No doubt you have heard many times, from advisers, mentors, peers, and science-career pundits, that collaboration is valuable for research. But most of the advice I have encountered over the years has been general in nature, not oriented to the early careers of scientists. It's true that scientific collaborations are some of the best and lowest cost opportunities for scientists to broaden their experiences, establish new professional connections, and cultivate job opportunities. But which collaborations offer the most value for the participants' careers? And how much collaboration should a graduate student or postdoc engage in? These are good questions. Read on for some answers.

Collaboration's raison d'être

One big advantage of a collaboration is that, compared with your relationship with your adviser, it puts you on a more even footing. In most collaborations, specific team members take responsibility for specific aspects of a scientific project. Instead of a single "boss" with her hands on the controls, handing out assignments to underlings, a scientific collaboration is much more like a network of peers--even if some of those "peers" are graduate students and others are faculty members. The relationships a young scientist develops in a collaboration tend to be more egalitarian, less conflicted, and generally more positive than the more hierarchical relationships that exist within a laboratory. Ironically, by being accountable to more than one person you become more independent. Instead of being a laborer at the whim of a single individual (who isn't you), you now have a seat at the table.

Scientific collaborations are a principal method by which young scientists can cultivate their professional networks and develop job opportunities. A successful scientific collaboration almost invariably results in the opportunity to do more work together later. Traveling (either literally or intellectually) from one lab to another exposes you to the members of your collaborators' labs and other people in their host institutions.

Even remote collaborations can be valuable. But physically traveling to your collaborator's lab is a very good idea. There's no better way to become known to others and have your talents appreciated than being physically present for a while and sharing what you know. No matter how good you become, you never completely outgrow your beginner's status at your home institution. But scientists at that other institution can see you as you are now, at once fresh and mature. You might come up first in their minds when an employment opportunity opens up.

Identifying opportunities for collaboration

You develop opportunities for scientific collaboration the same way you develop job opportunities: by networking and keeping your eyes and ears open. Still, once they arise, such opportunities do not develop by themselves: You have to develop them.

I find that the most fruitful opportunities come during technical meetings, at which it is possible to hear dozens of talks and examine hundreds of posters as the authors stand by, ready to answer questions. In just a few minutes, you can discover whether the scientific grounds, interest, and resources for a collaboration exist. You can also get a feeling for whether the presenter is someone you would like to work with. By approaching a potential collaborator in person in a friendly, direct, enthusiastic, professional manner, you can make a good first impression: This is a peer.

Once you have established that the collaboration opportunity is genuine and valuable, then it's time to contact your principal investigator and ask permission. Act first, apologize later. Just don't give away any trade secrets that might get you in hot water.

The best collaborator

Any colleague or laboratory can collaborate, but I have found that the best choices for early-career scientists have the following characteristics:

1. They're established. By collaborating with seasoned scientific professionals, you can benefit from their reputation, wisdom, and experience. You might end up spending most of your time with grad students and postdocs in the lab, but you probably will get to know the head of the lab. And even if you rarely work directly with the PI, you'll still reap many of the advantages of her experience and reputation.

2. They're in a different type of institution from your own. Research universities, national labs, medical centers, and industrial settings are very different research environments with different (and often complementary) resources and opportunities. If you are a postdoc at a national lab (like I was), your environment is rich in infrastructure and equipment but often thin in person power. In contrast, academia tends to be rich in labor (graduate students and postdocs) but limited in infrastructure. During my postdoc, I cultivated several scientific collaborations for which I provided access to or results from the "big tools" around me at the national lab while my academic colleagues provided human expertise and labor to crunch data and analyze results. Other types of asymmetric collaborations can also be fruitful. One great example is when basic scientists collaborate with clinical scientists, perhaps at medical schools, giving the basic scientists (indirect) access to human-subject and clinical-trial expertise and leading to more efficient "translation" of research results into real-world therapies.

3. Pairing methodologies with problems. A particularly fruitful type of interdisciplinary collaboration can occur between an expert in a methodology or scientific technique and an expert on a scientific problem. Statistics, applied mathematics, spectroscopy, numerical modeling, and other computational expertise can be applied to a wide variety of scientific problems. Applied fields such as climatology, neurology, geophysics, and cellular biology all address specific problems that can be attacked with a variety of tools. If you are problem focused, consider how the tools of your colleagues can be brought to bear on the important problems in your discipline. If you are an expert in a specific technique or body of knowledge, seek opportunities to apply those skills to problems in other disciplines. My postdoc opportunity arose out of a collaboration between me (with a methodology: solid-state nuclear magnetic resonance spectroscopy) and colleagues at Lawrence Livermore National Laboratory, who had a problem (understanding the physics of hypervelocity impacts).

How much of a good thing?

How much of your time in a year should be spent on collaborative activities? There's no single answer; it really depends on your circumstances. My advice is 30%.

Fiske's Rule: 30% of a young scientist's yearly research activity should be with people outside of his or her adviser's research group.

Many grad students I have spoken to are surprised and a bit concerned about how high this number is. How can a young scientist afford that much time indulging in scientific collaborations with others?

As you know from my previous articles, I believe in taking charge of your professional destiny. Even the most enlightened and supportive PI cannot care as much about your career as you do. The graduate school and the postdoc are intended to take a young scientist on a journey from apprentice to peer, but the laboratory has other, often contradictory, priorities. Scientific collaborations are an important means through which you can seize the initiative. As I noted in an earlier column, early-career scientists must seek opportunities to step out of the subordinate roles they play to their advisers and initiate and create on their own. There is no other way to become a scientist.

Images. Top: NIH. Middle: Kelly Krause

Peter Fiske is a Ph.D. scientist and co-founder of RAPT Industries, a technology company in Fremont, California. He is the author of Put Your Science to Work and co-author, with Dr. Geoff Davis, of a blog (at phds.org) on science policy, economics, and educational initiatives that affect science employment. Fiske lives with his wife and two daughters in Oakland, California, and is a frequent lecturer on the subject of career development for scientists.

10.1126/science.caredit.a0900005