Library
The Library
The public-domain books that operant conditioning was built from — Thorndike's puzzle boxes, Morgan's trial and error, James on habit, Yerkes's mice, Clever Hans, Darwin on expression — as complete texts, free to read here and published in formats that software and AI agents can use directly.
What this is
A free, full-text collection of the books, all in the public domain, from which the science of learning from consequences grew. Every page on this site cites primary sources; the library is where those sources live. Each book is published four ways: a readable page, one Markdown file for the whole book, one Markdown file per chapter, and plain text. A JSON manifest describes them all, and llms.txt lists them for AI agents.
The texts are transcriptions of the original printed editions, footnotes included, with page numbers, tables of contents and indexes removed. Nothing has been abridged or modernised.
In brief
- 9 books, 1,288,625 words, 127 chapter files. All public domain in the United States; seven of the nine are public domain everywhere.
- Start with Thorndike's Chapter II and Chapter VI, then Morgan on trial and error and James on habit. A full reading order is below.
- For software:
/library/library.jsondescribes every book and chapter; each book hasindex.md,book.txtand achapters/folder.
The books
Animal Intelligence
The puzzle-box experiments and the law of effect: the direct ancestor of operant conditioning.
Animal Behaviour
Morgan's account of learning by trial and error, instinct, and the limits of what animal behavior lets us infer.
The Principles of Psychology, Vol. 1
Volume 1 of James's masterwork, including the chapter on Habit that every account of behavior change still quotes.
The Principles of Psychology, Vol. 2
Volume 2: instinct, the emotions, and the will, including the ideo-motor theory of action.
The Dancing Mouse
A complete experimental study of one animal's learning: discrimination boxes, labyrinths, learning curves, and the efficiency of training methods.
Clever Hans
The horse that could not count: how Pfungst showed Hans was reading tiny movements of his questioners, and why blind procedures exist.
Animal Intelligence (Romanes)
The great anecdotal survey of animal minds that experimental psychology was invented to replace.
Animal Life and Intelligence
Morgan's first full treatment of animal minds, with his early chapters on habit, instinct and the senses.
The Expression of the Emotions
Darwin on habit: how serviceable actions become habitual, associated, and inherited.
A reading order for operant conditioning
The books total over a million words, and most of that is comparative psychology and philosophy of mind rather than learning theory. This is the route through them for someone who wants the origins of reinforcement, and nothing else.
- Start with Thorndike. Chapter II is the 1898 monograph: the puzzle boxes, the time-curves, and the case against anecdotes. Chapter VI states the law of effect and the law of exercise. Everything on this site descends from these two chapters. Chapter II. Animal Intelligence; an Experimental Study of the Associative Processes in Animals · Chapter VI. Laws and Hypotheses for Behavior
- Then Morgan on trial and error. Chapter IV of Animal Behaviour is the clearest pre-Thorndike account of learning from consequences, and the source of the rule of parsimony that keeps comparative psychology honest. Chapter IV. Intelligent Behaviour
- James on habit. Chapter IV of the Principles is the essay on habit that habit-formation writing still quotes, with the maxims at the end. Thorndike was in the audience. Chapter IV. Habit
- Yerkes on methods. Chapters XII to XVI show what a complete experimental programme on one animal's learning looked like in 1907: discrimination boxes, labyrinths, learning curves, retention, and a comparison of training methods. Chapter XII. Educability: Methods of Learning · Chapter XIII. Habit Formation: the Labyrinth Habit · Chapter XIV. Habit Formation: the Discrimination Method · Chapter XV. The Efficiency of Training Methods · Chapter XVI. The Duration of Habits: Memory and Re-Learning
- Pfungst on cues. Chapter V explains how a horse answered arithmetic questions by reading the questioner's posture. It is the founding case of stimulus control by unnoticed cues, and of the experimenter effect. Chapter V. Explanation of the Observations
- Background, in any order. Darwin's first three chapters on serviceable associated habits; Romanes's introduction, to see what the experimenters were reacting against; Morgan's earlier chapters on habit and instinct; James's chapters on instinct and will in volume 2.
For AI agents and software
Everything in the library can be fetched without an account, a key, or a browser. The formats are plain files at stable URLs, so an agent can read the manifest, pick a book, and pull exactly the chapter it needs.
Endpoints
https://operantconditioning.com/library/library.json— every book: author, dates, edition, copyright status, word counts, a summary of why it matters, and the URL of every chapter file.https://operantconditioning.com/library/llms.txt— the same catalogue as a short Markdown index, in the llms.txt convention.https://operantconditioning.com/library/<book>/index.md— the whole book as one Markdown file, with a metadata header, a table of contents, and footnotes after each chapter.https://operantconditioning.com/library/<book>/chapters/NN-slug.md— one file per chapter, each with a header naming the book, its position, and the previous and next files.chapters/index.mdlists them, as doeslibrary.json.https://operantconditioning.com/library/<book>/book.txt— the whole book as plain text, footnotes marked[n].
