Abstract: A system predicts the mental state of a user using a variety of contact or contactless sensors that measure heart rate, breathing or other data of the user. The mental state may be flow, overload, underload, stress, attention, motivation, valence, arousal, etc. While performing a game-playing task the difficulty of the task is adjusted based upon the predicted mental state of the user. A classifier for classifying a mental state is trained using event marking and ground truth data. Computer usage generates event markers. Game playing is used to simulate a mental state. Interruptions of a user are prevented when the user enters a flow, focused, or effortful attention state by classifying that state and indicating physically or electronically that the user should not be interrupted. Interruptions of a user are also prevented when the user enters a mind wandering state by classifying or predicting that state and indicating physically or electronically that the user should not be interrupted.