The way to learn a new subject from scratch is to read in a set order: a textbook or subject encyclopedia first, for the map and the words; then a recent review article, for the state of the research; then the handful of studies everyone cites. University library guides describe that order, and the research on how novices learn explains why skipping to the studies goes badly.
Self-study adds one job: checking that you understand, not just that you have read. This guide covers the order, the evidence on novices, the Feynman technique, "learn anything in 20 hours", free courses, and a four-week plan of our own.
EdCitation publishes this guide. Every page and study was read on 26 or 27 September 2026; where we read only an abstract, we say so. Find sources is where to find a field's review articles and most cited papers, Cite a source built our references from their DOIs, and Verify references is where to check a list before you rely on it.
How do you start learning a new subject from scratch?
You start with background sources, move to reviews, and only then read primary studies, testing yourself at each step. The table gives each stage.
| Stage | What to read or do | Why | Source |
|---|---|---|---|
| 1. Map the field | A textbook chapter, a subject encyclopedia or a handbook entry | Key terms, themes, names, and a short bibliography | University of Maryland Libraries (2026); CUNY Graduate Center Mina Rees Library (2026) |
| 2. Survey the research | One or two recent review articles | A summary of where the research stands, and its key studies | University of Texas Libraries (2023); New York University Libraries (2026) |
| 3. Read the key studies | The papers the reviews keep citing | The evidence itself, not someone's account of it | University of Reading Library (2026) |
| 4. Learn the methods | Worked examples before problems on your own | Novices are overloaded by unguided problem-solving | Sweller et al. (2019) |
| 5. Check understanding | Explain it in plain words, then self-test | People think they understand more than they do | Rozenblit and Keil (2002) |
| 6. Find and record sources | Search reviews and most cited papers; build each reference from the record | Every source real, and cited correctly | EdCitation's Find sources and Cite a source |
Where do you start learning a new topic? What library guides say
Start with a background source, not a journal article. The library guides below agree on the direction, from general to specific.
Textbooks, handbooks and subject encyclopedias
The University of Maryland Libraries (2026) education guide says reference sources help you define key terms, spot the main themes and people, and find keywords for searching, and that their bibliographies point to the important primary and secondary works. The CUNY Graduate Center's library calls encyclopedias a concise overview of "established" knowledge, each with a short bibliography of recommended further reading (CUNY Graduate Center Mina Rees Library, 2026). That bibliography is your first reading list.
Then a review article
A review article sums up the state of research on a topic. The University of Texas Libraries (2023) guide says reviews show who the main people in a field are, recent advances, gaps and debates, and, unlike research articles, are good places to get a basic idea of a topic. NYU's sociology guide recommends reviews, Annual Reviews among them, to get to know the trends and key ideas (New York University Libraries, 2026). What each kind of source is for is set out in types of academic sources.
Then the primary studies
The University of Reading Library (2026) guide advises starting broad and narrowing, and treats reviews as a way to find sources, not a substitute: go back to the original texts before you cite them.
How do you find a review article or a textbook's reading list?
Search with a review filter, or add the word "review" to your search. Oregon State University Libraries (2026) suggests both: add "review" as a keyword, as in "captive breeding zoos review", or use a database's own limit, such as Document type in Web of Science or Publication Type in Medline. The Texas guide shows PubMed's "Review" filter under Article Type.
For a reading list, check three places:
- The bibliography of an encyclopedia or handbook entry, as the CUNY and Maryland guides describe.
- A course's own readings. MIT's Introduction to Psychology on OpenCourseWare, taught by John Gabrieli in autumn 2011, comes with lecture videos, a free online textbook, readings and exams with solutions, built for independent study (Gabrieli, 2011).
- The reference list of a recent review. The studies it cites most are the field's core; a search sorted by citations in EdCitation's Find sources shows which of them other researchers lean on most.
A high citation count shows a paper matters to a field, not that it is right.
