A key component of expert and advice-giving systems is their ability to provide explanations to their users. Unfortunately, current systems assume that a single explanation is sufficient and that the user's only feedback will be requests for further elaboration. In reality, however, users frequently find fault with the provided explanations and respond with explanations of their own for why the system's response is unacceptable. The inability to understand or respond intelligently to this user feedback significantly degrades the value of existing systems. This project initiates a research program to explore how to extend expert and advice-giving systems to understand user response to their explanations and to revise these explanations accordingly. The object is an initial theory of how user feedback is understood and explanation revised in response to it, along with a prototype program embodying this theory. The proposed work will build on earlier research on using abstract planning knowledge to handle one specific class of user feedback.//