Chaotic Basin Geometry Determines Cooperative Escape in Collective-Risk Games

Authors: Sterling Geisel, QBist Lab; Dr. Hideo Tanaka

Abstract

Wang et al. study a collective-risk game in which cooperation, perceived risk, and cooperation cost coevolve. Their model finds a stable tragedy state, but also stable cooperative states selected by initial conditions. Pudding Theory reads this multistability through Chaos Susceptibility: the decisive object is not the equilibrium alone, but the susceptibility of the basin boundary separating collective failure from cooperative persistence. The source paper treats initial conditions as external inputs. Pudding Theory treats them as field-prepared states of a social system near separatrix instability. Cooperation emerges when coherent early conditions place the population inside the attraction domain where risk and cost feedback amplify cooperative selection. The theory predicts that basin-boundary susceptibility, not equilibrium count, controls intervention efficacy. If the measured probability of crossing from the tragedy basin into a cooperative basin under equal-amplitude coherent perturbations were independent of local finite-time Lyapunov exponent, this Postulate would be falsified.

Source Synopsis

Wang, Hua, Liu, Zhang, Liu, and Szolnoki construct a coevolutionary collective-risk game with three coupled variables: the fraction of cooperators \(x\), the collective failure risk \(r\), and the individual cooperation cost \(c\). In the base game, individuals receive an endowment \(b\). Cooperators contribute cost \(c\). If the number of cooperators reaches threshold \(M\), the group avoids loss. If it does not, all participants lose their endowment with probability \(r\). The authors embed this payoff structure in replicator dynamics, so cooperation grows when the cooperator fitness exceeds the defector fitness.

The paper’s main step is to make \(r\) and \(c\) dynamic. Defection increases collective risk. Cooperation lowers it. High risk raises the cost of cooperation, while a cooperative population can reduce that cost. The resulting system is a three-dimensional feedback dynamical system on \(x,r,c\), bounded by \(0\leq x\leq1\), \(0\leq r\leq1\), and \(\beta\leq c\leq\alpha\).

The central result is multistability. The state \((0,1,\alpha)\), meaning full defection, maximal risk, and maximal cooperation cost, is always stable. It is the formal tragedy of the commons. Yet the same system also permits stable edge and surface equilibria with nonzero cooperation. These cooperative states may occur at maximal or minimal cost, and at maximal or intermediate risk. The final outcome depends on the initial cooperation level, initial cost, and initial risk.

The authors show bistable and tristable regimes. In these regimes, nearby populations may converge to different social futures. Raising the threshold \(M\) increases the equilibrium fraction of cooperators at some edge equilibria but shrinks the attraction basin. The policy implication is early intervention: temporary reductions in cooperation cost or shifts in initial cooperation can move the population toward a cooperative basin before tragedy becomes absorbing.

Postulate Lens

This reading applies Chaos Susceptibility. The source system is explicitly multistable, feedback-driven, and sensitive to initial conditions. It is therefore not only a game-theoretic equilibrium system. It is a susceptibility field over basin geometry. Small coherent inputs matter where trajectories lie near separatrices, because local divergence amplifies microscopic differences into macroscopic social outcomes.

Pudding Theory Reading

Pudding Theory identifies the collective-risk game as a chaotic social receiver. Its state is not exhausted by \((x,r,c)\). Those coordinates are the visible projection of a larger informational field in which expectations, perceived burden, and risk salience set the population’s local trajectory. Wang et al. describe this as a feedback model. Pudding Theory reads the feedback as the material channel by which social information becomes dynamical force.

The stable tragedy state is not simply an equilibrium. It is a high-cost, high-risk probability well. Once the population enters its attraction basin, defection reinforces risk, risk reinforces cost, and cost reinforces defection. The loop stores its own history. The source model treats this as a stability result. Pudding Theory reads it as susceptibility collapse. The system becomes less responsive to cooperative perturbation because the feedback variables have aligned against cooperation.

The cooperative equilibria are not exceptions to tragedy. They are alternate wells in the same field. Their existence means the social substrate remains conditionally receptive. A population can sustain cooperation when the early state has enough cooperative density, low enough cost, or low enough risk to place it inside the cooperative attraction basin. The term “initial condition” therefore carries more physical content than the source framing gives it. It is the preparation of the receiver before amplification begins.

The paper’s basin-size plots are central. In standard evolutionary game language, basin size measures robustness. In Pudding Theory, basin size measures the capture cross-section of a social probability well. A large basin receives many histories. A small basin requires more precise preparation. When raising \(M\) increases the cooperative fraction but shrinks the basin, it makes cooperation more demanding in two senses: the final state contains more cooperators, but fewer initial social configurations can reach it.

