After a relatively painless month or so of writing, I’ve finished my thesis. I’m yet to be examined, so whilst I can’t comment on how to write a good thesis, I’ve had a few thoughts on the process of getting from start to finish. Given the surprising lack of information on writing PhD’s (which is baffling since everybody in science either has one or has to write one), I thought I’d collect together a few things I wish I’d known earlier.

Before you start

Take notes throughout

Your thesis discussion provides you with a tabula rasa on which to scrawl your thoughts about your field. Time and time again, I’ve found that the most valuable insights into the frailties of a particular approach come when you first learn about it. These little voices are sometimes subdued once you’ve worked in the field for some time. Operating within a certain framework for too long, you come to accept its assumptions more easily (perhaps because desires and beliefs are entangled?).

I wish I’d kept a single word document (or Evernote note) with my initial reactions to all of the ideas that would go on to form integral parts of my thesis. Once you’ve become expert, you could then sift through these naive impressions to separate out the ignorant from the insightful. These might then form the useful material for your discussion.

Similarly, in your introduction you will need to provide a reasonably approachable explanation of the key concepts in your thesis. Capturing the aspects that helped you understand the idea is one of the easiest ways to do this. Once more, capturing these key points as you go along would make your life much easier! Similarly, if you’ve spent some time teaching during your PhD, try to keep track of what works and what doesn’t – this might help separate the explanatory wheat from the chaff.

Read some theses

Most institutions have an online archive of theses; many authors also provide them on their personal websites. Reading the theses of people you admire and respect is a good idea, and even a pleasure (there’s a certain pseudo-intimacy which comes with reading the product of 3-6 years of somebody’s life). Reading around gives you an impression as to the diversity of lengths and structures that are acceptable, which is surprisingly large. You can even read quite far away from your own field: if the author has done a really good job, the introduction ought to be intelligible even if the material is unfamiliar (see e.g. David Mackay’s 89 pager).

If you’re in cognitive neuroscience and fancy reading a few, I’d recommend taking a look at Laurence Hunt’s (beautifully written, very deep introduction, largely focused on two experiments) and Demis Hassabis’s (written with an impressively confident and visionary tone [foreshadowing his future ambitions], focused on putting forwards a particular idea with numerous experiments in support of it). I also dipped into Peter Smittenaar’s and Helen Barron’s, both of which are excellent.

Sharpen your tools

Broadly speaking, there are three tools you’re going to need to use to write your thesis (assuming you’ve finished analysing your data): a word processor, a reference manager, and some sort of graphics program for your figures.

Word processing

In some fields LaTeX is the norm. In ours, we use Word, so that’s what I can offer advice about.

Thing is, you think you know how to use Word, right? I too considered myself pretty competent, and then proceeded to learn a new thing about Word every 3 days whilst writing my thesis. Here are the things you should master, before you really get cracking:

Styles

I always thought these were an irritation. They are actually super useful. You probably know the concept – tagging different pieces of text with a style that you can then alter in order to instantly change the formatting of all of the text associated with that style – but their function extends beyond that. In particular, they’ve very useful for grouping together similar elements (e.g. all of your chapter titles), which makes numbering + table of contents generation a dream. At the very least, you should set up:

• A series of headings styles (h1,h2,h3 etc.) to provide chapter headings, subheadings, etc. If you start with word’s built in ones, these will be usefully numbered.
  • TOP TIP: you’ll find that you want your Chapter headings to be effectively numbered (because you want the subheadings to be 1.1,1.2 for chapter 1, 2.1,2.2 etc. for chapter 2), but you probably just want them to read ‘Chapter 1: ….’ rather than ‘1. Chapter 1: …’. You can neatly hide the number associated with a heading by selecting the number, right-clicking, selecting ‘Font’ and then ticking the ‘Hidden’ button.
• A style for your references
• A style for your captions
• A style for tables
• A style for your footnotes

Captions

Similarly to the styles – the real utility of these is that they automatically number and you can swiftly generate a table of figures. You can set these up to number according to your chapters, so that they inherit the numbering from your Header-1 style (so chapter 1’s first figure is 1.1, chapter 2’s is 2.1). Slightly non-intuitively, on the Mac the caption insertion comes under the ‘references’ tab. You can label both your figures and your equations in this manner.

Cross-references

I had a few supplementary figures which I didn’t want to label like the figures, but I still wanted to include references to in the Table of Contents (TOC). You can achieve this by inserting Bookmarks and then inserting Cross References in the TOC. These will automatically update if the page numbers shift (as they frequently do as you add and remove things).

