Behavioral Inference at Scale: The Fundamental Asymmetry Between Motivations and Belief Systems - Dataset

Processed behavioral datasets and raw game logs from a large-scale study of LLM-based agent behavioral inference. Agents (Llama 3.1-8B) assigned one of 36 profiles (9 moral alignments x 4 motivations) navigate a procedurally generated dungeon environment. The deposit contains raw game logs (17,411 games, 1,575,377 sequences) and the legacy BiLSTM and Longformer training parquets used for the headline results in the associated paper. Associated paper: Starace and Soule, Behavioral Inference at Scale: The Fundamental Asymmetry Between Motivations and Belief Systems. Full documentation including cohort notes and the data processing pipeline is in README.md.