Skip to content

Multimedia Learning

Pattern · All patterns
Evidence · 85 claims (38 for, 45 mixed, 2 against) · 65 studies (36 causal, 11 review, 10 quant-synthesis, 3 theoretical, 2 associational, 2 design, 1 qualitative), q1–q4 · 18 of 65 report an effect size · 76 claims rest on one study

Description and scope

A reusable policy for deciding whether an explanation needs a picture, what the picture should show, where its words go (on it, beside it, spoken or written), and how the learner controls the pace. It instantiates the conditional principle. The study configurations below are evidence; the response-dependent policy is an untested design proposal. Use it to gather better evidence about a learner's understanding, not to certify a learning style or a latent state.

Inputs: establish the design brief

Field Record or elicit
Learner and starting observation What the learner can already say or do about this structure or process (an explanation, a labelled sketch, a prediction), with prior knowledge of this content, reading and listening access, and any uncertainty. Preserve unknowns.
Context and activities Medium and device, whether learners can pause, replay or re-read (system-paced or learner-paced), captioning or accessibility requirements, available time, and what the learner does after viewing.
Learner-valued goal Ask what the learner wants to be able to do with this and why. Record disagreement with the designer objective; negotiate rather than assume agreement.
Designer objective The capability intended (recall, explanation of a mechanism, troubleshooting, a procedure) and the knowledge type (concept, procedure, principle, complex-skill); whether the picture will be available when the capability is used.
Outcome Instrument, locally chosen criterion, immediate or delayed horizon (immediate-performance, delayed-retention, near-transfer, far-transfer), and any separate use outcome.

A request to "make it more engaging" or "add visuals" does not specify these inputs. Ask for them before choosing a format, a segment length or a predicted gain. Check first that the picture is correct and depicts what the words describe; a misleading diagram studied carefully is not progress.

Sequence and conditional policy

  1. Elicit briefly. Ask the learner to explain or sketch how the thing works, or to predict what happens when one part changes, before the lesson. Record the response and any help given; this probe is itself an exposure.
  2. Decide whether a picture has a job. If the explanation concerns a structure, process or spatial relation the learner cannot yet describe, consider adding a picture that shows that relation. If the content is a list of verbal facts, or the learner already explains it well, a picture may add little; the evidence below does not require one for every learner.
  3. Configure the candidate. Where the material has many named parts and the learner is new to it, name the parts briefly first (pretraining). Choose a single static picture to show structure and an animation to show change over time. Write labels and captions into the picture at the part they describe; do not present the same words as narration and as full on-screen text over a visualization; leave out decorative or interesting-but-irrelevant images; where the presentation runs continuously and the learner is new to it, consider pause points or learner control. Record exactly what was delivered.
  4. Observe and choose the next activity provisionally. Use the table below, varying one feature at a time where feasible; placement, pacing, channel and prior knowledge can confound each other.
  5. Reobserve. Use a new representative task under the agreed aids, criterion and horizon. Retain unresolved explanations. If the response disagrees with the expectation, revise the interpretation or configuration rather than adding more media.
  6. Across sessions, withdraw support as the learner explains unaided (proposal, untested as a schedule). Early sessions: pretraining, short segments of a few minutes with a pause or "next" between them, labels in the picture, narration for system-paced material, and explicit guidance on when to pause and replay. Once the learner answers a new "what if" question correctly without supports on two occasions, lengthen the segments, drop the pretraining and offer the picture with narration on request, then move to application tasks. Basis for the direction: segmentation helps less as prior knowledge grows [~M] and integration helps less for more experienced learners [~M]; the segment length and the switching criterion are proposals, and no claim in the wiki says how many sessions the withdrawal takes.
Observation Discriminating probe Proposed action, with its uncertainty
Can repeat the words but not explain how parts relate Ask for a "what if this part changed" prediction with the picture present, then withdrawn. If the prediction fails with the picture present, revise the picture to show the relation, or prompt the learner to explain the link between words and picture. If it fails only when withdrawn, the objective decides whether independence from the picture is required. Neither branch is a validated diagnosis.
Reports the screen as hard to follow; looks back and forth Re-present one segment with labels placed on the picture and no duplicated on-screen text; compare effort and the explanation produced. If integration helps, apply it across the lesson. If not, check pacing and language access before changing anything else. One comparison does not establish the cause.
Misses spoken content in a narrated animation Offer pause control or the same words as re-readable text with the same picture; check comprehension of the words alone. Keep narration where learners can follow it; where they cannot (language, hearing, noisy setting, dense terminology), keep written text or short key terms, which the evidence does not show to be harmful without a competing visualization. Captions required for access stay. For children with cochlear implants, seeing the speaker as well as hearing them raised sentence recognition for most [+W] (23 of 27), so keep the speaker's face visible where speech carries the content.
Agrees with whatever the picture seems to show Mix in statements that are false of the picture, and ask the learner to say what in the picture supports each answer. If false statements are accepted, the picture is lending credibility rather than information: graphs made students confirm statements regardless of truth [~W]. Prompt the learner to check claims against the picture before adding more pictures; one eye-tracking study, so treat as a hypothesis.
Likes the illustrated version but explains no better Remove decorative images, keep the informative picture, compare explanation quality. Drop images that do not inform. Liking, mood and interest are not learning outcomes, although they may matter to the learner.
Explains and predicts well from the start Give a changed-condition question without supports. Reduce segmentation and extra support provisionally and move to application; the reversal evidence is from one support in one study, so check again rather than assume.

Choosing configurations from evidence

Do not convert η² or a regression β into a probability for a learner, rank these unlike comparisons by effect label, or predict delayed use from an immediate test. A precise abstention names the missing comparison, target or observation and the next way to resolve it.

Fitting the design to a situation

Ask for four facts when the brief omits them: who controls the pace (system-paced video or lecture, or learner-paced pages), how complex the material is (how many parts must be held in mind together), the learners' prior knowledge and reading fluency in the language of instruction, and whether the outcome is retention or transfer. Each row changes step 3 (configure the candidate) unless it says otherwise.

If the brief says… Then change… Basis
A system-paced video, lecture or narrated animation Apply the configuration rules strictly (words beside the picture, narration rather than printed text, no extras); the design principles matter more when learners cannot pause Design principle effects are moderated by pacing and complexity [+M]; Pacing moderates the modality effect [+M]
Learner-paced pages, slides or a website Keep the words printed next to the picture; do not convert them to narration, which showed no benefit (slightly negative) when learners set the pace Pacing moderates the modality effect [~M]
Complex material: a process with many interacting parts, and learners new to it Integrate each label or sentence into the picture at the point it describes, present spoken words with the matching part of an animation, and name the parts briefly before the lesson Integration benefit is larger for high element-interactivity material [+M]; Integrating related elements gives a large average benefit for novices [+M]; Pretraining improves transfer [+M] (one VR experiment)
Learners already know the content well Cut explanatory text that repeats what the diagram shows, and reduce segmentation; check on a transfer item rather than assuming Integration benefits may reverse for more experienced learners [~M]; Segmentation benefits shrink with expertise [~M]; The integrated-illustration benefit was stronger for low-experience students [~M]
Second-language readers or less fluent readers, and the material has no picture or animation Give the words both spoken and written Spoken–written advantages are larger for L2 and nonfluent readers [+M]; Verbal redundancy does not help when images or animation are included [~M]
A narrated animation carries the full narration as on-screen text Remove the duplicated text, or keep only key terms beside the picture; keep captions where access requires them Redundancy principle [+M]; Verbal redundancy does not help when images or animation are included [~M]
Video in a second language, for language learners Add captions in the target language Captioning second-language videos produces large benefits [+M]; captioned video improved L2 listening, generalising to new sentences and speakers [+W] (reported second-hand)
The brief asks for engaging extras: stock photos, background music, anecdotes Leave out what the explanation does not need, and replace decorative images with ones that show the relation being explained Coherence principle [+M]; Interesting but irrelevant details can impair learning [~M]
The choice is between a static diagram and an animation Choose by what the learner must identify: an animation for how parts move, a single static picture for how they are arranged (not a strip of snapshots, which helped neither); if an animation is used, prefer it system-paced with spoken commentary. The integration rules apply equally to both Animation beats static pictures for identifying motions [+M]; a single picture beats animation for spatial arrangements [~M]; animation advantage larger when system-paced [~M] and with auditory commentary [~M]; Contiguity effects do not differ between static and dynamic presentation [~M]
A long continuous presentation, and learners are new to the content Break it into meaningful segments with a pause or "next" between them; allow for more study time Segmenting improves retention and transfer [+M]; Segmentation increases learning time [~M]
The plan relies on learners using pause and replay controls Build the pauses into the presentation, or tell learners when to pause and replay (after each step, when a term is new), rather than counting on them to use the controls unprompted Learner flow control over animations was rarely used [~W]; strategy guidance beat unguided learner control [+M] (immediate tests only)
Learners will study the animation in pairs or small groups Animation is more worth its cost here; avoid adding snapshot strips of the key steps for pairs Rebetez et al., one experiment, no effect sizes printed: transfer benefited from animation only for learners studying collaboratively [~W]; snapshots were marginally helpful alone and harmful in dyads [~W]
A digital storybook for young children Keep animated illustrations, sound and music that match the story; leave out hotspots, games and other interactive features Multimedia-only stories helped comprehension (g+ = 0.39); stories that also had interactive features did not [+M] (meta-analysis). The gain was larger in studies of children the authors classed as disadvantaged (g+ = 0.66): a difference reflecting circumstances, not a property of the children or a cause
Choosing between a video and an illustrated text For children and slower readers, consider a narrated video, and check comprehension rather than assuming it; for university students integrating several sources, choose on access and time, since the medium made no difference to understanding Fifth graders studied videos faster than equivalent illustrated texts [~W], most of all the slowest readers [+W] (studying rate, not learning); with 255 university students, medium had no effect on integrating complementary sources [~M] and videos took longer to process [~W] (d = 0.52)
Narration, animation or slides generated by an AI tool Check before use that the picture keeps the concept's defining constraint, that the synthetic voice reads numbers and pauses correctly, and that each slide element appears when the narration reaches it Educators' focus groups: generated animations can drop a defining constraint [~W]; expert reviewers of generated lecture videos flagged voice, reveal timing and layout [~W]. Neither measures learning
An immersive VR headset lesson Do not assume narration wins: pilot spoken and short anchored printed text on the brief's own test. Name the equipment's parts before the VR session Conflicting, as on the modality principle: the modality effect was larger with head-mounted displays [~M], but in a VR experiment with 61 learners printed text beat narration on recall [-M]; pretraining before VR improved real-task transfer [+M]
The outcome is transfer, not recall Expect smaller gains from adding written text to narration, and judge the configuration on a transfer item in step 5 Spoken–written advantages are larger for retention than transfer [~M]

The evidence by question

Further evidence, not yet read against this model

Claims linked to this page later go here until they are read against the model. - Dual Language Programs Gap Closure Longitudinal [+M] — (not about this page) read 2026-10-19: a second-hand report of programme-type outcomes in dual-language schooling; says nothing about words and pictures. - All ten multimedia treatment conditions produced significant pre-to-post achievement gains with medium or high effect sizes [+W] — read 2026-10-19: pre-to-post gains within each condition, no comparison between configurations, so it cannot choose one; the between-condition tests from the same study are used above. - Diagram-comprehension scores were already high before the treatment and showed no significant pre-to-post increase [-W] — read 2026-10-19: a 30-person evaluation of a tool for inspecting qualitative models, at ceiling; not a test of pairing words with pictures. - Anatomy images were more likely than evolution images to have direct placement only [~W] — read 2026-10-19: a content analysis of textbook images, with no learner outcome. - Readers judge a diagram's argumentative position above chance in all conditions except when both text and color are absent, and deleting graphics is more detrimental than deleting text [~W] — read 2026-10-19: perception of argument diagrams in 4-second judgments, not learning from an explanation; does not change the model. - Prior research failed to find clear evidence that animated graphics improve comprehension over static graphics [~W] — read 2026-10-19: an article's introductory characterisation of earlier work, no new data; the animation-or-static rule above rests on the experiments and meta-analysis. - Students adapted representation choices to task: the model was viewed more for the three-dimensional dihedral animation, the graph more for the two-dimensional bending and stretching animations [~W] — read 2026-10-19: ten chemistry students choosing among linked representations; multiple representations are scoped out of this page. - Learning with multiple representations imposes high demands and often fails to produce expected results because learners do not integrate the representations [~W] — read 2026-10-19: an article's motivating statement with no data printed; consistent with the scope note on multiple representations. - In Chun and Plass's CyberBuch studies, the text + picture annotation group significantly outperformed text-definition and text + video groups on a vocabulary test [+W] — read 2026-10-19: vocabulary glosses, reported second-hand; language-learning glosses are scoped out of this page. - In Yoshii's study, text + picture glosses beat text-only glosses only on a definition-supply test, with no L1 versus L2 gloss difference [+W] — read 2026-10-19: vocabulary glosses, reported second-hand; scoped out with the item above. - The segmentation advantage is maintained across repeated readings and is widespread across participants [~W] — read 2026-10-19: "segmentation" here is printing text in phrase units for children's reading, not dividing a narrated presentation, so it does not bear on the segmentation rule.

Illustrative design instance and observation record

Invented illustration, not a tested lesson. Adult apprentices in a workplace programme study a short narrated animation of a pump cycle. One apprentice says they want to work out why a pump on their own site keeps failing; the designer's objective is a stage-naming quiz. Elicit the difference and agree whether explaining a fault matters, then choose a representative task (predict what a stuck valve does to the cycle) and decide whether the diagram is available at test. Before the lesson, the apprentice names parts but cannot say what a valve does to the flow. The candidate configuration places labels on the diagram, speaks the explanation, adds pause points, and leaves out a stock photo of a plant room. If the apprentice still cannot predict the fault with the diagram present, revise the diagram to show flow direction before adding anything else; assess delayed application separately if it is a target.

Record: material and its configuration (picture, placement, channel, pacing, segments) → permitted and actual help → response or explanation → candidate interpretations and confidence basis → discriminating probe → chosen activity and justification → next observation at the stated horizon. Preserve the learner's stated purpose and any change in it. An agent may summarize this record, but must not fabricate missing inputs or assign numerical state probabilities without a calibrated model.

Elements and limits

Multimedia learning, multimedia instruction, digital learning, personalized pacing, self-explanation, pedagogical agent signaling and accommodations.

This pattern is scoped to explanations of structures, processes and relations presented through words and pictures. Learner-made representations, interactive simulations (a field study in undergraduate electrodynamics found sketching and simulation tasks alongside the formula [~M] beat calculation-only tasks, d = 0.40, while raising perceived cognitive load, and a review there finds a second visual representation helps only with the learner's skill at connecting representations), video modelling of skills and language-learning glosses need their own configurations and evidence. Most of the recorded evidence comes from short lab lessons with university students; the policy supports observation and design reasoning, and its predictive accuracy and the learner-significance mechanism remain to be tested with actual learners.

Claims

Examples

Key Sources

  • Mayer, R. E. (2009). Multimedia learning (2nd ed.). Cambridge University Press. doi:10.1017/cbo9780511811678
  • Kutbay, E. & Akpınar, Y. (2020). Investigating modality, redundancy and signaling principles with abstract and concrete representation. International Journal of Education in Mathematics, Science and Technology (IJEMST), 8(2), 131-145. https://eric.ed.gov/?id=EJ1255527
  • Marjorie Ceballos, Joyce W. Nutta. (2022). Emergent Bilinguals and Multimedia Instructional Design: Applying the Science of Learning Principles to Dual Language Instruction. Education Leadership Review, 23(1). https://eric.ed.gov/?id=EJ1380081
  • Mayer, R. E., Steinhoff, K., Bower, G., & Mars, R. (1995). A generative theory of textbook design: Using annotated illustrations to foster meaningful learning of science text. ETR&D, 43, 31–41. doi:10.1007/BF02300480
  • Rebetez, C., Betrancourt, M., Sangin, M., & Dillenbourg, P. (2010). Learning from animation enabled by collaboration. Instructional Science, 38(5), 471–485. doi:10.1007/s11251-009-9117-6
  • Berney, S., & Bétrancourt, M. (2016). Does animation enhance learning? A meta-analysis. Computers & Education. doi:10.1016/j.compedu.2016.06.005
  • Ploetzner, R., Berney, S., & Bétrancourt, M. (2021). When learning from animations is more successful than learning from static pictures: learning the specifics of change. Instructional Science. doi:10.1007/s11251-021-09541-w
  • Lin, Y.-C., Liu, T.-C., & Kalyuga, S. (2022). Strategies for facilitating processing of transient information in instructional videos by using learner control mechanisms. Instructional Science, 50, 863–877. doi:10.1007/s11251-022-09600-w
  • Takacs, Z. K., Swart, E. K., & Bus, A. G. (2015). Benefits and pitfalls of multimedia and interactive features in technology-enhanced storybooks: A meta-analysis. Review of Educational Research, 85(4), 698–739. doi:10.3102/0034654314566989
  • Delgado, C. Y., & Mayer, R. E. (2024). Implementing pretraining to optimise learning in immersive virtual reality. Journal of Computer Assisted Learning, 41(1). doi:10.1111/jcal.13099
  • Jaeger, A. J., & Fiorella, L. (2024). Metacognitive effects of instructional visuals: the role of cue use and judgment type. Metacognition and Learning. doi:10.1007/s11409-023-09370-x
  • Ögren, M., Nyström, M., & Jarodzka, H. (2017). There's more to the multimedia effect than meets the eye: is seeing pictures believing? Instructional Science, 45, 263–287. doi:10.1007/s11251-016-9397-6
  • Adesope, O. O., & Nesbit, J. C. (2012). Verbal redundancy in multimedia learning environments: A meta-analysis. Journal of Educational Psychology. doi:10.1037/a0026147
Page metadata
Field Value
type pattern
id multimedia-learning
title Multimedia Learning
description A reusable policy for pairing an explanation's words with an informative picture, choosing placement, channel, pacing and segmentation from the learner's observed response and the intended outcome.
status review
grain_size lesson
author multimedia learning tradition
moved_from principle/leader-criteria-sol-alignment-digital-materials, principle/prefer-narration-over-on-screen-text-middle-school
canonical True
generated by claude/unspecified on 2026-10-19
trust unverified — no one has confirmed this page's content
sources 2 — kutbay-2020, marjorie-ceballos-2022

Frontmatter is the page's machine-readable record — see the Schema & Guide.