How Cognitive Science and Neurodiversity Research Should Reshape the Way We Teach Complex Ideas
Open almost any online course, corporate training module, or educational slide deck in 2026 and you will find the same default gesture: dense content broken into bullet points. The bullet is the visual idiom of modern learning design. It signals clarity. It promises ease. For many of us, it is the first formatting move we make when a paragraph starts to feel “too long.”
Yet decades of cognitive science suggest that this default is often wrong — not slightly wrong, but consequentially wrong for the kinds of learning we say we care about most. The bullet is excellent at one thing (quick reference) and poor at something else entirely (building durable understanding of connected ideas). When we confuse these two goals, we produce materials that feel educational while failing to educate.
This article makes the case, from the research literature, for a more careful approach to formatting complex material — one that treats format not as decoration but as a cognitive variable that directly shapes what learners take away. We will look at what working memory can and cannot do, why prose and bullets operate on different cognitive systems, and what research on neurodivergent learners reveals about a common but mistaken assumption: that fragmenting information is always an act of accessibility. The truth, as is so often the case, is more interesting than the folk wisdom.
The Architecture of Working Memory
The foundation for any serious discussion of learning design is the architecture of human memory. The consensus model across decades of cognitive research describes three layers: sensory memory, which briefly holds incoming stimuli; working memory, where we actively manipulate information; and long-term memory, the vast storehouse where durable knowledge lives.
Working memory is the bottleneck. It is small, fast, and unforgiving. George Miller’s classic estimate of “seven plus or minus two” chunks has been refined downward — in the context of active processing, working memory appears to hold closer to two to four genuinely independent chunks at once, with a functional duration measured in seconds unless the information is actively rehearsed or encoded into a schema.
This matters for instructional design because every decision about presentation is also a decision about what working memory has to do to make sense of the content. John Sweller’s Cognitive Load Theory, first articulated in the late 1980s and continuously refined since, describes three kinds of demand competing for this limited resource.
Intrinsic load is the inherent difficulty of the material itself — the number of interacting elements a learner must hold in mind simultaneously. A vocabulary word has low intrinsic load. The relationship between monetary policy, inflation expectations, and consumer behavior has high intrinsic load, because the elements cannot be understood in isolation.
Extraneous load is the cognitive cost imposed by how information is presented, regardless of content. A confusing layout, a decorative image that adds no meaning, or a formatting choice that forces the learner to mentally reconstruct relationships the author should have made explicit — all consume working memory that could otherwise be spent on learning.
Germane load is the effort the learner devotes to building schemas — the mental structures that convert new information into durable knowledge. This is the productive load, the effort that actually produces learning.
The instructional designer’s job, under this framework, is to minimize extraneous load, manage intrinsic load through careful sequencing, and create conditions for germane load to take over. Formatting is not separate from this work. Formatting is this work.
Mayer’s Principles: Coherence, Signaling, Segmenting
Richard Mayer’s Cognitive Theory of Multimedia Learning builds directly on this foundation. Across more than two hundred controlled experiments over four decades, Mayer and his collaborators have developed evidence-based principles for how words, images, and structure combine to produce — or prevent — meaningful learning. Three of these principles are especially relevant to formatting.
The coherence principle states that people learn better when extraneous material is excluded. Decorative images, off-topic anecdotes, and what Mayer calls “seductive details” — interesting but irrelevant content — consistently reduce learning outcomes. The principle has a corollary that matters: removing material should not mean severing meaningful connections. There is a difference between stripping away the unnecessary and stripping away the connective tissue that makes ideas cohere.
The signaling principle holds that learners do better when cues highlight the organization of essential material. Headings, bold emphasis on key terms, and clear transitions all function as signaling devices. Importantly, signaling works because it accompanies prose, not because it replaces it. A well-placed bold term inside a paragraph of connected reasoning is a signal. A fragmented list of bolded terms with no surrounding argument is just noise.
The segmenting principle recommends breaking complex material into learner-paced chunks. But segmenting means controlling the rate at which new material arrives — not shattering the internal logic of any single idea. A well-segmented lesson presents one coherent concept fully, lets the learner process it, and then moves to the next. A poorly fragmented lesson delivers the same information as scattered disconnected bullets and calls it “chunked.”
This distinction — between segmenting and fragmenting — is where much modern learning design goes wrong. The bullet point has become a way to perform clarity without producing it.
What Bullets Actually Do, and When They Fail
The research on bulleted versus prose formatting is more nuanced than either camp sometimes suggests. Bullets genuinely support working memory for certain content. They pre-chunk information into discrete units, reducing the extraneous load of extracting structure from continuous text. For parallel, discrete items that share a common frame — ingredients, safety checklists, platform features — bullets outperform prose in both comprehension speed and retention.
But that efficiency advantage has a narrow domain. Bullets strip away the connective tissue of language: the because, and so, in contrast, which led to, however. These small words do enormous cognitive work — they tell the reader how ideas relate. Without them, the learner must infer the relationships, and inference is exactly the kind of effort that consumes working memory. When the relationships between ideas are the thing being taught, forcing the reader to reconstruct them from fragments is the opposite of help.
This is why research comparing bullet and prose formats for narrative, argumentative, and conceptual content consistently favors prose — especially prose with strong signaling through bold anchor terms, clear paragraphs, and visible structure. A 2018 review of presentation research concluded that bullet-point lectures often combine spoken words and displayed text in ways that actively decrease comprehension, because the learner’s working memory must reconcile two partially overlapping inputs rather than processing one coherent signal.
The test, then, is not “bullets or prose” but “what kind of content is this?” Three questions help. Are the items genuinely parallel and discrete, or connected steps in an argument? If connected, prose serves them better. Will the learner read this to learn, or scan it to reference later? Reference material can be bulleted; learning material usually should not be. Would the ideas still make sense shuffled into a different order? If yes, bullets are fine. If no, the order is carrying meaning — and meaning needs prose.
The Fluency Illusion and Desirable Difficulty
A subtler problem with the reflexive move toward bullets concerns the fluency illusion. When information feels easy to read, learners tend to believe they have understood and retained it — even when objective measures show they have not. Fluent-feeling formats produce overconfidence. They generate the subjective experience of learning without the underlying cognitive work that produces durable knowledge.
Robert Bjork’s research on desirable difficulties points in the opposite direction from our formatting instincts. Bjork’s work shows that certain kinds of effort — effort that feels slightly harder in the moment — consistently produce better long-term retention and transfer than the easier alternatives. Relevant here, coherent prose that requires the reader to track relationships outperforms pre-chunked fragments that do the tracking for them.
This does not mean harder is always better. There is no learning benefit to gratuitous difficulty. The research is specific: difficulty that engages the learner in constructing meaning produces durable learning; difficulty that merely obstructs meaning does not. A well-written paragraph that asks the learner to follow a causal chain is a desirable difficulty. A wall of text with no paragraph breaks or signaling is just bad design.
The practical implication is uncomfortable for course creators who want their material to feel effortlessly easy: some of what we should be optimizing for is not the learner’s ease in the moment but their understanding two weeks later. Those are not the same thing.
Designing for Neurodivergent Learners — What the Research Actually Says
Perhaps the most consequential — and most often misunderstood — area of this research concerns neurodivergent learners. Between fifteen and twenty percent of the population, by conservative estimates, is neurodivergent: ADHD, autism spectrum, dyslexia, dyspraxia, dyscalculia, or some combination. Some estimates run considerably higher. These learners are in every classroom and every course.
The folk wisdom about designing for this audience is partly right and partly wrong in ways that matter.
What is right: neurodivergent learners frequently struggle with long, dense, unbroken blocks of text. A 2024 focus group study published in PLOS One found that neurodivergent university students consistently reported that reading long-form online content required significantly more effort to stay focused, often requiring re-reading the same passage multiple times before comprehension. The same study found that neurodivergent students experienced higher cognitive load across almost every common online learning activity — reading, listening, watching, and note-taking — than their neurotypical peers.
What is wrong, or at least oversimplified: the conclusion that the solution is to fragment everything into bullets. The PLOS One research, and the broader neurodivergent learner literature, consistently points toward a different set of interventions: shorter paragraphs rather than longer ones; sans-serif fonts with adequate line spacing; clear signaling through headings and bold anchor terms; plain, literal language rather than metaphor and idiom; and — critically — multiple modalities, so that text can be paired with audio, visuals, or interactive elements rather than replaced by compressed lists.
The reason is that fragmentation without structure often increases cognitive load for neurodivergent learners rather than decreasing it. A long list of bullets requires the reader to hold the overarching frame in working memory while processing each item. For a learner with ADHD, whose executive function is already under strain, that frame can evaporate by bullet number five. Prose paragraphs with strong signaling provide the frame in the text itself — the connective words, the transitions, the causal language — which reduces the executive burden of keeping the structure in mind.
Readers with dyslexia benefit from short paragraphs, high-contrast sans-serif typography, and clear language. Readers on the autism spectrum benefit from predictable layouts, reduced sensory clutter, and literal phrasing. Readers with ADHD benefit from content that is easy to enter, easy to resume after interruption, and that offers genuine structural markers to re-orient after drift.
The design implication is nuanced: inclusive design is not the same as minimal design. It means making the cognitive structure of the material visible and predictable, so that the reader’s executive function is freed up for understanding the content rather than parsing the format.
Universal Design for Learning
The most developed framework for putting these ideas together is Universal Design for Learning (UDL), developed by the Center for Applied Special Technology (CAST) and now in its 3.0 iteration, released in 2024. UDL organizes inclusive design around three principles: multiple means of representation (the what of learning, how content is perceived and understood), multiple means of engagement (the why, addressing motivation and persistence), and multiple means of action and expression (the how, addressing how learners demonstrate understanding).
UDL is not a prescription for any one format. It is an argument for designed variability — for building flexibility into the material from the start rather than retrofitting it after the fact for learners who struggle. The philosophical move, and it is a significant one, is to locate barriers to learning in the design of the environment rather than in the deficits of the learner. When a learner fails to thrive, UDL first asks what in the materials and methods created the barrier — and then redesigns around it.
This reframing matters especially for the neurodivergent population, because the alternative — a single “standard” format that works for the notional average learner — systematically excludes a substantial fraction of any course audience. UDL’s insistence on multiple means is not an indulgence; it is an acknowledgment that learner variability is not the exception but the rule.
There is one further wrinkle worth naming. The expertise reversal effect, documented by Slava Kalyuga and colleagues across many domains, shows that instructional techniques which help novices can actively hurt more advanced learners, and vice versa. The heavy scaffolding that rescues the novice becomes redundant noise for the expert. The spare, elegant prose that lets the expert apply existing knowledge leaves the novice stranded. A single format cannot optimize for everyone. This is another argument for layered depth — a clear, well-signaled prose spine for core content, supplemented by optional detail, worked examples, and follow-up experiments that learners can engage with or skip.
A Working Framework for Course Creators
Pulling this research together:
Default to prose for conceptual, narrative, and argumentative content. Any time the relationships between ideas are themselves what you are teaching — causes, consequences, comparisons, processes with meaningful sequence — prose will serve your learners better than bullets. Keep paragraphs short, one idea at a time, with clear topic sentences.
Use signaling aggressively. Bold anchor terms inside prose. Descriptive headings that preview what follows. Short introductory sentences that orient the reader before dense content arrives. These devices reduce extraneous load without sacrificing the connective tissue that makes ideas cohere.
Reserve bullets for the content they actually serve. Parallel, discrete items. Reference material the learner will scan rather than read. Checklists and procedures where order must be unmistakable. When in doubt, ask whether the items would still make sense shuffled into a different order. If yes, bullets are fine. If no, prose is the honest choice.
