How does an individual's cognition change a system which is a collective behavior of individuals? Or, how does a system affect an individual's cognition? To examine the interplay between a system and individuals, we study a cognition-based network formation. When a network is not fully observable, individuals' perception of a network plays an important role in decision making. Assuming that a communication link is costly, and more accurate perception yields higher network utility, an agent decides whether to form a link in order to get better information or not. Changes in a network with newly added links affect individuals' perception accuracy, which may cause further changes in a network. We characterize the early stage of network dynamics and information dispersion. Network structures in a steady state are also examined. Additionally, we discuss local interactions and a link concentration in a frequently changing network.