The model includes different formulations how agents make decisions in irrigation games and this is compared with empirical data. These experiments are 5 person lab experiments with monetary incentives. Participants made investment and extraction decisions in 10 rounds. Participants varied in being upstream vs downstream. There were two treatments limited and full communication. With limited communication we found more inequality. The aim is to test different theoretical models, especially explaining effect of communication. We compare different null models (selfish, cooperative, random) and two theoretical models: conditional cooperation and social values.