Following up on the previous Nature Select team's empathy assessment method, I also reviewed their MAPO training approach. The purpose of this training method is to enable the model to maintain a stable "empathic persona" during multi-turn interactions and provide relatively consistent responses.

In other words, this methodology's focus is on the Agent's behavioral performance itself, rather than understanding a true psychological model, guiding human-computer interaction toward explicit "health/goodness."

Moreover, both papers lack an operationalizable definition of "empathy." In the text, it can roughly be understood as "the ability to demonstrate understanding, support, comfort, and non-aggressive responses in conversation." This definition is clear but notably lacks depth, and the conceptual ambiguity does not facilitate researchers in genuinely addressing the Agent's human-computer relational structure problem.

Original text: