How Predictive Minds Explain and Control Dynamical Systems

Roman Tikhonov, Sarah Marzen, Simon Dedeo

Keywords: prediction, explanation, control, predictive mind, LTSM, psychology, dynamical systems

TL;DR: Even ``predictive" minds might find it easier to explain, or control, a dynamical system, than to predict it.

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.

In-person Presentation: yes