Answers are grounded in this paper and cite their sources.
Exploratory computational experiment reveals wealth influences cooperation during shared threats.
Economic stratification organizes daily life but collapses under sufficiently severe shared threats. We propose a stratification–threat function S(τ) in which behavioral stratification first amplifies under moderate threat and then collapses above a critical threshold τ*, modulated by three variables: threat immediacy, escape excludability, and shared-fate visibility. We test this framework in an exploratory computational experiment: 18 autonomous LLM agents divided into three equal-sized economic classes face an escalating shared threat over 100 decision cycles across ten experimental conditions totaling 18,000 agent decisions. Lower-class cooperation is significantly lower than upper- and middle-class cooperation (64.3% vs. 95.0% and 97.3%; Cohen's d = 1.33, p < 0.0001), not from selfishness but because contributing threatens their survival. A random-action baseline confirms that 20.5 percentage points of the class gap are structural, while LLM reasoning widens it by an additional 10.2 points. Inequality deepens under shared threat (Gini: 0.470 → 0.563, d = 4.12, p < 0.0001), and when threat recedes, lower-class cooperation collapses to 30.8% while upper-class cooperation remains at 72.5%. These results suggest that the capacity for collective action is itself stratified by wealth, with implications for disaster policy, inequality research, and multi-agent AI safety.
No takes yet. Share an insight, caveat, or question.
Huynh The Dong Tran (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: