How Predictive Minds Explain and Control Dynamical Systems

Abstract

We study the relationship between prediction, explanation, and control in artificial ``predictive minds''---modeled as Long Short-Term Memory (LSTM) neural networks---that interact with simple dynamical systems. We show how to operationalize key philosophical concepts, and model a key cognitive bias, ``alternative neglect''. Our results reveal, in turn, an unexpectedly complex relationship between prediction, explanation, and control. In many cases, ``predictive minds'' can be better at explanation and control than they are at prediction itself, a result that holds in the presence of heuristics expected under computational resource constraints.