Conventions: chapter headings are ##, sections within a chapter ### and ####; footnotes use the [^n] syntax and are numbered continuously through each book; figures are  with the image files alongside; original tables are Markdown tables, or fenced text blocks where the layout could not be preserved. Budget roughly 1.3 tokens per word: Thorndike's Chapter II is about 50,000 tokens, James's chapter on habit about 13,000. The book pages also carry schema.org/Book data, and the whole site has a site-wide llms.txt and llms-full.txt (the library is not included in llms-full.txt, to keep that file usable).
What is not here, and why
B. F. Skinner. None of Skinner's books is in the public domain. Under the 95-year term that applies to works of that era, The Behavior of Organisms (1938) does not become free until 1 January 2034, and everything he wrote afterwards follows year by year. His work is quoted and cited throughout this site, but it cannot be republished here. The Skinner page has a selected bibliography.
Hull, Tolman, Guthrie, Hebb, Ferster and the rest of the twentieth century. Same reason. Under U.S. law a work published in 1931 enters the public domain on 1 January 2027, 1932 on 1 January 2028, and so on; the library will grow by a year every year. Skinner's first journal papers, from 1930 and 1931, are the earliest of his writings to cross that line.
Pavlov and Watson, and Thorndike's Psychology of Learning. These are public domain in the United States and belong here. They are not on Project Gutenberg, so they must be transcribed from scanned copies, which takes time. They are next:
- Ivan Pavlov, Conditioned Reflexes (1927). G. V. Anrep's translation (Oxford, 1927). Public domain in the United States; Pavlov died in 1936 and Anrep in 1955, so also public domain in life-plus-70 countries since 2026.
- Ivan Pavlov, Lectures on Conditioned Reflexes (1928). W. H. Gantt's translation (1928). Public domain in the United States; not yet in life-plus-70 countries (Gantt died in 1980).
- John B. Watson, Behavior: An Introduction to Comparative Psychology (1914). Public domain in the United States; in life-plus-70 countries not until 2029 (Watson died in 1958).
- John B. Watson, Psychology from the Standpoint of a Behaviorist (1919). Public domain in the United States; in life-plus-70 countries not until 2029.
- John B. Watson, Behaviorism (1924). Public domain in the United States; in life-plus-70 countries not until 2029.
- Edward L. Thorndike, Educational Psychology, Vol. II: The Psychology of Learning (1913). Public domain worldwide.
Copyright, sources and reuse
Every book here was published before 1931 and is therefore in the public domain in the United States. In countries that protect a work for seventy years after the author's death, seven of the nine are also public domain; the two exceptions are marked on their pages: Yerkes's Dancing Mouse (Yerkes died in 1956, so 1 January 2027 in those countries) and the English translation of Clever Hans, whose translator's dates we have not been able to confirm. If you are outside the United States, check those two before republishing them.
The texts are volunteer transcriptions distributed by Project Gutenberg, whose eBook number is given on each book's page. We removed the Project Gutenberg licence text, tables of contents and indexes, kept every footnote with its chapter, converted the formatting to Markdown, and left the words as transcribed, including nineteenth-century spelling and the occasional transcription error. Project Gutenberg is a registered trademark of the Project Gutenberg Literary Archive Foundation, which is not affiliated with this site and has not reviewed these editions.
Our editions add no rights of their own. You may copy, quote, print, mine, index, translate or republish these texts without asking. If you cite one, cite the original edition and, if it helps your reader, the URL here: for example, Thorndike, E. L. (1911). Animal intelligence: Experimental studies. New York: Macmillan. Full text: https://operantconditioning.com/library/thorndike-animal-intelligence/. The rest of this site is not public domain; only the library is.
Frequently asked questions
Are these the complete books?
Yes. Every chapter, every footnote and every figure that the source transcription contains is here. What is left out is the apparatus of the printed book: tables of contents (replaced by ours), lists of illustrations, indexes (which refer to page numbers that no longer exist), publishers' advertisements, and the Project Gutenberg licence. Where a transcription had a transcriber's note explaining a table, it was kept.
Why is there nothing by Skinner?
Because his books are still under copyright. Skinner published from 1930 to 1990, and U.S. copyright on works from that period lasts 95 years from publication. His first journal papers (1930–31) are only now crossing that line; The Behavior of Organisms follows on 1 January 2034. The library will add his work as it becomes free.
Can I use these texts to train or ground an AI model?
In the United States, yes: public-domain text can be used for anything. Outside the United States, the same is true for seven of the nine books; check the copyright note on the Yerkes and Pfungst pages for your jurisdiction. The Markdown and plain-text files are the formats to use, and library.json tells you what each file is.
I found a transcription error. What should I do?
Email hello@operantconditioning.com with the book, the chapter and the sentence. Corrections are made in all four formats at once and noted in library.json.