> ## Documentation Index
> Fetch the complete documentation index at: https://meta.niceshare.site/llms.txt
> Use this file to discover all available pages before exploring further.
# Concept Map
> Concept Map joins two ideas with a linking phrase, so each line is a claim you can check. Learn the origin, drawing rules, uses, and limits.
**Category**: Methods
**Type**: Knowledge-representation method
**Origin**: Joseph D. Novak and colleagues, Cornell University, 1972; grounded in David Ausubel's assimilation theory
**Also known as**: Concept mapping; labeled node-link diagram
**Quick Answer** — A Concept Map states knowledge as sentences: two concepts joined by a linking phrase. Joseph D. Novak's group at Cornell developed it in 1972, while following how children understood science, on the learning theory of David Ausubel. A floating keyword is not yet an idea. The idea is the claim on the line.
## What is Concept Map?
A Concept Map is a diagram of knowledge in which each concept sits in a box and each line carries a linking phrase, so the two boxes and the phrase read as one proposition.
> Concept maps are graphical tools for organizing and representing knowledge.
Joseph D. Novak and Alberto J. Cañas open their 2006 account of the method with that sentence, and they revised the account in 2008 at the Institute for Human and Machine Cognition. In their definition, a concept is a regularity you can name in objects or events. A proposition joins two or more of those labels with linking words into a statement that could be true or false. "Plants need light" is a proposition, and the words "plants" and "light" merely parked beside each other are only a neighborhood.
Picture a Saturday trip with a child. A [mind map](/methods/mind-mapping) can stay a colored spray of station names around one plaza. A Concept Map has to say that the red line connects Central to the museum, or that the museum closes before the train home. The demand for a verb is the everyday cue that the picture has become a claim.
[Issue tree analysis](/methods/issue-tree-analysis) splits a question into parts that should not overlap. A Concept Map may still cross-link those parts, which is often where a new idea shows up.
### Concept Map in 3 Depths
* **Beginner**: Pick one question. Write three concepts. Draw a line only when you can read a full sentence across it.
* **Practitioner**: Put the broader concept above the narrower one. After the first tree stands, hunt for one cross-link between branches. Check that sentence against a source, or against a person who was there.
* **Advanced**: The hierarchy lives in the question, not in the subject. Change the focus question and the same concepts should move. A beautiful map can still store a false proposition. The drawing is a scaffold for meaning, not a proof.
## Origin
Concept Map was invented so that a change in understanding could be seen, compared, and argued over. The occasion was a stack of child interviews that transcripts made hard to compare.
David Ausubel argued that people learn by assimilating new concepts and propositions into the framework they already hold. Novak and Cañas trace that claim to *The Psychology of Meaningful Verbal Learning* (1963) and *Educational Psychology: A Cognitive View* (1968). Rote learning can leave the old framework untouched, while meaningful learning changes it. The condition a teacher cannot force is the learner's choice to connect the new idea to the old one, instead of memorizing a sentence for the test.
Joseph D. Novak joined Cornell University in 1967, and his group wanted to follow children's science knowledge across school years. Interview transcripts swamped them, and a change in understanding was hard to point to on the page. In 1972 they began drawing that knowledge as concepts in a hierarchy, with a linking phrase on every line.
Novak and Cañas date the invention to that year, during the program whose twelve-year record Novak and Dismas Musonda published in 1991. Cornell's 2023 account agrees that the team had already published early work on the tool by 1972.
The 1991 paper in the *American Educational Research Journal* records the lessons, and it records the scores the maps made readable. Audio-tutorial science lessons went to 191 children in first and second grade, and a similar group of 48 children did not receive those lessons. At grades 2, 7, 10, and 12, the instructed students showed substantially more valid understandings and fewer misconceptions. Scores taken from the concept maps also favored the instructed group, which the authors read as a lasting effect of the early lessons, with the maps as the record.
Novak and D. Bob Gowin carried the tool out of the lab in *Learning How to Learn* (1984). Novak and Cañas later stated the drawing rules in a 2006 technical report, revised in 2008 under the title *The Theory Underlying Concept Maps and How to Construct and Use Them*. The rules are a focus question, a hierarchy, cross-links, and examples kept outside the concept boxes.
A 2023 Cornell account notes that Novak also advised organizations including Procter & Gamble, NASA, and the U.S. Navy. The list shows how far the idea traveled, and it remains short of a measured trial.
## Key Points
A Concept Map earns its keep when every line can be read aloud as a sentence someone could dispute, and the four moves below are that method. A poster of related words is a different object.
Write the question the map must answer before you draw a box. "How does a houseplant stay alive on a desk?" is a question, and "Biology" is only a label. Novak and Cañas note that the same concepts change rank when the context changes, which is what makes a hierarchy honest. If you cannot say what the map is for, stop adding nodes.
Put a short label in each box, usually one concept, and keep the full sentence off the box. On the line, write a linking phrase such as "need," "are made of," or "fails when," then read box, phrase, and box as one statement. If the reader has to guess the verb, the line is empty. A stated verb is what makes the diagram checkable.
Place more inclusive concepts above more specific ones, and read the map from the top down before you look across branches. A cross-link joins an idea on the left to an idea on the right, and Novak and Cañas treat those links as the usual site of a creative leap. One good cross-link beats a new branch of synonyms. If two propositions contradict, keep both visible until you know which one fails.
Particular examples can sit under a concept, and Novak and Cañas keep them outside the boxes, because an example is not itself a concept. After you study, close the source and draw the map again, which is [active recall](/methods/active-recall) with a structure. Compare the new map with the old one. The links you cannot rebuild are the lesson.
## Applications
Use a Concept Map when the risk is an unnamed relationship, in a setting small enough to finish on one page.
