Pathways to Research

The other day I had a slight disagreement with a colleague of mine about the best advice to give to new PhD students about how to tackle their research. Talking to a few other members of staff about it subsequently has convinced me that there isn’t really a consensus about it and it might therefore be worth a quick post to see what others think.

Basically the issue is whether a new research student should try to get into “hands-on” research as soon as he or she starts, or whether it’s better to spend most of the initial phase in preparation: reading all the literature, learning the techniques required, taking advanced theory courses, and so on. I know that there’s usually a mixture of these two approaches, and it will vary hugely from one discipline to another, and especially between theory and experiment, but the question is which one do you think should dominate early on?

My view of this is coloured by my own experience as a PhD (or rather DPhil student) twenty-five years ago. I went directly from a three-year undergraduate degree to a three-year postgraduate degree. I did a little bit of background reading over the summer before I started graduate studies, but basically went straight into trying to solve a problem my supervisor gave me when I arrived at Sussex to start my DPhil. I had to learn quite a lot of stuff as I went along in order to get on, which I did in a way that wasn’t at all systematic.

Fortunately I did manage to crack the problem I was given, with the consequence that got a publication out quite early during my thesis period. Looking back on it I even think that I was helped by the fact that I was too ignorant to realise how difficult more expert people thought the problem was. I didn’t know enough to be frightened. That’s the drawback with the approach of reading everything about a field before you have a go yourself…

In the case of the problem I had to solve, which was actually more to do with applied probability theory than physics, I managed to find (pretty much by guesswork) a cute mathematical trick that turned out to finesse the difficult parts of the calculation I had to do. I really don’t think I would have had the nerve to try such a trick if I had read all the difficult technical literature on the subject.

So I definitely benefited from the approach of diving headlong straight into the detail, but I’m very aware that it’s difficult to argue from the particular to the general. Clearly research students need to do some groundwork; they have to acquire a toolbox of some sort and know enough about the field to understand what’s worth doing. But what I’m saying is that sometimes you can know too much. All that literature can weigh you down so much that it actually stifles rather than nurtures your ability to do research. But then complete ignorance is no good either. How do you judge the right balance?

I’d be interested in comments on this, especially to what extent it is an issue in fields other than astrophysics.

15 Responses to “Pathways to Research”

  1. […] “The other day I had a slight disagreement with a colleague of mine about the best advice to give to new PhD students about how to tackle their research. Talking to a few other members of staff about it subsequently has convinced me that there isn’t really a consensus about it …” (more) […]

  2. There’s the story about a professor writing a famous unsolved mathematical problem on the chalkboard. A student can’t make the class but comes in later to write down (what he thinks is) the homework and turns it in the next day. I think this actually happened at least once (which a quick web search will verify; I’ll post a link if I have time if no-one else has). The moral of the story is the same as that in yours.

    IIRC Knuth also included some famous unsolved mathematical problem, without mentioning the fact, in the list of problems in one of his textbooks.

  3. With regard to the original question: I think it depends on theory/experiment, the field, the project, the student and (unfortunately) the funding agency. When I was at Jodrell Bank, I noticed that the students did a literature review at the beginning of their thesis work. I still think that’s a good idea in general. (However, keep in mind that at the time almost all theses at Jodrell Bank were observational, not theoretical. This is probably more often a better idea for non-theoretical work, while for theoretical work it might be good not to know too much.) Probably the most important thing is that the time is there if the student wants to do such a thing; in other words, there shouldn’t be pressure to get started right away, though perhaps that should always be an option.

    • telescoper Says:

      The problem with doing the literature review at the start like that is that there are so many papers these days, and some subjects change so quickly, that it might be out of date before you really get anywhere. I think you’re better off if you just try to keep up as you go along.

      • Again, it depends on the subject etc, but I think a literature of this sort shouldn’t include any “normal papers”, much less new normal papers, but rather books, review articles and a few classic papers, i.e. the same things you or I might read if interested in a field but not working in it. Things like Coles and Ellis. 🙂

      • The point of the literature review at Jodrell Bank is not so much the background reading. Much more important is the writing part. The sooner a student begins to write, the better. If the purpose was only getting up to speed on the research, doing hands-on (or brain-on, in case of theory) work would be better. For undergraduates it is called problem-based learning.

