The Role of Single PI Science….

I often get the question, especially when speaking to biologists, about whether there is still a role for the single principal investigator (PI) style of hypothesis-based research that was mainly the norm for the life sciences in the last century. My parents’ neuroscience lab at Cal Tech was certainly emblematic of that approach in the 1970’s.  In such research groups, a single senior person was the PI for a grant (or grants). There were well-defined hypotheses for each research project executed by a junior person and generally cross-laboratory collaborations were rare.

I’ve written about team science, fairly recently. And the advent of Internet-enabled open collaboration technologies such as Jupyter Notebooksalong with large-scale major research facilities that produce open data are certainly driving a trend. But I do think there is still a critical role for the small-scale single PI approach.

Why? First, from the standpoint of training new scientists, junior folks tend to get more intense mentorship is such groups. At least for some trainees, that’s clearly desirable. Second, there is a lot to be said for the classic Popperian experimental design. With a falsifiable hypothesis and with proper statistical analysis, I think it’s less susceptible to p-hacking. Finally, I would argue that massive scientific teams might tend to be more conservative, missing out on truly transformative research findings that might arise in the laboratory of a courageous single-PI.

Team science versus Single Investigator

Certainly across biosciences, a notable trend has been the lists of authors for single papers getting longer. This is especially true for the high impact journals and reflects the evolution of the practice of scientific research from individual investigator to large teams of scientists all working on various parts of a single question or problem. Part of this evolution is due to the need to use many methodologies to completely tell a single scientific “story” –in many cases considerably more techniques than any one single investigator can manage.

This team approach has been explicitly pushed in recent years, most saliently by the recently retired NIH director, Elias Zerhouni. There are some real problems however with the team approach. One of the most important is that maintaining quality control over the entire corpus of experiments that make up a team-authored paper becomes potentially challenging. An additional complication is that with large teams, who actually did what becomes opaque to the reviewer.
I’m not advocating a wholesale return to single PI science in biology–the subject matter has become too complex for many questions in the discipline. Rather, I’m urging a renewed appreciation for what can be accomplished in a single PI laboratory, where, in outstanding cases, a single creative mind can design an elegant set of experiments that like a fine gem, outshine the industrial output of large team labs.
Going further, it seems to me that with appropriate Science 2.0 sharing approaches, we may see a new renaissance of individual investigators re-using data produced by very large groups in imaginative ways that lead to real scientific progress.