Turn the chapter's question into the top of the page. From memory, write eight to twelve propositions. Then open the book and mark which sentences it actually supports. If a linking phrase only sounds fluent, say it in plain words with the [Feynman technique](/methods/feynman-technique).
In the first month of a role, ask how a real decision gets made. Put people, artifacts, and rules in boxes. Put the verb on the line: "approves," "stalls when," "never sees." A missing line is a better onboarding question than a longer glossary.
A school choice, a care plan, or a shared budget stalls when people trade slogans. Ask one question, such as what has to be true before a parent moves this year. Each proposal becomes a proposition the others can accept or amend. The map does not vote. It shows which claim is doing the work.
When a [systems thinking](/thinking/systems-thinking) discussion stays a list of parts, draw the parts as concepts and force one cross-link. "Overtime increases errors" and "errors increase overtime" are propositions. If you cannot name the relationship, you have a cast list, not a feedback story.
## Case Study
The clearest measured test of Concept Map pools many lessons, and it refuses to rest on one favorite classroom. The numbers below come from two published reviews of learning studies.
By 2006, studies of node-link diagrams disagreed about the size of the benefit. John C. Nesbit and Olusola O. Adesope gathered 55 of those studies in *Review of Educational Research*. The pool held 5,818 learners from grade 4 through postsecondary, in subjects such as science, psychology, statistics, and nursing. Concept maps were associated with better retention, and the size of that advantage ran from small to large, so the studies did not cluster around one number.
Noah L. Schroeder, Nesbit, Carlos J. Anguiano, and Adesope asked the question again with a larger net. Their meta-analysis went online on 21 March 2017 and was printed in *Educational Psychology Review* in 2018. They defined the diagram in Novak's terms: a node is a concept, and a link names the relationship between two concepts.
The pool mixed concept maps with knowledge maps, a close cousin that also names the link, and it held 142 independent effect sizes from 11,814 learners. In each study, the comparison was the other instruction used in that study.
Learners who used the maps scored higher by a moderate margin, a gap of Hedges' g = 0.58. Hedges' g counts the advantage in standard deviations, with a correction for small samples, and a standard deviation is a typical spread of scores, so 0.58 is a real step up. In a pool this large, luck is a weak explanation, and the reported probability was below 0.001.
The split that matters for practice is who drew the map. Learners who created the map gained more, g = 0.72, than learners who only studied a map drawn for them, g = 0.43, and both gains favor the maps. Building the propositions was the stronger use, in science and engineering fields and outside them.
What Concept Map explains is the gap between those two numbers. A map you only look at can still arrange a chapter, while a map you build forces each linking phrase into your own words, which is the meaningful-learning condition Ausubel cared about.
The result stays inside its pool. Nesbit and Adesope measured recall and transfer of taught material, from grade 4 through postsecondary, so a product review or a family budget was outside the evidence. Where the studies disagreed, a single g cannot promise your next hour, and a diagram with blank lines was outside the treatment they counted.
## Boundaries and Failure Modes
A Concept Map fails when the question is fake, when the line is blank, or when the drawing is treated as proof. Each failure shows up in the drawing itself.
The first boundary is the focus question. Novak and Cañas are explicit that hierarchy depends on context, so a map titled "the company," with no question, will rank concepts by habit, often by the org chart, and then look objective. Ask why onboarding slipped this month, and the same concepts should rearrange. If they do not move, you drew a logo.
The second boundary is scale. The 2008 account recalls George A. Miller's 1956 finding that working memory handles only a small number of units at once, about five to nine. Novak and Cañas draw a practical limit from that finding: in one pass, relate about two or three concepts.
A forty-box mural built in a single meeting overruns the limit, so people nod at the shape and remember the colors. Build the map in passes, or split it by question.
The common misuse is to borrow the name for a [mind map](/methods/mind-mapping). Unlabeled rays from a central picture are a useful burst of associations, and they stay associations until a linking phrase makes them propositions.
A second misuse is to treat a tidy map as evidence. "Ice cream sales cause drownings" is a well-formed proposition and a misleading one, and the format makes the claim easier to attack while leaving the claim's truth untouched. The Cornell study is easy to misread in the same spirit: the score gap favored children who had received early science lessons, and the maps recorded that gap.
## Common Misconceptions
Most weak maps come from three mix-ups about what the lines are supposed to carry.
A mind map radiates keywords from one center, and the branch often has no verb. A Concept Map can have several hubs. Every line has to carry a linking phrase that completes a sentence. If you delete the phrases and the diagram still seems finished, you were making a different tool.
Novak and Cañas treat valid propositions, a hierarchy that matches the question, and cross-links as the features that carry meaning. Ten boxes that restate one idea are a list in costume. Two boxes and one hard cross-link can be the whole insight. Add a box when it lets you say something new.
The map represents a cognitive structure. It is not the structure itself. On the Ausubel account that Novak and Cañas use, new propositions have to settle into what you already hold. A polished poster can sit beside a mind that still recalls the old misconception. Redraw the map from memory on another day. The links you cannot rebuild are the ones you do not yet hold.
## Related Concepts
These pages take up the jobs that sit beside a Concept Map: associating freely, remembering, explaining, and taking a system apart.
Radial keywords for a fast burst of associations, before any line has to become a sentence.
Close the source and rebuild the propositions, instead of rereading a map you already like.
Say the linking phrase in plain words. A foggy noun means the proposition is not ready.
Use a cross-link when the relationship runs in a loop, not only as a one-way definition.
Split a question into non-overlapping parts, then map how those parts actually affect each other.
Push the lower boxes until a proposition rests on something you could test.
## One-Line Takeaway
Read every line as a sentence, then redraw the map from memory before you trust it.
This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.