The increasing number of independent IEEE 802.11 WLANs owned and managed by autonomous users has led to increased interference, resulting in performance degradation and unfairness. Performance can be improved by allowing these networks to operate on different channels. Due to the autonomous nature of the networks, a suitable channel selection scheme should be distributed, adaptive and require no explicit coordination. In this paper, we model the channel selection of WLANs as a non-cooperative game in a learning setting. Using a novel method of acquiring a disruption factor value, we propose a class of socially conscious channel selection schemes based on game-theoretic learning. These schemes are distributed, adaptive and are able to improve fairness without explicit inter-network communication. These features allow the WLANs to coexist in an interference-limited but non-cooperative environment. They also have the advantage of not requiring any modification to the existing 802.11 sta...