What you should know at the beginning of your PhD
Part 2: Daily work
Feburary 10, 2025
This is the second part of my blog post series (see here for the first part), in which I provide some strategies that have helped me successfully complete my PhD. Of course everyone and every discipline is different, so not everything in here may apply to or work for you, but I hope some of the advice is helpful and inspires you to rethink your status quo of going about your daily life as a researcher.
Find out your working style
Some people are most productive in the mornings, while others are night owls. Some work best when the deadline is far in the future, whereas others need a bit of pressure to get things done. Some prefer to think things through before discussing them in a meeting, while others come up with their best ideas during direct interaction with colleagues. In other words, everyone has their unique 'working style.' By discovering how you work best, you can organise your daily PhD life accordingly.
Short-term and long-term To-do lists
I haven't met many academics who report being on top of things. Instead, most researchers constantly feel like they're behind schedule with everything. This is not unexpected. The tasks you take on as a researcher are often too much for one person in a full-time position, and there is a lot of pressure. Additionally, even senior researchers tend to underestimate how long certain tasks will take, and frequently, unexpected obstacles arise. So, how do you avoid either staying in the office very late every day or going home with a bad conscience?
To-do lists are very useful for organizing yourself, but they come with one big problem: they tend to get longer and longer. While you might cross off two tasks, five new ones are added. That’s a frustrating feeling. For this reason, I started using different types of three to-do lists.
- Longer-term to-dos: These include everything that needs to be done within the next few weeks to months. These should align with your PhD master plan (see part 1 of this series).
- Shorter-term to-dos: These cover tasks to be completed within a certain week. Here, I recommend considering the timeliness and importance of tasks, according to the Eisenhower Matrix.
- Daily to-do lists: At the beginning of each day, I write down what is realistically achievable that day, considering that unexpected events might occur. On some days, tasks may take longer than expected and will need to be carried over to the next day. But if this list is realistic, you'll have days where you go through it quickly and feel a sense of achievement and optimism regarding achieving bigger goals.
When you read, read with a goal in mind
There is an overwhelming amount of papers out there, which often leads to anxiety about missing something important. The instinctive solution might seem to be reading as much as possible. However, without a clear plan, reading numerous papers from start to finish may not bring you closer to your goal of earning your degree. In fact, it could increase your anxiety, especially since the literature often presents conflicting information and the more you read, the more messy everything seems to become.
If you have a concrete plan for completing your PhD—such as having your projects outlined (see part 1 of this series)—you can start drafting papers even before you have any data. Consider how the paper might look, what arguments you need to present in the introduction, and how you might interpret various possible outcomes. If you find that certain outcomes would be difficult or impossible to interpret, this exercise may help you identify ways to amend your project plan accordingly.
With a rough paper structure in hand, you’ll more easily identify what information is missing from your logical flow and what you need to extract from the literature. I often begin with a document where I list all the questions I want answered during my reading sessions, then I fill in the gaps, including the corresponding references.
Of course, it’s also enjoyable and beneficial to read papers out of pure interest. If you do so, that's great! But don’t expect your brain to retain everything you read. Take as many notes as you can. Later, during goal-oriented reading sessions, you won’t need to reread the entire paper—you’ll already have something written up about it.
Nowadays, AI tools allow you to quickly extract relevant information from papers and even create structured tables (one such tool is Elicit). These tools can help you become more efficient in extracting goal-oriented information from the literature. However, be aware that if you use AI in your research, you are still responsible for providing an accurate overview of the literature. So, don't blindly trust these tools—use them as an additional resource, not a sole source.
Learn about the methods before you apply them
It is easy to successfully (with “successfully” defined as not getting any error messages) click buttons in a software package and get some results. Often, supervisors push for quick results and papers, but please, remember that you are not a robotic arm who is doing work for someone else. The difficulty is understanding what is going on underneath the GUI, what algorithms are used and what they do, and this understanding is a prerequisite for interpreting your results and defending your choice of methods (“My supervisor told me to use this setting” does not count as good rationale when responding to a reviewer’s query or a question at a conference presentation). This does not mean that you have to understand every single equation that is implemented in the software, but to have an idea what the algorithms are doing conceptually and be aware of the main considerations you need to make when setting parameters. Most software packages have example data sets and good tutorials to go through, which will prepare you very well for your own data analysis. Remember that it is bad scientific practice to try out various parameters on your own data and pick the combination that gives you the “best result”.
Read recent methods papers
Most likely, your lab will have data acquisition and analysis pipelines that they have been “successfully” using forever. But methods are constantly developed and evaluated and the reviewer of your paper may be aware of the latest trends. So instead of running the risk of having to reanalyze all your data later (with getting different results), it may be worth checking the latest literature on your methods before relying on pre-established pipelines. Papers are not the only source of finding about the most recent developments, there are often also workshops, educational lectures, blogs or even Youtube videos.
Whatever you do, look at your data
Successfully running an analysis pipeline does not actually mean not having error messages, it means having done the intended process to your data. Look at the raw data, but also the processed data, and whatever QA-outputs are available by your software package. If something went horribly wrong, you will most likely spot it then, you may not necessarily spot it when you only look at the output of the last analysis step.
Learn how to script and how to do stats
Learning how to script is a bit of a time investment, but nothing compared to repeatedly doing things manually (there are probably more fun types of occupational therapy!)! You will save your future self a lot of time. Also, if you have your scripts structured well and documented, then you will be able to quickly find out what you have actually done when writing up your methods or when trouble shooting weird results. ChatGPT is actually not too bad at helping you learn how to code and writing simple script templates, so make use of the latest technology. Again, you are responsible for your code, so do not trust AI blindly.
Stats is something that most neuroscience relies on. If you refresh (or learn, depending on your background) your stats knowledge in the beginning, this will not only allow you to make better choices for your own analyses, but also to more critically evaluate other people’s work.
Live like you are going to die any minute: be organised
Don’t overestimate the capabilities of your brain, during your PhD it will be overloaded by information and it will have to forget a lot. The only solution to this problem is to be organised and write things down. This will also calm you down because you will know that even if you have a car accident and forget everything you have done so far, you won’t be lost. I highly recommend taking detailed meeting minutes, have a strict and intuitive file naming structure (if you don’t, then scripting will be very hard), write down which scripts do what and in what order to run them and what outputs are your main results. Ideally, also learn how to use version control and Git, so you can share your scripts and increase the transparency and impact of your research. I recommend living your PhD with the idea in mind that if you die suddenly, already the day after somebody else can take over your project and bring it to a well deserved successful end. If you do not die, you will thank yourself later for being so organised.
Before you start writing anything, do a scientific writing course
Even if you are a native English speaker who has won prices for writing good stories in high school, this does not mean that writing papers will be easy for you. Becoming a good scientific writer is a long process and takes a lot of experience. Expect that you will not recognize the draft of your first paper anymore once your supervisors have been through it with tracked changes. However, there are some general rules and guidelines that will make your first texts a lot better, so learn those before you start writing. You will do yourself and your supervisors a big favour. Even in the times of AI tools that help with writing, I still recommend that you do critically double-check everything they have rephrased or suggested. Often they make things sound “better”, but slightly change the logic of the text. In order to do so effectively, you still need to understand the characteristics of effective scientific texts.