How does your lab grow?

What's your lab's composition? In the recent issue of Neuron, Nobel laureate David Hubel argues that the shape of a lab has changed radically, and for the worse. In today's labs, he writes,

Leading each group is a senior scientist, a full professor, or an assistant professor who can expect to be promoted to tenure in a few years. Under the leader come post-doctoral fellows, graduate students, and a few research assistants. This form of organization is in sharp contrast with what prevailed a generation ago, when a group consisted usually of one person, or two or three, and there was less distinction, if there was any, between group members.

He goes on to say that as a result of today's division of labor, the PI's time is taken up with grants, committee work, teaching, refereeing, and lab management--rarely experiments. In contrast, when Hubel started working at Harvard,

Everyone in the group worked alone or with one other postdoc, on projects they thought up themselves and carried through independently.

Thus everyone was responsible for his own experiments, analysis, and writing, and would use other group members as critical sounding boards.

Hubel argues that the old way was preferable because:
a) Postdocs learned much more independence, as they had to think through projects and data themselves rather than being directed.

b) One thinks best about science when one does one's own experiments, as opposed to dividing the bench and grant chores between two people.

c) Scientists, who are selected for their skills at benchwork, get to continue doing benchwork.

He doesn't point out the corollary to (c), that because good scientists may be bad managers, this system avoids placing such people in positions of power over scads of postdocs. But to me this is a critical point. Numerous blog posts from multiple bloggers detail their frustrations with poor management, whether it comes in the form of the PI overseeing antibody supplies or being unhelpful with projects and manuscripts.

This is partly a rehash of an old argument: should PIs keep doing experiments? The axis of opinion ranges from PhysioProf's point that PIs are good at thinking big picture, so it's best for them not to waste time on experiments, to Ms. PhD's argument that those who can't do, shouldn't teach (= run a lab). (I do think that PIs who continue with experiments generally have a good grip on their labs, while those who don't have more heterogeneous track records.)

Still, the crux of Hubel's argument, and one that I find interesting, is that scientists should be given greater responsibility for their own research earlier on, as postdocs rather than as assstant professors. Too many postdocs must follow their PI's project "suggestions" and don't have an opportunity to design a project from scratch. This produces a lot of disgruntled postdocs and Mini-Me PIs.

Some institutions, like the Max Planck, do offer junior scientists independent research positions leading only one or two other people. My advisor held this type of position and says it was scientifically the best thing he ever did. But very few such jobs are available today.

Of course the current arrangement is probably higher throughput, because savvy PIs can lead squads of unimaginative postdocs through an exciting research program. And there are doubtless many people who prefer the current design, not least because it facilitates interdisciplinary work. But I agree with Hubel that more independent small research programs populated by relative youngsters (in their 30s!) who would otherwise be postdocs would be a welcome alternative.

I wish that more institutions offered Max Planck-style fellowships. For many, perhaps most, people, these jobs would serve as springboards to more traditional faculty positions, while fostering independence and creativity, and eliminating the "pedigree" aspect of the current postdoc system.

And perhaps a few scientists would discover along the way that they preferred it.