Towards a Neuro-Symbolic Cycle for Human-centered Explainability