Coordination between multiple autonomous agents is a major issue for open multi-agent systems. This paper proposes the notion of Behavioral Implicit Communication (BIC) originally devised in human and animal societies as a different and critical coordination mechanism also for artificial agents. BIC is a parasitical form of communication that exploits both some environmental properties and the agents' capacity to interpret each other's actions. In this paper we abstract from the agents' architecture to focus on the interaction mediated by the environment. To implement BIC in artificial societies two environmental properties are necessary: the "observability" of the software agents' actions and the "traceability" of the environment. The goal of this paper is to address the first property defining a model of observation mediated by the environment. From the viewpoint of the agents, a crucial distinction is proposed between the environment and the artifacts they can use.
Both the environment and the artifacts can affect the visibility of agents in several respects and facilitate different forms of implicit communication between them.
A typology of environments and examples of observation based coordination with and without implicit communication are described.