Why is a new subject so hard at first? What the research shows
A new subject is hard because you have nothing in long-term memory to organise it with, so every new idea competes for scarce working memory. That is the core claim of cognitive load theory.
Novices and working memory
Sweller et al. (2019), reviewing twenty years of cognitive load research, describe working memory as limited in capacity and duration for new information, but not for organised information retrieved from long-term memory. An expert handles as a single element what a novice must hold as many interacting pieces.
Worked examples before problems
The same review describes the worked example effect: novices learn more from studying complete solutions than from solving the same problems unaided. It also reports the limit, the expertise reversal effect: as expertise grows, help designed for novices shrinks in value, disappears, and can reverse. So study worked examples first, then fade them into problems. For mathematics, Barbieri et al. (2023) pooled 181 effect sizes from 55 studies and found a medium benefit of worked examples (g = 0.48), larger for correct examples than incorrect ones.
Prior knowledge: a head start, not a verdict
Prior knowledge predicts where you end up more than how much you gain. Simonsmeier et al. (2022) analysed 8,776 effect sizes from 493 studies and 126,050 participants. The correlation between what people knew before and after was high (r = .534); the correlation between prior knowledge and normalised gains was close to zero (r = .059), with a wide spread across studies. The authors conclude that this spread rules out both "knowledge is power" and the view that prior knowledge hardly matters. Starting from zero puts you behind; it does not stop you learning.
Does the Feynman technique work?
No study has tested the Feynman technique under that name, but its core steps, explaining in plain words and going back where you stall, draw on research that does hold up.
Where it comes from
The York library guide gives four steps (explain the topic as if to a child, find where you cannot answer "why", simplify, repeat) and says it was developed by the physicist Richard Feynman (University of York Library, 2026). The person who popularised it tells it differently: Scott Young introduced it in a free study bootcamp in 2011 (Young, 2011), and writes that it is only loosely based on Feynman's own practices (Young, 2024). We found no source in which Feynman described a method by this name.
What the evidence says
Two lines of research fit. Rozenblit and Keil (2002) called one the illusion of explanatory depth: people believe they understand how things work far better than they do. In their first study, 16 Yale graduate students rated their understanding of devices such as a zipper and a flush toilet, then wrote explanations, and their ratings dropped. Kobayashi (2019), pooling 28 studies (we read the abstract), found learning gains from preparing to teach (g = 0.35) and larger ones from teaching after preparing (g = 0.56), including on delayed tests. Self-explanation was rated of moderate utility in the review covered in how to study effectively.
Our verdict: a sound habit, tested only in its parts. Do it with the book closed.
Can you learn anything in 20 hours?
You can get past the frustrating start of many skills in about 20 hours of focused practice, according to Josh Kaufman, but that figure is Kaufman's own and has not been tested in a published study.
Kaufman set the idea out in a 2013 TEDx talk and in The First 20 Hours (Kaufman, 2013; TEDx Talks, 2013). In the talk, Kaufman traces the 10,000-hour figure to K. Anders Ericsson's research on people at the top of competitive fields, and argues that 20 hours takes a beginner to "reasonably good", not expert. The four steps: break the skill into parts, learn enough to correct yourself, remove barriers to practice, and practise for at least 20 hours. The talk cites no study for the number 20.
The 10,000 hours is shaky too. Macnamara et al. (2014) found that deliberate practice explained 26% of the differences in performance in games, 21% in music, 18% in sports, 4% in education and under 1% in professions: important, but not everything.
Our reading: the method suits a skill you can practise and watch improve, such as an instrument. A subject like economics is not one skill; spend its first 20 hours on the map and the worked examples.
Where can you learn a new subject for free?
Two large free collections are run by universities themselves.
- OpenLearn, from the Open University, launched in 2006 and describes itself as a free learning platform with hundreds of short courses of 1 to 24 hours, many adapted from parts of OU modules. Every course offers a statement of participation and some a digital badge; neither carries formal credit (The Open University, 2022).
