%% ### #todo - [ ] Andy Matuschak on "Are memory systems really necessary?" - 1. Naturalistic environments often don’t reliably or clearly surface conceptual gaps. And if you have a nuanced conceptual understanding at the moment, the naturalistic environment often won’t reinforce all aspects of it, even though those aspects may in fact be useful later. - 2. Memory systems can help you more rapidly “bootstrap” yourself to the point where you can use that material naturalistically. %% ### Goal - Design a system that improves long-term memory and “makes memory a choice rather than a haphazard event” ([Michael Nielsen](http://augmentingcognition.com/ltm.html)). ### Benefits - **Memory**: Efficiently improve long-term memory. - **Understanding**: Reveal knowledge gaps and incentivize understanding of a topic. - **Creativity and problem-solving**: Increase fluency and speed in associating and combining ideas, increase the ability to intuit patterns. ### Background - **[[The forgetting curve refers to the empirical finding that the probability of correctly recalling an item declines roughly exponentially with time unless the information is reinforced|Forgetting curve]]**: The probability of correctly recalling an item declines roughly exponentially with time unless the information is reinforced. - **[[Retrieval practice involves drawing ideas from memory rather than looking up the answer|Retrieval practice]]**: When you attempt to actively recall some knowledge from memory, you tend to reinforce those memories. - Retrieval practice is best on intellectual, highly factual, verbal domains, but may still work in many low-level domains. - Types of questions, in descending order of preference: 1. Free recall, 2. Short answers, 3. Multiple-choice, 4. Cloze deletion, 5. Recognition. - **Exponential backoff**: Revisit each piece of information at exponentially increasing intervals. This slows the decay of memory and gradually consolidates it. ### Design principles - **Optimize for emotional commitment**: Develop a strong positive emotional connection to the review sessions. Select material that is immediately relevant. - **Optimize for long-term goals**: Memorize things that serve you in the long term. - **Do not memorize without comprehension**: It’s far more useful to actually understand a topic. It also helps you select the right material to remember. - **Start with the big picture**: Build a picture of the whole topic first before breaking it down into simple items to memorize. - **Build on the basics**: Memory of the basics is often the largest barrier to understanding, and comparatively cheap to remember. - **Avoid overfitting**: Don’t remember irrelevant information or random noise. Identify exactly what you want to know and be as precise about it as possible. - **10-minute rule**: Select material that seems worth investing 10 minutes of your time in the future to memorize. - **Don’t reinforce wrong beliefs**: Add the source to whatever claim you want to remember. Don’t take it at face value. - **Use software that manages the review schedule**: E.g., [Anki](https://apps.ankiweb.net/). Keep it simple and use a small set of features – for instance, simple “question and answer” cards. - Anki keyboard shortcuts: - Cmd + N: Change note type - Cmd + Enter: Add card to deck - Cmd + Shift + C: Insert cloze - **Write good prompts**: Design prompts that, when used in aggregate, force you to use your memory as much as possible and fully “know” the material. - **Focused** on one detail at a time (atomic). - **Precise** about what they’re asking for, permitting exactly one answer. - **Consistent** in the answers they produce and memories they retrieve. - Inconsistent activations tend to erode your memory. - **Tractable** so you can answer them correctly almost always. - **Effortful** so you actually have to retrieve the answer from memory. - Desirable difficulties lead to stronger memories. - **Add cards iteratively**: When reading a paper, capture the core claims, questions, and ideas in 5-10 cards. Come back later and add more cards if needed. - **See it as a virtuoso skill, to be developed**: Building a great spaced repetition memory system takes time and practice, and yields substantial benefits. %%### Andy Matuschak's examples - **Attributes and tendencies**: What makes X an X? What’s always, sometimes, and never true of X? - How is X usually made? - **Similarities and differences**: What relates and distinguishes X from other adjacent concepts? - How is X different from Y? - **Parts and wholes**: What are some examples of X? Are there important “sub-concepts”? Is X part of some broader category? - Name at least 3 examples of X - **Causes and effects**: What does X do? What causes it to do that? What doesn’t it do? When is it used? - Why do people use X instead of Y to do Z? - **Significance and implications**: Why does X matter? What does it suggest? Make it personally meaningful. - What explains why Y is often better than Z? - **Procedures**: What are the important verbs, and when should you move between them? What are the key adjectives, adverbs, subjects, objects? - When preparing X, what should you do after doing Y?%% --- Type: #permanent Topics: [[Learning (Index)]], [[Knowledge management]] Links: - [Spaced Repetition for Efficient Learning – Gwern](https://www.gwern.net/Spaced-repetition) - [Augmenting Long-term Memory – Michael Nielsen](http://augmentingcognition.com/ltm.html) - [Twenty Rules of Formulating Knowledge – Piotr Wozniak](https://www.supermemo.com/en/archives1990-2015/articles/20rules) - [How to Write Good Prompts – Andy Matuschak](https://andymatuschak.org/prompts/) - [Rules for Designing Precise Anki Cards – Soren Bjornstad](https://controlaltbackspace.org/memory/designing-precise-cards/) - [Anki Essentials – Alex Vermeer](https://foggymountainpass.com/anki-essentials/) - [Spaced Repetition – LessWrong](https://www.lesswrong.com/tag/spaced-repetition)