The answer, as you might expect, was greater candor and its corollary, greater confidence and openness to learning. Good teams simply communicated better, and, in a learning environment, that meant surfacing and talking about mistakes.
In a discussion about the topic with HBS Working Knowledge, professor Edmondson summarized her findings thusly:
In well-led teams, a climate of openness could make it easier to report and discuss errors—compared to teams with poor relationships or with punitive leaders. The good teams, according to this interpretation, don't make more mistakes, they report more. When I suggested this to physicians involved in the study, they were skeptical. Their response was understandable: With a research grant for the purpose of identifying the error rate, this idea was decidedly unwelcome. My interpretation of the data suggested that we might not be finding the definitive error rate—and further errors might be systematically underreported in certain units but not others. Their skepticism forced me to work hard to develop ways to support my proposition, which ultimately they came to see as reasonable, if not obvious in retrospect.
Once again, we see that learning in adults means supressing instincts for self-protection, defying organizational incentives to conform and be "team players," and ignoring ingrained concepts like division of labor and roles/responsibilities.
This is from a 2006 working paper on the subject, "When Learning and Performance Are At Odds" from Professor Edmondson and her collaborator, Sara Singer:
...Effectively conducting an analysis of a failure requires a spirit of inquiry and openness, patience, and a tolerance for ambiguity. Such an inquiry orientation is characterized by the perception among group members that multiple alternatives exist, frequent dissent, deepening understanding of issues and development of new possibilities, filling gaps in knowledge through combining information sources, and awareness of each others’ reasoning and its implications(Argyris et al., 1978). Such an orientation can counteract common group process failures. Learning about the perspectives, ideas, experiences, and concerns of others when facing uncertainty and high stakes decisions, is critical to making appropriate choices.
Looking at this through the prism of innovation, you can see how using the whole disorderly team, how arguing and soliciting dissenting views is essential. Innovation means confronting the unknown, the complex, the ill-defined. Mistakes are to be expected, not avoided. Confronting, embracing failure, then gathering the entire teams's viewpoints on what didn't work and how to fix it, then stepping back and trying a different tack, is essential. Locating dead ends and understanding failure quickly and changing course leads to faster innovation development, lower cost and higher probability of eventual success.