---
name: Concept Map
status: draft
description: STRUCTURE/KNOWLEDGE family. A graph of concepts as labeled nodes and the labeled relationships between them — every node-link-node triplet reads as a sentence you can check.
sources:
  - title: Novak, Joseph D. & Gowin, D. Bob (1984), Learning How to Learn, Cambridge University Press
    url: https://openlibrary.org/works/OL2711859W
  - title: Novak, Joseph D. (1998), Learning, Creating, and Using Knowledge, Lawrence Erlbaum Associates
    url: https://openlibrary.org/works/OL2711856W
---

# Concept Map

## Why it matters

A concept map lays out what you know about a topic as a picture: the *concepts* are labeled boxes, and the *relationships* between them are labeled arrows — and the labels on the arrows are the whole point. "Photosynthesis → produces → glucose" isn't just two boxes near each other; it's a complete sentence, a claim you can read aloud and check. A map made of these claims shows the difference between someone who has connected an idea to everything around it and someone who has only memorized the words, because the connections are right there on the page, named, where you can see whether they hold.

For example: a medical resident is sure she understands acute kidney injury. She can recite the three categories — pre-renal, renal, post-renal — so it feels solid. Asked to draw a concept map, she puts the three categories at the top easily. Then she has to label the arrows, and the trouble starts: she can write "low blood pressure → reduces → kidney perfusion," but the arrow she *can't* draw is the cross-link from fluid status over to glomerular filtration rate over to the specific pattern of tubular injury. The blank space where that link should be is the thing she didn't actually understand. The list of three categories looked like knowledge. The missing cross-link was the gap.

- **What it shows.** A body of knowledge as concepts joined by *labeled* relationships, so that every connection is a stated proposition — and so the connections you can't name become visible as gaps.
- **When to reach for it.** When the meaning of a topic lives in how its ideas relate, and you want to externalize that structure to teach it, learn it, or check whether you (or someone) truly grasp it.
- **How to read it.** Trace a path of node → link → node and read it as a sentence; the more cross-links bridging separate branches, the more integrated the understanding behind the map.
- **What you'd miss without it.** The relationships you've been assuming but never made explicit — prose lets you glide over a missing connection; a concept map forces you to either draw the arrow and label it or leave a hole.
- **Where it misleads.** A map can look impressively full while its arrows carry vague labels like "relates to" or "is connected to" — the diagram's value collapses to a mind map's the moment the relationships stop being real, checkable propositions.

## How to read it

Start at the top. Concept maps are usually drawn as a hierarchy: the **most general concepts sit near the top**, and as you move down, the concepts get **more specific** — examples, special cases, fine details. So the top of an attachment-theory map might read "Attachment," and far below it sit "secure," "anxious," "avoidant," "disorganized." The vertical position is itself information: it tells you which ideas are broad and which are particular.

The boxes are **concepts** — usually nouns or noun phrases, one idea each. The arrows between them are **labeled relationships**, and this is what separates a concept map from every looser diagram. Each arrow carries a verb or short verb-phrase, chosen so that the **concept → label → concept** runs together as a grammatical sentence: "supply curves → slope upward in → competitive markets." Read that way, a concept map isn't a cloud of associated words; it's a stack of *propositions*, and a proposition is something that can be true or false. That testability is the feature. You can walk up to any arrow on the map and ask "is that actually right?" — which you cannot do with an unlabeled branch that merely says these two things are somehow related.

The arrows worth looking hardest at are the **cross-links** — the ones that reach sideways to connect concepts in *different* branches of the hierarchy. A vertical arrow inside one branch ("kidney injury → includes → pre-renal causes") is the expected, easy kind. A cross-link jumps between branches ("fluid status → determines → glomerular filtration rate," tying the volume branch to the filtration branch), and those are the arrows that only get drawn when someone genuinely understands how the parts of a topic hold together. So a richly cross-linked map is the signature of integrated knowledge, and a map that's all neat vertical branches with nothing crossing between them is the signature of knowledge still in separate, unconnected piles. This is why a concept map differs from a **mind map**: a mind map has one topic in the center with branches radiating outward, free and fast and *unlabeled*; a concept map insists on labeling every relationship, and that single discipline is what turns a picture of a topic into a checkable map of understanding.

