What is a PI responsible for?

Some further thoughts on all this STAP fraud, after reading posts and comments at Drugmonkey and Michael Eisen.

Faking data and committing fraud is the cardinal sin of science. It’s a really big deal, and I honestly think ORI is toothless when it comes to consequences (oh you robbed a bank? that’s cool, keep it, but we’re going to pay someone to follow you around and make sure you don’t rob another one for 5 whole years). Fraud occurs for many reasons, but I think that in many (or even most, it’s impossible to say, as I’m sure most fraud is never detected) cases, senior authors and co-authors can be mostly blameless. If someone is working in your lab and gives you convincing but fake results, it can be extremely hard to detect. Even harder if they are giving you a subset of real results, just those that support a hypothesis, for example. There is a lot of trust placed in people who are under enormous stress and career pressure. It’s bound to happen at some frequency.

On the other hand, it is clear from the online discussion (and the real life of any science trainee), that there are many labs in which cheating is enabled by a system of rewards and expectations, created by the PI, that certain results are what is necessary for praise and career advancement. Data fakers are still to blame, but obviously there is some additional culpability here.

Zooming out even further, the media “witch hunts” that result from high-profile fraud cases result from the absurdity of the hype that surrounds science in the first place. If journals, institutions, and the media (and we have to admit it, us scientists too) didn’t drive this machine of fame, celebrity all the attendant dishonest bullshit that comes with it, perhaps there wouldn’t be a need for such vicious recrimination when someone we just anointed a Great Science Hero turns out to be a flawed person. Despite university press releases, the “News” arm of the glamour mags, and the growing annual cycle of Cash Prizes and Top 10 Lists, science almost never happens in stunning leaps of individual genius, and progress never depends on any one person or lab.

Fraud–faking science results–should be punished. Jobs and grants are zero sum–everyone who gets one by cheating stole from an honest scientist (or a less able cheater, I guess). It should probably end your research career. Let’s not pretend that’s a “witch hunt” or a disproportionate response. We think doping athletes should be banned, embezzling bankers should be fired and lose their licenses. What’s the difference? Why should we bother or care about “rehabilitating” Marc Hauser? He had a chance to contribute and blew it, there are many, many others deserving of that chance.

But science is done in teams. If the faker is a trainee, it’s harder to tease apart what happened, and questions inevitably arise. Mistakes and negligence should have consequences too, but these are complicated: the PI and co-authors are somewhere along a spectrum of being innocent victims of the fraudster, of having been negligent in oversight, or having actively contributed through (non-fraudulent) bad leadership or research practices. However, anyone who reads Retraction Watch knows that except in cases where it is proven that the PI actively participated in the fraud, there are essentially no career or funding consequences for them. They may take a credibility hit for a while. Is that bad? I don’t know.

For me, it raises questions about our expectations of a PI. As I wrote, almost everything said about Obokata should’ve have raised red flags instead of garnering praise. Can someone running two labs and holding an administrative position, managing perhaps 20+ scientists and staff even fulfill the minimal expectations we have for 1) Instilling the right system of values and incentives that will minimize the temptation to commit fraud; 2) Be familiar enough with the people, experiments, and data in their lab(s!) to have a fair shot at noticing when something might be amiss?

These are management roles that, in my view, cannot be delegated if 1) You are the grant holder and 2) Your name is last on the paper. The inevitable accumulation of funds and trainees that comes from the kind of reward system we have—grants beget grants in a positive feedback loop—leads to very large labs that become the epicenters of disciplines and subfields. But is it even plausible that these PIs are competently doing their jobs and meeting their oversight responsibilities?

---