Using Prior Data to Inform Model Parameters in the Predictive Performance Equation

The predictive performance equation (PPE) is a mathematical model of learning and retention that attempts to capitalize on the regularities seen in human learning to predict future performance. To generate predictions, PPE’s free parameters must be calibrated to a minimum amount of historical performance data, leaving PPE unable to generate valid predictions for initial learning events. We examined the feasibility of using the data from other individuals, who performed the same task in the past, to inform PPE’s free parameters for new individuals (prior-informed predictions). This approach could enable earlier and more accurate performance predictions. To assess the predictive validity of this methodology, the accuracy of PPE’s individualized and prior-informed predictions before the point in time where PPE can be fully calibrated using an individual’s unique performance history. Our results show that the prior data can be used to inform PPE’s free parameters, allowing earlier performance predictions to be made.