Source-linked AI summary
Does strong heterogeneity promote cooperation by group interactions?
Matjaz Perc
TL;DR
The paper asks whether strong heterogeneity promotes cooperation in group-interaction games as it does in pairwise games. It studies public goods games on three graph types with uniform or exponential payoff distributions and finds that uniform distributions maintain higher cooperation, challenging the assumed generality of heterogeneity’s benefits.
Problem
The paper addresses limited evidence on whether strong heterogeneity promotes cooperation in group-interaction games as it does in pairwise games.
Method
The study simulates public goods games on square, triangular, and random regular graphs using uniform or exponential player-specific payoff scaling.
Results
Uniformly distributed public goods are more successful at maintaining high cooperation than exponentially distributed public goods.
Takeaways & Limitations
Group interactions may be less susceptible than pairwise interactions to cooperation promotion by strong heterogeneity, so strongly heterogeneous states warrant reexamination in other games.
Takeaways & Limitations
Quenched heterogeneities can create long-lived patches and power-law extinction, complicating classification of final stationary states.
Abstract
from arXiv · showhide
Previous research has highlighted the importance of strong heterogeneity for the successful evolution of cooperation in games governed by pairwise interactions. Here we determine to what extent this is true for games governed by group interactions. We therefore study the evolution of cooperation in the public goods game on the square lattice, the triangular lattice and the random regular graph, whereby the payoffs are distributed either uniformly or exponentially amongst the players by assigning to them individual scaling factors that determine the share of the public good they will receive. We find that uniformly distributed public goods are more successful in maintaining high levels of cooperation than exponentially distributed public goods. This is not in agreement with previous results on games governed by pairwise interactions, indicating that group interactions may be less susceptible to the promotion of cooperation by means of strong heterogeneity as originally assumed, and that the role of strongly heterogeneous states should be reexamined for other types of games.
1. Introduction
The paper examines whether heterogeneity that promotes cooperation in pairwise games does so in group-interaction public goods games. It motivates this question through the complexity of group reciprocity and studies heterogeneous payoffs across several interaction graphs.
- Motivation: Pairwise games are easier to reciprocate because each interaction involves only two players, whereas group interactions require tracking many players.The public goods game exemplifies group interactions, making reciprocal responses more difficult.
- Motivation: Cooperation requires sacrificing personal benefits for social welfare, while defection prioritizes individual fitness regardless of societal consequences.In the public goods game, cooperators contribute to the public good and defectors contribute nothing.
- Motivation: Failure to maintain cooperation produces a tragedy of the commons, although experiments show that humans cooperate more than standard expectations predict.This motivates identifying mechanisms that sustain cooperation in public goods settings.
- Related work: Additional strategies, including abstention, punishment, rewarding, and exploration, have been shown to promote cooperation.These mechanisms expand competition beyond cooperation and defection.
- Related work: Heterogeneity among players has become an important mechanism for promoting cooperation, especially through scale-free networks.Much prior work has focused on games governed by pairwise interactions.
- Research question: The study tests whether strongly heterogeneous states promote cooperation in group interactions using public goods games on square, triangular, and random regular graphs.Payoffs are distributed using either uniform or exponential scaling factors across graph structures with differing spatial and clustering properties.
2. Setup
The paper models public goods games on structured interaction graphs with overlapping groups and quenched player heterogeneity. Monte Carlo simulations compare payoff distributions and strategy updating while noting difficulties caused by slowly evolving heterogeneous states.
- Interaction structure: Players occupy square-lattice, triangular-lattice, or random-regular-graph sites and belong to G overlapping groups.Groups contain G players, and each player belongs to g = G groups.
- Game setup: Cooperators contribute 1 to each public good, whereas defectors contribute nothing.Players are initially assigned cooperation or defection with equal probability.
- Heterogeneous payoffs: Individual scaling factors introduce payoff heterogeneity through uniform or exponential distributions.The scaling factor h_x determines how heterogeneity is introduced, with χ sampled uniformly from the unit interval.
- Caveat: Quenched heterogeneity may produce Griffiths phases and patches where one strategy survives for very long times, complicating classification of stationary states.Patch-size effects can make extinction follow a power law and create technical difficulties in identifying final states.
- Caveat: Varying quenched heterogeneities on an extremely slow timescale can preserve conclusions while accelerating extinction.The authors present this variation as a possible way to alleviate analytical difficulties associated with quenched heterogeneities.
- Simulation procedure: Monte Carlo updates calculate two players’ payoffs and attempt strategy transfer using a Fermi probability with K = 0.1.The update sequence selects player x, selects neighboring player y, and attempts transfer from x to y.
3. Results
Across square, triangular, and random regular graphs, uniformly distributed public goods promote cooperation more effectively than exponentially distributed public goods. The advantage persists despite differences in spatial structure and clustering.
- Square lattice: At η = 0.01 and η = 0.1, the square lattice requires r = 3.74 for cooperators to survive, falling to r = 0.45 at η = 10 under uniform distribution.For η = 1.0, the threshold is r = 2.45; values below r = 1 become possible when some h_x exceed G.
- Square lattice: Uniformly distributed public goods produce higher stationary cooperator densities than exponentially distributed public goods on the square lattice.The exponential distribution also yields a substantially smaller pure-cooperator region and causes ρC ≈ ρD from η ≳ 0.7 within the mixed phase.
- Triangular lattice: On the triangular lattice, uniformly distributed public goods outperform exponentially distributed public goods despite the graph’s high clustering coefficient.The main phase-diagram features remain identical to those on the square lattice for the same heterogeneity type.
- Random regular graph: On the random regular graph, exponential distributions produce lower ρC values and no pure-cooperator region, making the failure of strong heterogeneity most obvious.The D → C + D transition line remains largely unaffected by the payoff distribution.
4. Summary
The study tests whether strong heterogeneity promotes cooperation in group-interaction games as it does in pairwise games. Across three regular graphs, uniformly distributed public goods maintain higher cooperation than exponentially distributed public goods, challenging the generality of the pairwise-interaction result.
- 4. Summary: The study examines whether strongly heterogeneous states promote cooperation in the public goods game, which is governed by group interactions.This question is motivated by prior findings that strong heterogeneity facilitates cooperation in pairwise-interaction games.
- 4. Summary: The authors compare uniformly and exponentially distributed payoffs on square, triangular, and random regular graphs.These graph types test the conclusion across different interaction structures.
- 4. Summary: Uniformly distributed public goods maintain higher cooperation than exponentially distributed public goods across the interaction graphs studied.This differs from previous results reported for pairwise-interaction games.
- 4. Summary: The findings question whether strong heterogeneity generally promotes cooperation in games governed by group interactions.The conclusion holds irrespective of the interaction graph type examined.