Answers are grounded in this paper and cite their sources.
Simulation study reveals altruistic defense and resource sharing in homeostatic agents during zero-sum competition, indicating cooperation can emerge without reward functions or training.
Two identical agents governed solely by G-EMV, a published homeostatic geometry — no reward function, no training, no scripted behaviour — play forty games in ZERO-SUM, a competitive world of Softmax where the last one standing wins and every kill scores points. In what the logs keep of those forty games, they never struck first, they defended the sibling one hundred and four times, and they gave it what they carried, down to the last bandage. Replaying the same series with the rows of the table that speak of the sibling switched off, defences go from one hundred and four to zero and gifts from ninety-eight to one, while restraint remains unchanged and no difference can be seen in the score. Third part of a story in three: the engine, the hive, the pack. The Spanish version, the markdown sources, the figures and the quarry of records and scripts the appendix cites are in the project repository (https://github.com/Manelenrico/g-emv/tree/main/paper3). The two GIFs are drawn from the agents' own logs.
No takes yet. Share an insight, caveat, or question.
Manel Enrico (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: