The Dual Coding Delusion
How dual coding theory became a victim of edu-mythology and why adding visuals won’t make your teaching more memorable.
Few ideas have spread as fast through staffrooms as dual-coding theory. Somewhere between Rosenshine and retrieval practice, it became one of those concepts that every school feels obliged to mention in CPD slides. Add a few icons to your PowerPoint, sketch a diagram next to your success criteria, and you’re supposedly harnessing the power of cognitive science. But as with so many ideas in education, what began as an elegant theory has metastasised into caricature.
Dual coding, as originally proposed by Allan Paivio in the 1970s, wasn’t a classroom gimmick. It was a theory of representation in human memory. Paivio suggested that we store information through two distinct but interconnected systems: a verbal system, which processes linguistic information such as words, and a non-verbal system, which handles imagery, spatial relations, and sensory impressions. When information is encoded in both systems, the two traces can reinforce each other.
Take something concrete like the word apple. If a learner reads or hears the word, that engages the verbal system; if they also picture the shiny red fruit, the non-verbal system joins in. Later, remembering either the image or the word can trigger recall of the other. Paivio called this dual coding because both codes contribute to retrieval.1
Over time, experimental psychology provided evidence that words associated with concrete imagery are easier to remember than abstract ones. This “concreteness effect” lent support to Paivio’s model and helped explain why pictures are often more memorable than text alone.2 Crucially, though, dual coding was never intended as an instruction manual. It was a descriptive theory of how information might be represented in memory, not a prescriptive guide to classroom practice. Its power lay in helping psychologists understand memory processes, not in telling teachers to draw cartoons.
Later researchers, most notably Richard Mayer and John Sweller, attempted to translate these cognitive insights into practical teaching principles. Mayer’s experiments, conducted from the late 1980s onwards, tested Paivio’s claim that dual representations aid recall, using short computer-based lessons that paired narration with diagrams or animations. His results led to the Cognitive Theory of Multimedia Learning (CTML), which holds that people learn more effectively from words and relevant pictures than from words alone, provided that the two forms of information are processed together in working memory.
He identified a set of design principles that govern when dual coding helps and when it hinders:
The modality principle shows that spoken explanation plus visual image outperforms on-screen text plus image, because it spreads processing across auditory and visual channels.
The coherence principle warns that decorative visuals, background music or redundant labels increase extraneous load, diverting attention from the core message.
The contiguity and signalling principles stress that learners benefit when related words and images are presented close together in space and time and when key relationships are made explicit.
In other words, dual coding only works when the visual and verbal materials are integrated and economical.
Sweller’s Cognitive Load Theory (CLT) complements this by modelling the limitations of working memory. Building on Paivio’s two-channel assumption, Sweller argued that any instructional method must balance three types of load: intrinsic (the complexity of the material itself), extraneous (poor design that distracts), and germane (mental effort that builds schema). Visuals can therefore lighten the cognitive burden by externalising complex information (E.g., showing a process flow rather than describing it verbally) but they can just as easily increase load if they are poorly aligned or redundant. CLT studies demonstrate that learners perform better when visual and verbal materials are coordinated to reduce split attention and to support schema construction rather than competition between channels.
Together, Mayer and Sweller operationalised Paivio’s insight into a set of conditional rules: dual coding enhances learning only when the dual materials are conceptually coherent, temporally aligned, and cognitively economical. In practice, this means that good multimedia instruction is not about adding images, but about designing for integration.
In short, dual coding’s success depends not on the presence of pictures, but on the quality of the mapping between words and images. When that relationship is tight, learning can flourish; when it’s loose or irrelevant, cognition fragments.
In classroom life, “dual coding” has been simplified into “add pictures to your teaching”. This is roughly equivalent to saying that because protein builds muscle, you should eat steak for every meal. What’s gone missing is the mechanism. Paivio’s work, and later extensions by Sweller and Mayer, imply that dual coding works when two channels of information reinforce the same concept without overwhelming working memory. The learner must be able to integrate the verbal and visual inputs into a single mental model. When that integration fails, the cognitive benefit evaporates.
