How To Read A Paper
Finding papers:
Use scholar.google.com to search for papers based on keywords, and create alerts. Alerts will send you new papers as they are published. If a paper is unavailable for free online, you can email the first or last author for a copy. A rather unorthodox but shockingly effective search engine for papers is the desktop application Harzing’s Publish or Perish. You may also consider a tool to keep your papers annotated and in one place, such as Papers.
Lab webpages will have their papers listed and available for download, e.g. https://cmngroup.princeton.edu/publications.htm
Step 1: Skimming the visuals
You need to check if the paper is relevant. Scroll through the paper to find images or equations to get a quick idea of the work presented in the paper. Note that when writing most papers, the authors create the figures and write the equations first, and only once those are finished do they begin writing the text.
Step 2: Read up to the first reference of the first visual
Now thoroughly read the abstract and the introduction and keep reading into the main text until you come to the first mention of Figure 1 or Equation 1. While reading, mark down some of the statements in the paper. Mark statements that make sense with a green highlighter, or a checkmark next to the statement, statements that are confusing with a yellow highlighter, or a question mark, and statements that seem wrong with a red highlighter, or an x. Don’t mark every sentence, just the ones that jump out at you.
Step 3: Research confusing or wrong statements.
Before moving on, look up all the words you didn’t understand and do some background reading on confusing statements. If you find some information hard to search for on the internet, the introduction of a paper has a large number of citations, you can go to these citations and see if any of those papers have the information you need. Repeat the steps in this document for that paper as well. If you find yourself 3 papers deep and looking at the citations again, abandon the search and find someone to ask another scientist for help.
If the authors of the paper have a Bluesky account, they may have posted about this paper at some point. You can often find these posts attached to the online version of the paper in the digital journal. These threads do a great job of explaining papers. If the author has given a recorded talk shortly before or after the paper was published, they may explain the paper in the recorded talk.
Step 4: Read the captions to the visuals and study the visuals
Now that you have read and understood the text up to the first reference of the first figure, it is time to look at the figures. Go to each figure and look at each panel, reading the caption that goes along with each panel. Do your best to note things that make sense, but don’t bother noting things that are confusing or wrong yet, the main text might explain those things. Go through all the figures, reading about each panel and trying to understand what each panel is showing you.
If the paper has equations, you can read these in-line during Step 5.
Step 5: Read up to the discussion/conclusion
Now read the paper through to the end of the discussion, again marking statements that make sense, statements that are confusing, and statements that seem wrong. In the text, every panel of every figure will be referenced, as each one is referenced, flip back to the figure and see if you understand now that you have read the main text. If you still don’t understand the panel, mark the panel as confusing or wrong.
If you are confused about something, you may want to check the methods, experimental procedures, and supplementary materials or videos that might clarify some information. Reading these portions of the paper isn’t necessarily important, but sometimes it can be useful.
Step 6: Read the discussion/conclusion
Read the Discussion/Conclusion sections closely, but not necessarily while taking notes. This section is where authors are summarizing the work, reiterating important points, drawing conclusions, and speculating about future discoveries. Up until this section the authors have tried to keep themselves out of the paper, presenting only the facts. In the discussion and conclusion, authors tend to speak in their own voice, which can make this text easier to follow and the key ideas of the paper really jump out. Very confusingly, this is also where authors are forced to say things they don’t really want to say by the reviewers. As a consequence of this disasterous culture, the discussion section is a bit of a russian roulette.
Step 7: Find the good
Now it’s time for the hardest part of reading the paper. Go through the paper, and figure out what is good about the paper. Why did the authors write this paper? What is useful about it? What hidden gems and brilliant ideas doesn’t it contain? Some creative and hardworking scientists worked on this paper, often for years, and deep inside the text lies wildly important information. Find the best bits of the paper and write about them enthusiastically in the margins to bring up during journal club.
Step 8: Find the bad
Now its time for the easiest part of the paper. Try to tear it down. What conclusions seem unwarranted? What figures or statements seem sloppy? Can you poke holes in the case the authors are trying to make? Do you trust the data? Have a debate against the paper and find other ways the data could be interpreted. Find reasons not to trust the data. Write down the best of these reasons in the margins, to bring up during journal club.
Step 9: Go talk with other scientists
Once you’ve read a paper, bounce your biggest ideas about the paper off of other scientists. This is the purpose of a journal club! If you don’t have a journal club that wants to cover the paper, find another scientist who has read the paper or is willing to read the paper. If you can’t find anyone to talk to, but have some burning questions, it’s ok to contact the authors directly and ask the first author to have a video chat or ask them questions via email.
If you don’t read
A major part of being a scientist, is reading the scientific work of others. Many, including myself, would argue that this is either the most critical or second most critical skill of being a scientist. If you don’t like reading papers, you are at a disadvantage that you should find a way around. Most of what is written here applies to any form of scientific consumption.
You can heavily attend talks and courses and conferences, but you risk only get a shallow level of information, mostly from charasmatic speakers, that is likely not much more correct, critical or insightful than what a chatbot could give you. If this is your strategy, try to attend talk series or courses where you can ask questions.
You can surround yourself with collegues who like reading papers.
More and more, you can listen to scientific podcasts, in which scientists talk about research papers. Note, while science communicators are an interesting resource, that is not what I am suggesting here, here I am suggesting you listen to scientists who are in the trenches and happen to be broadcasting their reading.
You can listen to audio versions of papers. If English is a barrier, you can even get a chatbot to translate a paper for you, and read it out loud to you.