Our agent protocol languages (MAP and LCC) are intended to guide the
interactions of rational agents. By design, these languages
do not
enforce any particular model of rational behaviour on the agents.
Nonetheless, it is important to appreciate the kinds of rational
behaviours that our agents may adopt. Therefore, in this
talk I will
outline three popular definitions of rational agency: reactive agents,
practical-reasoning agents, and deductive-reasoning agents. I will also
sketch the connection to agent protocols.