All interviewing in research roles tends to be based on a) a 1 hour seminar given by the candidate b) current people talking to the person 1:1 in conversations that usually sound more like peer-to-peer gossip than interviews. I am assuming here a PhD or postdoc level person, and that basic disciplinary skills, a decent publication record, and knowledge of the literature and trends in their field are a given. Suitably watered down, my answer applies to BS/MS people as well, taking on a research role.

For a decent, if somewhat obsolete look at the higher levels of research thinking, look at the slim little book, "A PhD is not Enough" by Peter Feibelman. Another great resource is Richard Hamming's fantastic talk at Bell Labs from 1986, "You and your research": http://www.chris-lott.org/misc/k... .... both shaped my thinking about research very significantly.

This whole answer is for "average" researchers (who will all nevertheless be extremely smart and creative by the standards of the general population). All rules can and should be broken if you spot a true genius.

There are 4 basic things to look for:

a) Novelty of vision: the person is thinking about interesting problems they want to work on that show some signs of originality and ambition and isn't entirely a process-oriented person who is just very good at a particular type of problem. No good researcher is every content to be merely a process tool for someone else's agenda, so if you find somebody who says "I know everything about machine learning and the methods of the field, but there is no problem I am interested in working on right now, give me something to do" something is wrong (the person may be a genius, like Paul Erdos, who'd show up at friend's places saying "my mind is empty" and work on anything the friend wanted to work on... but among the ordinary minds, "empty mind" is usually a sign of burnout or incompetence).

b) Collegiality: the person can get along, collaborate, is interested in others' research, and naturally comes up with potential ways to apply their skills to others' problems. This is CRUCIAL. Every researcher has ideal "perfect" problems they'd love to sink their teeth into. Most are practical enough to just keep them around for the right opportunity, and perfectly willing to get interested in the problems actually available to be worked on.

c) Pragmatism/resourcefulness: the person understands funding challenges and knows the basics of what it takes to get research funded in the particular environment, shows willingness to adapt their interests to the needs of funding sources, and has at least a passing familiarity with the major sources of money and preferably, people who control the money. But too much of this is a danger sign. If a person shows signs of knowing more about the intricacies of DARPA processes and current funding politics than about interesting research problems, paradoxically, they'll be BAD at raising money. The funding agencies don't like researchers who are too polished as "operators" at the expense of research ability. Ultimately, getting the money is just an important and time-consuming sideshow. The research is and should be the main act.

d) Drive: a sign of relentless energy, which can be hard to detect because many highly-driven researchers come across as soft-spoken, quiet and low-energy in face to face interactions. But they'll put in 80 hour weeks in the lab.

You should also get a read on what the person's basic external motivator is (outside of "I just want to do interesting work" which is true and tautological of everybody in research): papers? patents? building prototypes? winning prizes? getting new things to market? building a large program and managing others? working with cutting edge equipment? If they are purely internally motivated, they are likely too philosophical to survive. If they don't like at least one of the "external games" they may crash and burn bemoaning the crassness of it all. You need to find the candidate's external "addiction." Mine happens to be model-building (simulations).

If you can't offer that external motivator, the person is not a fit and will eventually leave/fail. R&D people can be surprisingly low on self-awareness, and in denial about what they actually want. There are a lot of people for instance, who'd rather be in academia writing papers who are sitting in industrial R&D labs being forced to put out fires downstream. If you can't find a way for these people to feel like academics while doing the R&D work the company needs, they won't be motivated. Conversely there are people sitting in academia who hate writing papers and would rather spend all their time engineering difficult prototypes, and acting like startup CEOs.