Sorry for the delay on chiming in - I've been cleaning up stuff after a conference .

Uboat, it sound to me like the collectivist-Individualist dichotomy you are using comes out of Geert Hofstede's work. The model he has come up with is generally aimed at producing models of national culture, and it is based on 1960's style anthropology (which he readily admits, BTW).

When we look at "cultures" of any type, defining what level we are looking at is crucial for a whole slew of reasons. So a national culture can be quite different from an organizational culture or a corporate culture.

One of the key differences is how a particular sub-national group organizes particular activities and how that organization is both accepted (and validated) at the national level and constructed as a separate "role cluster" (sorry, I'm using a lot of technical jargon here ). Put simply, certain tasks get grouped together and the people who perform them are allowed, and sometimes required, to act in ways that are different from the general social norms.

Usually, the decision to organize in a particular way evolves out of a) the groups' experiences with performing those tasks and b) the more general social perception of how well those tasks are performed. Think of it this way, you rarely see changes in organizational cultures until there is a failure crisis which calls the social legitimacy of the group performing them into question (Andrew Abbott goes into this really nicely).

Getting back to your specific question about inter-agency stuff, you might want to look at some of the material on cross-cultural communication. The simplest way to get some decent communications going across different organizational cultures is to treat a joint team as a project team and then use a series of team building exercise to construct a team culture.

Cheers,

Marc