Samuel Bird

Morale in research: Daily deflation vs Palpable perspective

Now that I think about it, my research today was a success.

I had an idea last night, about a way that I can very easily generalise a particular variational Ansatz to capture a desired effect, which I want to capture but am as of yet unable to. Knowing that most research ideas don’t pan out, I told myself to not get my hopes up and to remain level-headed, but, despite my protestations, my hopes did get raised. A mediocre nights sleep and some morning experiments later, the idea didn’t pan out. I played with it a bit, realised I had an error in my implementation, raised my hopes, and ran it again. Still no cake. This happened a few more times.

What happened afterwards was a bit of a blur. I think I brainstormed a bit, drew a logic tree, tried to distill what I learned from this result, and then derived a few possible next steps. One being to try a more complicated, but natural follow-up, Ansatz, working off of one hypothesis, and one being an alternative type of calculation which would verify a different but not independent hypothesis. By this point though I was tired, morale was running thin, and I ended up hacking away on autopilot at the alternative calculation without any definitive result or feeling of direction. In hindsight I still think the idea is solid but better left for a new day.

The scenario I’ve just described isn’t exactly a rare occurrence. This happens fairly often to me, and to others as well I hear. In fact, in a sense, it’s an ideal day of research. I came in with a concrete idea that I wanted to test, I tested it very quickly, and am now able to learn from it, and use the new information to guide my next step. This is pretty much what research is like.

And yet, I feel a little deflated.

Day to day deflation

Part of this is that I had my hopes raised. While not ideal, and one always hopes to maintain the cool stoutness of a good scientist, the fact is we have emotions and that’s not really avoidable. So what? I liked my idea. What can I do?

Another important reason for feeling deflated is simply fatigue. Spending hours thinking hard about something is very tiring. The more intense the concentration, the more tiring. This is why regular breaks, fresh air, and not working 24 hours a day are all important. Now, I didn’t go crazy today, but I did do 6 and a half hours of research, with about an hour of other work mixed in. Spending that long on hard thought typically doesn’t happen, at least not continuously. I’m also not sure it’s possible without breaking it up more, at least for me. As I said, the last hours were spent in a kind of blur, which was maybe productive but not so directed, and probably not very efficient. This is in contrast to the fast-moving and efficient earlier hours of the day.

Fatigue can develop over days as well though, not just within a single day. Taking regular days off helps a lot. In Littlewood’s miscellany I remember he writes about how he started taking Sundays off and remarkably found that he got more done in the week due to how many new ideas he had on Monday. He goes even further and decides to take Saturday off as well to similar effect.

Not independent from fatigue, there are certain characteristics of how the time is spent that contribute to a feeling of deflation at the end of the day. One is spending lots of time on autopilot, either doing something mundane for a long time in a kind of loop, iterating on top of yourself, with very little direction. This is when I spend ages on a task without really having a good reason why if someone were to ask. I am typically somewhat aware that I’m doing this as well, it’s unusual. I’ve gotten much better at snapping out of this, and forcing myself to brainstorm alternative tasks that are more concrete or take a break, but it still happens more often than I’d like. I believe this is fed by fatigue though, the friction in taking myself out of autopilot.

Accurate evaluation of research progress

Importantly, by my observation, feeling deflated at the end of the day typically has nothing to do with poor research progress. And yet this day-to-day deflation has a habit of dominating how I feel about the state of a particular project or my research generally. By it’s very nature, research progresses by testing ideas, most of which don’t pan out, and iterating. Thus, I think it’s important to try and detach the feeling of deflation from our mental evaluation of how a particular research project is going.

To do this, I find it helps to have a specific picture in mind of what I want to research progress to look like, and then check in with this when I’m feeling deflated. This forces me to acknowledge an accurate picture of progress, and not just dwell in a fatigue-induced distorted world where up is down and clocks melt off walls.

I think a better metric to evaluate my day-to-day research performance is based on 3 simple checks:

The last one in particular is easy to neglect. Sometimes an idea doesn’t go as expected, I feel deflated, and then I’m no longer in the right headspace to reflect carefully and distill valuable lessons. In this case, it’s better to take a break than do a half-baked job. This is actually the most important part of the metric along with the concreteness of new ideas, since they distinguish fast-paced but crazed stumbling from something trying to be better. As a PhD student, I don’t expect to have the same skill at these as a more experienced researcher. However, I can do my best to check that all my ideas are concrete and tied into project goals, and to write down what I learn from each test. This way I can hope to improve with practice.

Perhaps, however, one advantage a PhD student does possess is the flexibility of time to be quick at iterating on ideas by implementing and testing fast. This doesn’t just mean working long hours though: I think some of my fastest idea testing cycles have corresponded to when I’ve worked relatively few hours, and so had the clarity of mind and curiosity to brainstorm frequently, hence make good decisions, minimise wasted effort, and then implement each step asap.

The truth is that progress generally doesn’t feel like victory, mostly it feels like a lot of confusion and struggling to decide between which ideas to pursue. But if you take the time to rest and put things in perspective, these days can feel more like important steps in the right direction. Much like an explorer charting a map, sure the city of gold is the ’end destination’, but uncovering large swathes of the map is vital and terribly productive work. Who knows? You may even find some treasures you weren’t looking for in the form of new and surprising research questions or nuggets of knowledge.

The bigger picture: Morale

What I’ve been talking about thus far are fluctuations in Morale on the timescale of a day or two. As I’ve described, these are typically a result of day-by-day fatigue from working too much or not sleeping enough, combined with a loss of perspective about what proactive research should look and feel like. I’ve also discussed how having a specific idea of what I think effective research looks like (which may vary person to person or year by year) helps to regain this perspective.

Morale also includes fluctuations on longer timescales, and for a variety of reasons.

These might be work-related. For example, finding myself without a main project is both exciting, because I have the opportunity to dive into a new idea, and can be a bad hit to Morale, because it feels instinctively like zero progress is being made every day, leading to a prolonged deflation. My approach for this is to specify a spin on the metric above but instead thinking about what day-to-day progress looks like during this stage. E.g.

Other work-related reasons might be stress from working too much, pressure from job applications (or so I hear), etc. While each reason for low Morale deserves to be given it’s own attention to quell, there are a few (painfully) generic ways to help with these:

Other reasons for Morale fluctuations on longer timescales might be personal in nature and, depending on the reason, these might need to be dealt with separately first.