Adding relevant pictures or graphics to text improves learning from it, with a small pooled effect on reading comprehension, though whether dual coding explains the benefit is disputed¶

Claim · All claims
Evidence · 3 studies · 1 causal r?, 1 quant-synthesis r2, 1 review r? · q2–q4 · i1 small
Currency · newest study 2025 · 1 of 3 entries checked against their sources (automated) · not reviewed by a person

Pairing text with corresponding pictures or graphics improves learning from it: a meta-analysis of 39 experiments found a small-to-moderate benefit for reading comprehension (g = 0.39), and a narrative review concludes that carefully constructed illustrations improve text-dependent outcomes. The usual explanation, that verbal and visual processing create two memory traces rather than one (dual coding theory), is contested: Higdon et al. (2025) found the picture-superiority effect in memory followed physical distinctiveness instead. The evidence recorded here measures comprehension and text-dependent performance more than recall as such.

Subclaims¶

q4 i1 Adding graphics to text had a small-to-moderate positive effect on reading comprehension (g = 0.39) across 39 experiments with learners from elementary school to adults. The outcome is comprehension broadly, not recall specifically. → Guo et al. 2020

q2 i? A narrative review concludes that carefully built text illustrations generally improve learners' performance on text-dependent outcomes, and that research through the 1990s still supports this. The abstract gives no pooled effect size. → Carney and Levin 2002

q3 i? This study qualifies the mechanism, not the effect: the picture-superiority effect in recognition and free recall followed physical distinctiveness, and it disappeared when words were made as distinctive as black-and-white pictures. The authors conclude that dual coding does not explain it. → Higdon et al. 2025

Evidence¶

Guo et al. 2020¶

Guo, D., Zhang, S., Wright, K. L., & McTigue, E. M. (2020). Do you get the picture? A meta-analysis of the effect of graphics on reading comprehension. AERA Open, 6(1), 2332858420901696. doi:10.1177/2332858420901696

q4 · meta-analysis (random effects, 39 experimental studies) · i1 · small-to-moderate effect, g=0.39 · n=39 studies (2,103 participants) · quant-synthesis · r2

The meta-analysis pooled 39 experimental studies published between 1985 and 2018. Each compared reading text with graphics against reading the text alone. The pooled random-effects estimate favoured graphics (Hedges's g = 0.39, 95% CI [0.26, 0.51]), with moderate heterogeneity (I² = 45.91%) and no significant publication bias on Egger's test. The effect held regardless of grade level, and pictures, pictorial diagrams and flow diagrams did not differ significantly. Graphics helped more on open-ended and mixed-format assessments than on true/false tests. The authors frame the work with dual coding theory. However, the outcome they pool is reading comprehension, and some included studies found diagrams reduced verbatim recall. So this supports the claim's direction for meaningful comprehension more than for recall as such.

Carney and Levin 2002¶

Carney, R. N., & Levin, J. R. (2002). Pictorial illustrations still improve students' learning from text. Educational Psychology Review, 14(1), 5–26. doi:10.1023/a:1013176309260

q2 · narrative review · i? · no pooled effect size in the abstract · n=unreported (review of 1970s–1990s research) · review · r?

This narrative review covers research on pictures added to text. It concludes that carefully constructed illustrations generally improve learners' performance on a range of text-dependent cognitive outcomes, and that 1990s research still strongly supports this conclusion from the 1970s and 1980s. It also covers when, why and for whom pictures help, including less conventional illustrations and pictorial mnemonic strategies, and it offers practical "tenets for teachers". Only the abstract was read, so no numbers are reported here.

Higdon et al. 2025¶

Higdon, K. F., Neath, I., Surprenant, A. M., & Ensor, T. M. (2025). Distinctiveness, not dual coding, explains the picture-superiority effect. Quarterly Journal of Experimental Psychology, 78(1), 180–191. doi:10.1177/17470218241235520

q3 · peer-reviewed experiments · i? · no effect size in the abstract · n=unreported in the abstract · causal · r?

The experiments tested whether pictures are remembered better than words because they carry a verbal and a visual code, as dual coding theory holds, or because they are more physically distinctive. They varied the look of the stimuli: half the words appeared in varied fonts, sizes, colours and capitalisation, and half the pictures appeared in black and white. Earlier recognition results already favoured distinctiveness, and these experiments extended them to associative recognition and free recall. The authors argue that dual coding is "no longer a viable explanation" of the picture-superiority effect. This challenges the separate-retrieval-routes mechanism the claim states. It does not test learning from combined verbal and visual material.

Studies bearing on this claim¶

Does pairing the words with a relevant visual of the same idea improve learning over the words alone, and for whom? This table gathers the studies behind this claim and the 7 narrower claims that bear on it: 17 distinct studies (9 causal, 4 quant-synthesis, 4 review), q2–q4, 6 of 17 with an effect size coded. One row per study, codes as each entry records them; no effect is averaged and no verdict is drawn. Built by scripts/synthesize_claim.py from eval/synthesis/; a row that looks wrong is fixed on the claim it comes from.

Discussion¶

Mechanism. Dual coding theory (Paivio) holds that verbal and pictorial information are processed in partly separate cognitive channels; presenting both creates additive memory codes and additional retrieval routes (see Dual coding theory). This overlaps with, but is distinct from, the multimedia principle in Cognitive Load Theory — dual coding predicts benefits from two coordinated representations, while load theory warns that any added material must not overload working memory (see Cognitive overload degrades learning and Cognitive load reduction improves learning).

Boundary conditions. The benefit depends on the visual being corresponding and relevant: decorative images that do not map onto the verbal content can hurt learning, consistent with the coherence principle (see Coherence principle: irrelevant material hurts learning). Benefits are typically strongest for novices, who lack mental models to reconstruct imagery from text alone, and may reverse for advanced learners for whom pictures are redundant — the expertise reversal pattern described in Expertise reversal effect. Learners must also actually attend to and integrate both representations; simply adding a picture without prompting referential connections yields weaker effects. Where the visual itself is complex, pairing it with chunking or a structured advance organizer helps keep combined load manageable.

Open questions. Most supporting work comes from short laboratory or classroom interventions in well-structured domains (science, mathematics). The durability of effects over long retention intervals, and their magnitude in ill-structured domains where imagery is harder to define, remain under-established. Three entries are recorded: a meta-analysis on reading comprehension (Guo et al. 2020), a narrative review read from its abstract (Carney & Levin 2002), and memory experiments that dispute the mechanism (Higdon et al. 2025). Missing are studies of long retention intervals and of recall from combined verbal and visual instruction specifically.

Merged from “Dual Coding Improves Recall” (dual-coding-improves-recall): The mechanism is usually attributed to dual coding theory: verbal and pictorial information are processed in partly distinct cognitive systems, so presenting both creates two memory traces and, when the two are referentially linked, richer retrieval pathways at test. This places the claim squarely in the multimedia learning tradition, where the coherence principle shows that irrelevant material hurts learning — meaning the visual channel must carry relevant, explanatory content rather than decoration for the dual-coding benefit to appear (see Coherence principle: irrelevant material hurts learning.).

Boundary conditions follow from the same theory. Because both channels draw on limited working-memory resources, adding a second representation can overload learners if the material is already complex or the visual is poorly integrated — see Cognitive overload degrades learning. and Chunking reduces working memory load.. The benefit is therefore expected to be strongest for content that is naturally visualizable, for novices who have not yet formed mental models, and for visuals that are explanatory rather than decorative. Abstract or purely symbolic content, and visuals that merely repeat the text verbatim, are expected to show little or no gain. Learner-generated visuals (e.g., drawing or diagramming activities) may deepen the effect by forcing referential connections, but they also impose generation demands that could offset the benefit for novices.

The claim is also modulated by expertise: as learners gain domain knowledge, a second representation can become redundant and stop adding retrieval routes, mirroring the expertise-reversal pattern documented for worked examples and other scaffolds (see Expertise reversal effect). Designers should therefore treat dual coding as a novice-oriented support to be faded rather than a universal default.

Open questions: how durable the recall advantage is over long retention intervals, and how it interacts with learner-generated versus instructor-provided visuals, remain under-specified in the current evidence base. The entries recorded above support a benefit of adding graphics for comprehension, not a recall advantage as such, and one of them (Higdon et al. 2025) disputes that dual coding is the mechanism for the picture-superiority effect in memory.

