A Conversation with David Plaut

From Apprentice to Mentor




JF:
We had a student lunch with Barry Stein and we had an interesting discussion on the relationships between advisors and graduate students. There were two contrasting approaches we talked about. One was the advisor who checks with the student every day and who talks about how things are going. As well as the approach that Stein's own advisor used, which was to show him the lab and to show him the grant and then let him sink or swim. I was interested in how was your relationship with your advisor?
DP:
Well, it shifted and I think that worked for me. My relationship with Geoff, Geoff Hinton, actually changed a lot over the course of my graduate degree. Early on it was a bit like what you describe, where he had written a grant, there was a lot of ideas, he said: "read these couple of chapters and see what you are interested in." He had a connectionist simulator he had written, "play with this a bit." There was actually another graduate student of his entering at the same time. He sort of carved up a set of projects, "Why don't you go work on this and you can go work on that." Very hands-on, a lot of interaction with him and meeting with him all the time. Very much kind of an apprenticeship, where he really was leading the issues and I was understanding the approach. And was for the first year or so, that was very much the style. It actually was useful, because it allowed me to see how he structured a research project and how he decided whether an approach was or wasn't worthwhile. I would come back and talk with him about results and he would shape the direction of things.

But, gradually, as I gained more competence, as I understood some of the issues, and as my own interests diverged a bit from what he had laid out, he sort of eased away from that, so that our meetings actually changed in their character, with less of almost lecture-style, of his explaining to me what the issues are, and moved toward discussing and arguing what the issues are. And with Geoff Hinton, who I think I said is one of the most brilliant people I know, to get to the point where I could actually come up with an idea he didn't have already, it took me... I forget, but certainly more than a year. I can remember that point of raising some issue that he hadn't thought of as a great success. But it very much shifted from an apprenticeship to an equal-level colleague kind of interaction, where I shaped more of the direction of the work and what is interesting and he more advised me, in the sense of giving me feedback, rather than directing, in the sense of really dictating or really strongly indicating what direction things should go in. And that leads to the point, in the later stages of my work, where I was mostly telling him what was going on. I would come to him and explain to him what the issues were and I had moved actually in a direction that was somewhat far removed from his specific sort of burning research questions. And so, at that point, I really was operating fairly much as an independent researcher. But it was very good for me that that was a gradual transition. It wasn't that he expected me to be independent from the beginning, nor that he expected to micro-manage my research in the later stages.


JF:
How do you see yourself now in the role of the advisor? Would you like a similar kind of pattern?
DP:
I certainly think that that style is sort of my priors, it's where I start from. But I think one of the important aspects of advising is that every graduate student is different. Not only because they may come with different backgrounds but because their own personal style of work is very different. Some people may come with a tremendous amount of expertise in some area they're going to work in and it would really be inappropriate to be telling them what to do - they are already competent in deciding what to do in that. Some people come with expertise in a different area than what they are now going to work on and they really do need more hands-on guidance in learning the ropes of that methodology. So, it certainly depends on where someone is starting from in their own development.

But also you sort of have to learn what's most comfortable and what's most productive for someone, that some people really do want to be left to play and want to come back with questions, but pretty much feel comfortable exploring on their own and can shape that process sufficiently to be productive. It's one thing to say: "Oh yeah, I'll go try all these things", but that is not what a PhD is. A PhD is actually learning how to do research, learning to put together a research program, and learning to be an independent scientist. And that involves a kind of discipline as well as a kind of creativity. In fact, I would say that it's by and large mostly discipline, it's really taking a problem and really pushing it through. So, some people really can carry that off more or less by themselves, or to a greater degree by themselves. Some people, maybe people who haven't had much research experience to begin with, start off wanting to or benefitting more from more structure early on. Maybe starting with a project that has already been fairly well developed and pushing that through; then maybe generalizing it for themselves. It's not really a matter of ability or skill, it's really a matter of where someone is starting from and what kind of interactions they're comfortable with. So, in the interactions I have with students, I try to gauge my level of involvement with their own level of development and their own style and to the level of the work that's being done. Early on in a project maybe there needs to be more detail and guidance and later on I try to give them more room to develop it themselves.


JF:
Now, as one who is pretty close to the experience, I want to know what advice you would give to someone who has just finished their PhD.? What is out there? What opportunities for funding? What kind of strategies should a starting scientist take in order to carve out their place in the world?
DP:
I think that it is very important, particularly in interdisciplinary work (which I certainly find the most interesting and the most fruitful) - there is a danger in interdisciplinary work, which is to become a jack-of-all-trades, to become a little bit versed in a lot of things without having complete mastery, or at least substantial mastery, of a particular one - I think that it is very important to start with one methodology that you can fully employ, which you can use to get answers to questions. So, you can take some domain and, whether it's computational modeling, or cognitive experimentation, or neurophysiology, or whatever, I think it's important to start with a methodology that you have full mastery over, so that you can ask questions and you and others can believe the results you get. And then from that you can go on to develop other methodologies, other skills, that complement your current skills; extend your own understanding of the literature in other areas that might relate to the kinds of questions you can already ask. And start from a foundation of strength and develop the breadth, rather than try initially, early on, to cover a lot of things. It's not to say that that can't be done, but it's much harder and runs the risk of putting you in a position where you are not quite on top of any of the methodologies. It is critical to not be in that position because - exactly for some of the points you raised - to be able to develop a research program, to get funding, to publish in good journals, you really have to be methodologically sound. And that, I think, needs to be in place in at least one method before going on to others. It also gives you a theoretical starting place from which to evaluate other work, other approaches, to tie into your current work. If you try to be too broad initially, you run the risk of going-with-the-wind, making it harder to critically evaluate whether some piece of work really is new. How does it relate to past work? Should you believe the results of this particular paper? Can you really go through the methods section and evaluate the validity of the work? I'm a firm believer of interdisciplinary work, but, for me at least and I think that this is true generally, I think it needs to be grown from, grounded in, a particular methodology.
Back to the index. Back to the index.

Got a suggestion?
Send us a note! 		Have and idea to share?
Post it to the CNBC Newsgroup!