Adjustment for cognitive interference enhances the predictability of the power learning curve

Learning curves, which express performance as a function of the cumulative number repetitions when performing given task, have long tradition supporting managerial decisions in production and operations management. Performance generally improves task increases, with latter being primary proxy to reflect experience. A learning curve is usually maximum-likelihood trend-line that best fits raw data points by splitting them above below it. However, its curvature does not always accurately capture scatter around it, reduces accuracy. This paper advocates for an improved curve, one accounts variable degree cognitive interference occurs while moving from repetition next. To this phenomenon, memory traces measure residual (interference-adjusted), nominal (maximum), Two alternative models were developed. The first model aggregates experience each fitting data. second approximate expression and, continuous model, much easier implement. tested against different environments (such assembly), alongside more traditional power (log-linear) form plateau version. results show interference-adjusted fit very well, such they can serve valuable tools