This paper presents a computational model for simulating how people learn from serious games. While avoiding the combinatorial explosion of a games micro-states, the model offers a meso-level pathfinding approach, which is guided by cognitive flow theory and various concepts from learning sciences. It extends a basic, existing model by exposing discrete-time evolution, allowing for failure, drop-out, and revisiting of activities, and accounting for efforts made and time spent on tasks, all of which are indispensable elements of gaming. Three extensive simulation studies are presented involving over 10,000 iterations across a wide range of game instances and player profiles for demonstrating model stability and empirical admissibility. The model can be used for investigating quantitative dependences between relevant game variables, gain deeper understanding of how people learn from games, and develop approaches to improving serious game design.