Seminar on Norms and Agreement Technologies

Start Date: 2009-03-15

End Date: 2009-03-20

Location: Dagstuhl

Country: Germany


The context of NorMAS WS “Normative Multi-Agent Systems” organized in Dagstuhl¬†was the perfect context for the meeting of our WG2 on Norms.

First, there was a lot of people, including very outstanding scholars in this domain and in various disciplines; but also including a lot of young scholars presenting their research. The level of the papers and of the discussion was very qualified. Second, the participation was remarkably interdisciplinary (not only logicians and computer scientists, but also philosophers, cognitive scientists, etc.); this is a fundamental feature and objective of our WG2. Third, the participation to our initiative (a panel with an open discussion) has been very successful; the large majority of people was there: around 40 people.

We have started with a presentation (by Cristiano Castelfranchi and Pablo Noriega) of our COST Action in general, and in particular of the part on Norms. After this we started our Panel. The participants were 5 of the official members of the WG: Thomas Agnotes (No), Guido Boella (It), Cristiano Castelfranchi (It), Pablo Noriega (Es), Timoty Norman (UK). The central questions to the Panel (very central for our COST Project and WG) were the following:

1. Architectural

Does a norm just select a-posteriori, just cut, the desires of the agents, or its intentions? Should we consider it as some sort of filter? Or does it just enter the decision process (transition from desires to intentions) by adding costs? (Possible sanctions)Or does the norm generate new imported goals (candidate intentions) that are not at all “desires”: spontaneous, endogenous, pleasant? Are norms a new¬† (social) source and origin of mental goals? Are norms aimed at generating goals (and ‘how’), and then an intention, in the agent? (Prescriptions, imperatives). Which is the right architecture for a ‘normative agent’?

2. Norm management

Do we need artificial systems able to interpret and rigorously apply our norms and rules, for surveillance and monitoring of human behavior in organizations. Wouldn’t this be too rigid?

Human organizations require some flexibility in interpretation and application. They even require some ‘functional violation’ and some executive autonomy, in case of conflict between norms or of conflict between the norm and the problem to be solved.

This is also one pressure for norm evolution and organizational adaptation.

How can we guarantee, with our agents, room for flexible interpretations, for negotiation about the right application, for some intelligent violation and its recognition, or for re-negotiating about norms and rules?

3. Interface

How to enhance user’s awareness of (and control on) norms expressed in formal, non-ambiguous, and machine understandable languages through suitable interfaces and advanced norm explanation techniques?

4. Emergence

How norms (and rules, and roles) do emerge from spontaneous self-organizing practices and tacit conventions? Should they always become formal and explicit?

Are we able to support this normative emergence and tacit agreements?

There has been an interesting discussion, with 8 interventions from the audience. The interest for our WG has been definitely improved; and we elicited several request for participation to our activities.