I wish there was a name for this overall approach to studying, and that it was easier to describe. 'Study tips' and 'active learning' articles often give generic advice without actually helping you connect with your own mind, which is the real challenge.
Predictive learning might work.
I highly endorse the idea you're pointing to. You can do the same thing with papers if you want to understand them. Read the abstract and try to predict the introduction. Predict further content from that. E.g. predict their definition of corrigibility based off the 4 desiderata they describe. Predict interesting results based off the definitions and motivation. Predict the formal statement of a theorem based off an informal description. You can practice all sorts of sub skills by doing this sort of thing.
Thanks for the suggestion! I took it to ChatGPT to explore alternatives, and "Guess-and-Check" hit the mark. "Predicive learning" was a good starting point.
Using this technique in scientific writing is a game-changer. It helps you approach a paper with a curious and critical mindset, asking questions like:
"""
That's when I discovered more effective ways to approach reading, including what I'll call "Guess-and-Check," the technique of scanning and making predictions. Instead of trying to read every word in a textbook, in Guess-and-Check you scan the material and make predictions about what you think the text is saying. This active reading process can help you better engage with the material and activate your prior knowledge. After making your prediction, be sure to confirm or correct it by checking it against the text.
"""
This is similar to the way GPT-3 was trained! Pretty cool that you also found it effective!
After a long semi-hiatus during my MS degree, I am eager to revisit my sequence on scholarship, particularly the topic of learning new subjects from textbooks. This article provides a general overview based on my own introspection and experiences, rather than an extensive literature review. While I have been influenced by academic concepts and blog posts on learning and scholarship, my aim is not to connect with wider literature, but to express my thoughts and intuitions, in the hopes that they resonate with and assist some readers.
When I started studying engineering, I realized that reading and absorbing information from textbooks was a challenge. Sometimes, I found myself staring at a page without actually comprehending the words, while other times I would scan through material without understanding it. I tried forcing myself to read every sentence carefully, but found it to be unsustainable and too slow.
That's when I discovered more effective ways to approach reading, including what I'll call "Guess-and-Check," the technique of scanning and making predictions. Instead of trying to read every word in a textbook, in Guess-and-Check you scan the material and make predictions about what you think the text is saying. This active reading process can help you better engage with the material and activate your prior knowledge. After making your prediction, be sure to confirm or correct it by checking it against the text.
Scanning the material means looking at the key words and phrases in each sentence rather than trying to read every word. Focus on the main points and identify any areas where you're confused or need more information.
After scanning and absorbing material with Guess-and-Check, it's important to take a higher-level perspective to understand the relationships between concepts and ideas. By stepping back and considering how individual sections fit into the larger context of the textbook, you can better appreciate the author's purpose and intention. For instance, you might try explaining why Chapter 4 covers the material it does, and how it sets the stage for Chapter 5. By examining how concepts build on each other or diverge, you can enhance your understanding of the material and develop a more complete picture of the subject matter at hand.
As you transition from studying individual concepts to synthesizing a higher-level understanding of the overall progression of the material, you may find that certain concepts are more challenging or harder to remember. You might also encounter sections where you don't fully grasp why certain topics are presented in a particular order. These moments of confusion can be particularly valuable, as they highlight areas where you have the opportunity to deepen your knowledge and understanding. To tackle these challenges, consider revisiting those sections and apply the same scanning and predicting method you used earlier. By approaching the material in this way, you can clarify the points of confusion and deepen your overall comprehension of the subject matter.
Many study techniques emphasize the importance of practice problems, and while they are undoubtedly valuable, I believe that it's even more important to first develop a deep understanding of the individual concepts, as well as how they relate and progress throughout the material. This way, when you finally get to the practice problems, you won't just be mindlessly crunching numbers, but instead be able to see the problems in the larger context of what you've learned, and how they build on one another to push your understanding forward. So, don't rush to practice problems too quickly, instead take the time to truly understand the concepts and their interrelationships, and the problems will become much more interesting and engaging.
Using this studying-how-to-study strategy might look like a lot of scanning and thinking, and less of the traditional practice-problem approach. Instead, you spend time searching for points of confusion and gaps in your understanding while scanning the textbook, followed by addressing these issues through focused re-reading. By delaying the practice-problem phase until you've built a solid conceptual foundation, you gain a deeper understanding of the material and a greater appreciation for the significance of the problems. Ultimately, this approach can help you achieve a more profound mastery of the subject matter.
At a deeper level, the studying-how-to-study strategy is about learning how to pay attention to and control your own mind. You're learning to manage your memory, visualization skills, and ability to spot patterns and connections. This means that you're not just reading the material, but actively engaging with it. One way to do this is by translating complex concepts into visual images. For example, if you're studying statistics, you might create a mental picture of a bell curve and manipulate it in your mind in response to algebra that indicates a Normal distribution.
Another important skill that this strategy helps you develop is the ability to simplify complex information. When you encounter a long, technical description of a process, it can be overwhelming to try to remember all of the details. Instead, you can break the information down into smaller, more manageable chunks. Aim for no more than three items per chunk, and use simple language to describe each step.
Here's an example of how to go about chunking a technical description to make it more manageable:
Original description: "The process of converting a digital image into a printed version involves several steps, including color separation, halftoning, and printing. First, the image is separated into its individual color components using color separation. Next, halftoning is applied to each color component to convert the continuous tone image into a set of dots. Finally, the halftoned color components are printed using a printing press or other printing device to produce the final printed image."
