What goes on in a closed-door defense session?

Posted November 10, 2015 by Mike Wiser in Graduate School, Information / 0 Comments

Source: https://xkcd.com/1403/

Last week, I defended my dissertation. While I knew how to prepare the written dissertation, and the public talk, the third component – the closed-door session in which my committee examined me – was mysterious. I’d asked about 8 people over the years what their closed-door sessions were like, and all of them told me that they couldn’t remember. So, while my experience might not be typical, it is at least an experience, and thus potentially a useful data point. Your mileage may vary.

To put this into context, I’ve been a member of a biology department, and a program in Ecology, Evolutionary Biology, and Behavior. As of my final committee meeting, my committee wanted a 3-chapter dissertation; I ended up writing a 4th chapter as well. The first 3 chapters were within a wet lab system (the LTEE in the Lenski lab), while the 4th extended the analyses in a computational system (Avida). Chapters 1 and 2 had been previously published; chapters 3 and 4 had not. The fact that 2 of my chapters had been published likely influenced the whole proceedings substantially, as it’s less likely that a committee will want to interrogate work that has already undergone peer review. In my talk, I essentially just discussed 2 of the chapters: one published, and one not.

I’m quite comfortable with public speaking, which turns out to have been more helpful than I anticipated: PowerPoint crashed on me multiple times during my talk (which it hadn’t during my practice talks) but I was able to keep discussing the material while reloading the presentation and getting to the right place. Immediately after the public questions, my advisor recommended we take a few minute break, so I left the room to get something to drink and hit the restroom. When I came back, my committee members were talking in a small circle, and told me that they’d just been discussing that there was no particular topic they felt that they had to specifically grill me on (which implies that in some defenses, there are), and so the questions would be more open-ended.

Things I learned about my closed-door session from doing it:

It can be more of a conversation than an exam. Several years ago, I felt like my committee meetings shifted in an important way: I got the sense that I was being asked questions to elicit my thoughts and opinions about a topic because my thoughts were considered valuable. Previously, I’d had a sense of being asked questions to determine if I could come up with the right answer. This makes sense to me. My committee members are experts in their fields. But on the specifics of my dissertation, I have likely thought about this topic more intensely over the past several years than they have (even than my advisor, if for no other reason than that he’s overseeing multiple students, each with their own topics. Since my work is a subset of what he’s been working on, it makes sense that I’ve been more intense on those topics recently than he has). When it goes right, the PhD process turns a student into a colleague. By the time of my defense, I’d gone through the transformation from learning what other people have found to finding something new myself, and having thought about what that means.

It’s easy to over-prepare. I made a separate PowerPoint file for each of the chapters, consisting of all of the figures from that chapter, in order. I also found previous talks I’d given about each of the chapters – primarily, conference presentations – which I reviewed in the days before the defense and had available in case my committee wanted me to go through them. During my closed-door session, I didn’t project any slides at any point, nor did I give any presentation at all. I was asked questions, and did my best to answer them, but it was markedly more informal than I thought it might be.


Source: Source United States Space Command: Vision for 2020 (public domain)
Source: United States Space Command: Vision for 2020 (public domain)

You don’t always have to shoot flies with cannons. The running joke in my lab for years has been that my approach to lab work isn’t so much shooting flies with cannons, but obliterating them with orbital space lasers: overwhelming amounts of data and controls to take into account all of the possible objections I can foresee. I was ready with a number of analyses I did on some of the projects but which didn’t go into the published manuscripts due to space limitations. I was ready with alternative versions of graphs for the unpublished data, showing new analyses I’d done since sending my committee my written dissertation. None of those were needed. Now, maybe it’s possible that none of them were needed because I’d done all of them, and my committee knew me well enough to know that this is my standard operating procedure. But I think it’s more likely that questions about the big picture are just more interesting to most people than questions about the specific details, and my committee would rather spend their time talking about things which interest them.

Admitting you don’t know something is OK (just like in comps/quals). At one point, my advisor asked a follow-up question to one posed by another professor, essentially trying to get me to talk about a specific other paper. I blanked on what paper it was, and admitted it. When he mentioned the author, I was then able to launch into an explanation of what that study had done, but I had missed the specific connection he was trying to get at. None of my committee seemed to mind this, even though I was self-conscious about not having known the answer.

In the end, I was mostly asked about the context of my work, and extensions of it. Topics that I can recall questions about at the moment:

  • Connections between some of my findings and certain engineering applications
  • Whether I’d like to see similar studies funded. This led to a discussion about the relative value of replicating findings in similar circumstance versus making choices in other systems that would allow us to the extend the generalizability of findings (haploid vs. diploid, sexual vs. asexual, single cell vs. multicellular, etc).
  • A variety of simple numerical simulations we could test beyond the one I’d already identified as a possible future direction
  • Applying the general patterns I’d seen to fossil data, likely in response to a question I’d gotten during the talk
  • Using related but somewhat distinct mathematical functions for different data sets, again in response to a specific question from the audience
  • Why I’d chosen to do statistical transformations in some but not all cases in my data, and whether it might be worth reanalyzing the cases where we hadn’t transformed anything, so that we’d have a more general case

I had opinions about these topics, expressed them, and explained why I felt the way I did. I don’t know if my committee agreed with me, but I do know that they seemed to accept these views as both reasonable and sufficiently grounded in the field to be acceptable as answers. And that, fundamentally, is what I think is necessary for a defense.  I demonstrated that I: had identified unanswered questions in my field, had designed and run experiments to address those questions, and knew how my data fit in with what we already knew in the field.  Next step: getting the written document past the formatting review from the graduate school.

Mike Wiser

Mike is completing his PhD in Dr. Rich Lenski's lab, studying the predictability and repeatability of the evolutionary process, using experimental populations of E. coli. He has also been collaborating with several researchers in the Devolab to study the same processes in Avida.

More Posts

Leave a Reply