Dual coding: why combining images with words doubles your memory retention

In 1971, Allan Paivio published Imagery and Verbal Processes, reporting a decade of experimental work on how the mind represents information internally. His central finding — that concrete words (apple, bicycle) are consistently recalled better than abstract words (justice, democracy) — was not new. What was new was his explanation: two separate representational systems, operating in parallel, each capable of storing and retrieving information independently.

Paivio called this dual coding theory. Its implications for memory and learning are extensive, and its application to mnemonic technique is direct: any memory tool that activates only one representational system is operating at half capacity.

Two systems, not one

Paivio proposed that cognition relies on two interconnected but separate systems:

The verbal system processes and stores language-based information — words, sentences, propositions, logical structure. It operates sequentially, left to right, one word after another. Its currency is the word and the logical relationship between words.

The non-verbal (imagistic) system processes and stores mental imagery and spatial information — visual scenes, sounds, smells, textures, spatial arrangements. It operates simultaneously rather than sequentially: a mental image contains all its features at once rather than one after another. Its currency is the image and the spatial relationship between elements within it.

These systems are distinct but interconnected. The word "apple" can activate its visual referent (the non-verbal system), and a mental image of an apple can activate its name (the verbal system). When both systems encode the same information simultaneously, two independent retrieval pathways exist. If one pathway fails — if the verbal trace has decayed or is momentarily inaccessible — the other can complete the retrieval. This redundancy is why dual-coded information is remembered more reliably.

The concreteness advantage

Paivio's most robust finding was the concreteness advantage: concrete words are recalled better than abstract words, across word lists, sentence memory tasks, and prose comprehension tests. The magnitude is substantial — concrete words are typically recalled at roughly twice the rate of abstract words in free recall tasks.

The mechanism is dual coding. Concrete words (apple, bicycle, cathedral) automatically activate both the verbal and non-verbal systems — the word triggers a visual image without deliberate effort. Abstract words (justice, democracy, momentum) activate primarily the verbal system, with weak or absent imagistic activation. This gives concrete words two retrieval pathways and abstract words one.

The implication for mnemonic design is direct: when building an acrostic phrase, choose concrete words — words that evoke clear, specific sensory images. "Plastic meat" is concrete; "preliminary mechanisms" is not. "Every" and "some" are function words with minimal imagistic content. "Elephant" and "volcano" are maximally concrete.

Mayer's multimedia learning: dual coding in education

Richard Mayer's cognitive theory of multimedia learning (2001) applied Paivio's framework to educational design. Mayer proposed that learners have separate processing channels for verbal and visual information, that each channel has limited capacity, and that meaningful learning involves building connections between verbal and visual representations.

Mayer's research programme tested the multimedia principle across dozens of studies using a consistent methodology: learning from words + relevant pictures versus learning from words alone. The consistent finding: multimedia learning outperforms verbal-only learning, typically by moderate-to-large effect sizes. This effect generalises across learner ages, content types (science, mathematics, language arts), and media (text + diagrams, narration + animation).

The critical qualifier in Mayer's research: the images must be relevant, not decorative. Adding images that carry the same information as the text creates dual coding. Adding images that are unrelated or merely decorative imposes extraneous cognitive load without creating dual coding — and can actually impair learning. The principle is not "add pictures" but "encode information twice, in different systems simultaneously."

Applying dual coding to your mnemonics

Every mnemonic technique benefits from deliberate dual coding. The procedure is consistent across technique types:

For first-letter acrostics: After building your phrase, spend 20–30 seconds generating a specific mental image of the scene described. "Plastic Meat Aint Tasty" — where are you? What does the plastic meat look like, what colour, what texture? Is it on a plate or loose? Who is it for? The image should be specific, animated, and slightly unusual. Once you have the image, holding the phrase in mind will simultaneously activate the visual scene — and vice versa.

For method of loci: Each image placed at a memory palace location should be interactive and vivid — not a static object but an action occurring in the specific space. The method of loci is already a dual-coding technique by design: the route provides spatial encoding, and the images provide visual encoding. The verbal layer (the name of the item) is then retrieved by traversing the spatial structure and identifying the image.

For any vocabulary or factual memorisation: When learning a new term, immediately generate both a verbal association (what does this sound like? what other words does it connect to?) and a visual association (what image can represent this concept, even abstractly?). The keyword method (Atkinson & Raugh, 1975) formalises this for foreign language vocabulary: find a word in your native language that sounds like the target word, then generate a vivid image connecting the sound-alike to the meaning.

Why "visualise your mnemonic" is non-optional

The most common failure mode in mnemonic building is stopping at the verbal phrase. The learner constructs "Plastic Meat Aint Tasty," confirms the letters match PMAT, and considers the mnemonic built. This is encoding only the verbal system — achieving one pathway instead of two.

The 20–30 seconds of deliberate visualisation after phrase construction is not a refinement or optional enhancement. It is the step that activates the second representational system and creates the dual-coded trace. Without it, the mnemonic is operating at half its theoretical capacity.

The practical test: can you describe the scene in your mnemonic to someone who has never heard it? Where is it set? What can you see, hear, smell? What is happening? If you cannot answer these questions specifically, the visual encoding is incomplete.

Try it now: The Mnemonic Builder prompts you to visualise your phrase with dual coding in mind. After building your acrostic, you are prompted to note the visual scene before testing recall — implementing the dual-coding step that most mnemonic guides omit. Free, no account required.

The Mnemonics & Pattern Memory course covers dual coding in Lesson 5, including Paivio's experimental evidence, Mayer's multimedia learning research, and the keyword method for vocabulary acquisition. Six evidence-based lessons, free, no account required.

Further reading in this series

• What are mnemonics? The evidence-based guide to memory patterns
• First-letter mnemonics: acronyms, acrostics, and how to build them
• Method of loci: how to build a memory palace
• Chunking and pattern recognition: Miller's Law applied to learning
• Mnemonics for studying: how to use memory techniques for exams

References

• Atkinson, R. C., & Raugh, M. R. (1975). An application of the mnemonic keyword method to the acquisition of a Russian vocabulary. Journal of Experimental Psychology: Human Learning and Memory, 1(2), 126–133. https://doi.org/10.1037/0278-7393.1.2.126
• Clark, J. M., & Paivio, A. (1991). Dual coding theory and education. Educational Psychology Review, 3(3), 149–210. https://doi.org/10.1007/BF01320076
• Mayer, R. E. (2001). Multimedia Learning. Cambridge University Press.
• Paivio, A. (1971). Imagery and Verbal Processes. Holt, Rinehart and Winston.
• Paivio, A. (1986). Mental Representations: A Dual Coding Approach. Oxford University Press.