      • telescoper Says:

        That’s a good point.

        I always tell my students that a PhD Thesis has to have both analytic and synthetic aspects: you have to tackle a problem yourself and produced publishable results, but you also have to put that work in the context of the rest of the field. It’s the second part at which most students struggle, so getting them to learn how difficult it is early on is a good thing.

  4. Bryn Jones Says:

    I didn’t ever supervise a PhD student myself, but if I had done, I would have directed the student to some solid small research project at the outset, to be carried out alongside a literature review but with a slightly higher priority. My view is that an early small research project introduces the student to research skills at the outset. It is a learning process, and might also contribute a useful chapter to the thesis. Many practical research skills are best learnt in the context of doing research.

    My own experience as a PhD student was unfortunate. I was put on to literature reviews in two areas at the outset, but data for those projects did not materialise and I had to begin from scratch in a different field right at the end of my first year. Virtually nothing that I did in my first year contributed to my thesis work: it was effectively a write off. I knew a lot about those two fields I’d studied, but they were of no relevance to my PhD, and I did not research in those two areas in my subsequent career.

    The need to make early progress is particularly important in those countries that try to condense a PhD into three years, such as Britain. There is no time to waste.

    • I agree with Bryn (clearly all those years sharing an office must have had some effect). The ‘small starting project’ approach has been what I’ve done for all my PhD students. So far it seems to be working, and it gets them an early paper. Literature reviews in isolation don’t work effectively in my experience; people need some concrete thing to attach them to…

    • Bryn Jones Says:

      We shared an office for only about three years!

  5. John Peacock Says:

    For long-term success in research (which means changing field one or more times, probably), a deep technical foundation is essential, so the question is how to acquire this early enough that you can start research while you are young and particularly creative. When I was a student, I think the UK system did pretty well in this regard: there were plenty of grammar schools putting well-prepared kids into university, so that you could emerge at age 21 knowing a lot and ready to get stuck into research. I remember in those days we used to talk dismissively of the US and Continental systems where people spent so many years doing courses that they were dangerously close to 30 before getting down to serious research. This seemed pointless: I was given postgraduate lectures, but no-one got much out of them because they weren’t rigorously examined. So once you stop learning under the pressure of undergraduate-style exams, better to just get on with research, and realise you have to take responsibility for filling the holes in your education.

    But today the problem is that (in the UK, at least) the requisite technical foundation is no longer there. Even with the extra MPhys year, I think present undergraduates are probably less well trained than those entering a PhD a generation ago – and there is more to know than there was then. To achieve the same readiness today to undertake PhD research that you or I had, one probably needs to continue courses with detailed examination right through the MPhys year and a further year beyond. OK, so that’s never going to happen, but a good second best would be to scrap the MPhys project – I know the students like it and prefer it to more courses, but in the long term it isn’t good for them. Now you often get PhD students who only want to work extending their MPhys projects, because they lack the confidence in fundamentals to strike out in other directions. In this respect, I think the pendulum has swung, giving a big advantage to the US graduate school and its insistence on continuing rigorous examination. But the only way of getting that in the UK would be to lengthen the PhD, which will never happen while the emphasis is on training many more students than academia needs.

    But taking as a given the fact that the UK system doesn’t have a big programme of lectures with detailed examination (well, in astronomy; particle physics theory is probably different), the best you can do is start students off on a ‘learning by doing’ introductory project and keep emphasising to them that it is their responsibility to list the things they don’t know and make a point of reading up about them. But this is hard when there is so much of such material. I’m starting to suspect that the only solution is to add the extra year as a separate MSc diploma in between undergraduate and PhD (the very successful German system, which they ironically seem to be abandoning).

    • telescoper Says:

      I’ve long advocated the 3+2+3 Bologna approach to BSc – Masters – PhD. Indeed, I think we should have many fewer PhD students and many more masters students.

