AbstractThis paper formalizes the communication of agents with modal operators in arrow logic. A communication between agents consists of an agent's utterance and the other agent's perception, thus, both of the utterance and the perception are regarded as parts of a communication channel between agents. Information is regarded as a propositional content of a sentence. An information channel where information flows can be considered to be a program, in the sense that it gets an utterance as an input and puts an output to be a perception of some agent. In the real situations, th ere are so called miscommunications. Thus, the communication channel as a program may add some noise on information indeterministically. We implement the noises are some modal operators on information. We try to formalize the communication channels in arrow logic. In that, we especially pay attention to the following three problems: channel bottleneck, unreliable channel, and reverse information. This paper's contribution is two-fold. First, we formalize the theory of information flow, based on situation semantics, in terms of arrow logic. Secondly, we propose the theory of communication channels between agents by using arrow logic, where, classical modal operators like knowledge, belief, and perception are distributed on various places on the communication channel. We discuss the satisfiability and the applicability of our formalization, using the test principles by Barwise on this information flow model.