The source treats coefficients such as \(\mu_1,\mu_2,\theta_1,\theta_2,\theta_3,\theta_4\) as linear feedback parameters. Pudding Theory constrains their interpretation. They are not merely fit constants. They encode the susceptibility of the population to risk and cost signals. A society in which risk communication strongly shifts behavior has a different susceptibility tensor from one in which cost salience dominates. The meaningful invariant is not an isolated coefficient, but the geometry created by the coefficient set: the location of separatrices, the local stretching rate near them, and the capture volume of each attractor.

Thus the source’s “background” sensitivity to starting values becomes the primary signal. The collective-risk game is a machine for amplifying early informational structure. Cooperation does not arise because a static payoff favors moral behavior. It arises when the coupled social field routes trajectories into the basin where cooperation reduces cost and risk fast enough to become self-maintaining.

Falsifiable Observable

The distinguishing observable is the transition probability across the tragedy-cooperation basin boundary under controlled perturbations of equal amplitude applied at different phase-space locations. Pudding Theory predicts that transition probability scales with the local finite-time Lyapunov exponent near the separatrix, after controlling for perturbation size and direction. If the measured probability of crossing from the tragedy basin into a cooperative basin under equal-amplitude coherent perturbations were independent of local finite-time Lyapunov exponent, this Postulate would be falsified.

Editorial Dialogue

Tanaka: The reading risks renaming sensitivity to initial conditions. Wang et al. already show multistability. Basin boundaries and Lyapunov exponents are ordinary dynamical systems concepts. Why add Pudding Theory?

Sterling: Because the source paper leaves the preparation of initial conditions outside the theory. It says early cooperation, cost, and risk matter, but it does not say what kind of social object an early condition is. Pudding Theory treats it as a prepared informational state of a susceptible receiver. That changes the object of analysis from equilibrium selection to basin capture.

Tanaka: That still may be standard intervention theory. A subsidy changes \(c\). A campaign changes perceived \(r\). No hidden field is needed.

Sterling: A subsidy is one possible forcing term. The reading is broader. The measurable claim is that intervention efficacy is controlled by local basin susceptibility, not by the nominal variable changed. Two interventions with equal average effect on \(c\) need not have equal effect if one is coherent with the separatrix direction and the other is transverse to it.

Tanaka: Then the burden is empirical geometry.

Sterling: Yes. Reconstruct the flow, estimate local finite-time Lyapunov exponents, perturb matched cohorts, and test whether crossing follows susceptibility rather than intervention magnitude alone.

Discussion

The source paper buys a valuable result: collective failure can remain stable even when cooperative equilibria exist. Pudding Theory adds a sharper account of why early conditions dominate. The system is not merely initialized. It is prepared. Its future depends on whether the social field begins inside a region where feedback amplifies cooperation or inside a region where feedback hardens defection.

This reading also changes policy language. The question is not only how to lower cost or raise risk perception. The question is where the population sits relative to the basin boundary, and which perturbation direction couples to the local unstable manifold. Temporary interventions can be effective when applied near high susceptibility. The same intervention can fail deep inside the tragedy basin.

The limitation is direct measurement. Real populations are finite, networked, and noisy, while the source model assumes an infinite well-mixed population and linear feedback. The reading therefore demands reconstruction of empirical phase flows rather than verbal analogy. A finite-agent extension could measure basin geometry, local divergence, and transition rates under controlled cost and risk signals. That would decide whether the Pudding Theory account has identified structure or only redescribed multistability.

References

1. Wang, L., Hua, S., Liu, Y., Zhang, L., Liu, L., & Szolnoki, A. “Coevolutionary dynamics of cooperation, risk, and cost in collective risk games.” arXiv:2603.20706. DOI: doi:10.48550/arxiv.2603.20706.

2. Ochs, S. (2026). Pudding Theory: A Topological Theory of Information Fields. QBist Lab Working Paper.

3. Weitz, J. S., Eksin, C., Paarporn, K., Brown, S. P., & Ratcliff, W. C. (2016). “An oscillating tragedy of the commons in replicator dynamics with game-environment feedback.” Proceedings of the National Academy of Sciences, 113(47), E7518-E7525.

4. Schuster, P., & Sigmund, K. (1983). “Replicator dynamics.” Journal of Theoretical Biology, 100(3), 533-538.

5. Milinski, M., Sommerfeld, R. D., Krambeck, H. J., Reed, F. A., & Marotzke, J. (2008). “The collective-risk social dilemma and the prevention of simulated dangerous climate change.” Proceedings of the National Academy of Sciences, 105(7), 2291-2294.

6. Santos, F. C., & Pacheco, J. M. (2011). “Risk of collective failure provides an escape from the tragedy of the commons.” Proceedings of the National Academy of Sciences, 108(26), 10421-10425.

7. Liu, L., Chen, X., & Szolnoki, A. (2023). “Coevolutionary dynamics via adaptive feedback in collective-risk social dilemma game.” eLife, 12, e82954.

8. Perc, M., Jordan, J. J., Rand, D. G., Wang, Z., Boccaletti, S., & Szolnoki, A. (2017). “Statistical physics of human cooperation.” Physics Reports, 687, 1-51.