Cognitive World Models for Process-Level Social Influence Evaluation
Minghui Ma, Bin Guo, Han Wang, Mengqi Chen, Jingqi Liu, Yan Liu, Zhiwen Yu · 2026
Read Cognitive World Models for Process-Level Social Influence Evaluation by Minghui Ma, Bin Guo, Han Wang, Mengqi Chen, Jingqi Liu, Yan Liu, Zhiwen Yu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-29495.txt.
Description
Ma and six colleagues argue that judging social influence dialogue takes more than checking final answers. Persuasive conversation works by shifting a user's beliefs, desires, intentions, and emotions over many turns. Surface text scores such as BLEU and ROUGE, and single-score judge models, cannot track that inner movement.
Their Cognitive World Model, CogWM, reframes evaluation around how the user's internal state evolves. It predicts BDI/E states, short for beliefs, desires, intentions, and emotions, alongside the user's next utterance. The same model serves both as a user simulator and as an evaluation platform, with a three-tier scoring framework. Training used a Summarize-and-Allocate annotation pipeline over more than one hundred fifty thousand user turns from four scenarios. CogWM reached high emotion accuracy, more than double a strong commercial baseline. In thousands of trials it also ranked six commercial agents by their cognitive influence.
Catalog record
| CREATOR | Minghui Ma, Bin Guo, Han Wang, Mengqi Chen, Jingqi Liu, Yan Liu, Zhiwen Yu |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 89 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.29495