TOC’s

Somewhere near the end you’re going to need to insert a TOC and perhaps a list of figures. This is a breeze if you’ve set up your styes and captions appropriately. Again on the Mac, TOC’s are listed under the References tab on the toolbar.

Track changes

Well no. Don’t track changes. With a thesis of 200 or so pages and several swimming pools’ worth of blood, sweat, and tears, tracking changes is a bad idea. Poor old Word can’t cope with the complexity of a month or so tracked changes. It’s a much better idea to just save lots of different versions as you go along. I had a terrifying experience where Word started refusing to save the twelfth draft of my thesis. Track changes was the culprit.

Reference manager

I had written a few papers during my PhD, so I was fairly familiar with my reference manager of choice (Papers by Mekentosj). However, there were a few hiccups along the way that are worth reflecting upon:

Graphics

For figures from data, my advice (inspired by Dave Redish’s dictum to new PhD students, as reported by Zeb KN) is that you try and rely upon Matlab or Python or R or whatever you’re analysing in to do all of the grunt work. Fiddling in Illustrator seems like a good idea, but if you need to change anything, you have to do everything again (side-note: for a glimpse of what algorithmic graphics might look like, check out Stop Drawing Dead Fish by the incredible Brett Victor).

In Matlab, check out this tutorial on making publication-ready figures. I’d advise editing your startup file to alter the default figure properties to make them better looking and easier to read (e.g. changing font sizes, get rid of horrible grey background). For complex multi-panel figures, I found Panel to be a useful toolbox.

For other figures, you can use Illustrator or GIMP etc.. I don’t have anything particularly useful to say here, but some time savers:

• Make all of your figures the same width. That way they won’t shrink or grow when you put them into a word document and adjust them to fill the page, and so your font sizing will remain consistent.
• Some useful illustrator shortcuts (on Mac):
  • Cmd K allows you to adjust the ‘nudge’ size i.e. the distance something moves when you press the arrow key. Very useful for shifting between gross and detailed work
  • Cmd- and Cmd+ to zoom in and out
  • Hit V to change the current tool to ‘Move’
  • Cmd alt ; locks/unlocks your guides [NB you should use guides: here’s a guide-guide]
  • X toggles whether you’re selecting the fill or stroke of an object
  • Shift-X switches the stroke and fill (so if you have black stroke, white fill, you’ll have the opposite after hitting shift-X)
• Illustrator also has text styles – like Word, this can be useful for swiftly altering lots of text (e.g. your axis labels or sub-figure numbering)

Actually writing your thesis

Hopefully by this point in your PhD you’ve got a reasonable idea about how best to sit down and work. I found that the same principles applied here. Some people like to use micro-scheduling methods e.g. the Pomodoro technique, in which you complete a series of short work-rest bouts designed to ensure total focus whilst working. I generally don’t find these useful unless I’m doing something extremely boring. Most of the time, I found that chunks of about 2 hours worked better for me, and that the earlier I started in the day, the better.

One of the things I thought about a bit were ideas about fatigue and disengagement (I read Rob Kurzban’s superbly titled ‘Why everyone (else) is a hypocrite’ a while ago and it’s great; here’s a related review paper he wrote with Joe Kable). I figured that increasing the value of work, and decreasing the value of alternatives would make life easier. The concrete implementation of this was putting on the most interesting music I could afford to listen to whilst still working reasonably well. For deep work this meant Bach or, if I was falling asleep, Glitch Mob; for lighter stuff (editing figures, checking references) I listened to stuff with lyrics that I knew well (e.g. Bon Iver). Cups of tea are also useful for increasing the momentary value of a state; if you’re drinking a lot of tea, green tea is good to prevent becoming totally wired. If I was still working at 7pm, I often tried to shift to more discursive work and had a glass of wine or a beer (if you’re in Bristol, buy this).

If I was having difficulty focusing, I sometimes used Noisli for pleasant woodland sounds. It’s a good way to persuade yourself that you’re in the Alps when really you’re in Bloomsbury.

The darker side of ‘decreasing the value of alternatives’ – I wrote my thesis at home with my parents, and didn’t tell any of my friends that I was home. So I really had very little to do other than writing a thesis and other wholesome things (chop firewood, run, swim, and play the piano). I think this made it easier to focus.

Finally…

One of the last things you need to do is write your acknowledgments. I found this to be a very enjoyable and fulfilling experience (I think I now appreciate why practising daily gratitude is good for life satisfaction). It’s a chance to think about the people who helped you along the way, and the journey you’ve been on. Its the warmest, most personal part of the thesis – and probably the only bit that your friends will read :).

Good luck, and let me know about your own experiences + advice.