- MIT OpenCourseWare describes a free and open collection of material from more than 2,500 MIT courses, with no enrolment, under a Creative Commons licence; MIT offers no credit or certification to its users (MIT OpenCourseWare, n.d.).
Free and open textbooks are covered in where to find free textbooks legally.
A four-week plan for a new subject (ours)
This plan is our own synthesis of the sources above; nobody has tested it as a package.
- Week 1, the map. Choose one introductory textbook or free course and read its opening chapters with the routine in how to remember what you read. Keep a glossary of the terms you meet, and end each session by writing what you remember with the book shut, then checking, as in retrieval practice.
- Week 2, the reviews. Find two or three recent review articles in EdCitation's Find sources, with the Review article box ticked or "review" in the search. Read each abstract, introduction and conclusion, and list the studies more than one of them cites.
- Week 3, the key studies and the methods. Read three to five of those studies, using the passes in how to read an academic paper fast. Work through worked examples before trying problems alone. Build each reference in Cite a source as you go.
- Week 4, explain and test. Write a one-page explanation of the subject from memory for someone new to it; mark every place you stalled and go back to the source. Test yourself on the week 1 glossary without looking, and list the questions still open.
How does EdCitation help you learn a new field?
EdCitation is the best place to find a field's reviews and most cited papers, because every result is a published record and every reference is built from the publisher's record, never written. A chatbot can suggest a reading list; it cannot show you that each paper exists. EdCitation never writes any part of a student's work: the reading, the understanding and the credit stay yours.
Find the reviews and the classics
Find sources searches about 300 million published works, by topic or by the claim a sentence needs, with filters for the kind of source, years and open access, and a Most cited sort. On 27 September 2026, "cognitive load theory review", sorted by Most cited, returned 9,463,644 results from the back-up index, built on Crossref's records. First came Sweller's 1994 paper "Cognitive load theory, learning difficulty, and instructional design" in Learning and Instruction, cited 2,659 times. With the Review article box ticked instead, "cognitive load theory" returned 8,449 reviews from the main index, the first a 2016 systematic review of breathing changes under cognitive load in Neural Plasticity, cited 237 times. On that run the order held for the first 18 results only, and reviews cited 2,138 and 4,044 times sat further down, so read the whole first page.
Cite and keep what you read
Cite a source builds a reference from a DOI, title, PubMed ID, ISBN or web address in APA 7, MLA 9, Chicago 18 author-date, Harvard, IEEE and Vancouver, and screens for retraction. From 10.1007/s10648-019-09465-5 it returned, with no retraction notice:
Sweller, J., van Merriënboer, J. J. G., & Paas, F. (2019). Cognitive architecture and instructional design: 20 years later. Educational Psychology Review, 31(2), 261–292. https://doi.org/10.1007/s10648-019-09465-5
Verify references checked this guide's 20 references: 19 verified, and the OpenLearn page, which turns automated readers away, came back unchecked rather than "not found"; we had read it ourselves. Finding, citing and verifying are free with no account. With Max, $24 a month, the Library lets you read or borrow from more than 5 million books in one place. See Pricing.
Quick questions
What is the fastest way to learn a new subject?
Read in order: a textbook or encyclopedia entry, then a recent review, then the studies it cites, testing yourself as you go.
Does the Feynman technique actually work?
No study has tested it by name. Its parts are supported: explaining exposes gaps people did not know they had (Rozenblit and Keil), and preparing to teach improved learning across 28 studies (Kobayashi).
Can you really learn anything in 20 hours?
Twenty hours is Josh Kaufman's own estimate for getting reasonably good at a skill, not a tested finding, and it promises basics, not expertise.
How do I find a review article on a topic?
Add "review" to your search, or use a review filter. In EdCitation's free Find sources, tick Review article under Source type and sort by Most cited.
Is it too late to learn a subject if I know nothing about it?
No. Simonsmeier and colleagues found prior knowledge predicts where learners end up but barely predicts how much they gain.
