Small world yields optimal public goods in presence of both altruistic and selfish cooperators.

Empirical studies have shown that individuals' behaviors are largely influenced by social conformity, including punishment. However, a coevolutionary theoretical framework that takes into account effects of conformity on individuals' punishment behaviors has not been put forward yet. Herein we propose a coevolutionary game model to extend the theory of cooperation with conformity in spatial public goods game by considering pool punishment, as well as two converse feedback modes of conformity that strongly affect cooperators' punishment behaviors. We focus on how different parameters and spatial structures govern evolutionary dynamics on three different kinds of networks by employing mean-field analysis based on replicator dynamics and Monte Carlo simulations. On regular lattices, defectors are overall extincted since cooperators, especially selfish cooperators, have great evolutionary advantages due to strong network reciprocity, and at the same time the number of altruistic cooperators decays. Conversely, abundant shortcuts in regular random networks lead to the prevalence of altruistic cooperators, but cooperators suffer from free-riding behaviors of defectors. Of particular interest, we find that small-world topology can simultaneously help cooperators successfully outperform defectors by means of strong network reciprocity, and enable rich contacting opportunities with defectors to facilitate the expansion of altruistic cooperators. Therefore, we clarify that small world is the optimal topology subject to the dominance of altruistic cooperators.