Evolution of cooperation in the snowdrift game among mobile players with random-pairing and reinforcement learning

• Cooperation is researched in the evolutionary PDG among mobile players.
• A player interacts only with the nearest neighbor each turn.
• Players make decision based on expected utility.
• Cooperation is promoted in a large parameter range.
• Moving ability has five types of influences on cooperation level.

The evolutionary spatial game in a mobile population has attracted many researchers of biological, social and economic sciences. Considering some facts observed in the real world, this paper proposes a novel spatial evolutionary snowdrift game model with movable players. In this model, one player interacts only with the nearest neighbor in each turn, and makes decision in a reinforcement learning way. In a very large range of the parameters moving ability enhances cooperation, but under some special condition, velocity heavily depresses cooperation. Some explanations have also been given out by investigating the strategy-change behavior of players. The findings may be helpful in understanding cooperative behavior in natural and social systems consisting of mobile agents.