Simplified description: "To convert a digital image to a printed version, you need to do three main things: separate the colors, turn the image into dots, and print it. Here are the steps:
By breaking down the process into three steps, and then further explaining each step in simple language, it becomes much easier to understand and remember the key elements of the process. You don't have to write all this down. You can also chunk it in your head.
One potential objection to this strategy is that it may not be the most efficient way to learn for everyone. Some people may find that they learn better by working through practice problems first, rather than focusing on understanding the concepts before attempting problems. Additionally, some subjects may require a more hands-on approach, with more emphasis on practicing and applying the concepts in real-world scenarios rather than simply reading and understanding the material. Therefore, it's important to consider individual learning styles and the requirements of the subject matter when deciding on a study strategy.
Many students turn to flashcards, such as Anki, to learn complex technical material. While Anki can be an effective tool, TurnTrout's excellent article on using it to study mathematics and retain it long-term being a great example, I personally find it difficult to maintain the habit. I believe this is because I prefer to use the method I'm describing here to construct a stronger, holistic mental map of the subject first. Anki's approach is too narrow, focusing on individual pieces of information rather than synthesizing a coherent understanding of the subject. To truly comprehend the subject, I want a comprehensive mental picture of its concepts and their relationships. Anki is helpful for drilling specific points over time, but I think it's only effective after I have developed a solid understanding of the material using the methods I'm advocating.
While there are software tools available for iterative reading that resemble the approach I'm describing, I believe that the effectiveness of this method is not solely determined by what you're reading, but by what's happening in your mind - developing the ability to describe, observe, and regulate your own conscious mental processes. A software program can't access that information, only you can. Therefore, I prefer to focus less on software tools and more on developing your mental faculties for identifying confusion, recognizing patterns, recalling information from memory, visualizing abstract concepts, and so on.
The biggest challenge with this learning strategy is that much of the learning process is introspective and highly individual. Everyone relates to their own mind in their own unique way, making it difficult to model and share effective study techniques. Some people may not visualize information easily, others may not have a clear inner monologue, and so on. Furthermore, many people who have a growth mindset for a particular subject may have a fixed mindset about how they learn and think. There is little shared vocabulary to describe this sort of thinking, and even if there were, it can be highly technical and jargon-heavy, making it difficult to communicate practical study strategies effectively.
Based on my personal experience, I believe that the introspective practice I've described is not innate, and can go undeveloped for years. However, it is a skill that can be learned and is incredibly impactful once mastered. It's often overlooked because it's difficult to observe someone modeling the techniques directly, as studying is a private activity and this set of skills is introspective in nature.
I find this approach to studying more enjoyable, relaxing, and rewarding compared to other methods I've tried in the past. It feels organic and human, as if I'm building a relationship with myself, and it motivates me to continue. However, one challenge I've encountered is that textbooks often contain a lot of material that I don't need to read, such as filler, examples I've already worked through, or information that's too advanced for me. The specific bits of information that I need to focus on are scattered throughout the text, making it difficult and distracting to locate them. Additionally, studying using this method can feel unfocused and unstructured since I don't have a concrete goal in mind from moment to moment, like "read the next paragraph." There are various tasks to perform, such as reading, visualizing, remembering, and forming connections between different pieces of information. While it's not quite analysis paralysis, it can be distracting and hinder my progress.
A deeper issue with this approach is that it relies heavily on your ability to introspect about your own confusion, invent patterns, and analyze the text independently. If you struggle to find those patterns, articulate what you find confusing, or simply lack the motivation to truly understand the textbook, this method may not work well for you. It requires a shift away from the mentality of merely checking off items on a to-do list or minimizing study time while maximizing grades. Instead, you must approach the textbook with a deep desire to learn and understand the material thoroughly, even if it takes more time and effort. This means setting aside distractions and prioritizing the learning process, even when life gets busy or when other options seem more appealing.
I've primarily used this approach to study independently, relying solely on the textbook (in PDF format, as it allows for easier navigation, copying, and is more ergonomic). However, I'm now exploring the possibility of leveraging ChatGPT to aid in my introspection. While it's not always reliable for factual answers, I've found it to be effective in helping me clarify my areas of confusion through dialogue. Rather than using it solely to obtain direct answers, I believe it may be more beneficial in guiding me towards the information I need to learn.
I wish there was a name for this overall approach to studying, and that it was easier to describe. 'Study tips' and 'active learning' articles often give generic advice without actually helping you connect with your own mind, which is the real challenge. They tell you what to do, not how to do it. While this article suffers from the same problem to some extent, it's because the specifics of things like 'how to visualize and when to do it, and what to do if you're feeling stuck' deserve their own dedicated articles. I have so much I want to say on this subject, but so little time to do it, that I feel like I have to cram it all into one blog post. Paradoxically, this can result in generic study advice. By leveraging ChatGPT to help me turn my intuitions into clear blog posts addressing specific pieces of the puzzle, I hope to create a useful corpus of literature on this subject.
I apologize for the lack of citations and links in this article. It's based primarily on my personal introspection and experience over the years. While I have read various articles and scientific literature on learning how to learn, the ideas presented here are a culmination of my own understanding and reflections. Please take this article as one person's perspective and not as an authoritative source. My views may evolve over time, and I hope to refine and deepen my understanding of this topic further.
In the future, I plan to write more in-depth articles on several subtopics related to this approach, including: