Harry et al. 2026

Beyond fixed psychological personas: state beats trait, but language models are state-blind. Harry, T., Ngong, I. C., Nweke, C., Feng, Y., and Near, J. (2026). Findings of ACL 2026.

Relationship to this project — Between/within adjacency

Shows most psychological variation is within-person/state while LLMs respond weakly to state. Direct adjacency for this paper's between/within decomposition — but it does not study game-theoretic incentive response. Here, a trait-persona panel places most observed variation between prompt configurations while remaining weakly responsive to an economic lever.