Related Claims¶

• Coherence principle: irrelevant material hurts learning — added visuals must be relevant, or they impair rather than support learning
• Cognitive overload degrades learning — dual representations help only when combined load stays within working memory limits
• Cognitive load reduction improves learning — the load-management counterpart to adding a second representation
• Chunking reduces working memory load — a complementary strategy for managing representational demands
• Advance organizers improve learning — visual organizers are a common dual-coding implementation
• Analogical reasoning improves transfer — visual analogies pair verbal and pictorial codes to support transfer
• Hispanic adults valued culturally relevant visual design — bright colors, Hispanic representation, and realistic imagery — in lung cancer screening educational materials — a narrower finding that bears on this claim
• Decorative Illustrations Do Not Improve Learning — related
• Multimedia Principle Improves Learning — a broader claim this one bears on
• Learner-constructed graphic organizers are not shown to outperform provided ones: the one direct test, with college readers, favoured provided organizers on transfer — related
• Graphic Organizers Support Novice Comprehension — related
• Presenting words as spoken narration rather than on-screen text alongside graphics improves learning — related
• In one 37-student classroom experiment, recall and retention differ across media combinations in the direction Dual Coding Theory predicts — related
• Combining textual and pictorial glosses improves incidental vocabulary learning more than single-mode glosses — a narrower finding that bears on this claim
• A practice article reports that the studies it cites found better post-test outcomes from words with relevant pictures than from words alone, especially with spoken words — related
• Activation improves learning. — prior knowledge determines whether a second representation adds retrieval routes
• Dual coding theory — the theoretical account of separate verbal and visual memory systems
• Expertise reversal effect — added representations can become redundant for advanced learners
• Typical verbal achievement tests may underestimate students' subject-matter knowledge because of a mismatch between the test's symbolic form and students' symbolic encoding — related
• Dual coding elucidation raises adolescent EFL learners' awareness of lexical errors and correct usage more than verbal-only elucidation — a narrower finding that bears on this claim
• Viewers seek visual complexity as well as its reduction, so immediate perceptual closure is not needed in all graphic images — related
• A WestEd research bulletin reports a curriculum revision that replaced a decorative photograph in a math problem with a problem-relevant sketch, on the developers' view that such photographs distract students — related
• Participants valued multimodal experiences combining visualization with hands-on interaction, while reporting challenges with complex or unclear visualizations — related
• Retrieval context has a sweet spot at Top-K = 10: raising it to 15 did not improve recall but significantly worsened context precision via the distraction issue — related
• Representation-alignment tension between expert externalization and novice comprehension — related
• Working memory includes partially independent subcomponents for auditory/verbal and visual/2- or 3-dimensional information — related
• Concreteness, imagery, and verbal associative processes play major roles across multiple educational domains — a broader claim this one bears on
• Basic DCT mechanisms can be used to model diverse educational phenomena — a broader claim this one bears on
• Learning from visual displays often yields disappointing outcomes because displays pose substantial demands on learners — related
• Low-WMC readers recall fewer correct causes when text contains seductive illustrations compared with nonillustrated text — related
• People acquire different knowledge from different media, the closest indirect evidence for symbolic encoding specificity — related
• Iconic visual scaffolds in chemistry visualizations helped low-prior-knowledge high-school students but not higher-prior-knowledge ones, with the effect eliminated for younger middle-school children — related
• Material-oriented interventions such as complexity reduction, cueing/signaling, and text-picture integration are empirically tested support procedures for visual displays — related
• Working memory capacity interacts with illustration type: seductive illustrations impair comprehension only for low-WMC readers — related
• Persistence of inscriptions in the VMT environment supported students explaining postings and establishing collaborative activity — related
• Generative processing principles: words with pictures and conversational-style words improve transfer over words alone and formal style — related
• Congruent pictures paired with text facilitate reading comprehension better than text alone or pictures alone — possibly the same claim (merge candidate)
• The Multimedia Effect in Problem Solving meta-analysis was retracted in its initial published form after an anonymous reviewer's comments, with the authors stating the coding and interpretation issues do not affect the major findings. — related
• Stop light colors and the combination visualization each improved memory of uncertainty values compared to no uncertainty visualization — a narrower finding that bears on this claim
• Visualizing support—provided diagrams or learner-generated drawings—facilitates higher-quality explanations and better comprehension than explaining text alone — a narrower finding that bears on this claim
• In literature teaching, a method using a sympathetic approach with opportunity for picturing and imaging the experiences in the selection was found superior to one stressing technique, analysis and factual detail — related