There is a promising-sounding article in the February 2008 Communications of the ACM, Sharing Knowledge, by Peter Marks, Peter Polak, Scott McCoy, and Dennis Galletta.
How managerial prompting, group identification, and social value orientation affect knowledge-sharing behavior.
This is a simulated study of how these factors+ affect knowledge sharing in a lab setting amongst undergraduate business students. The goal is to say something about how to improve knowledge sharing with a slant toward knowledge management systems -- systems where one "deposits" their knowledge. I wonder if the same factors apply when the "system" is more of an organizational guideline to share with one another, rather than with a system.
The study design was purely around testing people's behavior with varying probabilities of return for themselves and the group over many "rounds." They also added "managerial prompts" for more sharing, and they factored the participants into one group with an affinity motivation and another without.
Overall, the authors demonstrated that managerial prompting creates additional sharing. For the other factors, the results were not as clear, partially because they did not recruit evenly on the social-value-orientation axis.* The results do seem to show that pro-social people tend to share more, which makes sense.
One thing I would love to see is a simulation of how people respond to the behavior of others in their group. For instance, along with being given information about how you can help other group members, wouldn't it be good to know how they had helped you in previous rounds? People are social critters, so if they've been helped by their colleagues in the past (even if that colleague has shared generally, not directly), I suspect they would be more likely to share in return.
+ Managerial prompting is the corporate / manager reminders that knowledge sharing is important. Group identification is the cohesiveness of the group that is expected to share knowledge. And social value orientation is the general sense that individuals have for individualistic or collective behavior.
* A funny note is that while they were looking for people with both pro-self and pro-social tendencies (the "social value orientation") the motivation to get students to participate was extra credit on their grades: a pro-self selector. And they found that most of their participants were pro-self. Surprised?