compare themselves to others in their age range could be beneficial.