Multimodal Learning and the Learning Styles Myth

Jun 17, 2026

If you ever tried to dissuade someone of the learning styles idea, you’ll have probably heard its number one defence; an eyewitness account. It usually takes a familiar shape: someone watched a colleague struggle with a written procedure, saw them understand it the moment it was demonstrated, and concluded they’d found living proof of a kinaesthetic style. The observation is honest, and the frustration behind it entirely understandable, because something did happen; the trouble is the explanation reached for to account for it. A talk Joseph Devlin gave at IDTX: Evidence Informed Practice, prompted me to set out exactly how we might be able to handle this, what we may well be observing if it’s not learning styles, and take a deeper look at multimodal learning in general.

Learning styles, as a matching theory, claims that each person has a preferred mode and learns best when teaching is tailored to it. Multimodal delivery claims that presenting material through more than one sense, and more than one form of representation, improves understanding and memory across the board. The first claim doesn’t survive testing; the second is among the better-supported findings we have. They look similar from the outside, and the perceived power of the first is borrowed, almost entirely, from the well-understood effects of the second.

Paid subscribers gain access to an additional fortnightly omnibus edition containing an additional 4-5 articles, including practical how-to’s, questions from subscribers, and deep dives into industry research, all that for just £35 a year!

Get 5% off forever

A note on terminology

Four words circle this topic and they’re not interchangeable, even though they’re often used as though they are.

“Multimedia,” in the sense Richard Mayer’s research uses it, means combining words and pictures, the verbal and the pictorial (Moreno and Mayer, 2007).

“Multisensory” means engaging more than one of the senses, sight, hearing, touch, and movement among them (Shams and Seitz, 2008).

“Mode” is the form a message takes, verbal or non-verbal, while “modality” is the sense through which it arrives, which is why Mayer and Moreno are careful not to conflate the two (Moreno and Mayer, 2007).

“Sensory” sits underneath all of this as the raw channel, the eye and the ear and the hand.

For this article, I’ll use “multimodal” as the umbrella term for the whole family, meaning the deliberate use of more than one sense, and more than one form of representation, to carry the same idea; the concepts are close enough, and overlap enough in their mechanisms and their practical implications, that grouping them serves the argument without distorting it. Where a distinction changes what you’d do, I’ll be sure to draw attention to it.

What you saw

That observation, the colleague who understood the demonstration after struggling with the document deserves an explanation. Several well-evidenced things could have produced exactly what was seen, none of which needs a fixed personal style:

An accessibility barrier. One format may have carried an obstacle the other removed, an undiagnosed visual difficulty, dyslexia, or a badly laid-out document, so changing format cleared an access problem rather than matching a preference (CAST, 2018).
Dual coding. Information encoded as words and as imagery at once leaves two linked memory traces rather than one, giving more routes back to it later; the demonstration added a visual trace the document lacked (Paivio, 1986).
The modality effect. Spreading information across the visual and auditory channels can ease the load on either one, so a spoken explanation alongside something seen is held more easily than a wall of text alone (Moreno and Mayer, 2007).
Prior knowledge. The format that helps depends on what someone already knows, not on who they are; guidance that supports a novice becomes redundant, and sometimes obstructive, for an expert, an interaction known as the expertise-reversal effect (Kalyuga et al., 2003).
Retrieval. If the second encounter involved recalling or doing rather than rereading, retrieval itself strengthened the memory, whatever format it arrived in (Roediger and Karpicke, 2006).
Spacing. The second format usually came later, and that gap alone improves retention, so the demonstration may have benefited from timing as much as medium (Cepeda et al., 2006).
Variation. Encountering the same idea in more than one form creates the kind of productive difficulty that tends to deepen memory, compared with a single smooth pass through one format (Bjork and Bjork, 2011).
Design quality. Often, one version was better made than the other, clearer and better sequenced, which has nothing to do with the senses involved (Pashler et al., 2008).

A few softer possibilities sit alongside these, harder to isolate in any single study: a different format can be more novel in the moment and so draw more attention; and a second attempt always follows a first, so re-exposure, and the ordinary tendency for a poor first effort to be followed by a better second, both inflate the apparent gain.

What none of these needs is the idea that the colleague is a type of person who learns through one channel; that’s the single explanation the evidence has repeatedly failed to support, including in studies using realistic materials outside the laboratory (Pashler et al., 2008; Rogowsky et al., 2015).

Thanks for reading Instructional Design Tips! This post is public so feel free to share it.

