A Proposed Preliminary Framework for Conceptual Modeling of Learning from Surprises