Learning to Control a Dynamic Task: A System Dynamics Cognitive Model of the Slope Effect

We developed a system dynamics model for a simple, but important stock and flows task where the objective was to control the water level in a tank within an acceptable range of the goal, over a number of time periods, in the presence of an unknown environmental inflow and outflow. We also report how this model accounts for human behavior, using behavioral data we collected from human subjects in the task. This exercise helped us understand the strategy and mechanisms our participants used in the simple stock and flows task and develop a model on the task. The model provides an integrated explanation on how the variation in the parameters of the model affects the performance and learning for the participant’s task. Finally, we present the model’s validity and predictions derived by looking into how the human data fits different learning conditions.