Source-linked AI summary
Information sharing promotes prosocial behaviour
Attila Szolnoki, Matjaz Perc
TL;DR
Information sharing may help address consequences of information being withheld or poorly integrated. The paper lets players on different networks share strategy information that changes their propensity to switch, and finds that sharing promotes cooperation beyond traditional network reciprocity.
Problem
The paper asks whether making information available can help mitigate unwanted consequences of information being withheld or not integrated into a broader picture.
Source: §1 · p. 2Method
Players share strategy information across networks, and matching strategies in the other network reduce the receiver’s propensity to change without altering its utility.
Source: §1 · pp. 2–3Results
Information sharing raises the defection temptation at which cooperators die out, with group-based sharing further improving cooperation in the prisoner’s dilemma; its effect is more modest in the public goods game.
Source: §3 · pp. 5, 11, §4 · p. 12Takeaways & Limitations
Strategy information alone can promote prosocial behaviour through the synergy between network reciprocity and correlated behaviour across networks.
Source: §4 · p. 12Information from large groups may be diluted and unable to guide strategy changes relevantly.
Source: §4 · p. 12
Abstract
from arXiv · showhide
More often than not, bad decisions are bad regardless of where and when they are made. Information sharing might thus be utilized to mitigate them. Here we show that sharing the information about strategy choice between players residing on two different networks reinforces the evolution of cooperation. In evolutionary games the strategy reflects the action of each individual that warrants the highest utility in a competitive setting. We therefore assume that identical strategies on the two networks reinforce themselves by lessening their propensity to change. Besides network reciprocity working in favour of cooperation on each individual network, we observe the spontaneous emerge of correlated behaviour between the two networks, which further deters defection. If information is shared not just between individuals but also between groups, the positive effect is even stronger, and this despite the fact that information sharing is implemented without any assumptions with regards to content.
1. Introduction
Information sharing can help address unwanted consequences of information being withheld or poorly integrated. The paper proposes sharing strategies between networks to affect players’ willingness to change, without changing cross-network utilities or evaluating the transmitted information.
1. Introduction
- Information availability supports data-driven research across social and natural sciences, while withholding or failing to integrate it can have unwanted consequences.The introduction gives examples including mobility, disease spread, online popularity, social movements, and political campaigns.Source: §1 · p. 2
- Network interdependence can make information in one network affect behaviour in another, beyond the cooperation-promoting role of network reciprocity within a single network.The introduction frames this as a possible source of additional support for cooperation across network layers.Source: §1 · p. 2
- Sharing strategies from another network scales a player’s probability of changing strategy, while utilities remain determined within its own network.A player becomes less likely to adopt a neighbor’s strategy when corresponding players in the other network already use the player’s current strategy.Source: §1 · pp. 2–3
- The information is strategy-neutral: cooperators and defectors can transmit it, and the model does not filter or evaluate its content.The authors state that sharing strategy information alone is sufficient to promote prosocial behaviour.Source: §1 · p. 3
2. Model
The model studies evolutionary games on two networks where information from one network adjusts players’ propensity to change strategy in the other. It compares no sharing, sharing from a corresponding player, and sharing from a group, across prisoner’s dilemma and public goods settings.
2. Model
- Players play the prisoner’s dilemma with four nearest neighbours on a periodic square lattice, with mutual cooperation yielding R = 1 and temptation T > 1 rewarding defection against a cooperator.Players begin as cooperators or defectors with equal probability; the same parametrization is used on both networks.Source: §2 · pp. 3–4
- Players imitate a randomly chosen nearest neighbour within their own network, with adoption probability determined by a Fermi function and scaled by w_x.The scaling factor w_x depends on strategies in the other network and incorporates information sharing.Source: §2 · p. 4
- Model S sets w_x from the strategy of the directly corresponding player, while model G uses that player and its neighbours in the other network.In model S, w_x is minimal for matching strategies and maximal when they differ; model G varies w_x linearly with the number of matching players in the reference group.Source: §2 · p. 4
- The public goods game tests robustness to group interactions: cooperators contribute 1, defectors contribute nothing, and group contributions are multiplied by synergy factor r > 1.The lattice uses overlapping groups of size G = 5, with each player in five groups.Source: §2 · pp. 4–5
- In the public goods game, each player’s total payoff is the sum of payoffs from the five groups in which they participate.Each player is surrounded by k = G − 1 neighbours and belongs to g = G groups.Source: §2 · p. 5
- Simulations use random sequential updating, system sizes L = 800 to 6400, and equilibration of up to 10^6 Monte Carlo steps.Each player on both networks has one opportunity on average to pass its strategy per Monte Carlo step.Source: §2 · p. 5
3. Results
Sharing strategy information across networks expands the prisoner’s dilemma conditions supporting cooperation, while correlated behavior and cooperative clustering help sustain it. The effect also appears in the public goods game, but is weaker under group interactions.
3. Results
- Information sharing raises the temptation to defect at which cooperators survive and dominate, while shrinking the mixed cooperator–defector phase.Models S and G outperform the unshared model I on these cooperation thresholds.Source: §3 · p. 5, Figure 2 · p. 6
- Cooperation initially declines more in S and G than in I, with a longer delay in G before recovery and dominance.The near-extinction dip in G requires sufficiently large systems for reliable results near phase transitions.Source: §3 · pp. 5–6
- In models S and G, excess cross-network cooperator pairs work with within-network clustering to overcome defection beyond network reciprocity alone.The excess correlation is measured as P(CCe) − P(C)²; model I has none.Source: §3 · pp. 6–8, Figure 3 · p. 7, Figure 4 · p. 8
- In model G, vulnerable defectors rise after aggressive spreading loses cross-network support, while cooperators advance slowly and compactly across both networks.The rise in vulnerable defectors is followed by an overall increase in cooperation.Source: §3 · pp. 9–10, Figure 6 · p. 10
- Information sharing also promotes cooperation in the public goods game, but its impact is modest because group interactions limit spatial fluctuations and vulnerable defectors.The effect is especially modest from single-player sharing in S to group-based sharing in G.Source: §3 · pp. 10–11, Figure 7 · p. 11
4. Discussion
Information sharing promotes prosocial behaviour beyond traditional network reciprocity through correlated behaviour across networks. Group-sourced information further improves cooperation odds, though group interactions and very large information sources constrain its effectiveness.
4. Discussion
- Information sharing promotes prosocial behaviour beyond traditional network reciprocity through synergy with spontaneously correlated behaviour across networks.Source: §4 · p. 12
- Group-sourced information further improves cooperation odds, but information sharing is less effective in games governed by group interactions.Source: §4 · p. 12
- Very large information sources may dilute guidance, while local information sharing resembles reputation-based decisions.Source: §4 · p. 12