One of the most common culprits is split attention. If words and images are not properly coordinated - either spatially or conceptually - working memory becomes overloaded. Imagine a a biology lesson where a teacher projects a complex diagram of the human heart while explaining blood flow. Pupils hear “the pulmonary vein carries oxygenated blood to the left atrium” but, faced with a tangle of unlabeled red and blue vessels, cannot tell which is which. Their minds dart between listening, looking, and guessing, leaving little capacity for actual understanding. Instead of supporting meaning, the diagram becomes one more thing to juggle.
A subtler failure occurs when visuals and words point in different directions. In a literature lesson on Macbeth, a teacher might show a still image of the murder scene while analysing the soliloquy, “Is this a dagger which I see before me?” The picture captures the act; the text explores the thought. Pupils’ attention is drawn to the physical violence rather than Macbeth’s psychological disintegration. The image and the language represent related but incompatible ideas, and the tension between them fractures the intended mental model: they remember the stabbing but not the self-doubt.
In both cases, the underlying problem is misalignment. For dual coding to work, the visual and verbal materials must converge on meaning: each should illuminate what the other obscures. When they conflict, duplicate, or distract, learners waste their limited capacity reconciling - or resisting - them. Instead of forming a coherent schema, they’re left with fragments. Dual coding’s promise becomes its own undoing: an elegant theory reduced to cognitive noise.
To reiterate, if images and words pull in different directions, or if graphics are merely decorative, the extra channel becomes noise rather than signal. The working memory system, already limited, must now juggle irrelevant details. Instead of deepening understanding, the visual distracts from it. This is why the classic PowerPoint stuffed with icons, arrows and stock photos often leads to less learning, not more. Teachers, in an earnest bid to “dual code,” raid sites like the Noun Project, pasting generic icons (a lightbulb for “idea,” a handshake for “agreement,” a brain for “thinking”) as if clip-art could stand in for cognition. The result is a visual Esperanto of symbols so overused they signify nothing. Instead of helping pupils think, slide are reduced to a caricature of an impenetrable instruction manual.
In theory, dual coding is meant to help learners build schema, rich networks of interconnected knowledge that can be flexibly recalled and applied. The active ingredients are:
Complementarity: the visual and verbal information depict the same idea.
Integration: the learner connects the two codes, rather than processing them separately.
Cognitive economy: the combined presentation reduces, rather than adds to, cognitive load.
Generativity: the dual representation supports recall and transfer beyond the immediate task.
When these ingredients are present, dual coding can be powerful. Think of a diagram of the water cycle annotated with concise labels, or a sentence diagram showing grammatical relationships. The image is not decoration; it’s a scaffold for abstract thought.
But in schools, the idea often mutates lethally. It becomes:
Visuals are always good. (They aren’t.)
Every concept must include a diagram or an icon. (Most shouldn’t.)
If it looks accessible, it must be effective. (Not necessarily.)
Cartoons make content memorable. (Only if the cartoon helps you think about the content.)
This is an all too familiar pattern. Research findings leak into schools stripped of nuance, repackaged as universal strategies. Consultants offer “dual coding” training by showing teachers how to draw stick figures on mini-whiteboards, as if the act of sketching alone were cognitive magic. In the process, the theory’s explanatory core (how humans process and integrate information) gets replaced by silly pictures.
Teachers, seeking to make content ‘more visual,’ may inadvertently strip away the linguistic challenge that builds long-term knowledge. Students remember the picture of the Battle of Hastings but forget why it mattered. They can point to the arrows in a PEE paragraph poster yet can’t write a coherent analysis. The visual replaces rather than prompts thought.
The danger of over-simplified dual coding lies not in using visuals, but in forgetting what they’re for. A visual aid should carry meaning. It should highlight relationships that words alone make harder to see. It should reduce abstraction, not add entertainment. Most importantly, it should be integrated with instruction. If you have to explain the picture separately, the benefit is lost.