The evidence

Multimodal delivery proves its value through the architecture of perception and memory. Our experience of the world is constantly multisensory, and the human brain seems to have evolved to learn best under those conditions; training that sticks to a single sense doesn’t engage the mechanisms multisensory experience recruits, and so tends to be less effective than training that combines them (Shams and Seitz, 2008). In controlled studies, pairing a sound with a visual signal produced faster learning within a session and stronger retention across sessions than the visual signal alone, and that multisensory experience went on to sharpen later single-sense performance (Shams and Seitz, 2008).

Dual coding theory holds that verbal and visual information run through linked but distinct systems, so material encoded in both leaves richer, more retrievable traces than material encoded in one (Clark and Paivio, 1991). Working memory is similarly divided, with partly separate stores for verbal and for visual and spatial material, so presenting an idea through both draws on more capacity than overloading either alone (Baddeley and Hitch, 1974). The instructional consequence has been tested many times over: people take more from words and pictures together than from words alone, and more from spoken words paired with a visual than from on-screen text paired with the same visual (Moreno and Mayer, 2007).

When adults and children learned foreign-language vocabulary with congruent gestures and pictures, rather than by hearing or reading the words alone, they remembered more, and the advantage held for months rather than minutes, with the non-verbal encoding carrying more of the memory than the verbal (Macedonia and von Kriegstein, 2012; Andrä et al., 2020). These are vocabulary studies, not workplace ones, so what carries across is the principle: an idea delivered through more than one congruent channel is held better than the same idea through one.

Getting it right

The benefit comes from congruent cross-modal encoding, and that phrase is enough to send me reaching for a slightly stronger drink, so in plain terms it means presenting the same idea through more than one sense at once, with the channels agreeing and reinforcing the same meaning rather than competing for attention. The gesture that matches the word, the diagram that shows what the narration describes, the demonstration that enacts the procedure being explained; the senses are saying the same thing.

This is also where older multisensory thinking went wrong, and where a fair challenge to learning styles has to watch its own footing. Practices like assigning someone a sensory channel and teaching them through it, or having people trace letters in sand because the tactile route is supposedly theirs, run on the same matching logic as learning styles and inherit the same lack of support (Pashler et al., 2008). What the evidence backs is congruent, meaningful encoding across channels; sensory activity chosen to fit a supposed personal channel is the part that fails. Restraint matters too, because piling duplicate streams onto one channel, narration read word for word from dense on-screen text, say, adds load rather than removing it, and the benefit shrinks or reverses as people gain expertise and need less support (Moreno and Mayer, 2007; Kalyuga et al., 2003).

In practice

Translating this into workplace material comes down to making the channels you use agree, chosen to fit the task and the audience. The principle looks different across the formats an L&D team is asked to produce, and a sensible programme uses a spread of them instead of defaulting to one.

Take the GDPR compliance module almost everyone’s had to build at some point. Instead of narrating dense on-screen policy text word for word, you might show a simple animated diagram of how personal data moves through a system while a voice explains each step, so the visual and the spoken account carry the same meaning together; a worked example of handling a subject access request, shown as an annotated flow with spoken reasoning, will land better than the same content delivered as text to read and identical narration to hear at once.

A workshop gives you channels e-learning can’t. Working through whether incidents are reportable, you might have participants sort anonymised case cards into “report” and “no report” piles while talking the reasoning through aloud, so the handling and the talk encode the same decision rule, reinforcing one another rather than sitting in separate slides.

A job aid works under different constraints again, because it’s used at the moment of the task, often under pressure. A decision flowchart that pairs short, plain wording with clear icons and a visible path lets someone follow the same logic through two channels at once, the words and the structure agreeing, which keeps it usable when there’s no time to read a manual.

A recorded walkthrough or short video earns its keep when the action and the explanation can run together: demonstrating the steps on screen while narrating why each one matters makes the visual of the action and the spoken reasoning congruent, closer to the in-person demonstration that worked in the first place than any document could be.

The through-line across all four is the same: congruent encoding across more than one channel, chosen for the content rather than for any person, and used with enough restraint that the channels reinforce each other rather than crowd them out. Sequence these formats across a programme so people meet an idea more than once, in more than one form and with gaps between, and you buy the spacing and retrieval benefits too, a fair amount of evidence-informed value from decisions that mostly cost nothing extra to make.

The colleague who finally understood the task once it was demonstrated was telling the truth about what they experienced; they were wrong only about the cause. Deliver the same idea through more than one sense, make the senses agree, and you help everyone, a far steadier foundation for design than sorting people into types.