      I don’t really think an extra Masters year is needed for most astronomy PhD projects, but in the more theoretical bits of cosmology and particle physics it probably is necessary. The problem of course, is funding and I don’t see any serious thinking at the level of government in postgraduate education, so I don’t see a satisfactory solution coming along in the near future.

  6. John: “I’m starting to suspect that the only solution is to add the extra year as a separate MSc diploma in between undergraduate and PhD (the very successful German system, which they ironically seem to be abandoning).”

    The old German system had the master’s degree (Diplom) as the first real degree. The Vordiplom covered about the same stuff as a Bachelor’s degree, but was a purely internal examination and a requirement for higher exams. The normal time for the Diplom, was 5 years, 4 years of lectures etc then a one-year thesis (in my experience, more work than a typical MSc, but I understand there are various types of MSc).

    Peter: “I’ve long advocated the 3+2+3 Bologna approach to BSc – Masters – PhD.”

    The Bologna thing looks similar, with the first 3 being the Vordiplom, the 2 being the higher-level courses and the thesis and the 3 being the doctorate. (Actually, the Vordiplom was closer to 2, making it 2+3+3.) However, as John notes, it is being abandoned. The problem is not the number of years in each phase, but what is done during this phase. The way Bologna is being implemented in Germany (though it varies widely, made worse by the fact that the states, not the national government, are in charge) has a number of disadvantages: many more exams and a more time-tabelled schedule. One could have kept the old system and given it some new Bologna clothes, essentially making the Vordiplom the Bachelor degree. For a variety of reasons, this didn’t happen.

    To sum up: Yes, the old system is being abandoned, and yes that is bad, but the new system isn’t that different in the way it is structured. Rather, the contents of each phase has changed for the worse.

    The main problem with the new system is the official Bachelor’s degree. Does anyone need it? Industry seems to prefer the old Diplom, which of course represents more work. Is a Bachelor-level degree in physics meaningful? Certainly not for research. Those wanting a more hands-on, industry-usable approach would be better off going to a polytechnic (a Fachhochschule in Germany). With the old system, there was more flexibility in the schedule, one could take advanced courses before getting the Vordiplom (or put off the Vordiplom exams until a convenient time. Since the Bachelor’s degree is now external, and not just internal to let the university decide if the student can continue, there are more exams and a stricter schedule.

    German universities are, or were, in general quite good. Why did they abandon the successful system? Mainly political pressure. In other words, if they didn’t, the Sun would have a headline: Germany going its own way. Again. (This is certainly not beyond the imagination of a newspaper which had a headline declaring Oskar Lafontaine the most dangerous man in Europe.) More “serious” politicians are just more discreet. Same with the Euro: a plebiscite would never have gotten a majority, but apparently there was no choice: joining the common currency was apparently a requirement for the WWII victors to agree to re-unification. German political parties tend to be more consensus-conscious than the general population, probably because they have to deal with the reaction of foreigners. (The recent circumcision debate highlights this: there is a large majority of the population which think it should be banned, but only the Linkspartei (i.e. communists) among the parties agreed. When rabbis publicly say that this is the greatest threat to Jews in Germany since the Holocaust, most politicians tend to accomodate them.)

  7. re “…especially to what extent it is an issue in fields other than astrophysics”

    “… There are several phases in the process leading to “find” new math. and while the “checking” phase is scary and involves just rationality and concentration, the “creative” first phase is of a totally different nature. In some sense it requires a kind of protection of one’s ignorance since there are always billions of rational reasons not to look at a problem which has been unsuccessfully looked at by generations of mathematicians”.

    Alain Connes, `Advice to the beginner’

  8. I seem to recall someone (I think it was Nathalie Deruelle) would given beginning graduate students a long list (I want to say of the order of 100, but I might be mis-remembering) of problems on the topic that the student was to work on (e.g. black holes). The problems were graded in difficulty and complexity so that the initial problems were textbook problems or problems found in published papers, and the final problems were research problems. Some of the questions included references if they were needed. This seemed to me to be a nice way to get a student of theory from the textbook/classroom type problem having a well-defined answer, to the more open ended research problem by way of the literature. It seemed to serve both purposes being discussed here.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: