Ever since the seminal work of Searle, two components of interaction protocols have been identified: constitutive rules, defining the meaning of actions and regulative rules, defining the flow of execution, i.e. the behavior the agent should show. The two parts together define the meaning of the interaction. Commitment-based protocols, however, usually do not account for the latter and, when they do it, they do not adopt a decoupled representation of the two parts. A clear distinction in the two representations would, however, bring many advantages, mainly residing in a greater openess of multi-agent systems, an easier re-use of protocols and of action definitions, and a finer specification of protocol properties. In this work we introduce the notion of behavior-oriented commitment-based protocols, which account both for the constitutive and the regulative specifications and that explicitly foresee a representation of the latter based on constraints among commitments. A language, named 2CL, for writing regulative specifications is also given.