Dual Coding: How Pairing Words With Pictures Helps You Remember
Combining words with relevant visuals gives memory two routes to the same idea. The research behind dual coding, Mayer's multimedia principles, and how to apply them to notes and cards.
Dual coding means learning an idea through both words and a relevant visual, so it is stored in two forms that support each other. The theory comes from psychologist Allan Paivio, who proposed in the 1970s that the mind has separate but linked systems for verbal and visual information. It is well supported: people generally remember pictures better than words (the picture superiority effect), and Richard Mayer's research on multimedia learning found that words plus relevant graphics usually teach better than words alone. The catch is "relevant." Decorative images, or the same words shown and spoken at once, can make learning worse.
(This explainer is from the Nebulearn blog. Nebulearn makes a flashcard app with image cards and image occlusion, which is one way to apply this.)
Dual coding is not "visual learner" theory
This is worth separating clearly. The learning styles idea says some people are visual learners and should get visuals while others should not. That claim has not held up in research. Dual coding says something different: combining words and visuals helps almost everyone, because two forms of the same idea are better than one. The learning styles myth covers why the first claim fails while the second holds.
What the research shows
| Finding | Source | What it means for you |
|---|---|---|
| Concrete words are remembered better than abstract ones | Paivio's work, 1960s and 70s | Words that call up an image ("apple") get coded twice; abstract ones ("justice") often only once |
| Pictures are remembered better than words | The picture superiority effect, replicated many times | A good diagram of a process often outlasts a paragraph about it |
| Words plus graphics beat words alone | Mayer's multimedia principle | Add a relevant visual to explanations, especially of processes and structures |
| Visuals placed next to their labels work better | Mayer's spatial contiguity principle | Put labels on the diagram, not in a legend or caption far away |
| Extra, decorative material hurts | Mayer's coherence principle | Leave out "nice" images that do not explain anything |
| Narration plus identical on screen text can hurt | Mayer's redundancy principle | When you explain a diagram aloud (in a study video or to a group), let the picture and your voice carry it, rather than also showing the same sentences as text |
Which visual for which content
The best visual depends on what kind of idea you are learning.
| Content type | Visual that fits | Example |
|---|---|---|
| Physical structure | Labeled diagram | Parts of the heart, a cell, an engine |
| Process or cycle | Flowchart or cycle diagram with arrows | Krebs cycle, the water cycle, a legal procedure |
| Sequence in time | Timeline | Causes of a war, stages of development |
| Comparison | Table or Venn diagram | Mitosis vs meiosis, two theories |
| Hierarchy | Tree diagram | Taxonomy, an organization chart, a classification |
| Relationship between quantities | Graph | Supply and demand, dose response |
| Location | Map | Geography, battle sites, wine regions (see geography map flashcards) |
| Abstract concept | Concrete example or analogy sketch | "Opportunity cost" as a fork in a road with a price tag on each path |
Tables sit between words and pictures: the content is verbal, but the grid layout carries meaning on its own. A comparison table is often the fastest way to add structure to a set of facts.
How to dual code your notes
- After reading a section, draw it. Not artistically: boxes, arrows, stick figures. If you cannot draw how the parts relate, you have found a gap.
- Explain the drawing in words. Write two or three sentences under it, or say it out loud. Now you have both codes, and they check each other. The Feynman technique is the spoken version of this step.
- Redraw from memory later. Close the book and redraw the diagram. This combines dual coding with retrieval practice, which is stronger than either alone. Blank sheet retrieval practice works with drawings too.
- Keep labels on the image. Write the names directly on the parts, with short lines pointing to them, instead of a numbered key elsewhere.
How to dual code your flashcards
| Card type | How it dual codes | Best for |
|---|---|---|
| Image occlusion | Hide one label on a diagram, recall it from its position | Anatomy, maps, structures |
| Picture front, word back | Recognize from the visual | Species, signs, artworks, faces |
| Word front, sketch back | Recall the word, check against the image | Processes, structures |
| Text card with a small diagram on the back | Answer in words, see the visual as feedback | Concepts that have a picture but are tested in words |
| "Draw it" card | Front says "sketch the stages of mitosis," you draw on paper, then compare | Processes you must reproduce |
Image occlusion is particularly efficient: one labeled diagram becomes many cards, each testing one part in context. Encoding diagrams effectively covers how to set it up without the common mistakes, like learning a label's position on the image instead of the structure itself.
Mistakes that undo the benefit
- Decorative images. A stock photo of a smiling student on a card about enzyme kinetics adds nothing and takes attention. Only use visuals that carry the content.
- One image forever. Seeing the exact same picture every time can teach you that picture rather than the concept. For identification tasks, vary the images. Bird and plant identification flashcards covers this.
- Labels far from the parts. Legends and numbered keys make you look back and forth, which costs working memory.
- Visual without words. A diagram you never put into words can be recognized but not explained. You need both codes.
- Too much on one image. A diagram with 40 labels is a wall of information. Break it into regions, one idea per card, as in the atomic encoding rule.
Where to read more
Allan Paivio's Mental Representations: A Dual Coding Approach (1986) is the theoretical source. Richard Mayer's Multimedia Learning lays out the design principles with the experiments behind them. The Learning Scientists, a group of cognitive psychologists, have free posters and blog posts on dual coding aimed at students and teachers, and Oliver Caviglioli's Dual Coding with Teachers is a practical guide full of examples.
In Nebulearn, you can add images to cards, use image occlusion on diagrams, and open a textbook PDF beside the card editor while you work, with FSRS handling the reviews.
Two codes beat one
Pairing words with a relevant visual stores the idea twice, so there are two routes back to it. Choose the visual that matches the content (diagram for structure, flowchart for process, table for comparison), keep labels on the image, skip decorative pictures, and always put the visual into words as well. Then test yourself on both, by redrawing from memory and by explaining it aloud.