How to Read Research
Many students “hit a wall” the first time they try reading academic research. I have some more general advice about this problem, maybe it applies to your situation. I find it applies pretty broadly to learning technical material in many subjects.
- Academic papers are self-contained for those who are already familiar with whatever niche they are published in. For example, if you have already read 100 papers on machine ethics, when you encounter the 101st paper in this subfield, you will probably be able to understand a lot without consulting a textbook.
- Those not familiar with the niche will need to consult outside resources and build themselves up to understand what’s going on. This is almost universally true.
- My recommended approach is: a0. Get ready to go slowly. I can be impatient, and I find I have to relax into the right mindset when learning a challenging topic. If your brain is in “eager-for-results” mode, the necessary steps will feel gratingly slow. Get into a meditative, patient, relaxed mindset. This is challenging, but you are capable. a. Start reading. b. You will run into a term you don’t know. c. Look up the term. Resources for this step include: a. Textbooks b. Wikipedia b. Stackoverflow/Mathoverflow c. ChatGPT or other chatbots d. knowledgeable friends e. ??? be resourceful :) d. You might find that you understand the definition for this preliminary concept. If so, go to step e. Alternatively, you might find that even the preliminary concept is overwhelming. Repeat steps a-c, working backwards from where you started to a place that you’re familiar with. Eventually you will hit a concept that you’ve seen before. At this point, advance to steps e and onwards. You might want to leave a breadcrumb trail in your notes so you can retrace your steps. e. When you understand the definition for a single term, write it down in a way that you can consult later. I like flashcards. People with really exceptional memory can skip this. For people with normal memories, this step is crucial. You WILL forget the information otherwise. f. Go over your notes and flashcards regularly, until you can recall the basics without consulting a textbook. “Without consulting a textbook” is important. There is only so much room in your working memory. To understand combinations of basic concepts, the basic concepts themselves need to become embedded in your mind to the point that they feel natural and obvious. When you have become familiar with something to this level, it takes up much less mental space than it did when you first learned it, and as a result you have more room to play with combinations of ideas. I like drilling flashcards to get recall started, and talking to friends or writing to help the concept settle into place. YMMV. g. When you feel comfortable with a concept, take a look at the breadcrumb trail from step d. and go one step forward. Repeat the process layer by layer until you get to the point where you started in step b.
For step d, when you are working backwards to find out how some crazy concepts connects to something you already know, chatbots can be very useful. I often do something like:
-
Ask the question.
”I’m reading this paper and I saw <some term I don’t know>. Could you explain
?” -
The model will respond. Skim this response for things that look familiar. If you don’t see anything familiar, try something like:
“I have a background in
. I’ve taken <HS/undergrad/grad> classes on . What concepts should I be familiar with to understand ? -
Look through the output for something you’re familiar with and start there, slowly adding complexity until you get back to the place where you first got stuck.
For mathematical notation specifically, you might break down the task into: a. Understand the concepts that the notation stands for. The above advice applies. b. Understand the notation itself (which symbols map to which concepts), including any quirky notation choices made by one particular paper.
For part b (understanding the notation), it might help to follow this hypothesis loop approach:
- Write down in your notes what you think a certain symbol could mean.
- Write down implications. What else would have to be true, if that symbol has the meaning that you think it does?
- Check the hypothesis based on the implications. Look around in the rest of the paper for evidence that would rule out your hypothesis.
- Repeat until you are confident in your interpretation of each symbol.