## When to use it

The concept map belongs to the **STRUCTURE/KNOWLEDGE family** of diagrams — the ones whose job is to make the architecture of a body of knowledge visible. Within that family it is the *propositional* tool: the one you reach for when the relationships between ideas, named and checkable, are the thing you're trying to capture. Reach for it to **externalize and connect a body of knowledge** (get a whole domain out of your head and onto one page where its structure is explicit), to **surface gaps** (the arrows you can't label are exactly where your understanding is thin), or to **check understanding** (your own or a learner's — building the map *is* the test, and reviewing it is the grade).

Knowing the neighbors is how you pick the right tool:

- A **mind map** is the looser cousin — one central topic, branches radiating out, captured fast for brainstorming and recall. It does not label its branches, so it cannot state propositions. Use it when you want breadth and speed; use a concept map when you want the relationships made explicit and testable.
- The **Relationship Mapping** mode produces a network view — entities as nodes and edges between them — but its edges typically show *that* two things are connected, not the labeled proposition saying *how*. Reach for it to see the web of who-connects-to-what; reach for a concept map when each connection needs to be a named, readable claim.

Reach for a concept map when meaning lives in the relationships, when you need to teach or assess understanding, or when you want gaps in thinking to become impossible to skip past. Skip it when you only need fast, free idea-capture (a mind map), when the structure is a procedure rather than a web of relationships (a flowchart), or when the question is which option to choose rather than how ideas connect (a decision tool). The concept map is the instrument for the *structure of understanding*, not for speed and not for choice.

## How Ora builds it

Ora produces a concept map from a **semantic spec** — a structured description of the **concepts** (the nodes, each one idea), the **labeled relationships** among them (each a subject-verb-object triplet, so every edge reads as a proposition), the **hierarchy** (which concepts are general and sit high, which are specific and sit low), and the **cross-links** (the relationships that bridge separate branches, tagged so they can be drawn differently). That spec is then rendered to a diagram: a layout engine arranges the general-to-specific hierarchy top-down — Graphviz in the established CMap-style convention — with edge labels turned on so the propositions are visible on the arrows, and with cross-links drawn in a distinguishing style (a curved arc or a different color) so the sideways integration stands out. An outline view and per-node alt-text accompany it, since the arrows and their labels carry meaning a screen reader has to be told about.

The diagram is the visual face of Ora's knowledge-structuring work: when you ask it to *map this domain and show me how the pieces connect*, the **Synthesis** mode does the work of pulling a body of knowledge into concepts and propositions, and this artifact is how it shows that structure. It is the natural partner to the **Relationship Mapping** mode — where that mode draws the network of entities and edges, the concept map adds the labeled-proposition discipline that turns a network into a testable map of understanding.

The technique is the work of **Joseph D. Novak**, who developed concept mapping with his research group at Cornell in the 1970s and set it out with D. Bob Gowin in *Learning How to Learn*. Its theoretical ground is **David Ausubel's** theory of *meaningful learning* — the thesis that new knowledge is learned meaningfully only when it is consciously connected to what you already know through explicit relationships. The concept map is the externalization of exactly those relationships, which is why building one is itself an act of learning and reviewing one is an act of assessment. Novak set out the mature framework in *Learning, Creating, and Using Knowledge*; the long-running open-source toolkit and the CXL interchange format come from the IHMC CmapTools project led by Alberto Cañas.

## Related

- **Fishbone Diagram** — a sibling in the structure-of-knowledge instinct, but pointed at one problem's causes: where a concept map links concepts with labeled propositions, the fishbone sorts candidate causes into categories under a single defined effect.
- **Flowchart** — the artifact for when the structure is a *procedure* (steps and decisions over time) rather than a web of relationships between concepts.
- **IBIS Argument** (Issue-Based Information System) — the relational map for *deliberation*: questions, positions, and the pro/con arguments among them, where a concept map maps settled knowledge rather than open argument.
- **Relationship Mapping** and **Synthesis** (modes) — the network view of entities-and-edges and the knowledge-structuring operation this diagram renders, with the concept map adding the labeled-proposition layer on top.

## Sources

- [Novak, Joseph D. & Gowin, D. Bob (1984), Learning How to Learn, Cambridge University Press](https://openlibrary.org/works/OL2711859W)
- [Novak, Joseph D. (1998), Learning, Creating, and Using Knowledge, Lawrence Erlbaum Associates](https://openlibrary.org/works/OL2711856W)