References
- Barbieri, C. A., Miller-Cotto, D., Clerjuste, S. N., & Chawla, K. (2023). A meta-analysis of the worked examples effect on mathematics performance. Educational Psychology Review, 35(1), Article 11. https://doi.org/10.1007/s10648-023-09745-1
- CUNY Graduate Center Mina Rees Library. (2026, August 24). Encyclopedias. Beyond Wikipedia: Background and reference sources. https://libguides.gc.cuny.edu/background/encyclopedias
- Gabrieli, J. D. E. (2011). 9.00SC Introduction to psychology [Course materials]. MIT OpenCourseWare. https://ocw.mit.edu/courses/9-00sc-introduction-to-psychology-fall-2011/
- Kaufman, J. (2013). The first 20 hours: How to learn anything . . . fast. Portfolio. https://www.penguinrandomhouse.com/books/312040/the-first-20-hours-by-josh-kaufman/
- Kobayashi, K. (2019). Learning by preparing-to-teach and teaching: A meta-analysis. Japanese Psychological Research, 61(3), 192–203. https://doi.org/10.1111/jpr.12221
- Macnamara, B. N., Hambrick, D. Z., & Oswald, F. L. (2014). Deliberate practice and performance in music, games, sports, education, and professions: A meta-analysis. Psychological Science, 25(8), 1608–1618. https://doi.org/10.1177/0956797614535810
- MIT OpenCourseWare. (n.d.). About OCW. Retrieved September 27, 2026, from https://ocw.mit.edu/about/
- New York University Libraries. (2026, August 25). Reviews of research. Sociology research guide. https://guides.nyu.edu/sociology/reviews-of-research
- The Open University. (2022, October 25). OpenLearn. OpenLearn. https://www.open.edu/openlearn/get-started/about-us
- Oregon State University Libraries. (2026, September 21). Find articles. Researching the literature review. https://guides.library.oregonstate.edu/c.php?g=285832&p=1905947
- Rozenblit, L., & Keil, F. (2002). The misunderstood limits of folk science: An illusion of explanatory depth. Cognitive Science, 26(5), 521–562. https://doi.org/10.1207/s15516709cog2605_1
- Simonsmeier, B. A., Flaig, M., Deiglmayr, A., Schalk, L., & Schneider, M. (2022). Domain-specific prior knowledge and learning: A meta-analysis. Educational Psychologist, 57(1), 31–54. https://doi.org/10.1080/00461520.2021.1939700
- Sweller, J., van Merriënboer, J. J. G., & Paas, F. (2019). Cognitive architecture and instructional design: 20 years later. Educational Psychology Review, 31(2), 261–292. https://doi.org/10.1007/s10648-019-09465-5
- TEDx Talks. (2013, March 14). The first 20 hours -- how to learn anything | Josh Kaufman | TEDxCSU [Video]. YouTube. https://www.youtube.com/watch?v=5MgBikgcWnY
- University of Maryland Libraries. (2026, August 25). Background info, encyclopedias, and handbooks. Education research guide. https://lib.guides.umd.edu/education/background
- University of Reading Library. (2026, September 4). Starting your literature review. Literature reviews. https://libguides.reading.ac.uk/literaturereview/starting
- University of Texas Libraries. (2023, August 28). Review article. Finding journal articles 101. https://guides.lib.utexas.edu/journalarticles101/reviewarticles
- University of York Library. (2026, September 2). The Feynman technique. Study and revision: A practical guide. https://subjectguides.york.ac.uk/study-revision/feynman-technique
- Young, S. H. (2011, September 1). Learn faster with the Feynman technique. Scott H Young. https://www.scotthyoung.com/blog/2011/09/01/learn-faster/
- Young, S. H. (2024, March 26). 5 keys to get the most out of the Feynman technique. Scott H Young. https://www.scotthyoung.com/blog/2024/03/26/5-keys-feynman-technique/