I've always been fascinated about learning new things and keep progressing - that's my thing - on stuff I had already grasped the basics. Be it skills - intelectual or motor -, niche topics, trivia, etc.
For the past 2-3 years I've been nerding out about learning how to learn - meta learning -, so I'll try to summarise some of the consensual scientific evidence on this topic as well as some extra stuff.
Index:
- Types of Knowledge
- The Feynman Technique
- STic Framework
- DiSSS System
For the nerds out there, I will also leave some material throughout the post - blogs, podcasts, people - from which you can learn more stuff about learning - how meta is this?
Types of Knowledge
There are not really consensual or agreed upon list of types of knowledge, but we can more or less define 3 pairs of 2 antithetical types of knowledge defined in Epistemology - the theory of knowledge:
A priori vs A posteriori Knowledge: a priori knowledge also known as non-empirical reasoning, derives from the world without having to experience it, even though it requires a little amount of experience in order to be applied, i.e., mathematical knowledge. A posteriori knowledge, also known as empirical reasoning, applies logic and reflection to, inductively, through experience, gain knowledge.
Explicit vs Tacit Knowledge: explicit knowledge is comprised of organized and easily communicable and transmittable elements, held in books, databases, etc; tacit knowledge on the other hand is very difficult to transmit and must be acquired through experience (example: learning how to master a song in the piano).
Propositional or Descriptive vs Non-propositional or Procedural Knowledge: propositional knowledge is simply knowledge which can be expressed through declarative sentences or propositions, it is knowing something or about something; non-propositional knowledge conversely is acquired by having done something.
As you might’ve noticed, these 3 pairs of theoretical concepts are really similar in that a priori, explicit and propositional knowledge are quite alike as well as a posteriori, tacit and non-propositional knowledge are.
#1 The Feynman Technique
The Feynman Technique was created by Richard P. Feynman a Nobel Prize in Quantum Electrodynamics in 1965.
This method covers four simple steps which will make you comprehend more deeply what you're trying to learn, and will make it stick for more time. It's important to mention that it's better suitable for concepts, complex or not.
First you should write down the title or topic you're going to learn, then you explain the concept in a very simple language with one or two examples of its application or analogies. This was basically it, but (!) here is where the excels: you minuciously identify subareas or little steps you didn't get totally right (ideally frozen on a piece of paper) and review them. The forth and last step is also identifying subareas or little steps in your explanation of the concept, but in which you used complex language (because you were not able to simplify it) and simplify it. Note: great article on this technique.
Summarising, and also trying to apply this technique to my explanation:
Explain it simply, identify what you haven't yet mastered and reexplain it.
⚠️
STiC Framework
This is probably the one I use the most often and the most backed up by the existing scientific evidence.
The four pillars of this framework are Active Recall, Spaced Repetition, Interleaving and Categorizing. I’ve first come in contact with the formal concepts of Active Recall and Spaced Repetition through Anki - a brilliant program at implementing these two techniques -, when learning German. The STic Framework was acronymized by Ali Abdaal. Let’s dive into it:
# 1 Active Recall
Active Recall or Retrieval Practice consists fundamentally of testing yourself before, during and after studying a topic.
It’s predictable that the testing pre-studying might constitute an objection - how would it be possible to test myself in a topic I haven’t even studied yet? However what this does is priming our brain for when we again see or hear an answer to retain it, more importantly if we fail the question. Active Recall also allows for the hypercorrection effect to play in - when we’re quite confident on an answer and get it wrong, the next time we are much more likely to get it right. A crucial aspect here is: testing shouldn't feel (too) easy. The harder we are trying to retrieve a certain a piece of information, that harder it will stick and get "wired" in our brain.
Testing yourself once is better than re-reading the same thing 4 times
#2 Spaced Repetition:
Spaced Repetition or just Spacing is simply deliberately not testing yourself, not practicing. This idea is based on the concept of The Forgetting Curve, created by Hermann Ebbinghaus, which hypothesizes that over time we tend to forget the stuff we learn unless we re-retrieve them. This Retrievability - how easy it is to remember something from memory - will, among other factors such as emotions, mnemonics, etc., depend on time and stability of memory - how rapidly the retrievability falls without practice. Each learning session or repetition increases the optimal interval before the next learning session, which will flatten out the curve of forgetting.
Example: in a school a group of students was divided into 2 groups, Group 1 had 8h in 1 day to learn Spanish vocabulary and Group 2 had 4h in 1 day and 4h 1 month later. Eight years later, they tested the two groups and found out Group 2 remembered 2.5 times more information than Group 1, both without additional studying. This reinforces the power of Spaced Repetition in Retrievability.
#3 Interleaving:
Interleaving or varied/mixed practice is basically mixing up topics we want to learn, it's the opposite of blocked practice, in which we learn a single topic at a time. This is anchored in 2 principles: contextual interference and discriminative contrast. Contextual interference - a certain variation of topics when learning information or a skill can yield better retention and transfer performance, even though it might lead to poorer performance while doing the task. Discriminativecontrast - mixed practiced allows for the establishment of better connections, permitting the transfer of knowledge across domains, finding differences and similarities between concepts and topics.
An example of this is the following: 4 given types of math problems (A, B, C and D), 5 exercises for each type of problem. Instead of solving the 5 exercises of type A and then 5 of type B, etc, interleaving would be mixing them together - it would be more difficult and take more time, but it would result in learning better how to match a problem to one type of strategy.
#4 Categorization System:
Categorizing consists of building a structure around a certain area of information instead of simply trying to learn it in a non-organized view, simply memorising it. Hermann Ebbinghaus also theorised that a given piece of information is much more likely to be retrievable if it is connected to another piece of information, i.e. categorising not only helps to memorize more information but also establishing connections and building knowledge upon knowledge we already have.
Note: David Epstein's great podcast episode on this.
DiSSS System
This System was popularized by Tim Ferriss and I've used it to learn Arabic, it’s working pretty well and making sure I stick to what I’ve committed to.
Deconstructing → break down what we want to learn into the minimal learnable bits, the LEGO blocks. Find small blocks which are easy to learn.
Selection → This is one of the most useful principles and applies to almost all areas of life - economics, sports, habits, mathematics -, the Pareto Principle aka The 80-20 Rule, which postulates that approximately 80% of the results (outputs) come from 20% of the causes (inputs). It was first conceptualized by Joseph Juran, a Romanian-American Engineer and named after Vilfred Pareto, an italian economist, who noticed 20% of the pea pods in his garden provided 80% of the peas. The production of peas was then shifted to mantaining and taking care of these 20% of the pea pods.
Sequencing → In what order should these blocks be learned? This is hierarchizing and setting an order of precedence, which also serves as planning, reducing the friction to start and keep on learning.
Stakes → This part has to do with commitment. Setting up stakes to create consequences and making sure the program/system is being followed.
Note: Tim Ferriss's #191 Podcast Episode: The Art and Science of Learning Anything Faster
Feel free to utilize some of these techniques and concepts and apply them to whatever you’re learning or want to learn.
Soon I will expand on this topic, exploring other concepts behind motor skill learning and the use of post-learning idleness to your advantage.