Many existing studies on human learning pay almost exclusive attention to how individuals learn. Unlike those studies, we examined influence of social structures on knowledge acquired by societies using computer simulations. We compared four types of social networks, namely regular, random, small world, and scale-free networks. When individual differences and the principle of homophily (i.e., people who have similar beliefs tend to have close relationships with each other) exist in societies, the societies would acquire pareto-optimal knowledge. We also investigated influences of highly connected individuals on knowledge acquired by societies. The results inarguably indicate that highly connected individuals play important roles in social learning, setting the standards for what type of knowledge to be acquired by societies.