Post no less than a paragraph addressing the following questions:

In what ways might integrating connectionist methods (e.g., neural networks) into symbol-based planning systems improve explainability?