Write for the modal learner, but layer for variability. A clear prose spine accessible to your primary audience, supplemented by optional depth, worked examples, glossaries, audio, and follow-up experiments. This honors the expertise reversal effect without abandoning learners on either end of the knowledge spectrum.
Treat accessibility as structural, not cosmetic. Sans-serif fonts, high contrast, short paragraphs, plain language, and multiple means of accessing the same content. These choices serve neurodivergent learners, English Language Learners, and anyone reading under fatigue or stress — which, over a course’s lifetime, is most learners at some point.
Test your assumptions with the people affected. No amount of research substitutes for feedback from actual learners, particularly those whose processing needs differ from your own. A layout that looks clean to a designer may feel disorienting to a reader with dyslexia. A page that feels efficient to a product team may be exhausting to a reader with ADHD.
Formatting as Respect
There is a final argument to make, and it is not entirely empirical. The way we format learning material is, at some level, a statement about what we believe learners are capable of.
When we reflexively reduce every idea to a bullet list, we communicate — whether we intend to or not — that we do not trust our learners to follow an argument, track a causal chain, or sit with a paragraph long enough to understand it. We optimize for the appearance of ease rather than the reality of understanding.
The alternative is not denser text or harder prose. It is more deliberate prose. It is respecting the learner enough to give them connected ideas rather than fragments, and respecting them enough to make the connections visible through careful signaling. It is designing for the full range of learners who will encounter the material — neurodivergent and neurotypical, novice and expert, stressed and rested — without collapsing that variability into a single false default.
Cognitive science has been telling us for forty years that working memory is small, that schemas matter, that prose carries connective tissue that bullets strip away, and that genuine accessibility means designed variability rather than minimum viable formatting. The bullet point is not the enemy. The reflex is. Learning design deserves the same craft we bring to any other creative discipline — which means choosing format on purpose, for the content at hand, and for the learners who will meet it.
The medium of learning is not only what is said. It is the shape the saying takes.
References
Bjork, R. A., & Bjork, E. L. (2011). Making things hard on yourself, but in a good way: Creating desirable difficulties to enhance learning. In M. A. Gernsbacher, R. W. Pew, L. M. Hough, & J. R. Pomerantz (Eds.), Psychology and the real world: Essays illustrating fundamental contributions to society (pp. 56–64). Worth Publishers.
CAST. (2024). Universal Design for Learning Guidelines version 3.0. Wakefield, MA: Author. https://udlguidelines.cast.org/
Kalyuga, S. (2007). Expertise reversal effect and its implications for learner-tailored instruction. Educational Psychology Review, 19, 509–539.
Kalyuga, S., Ayres, P., Chandler, P., & Sweller, J. (2003). The expertise reversal effect. Educational Psychologist, 38(1), 23–31.
Mayer, R. E. (2021). Multimedia learning (3rd ed.). Cambridge University Press.
Mayer, R. E. (2024). The past, present, and future of the Cognitive Theory of Multimedia Learning. Educational Psychology Review, 36(8).
Mayer, R. E., & Moreno, R. (2003). Nine ways to reduce cognitive load in multimedia learning. Educational Psychologist, 38(1), 43–52.
Miller, G. A. (1956). The magical number seven, plus or minus two: Some limits on our capacity for processing information. Psychological Review, 63(2), 81–97.
Nerantzi, C., Chatzidamianos, G., Stathopoulou, H., & Karaouza, E. (2024). Neurodiversity and cognitive load in online learning: A focus group study. PLOS One.
Paas, F., & Sweller, J. (2012). An evolutionary upgrade of cognitive load theory. Educational Psychology Review, 24(1), 27–45.
Sweller, J. (1988). Cognitive load during problem solving: Effects on learning. Cognitive Science, 12(2), 257–285.
Sweller, J. (2010). Element interactivity and intrinsic, extraneous, and germane cognitive load. Educational Psychology Review, 22(2), 123–138.
Sweller, J., van Merriënboer, J. J. G., & Paas, F. (2019). Cognitive architecture and instructional design: 20 years later. Educational Psychology Review, 31, 261–292.
Tobin, T. J., & Behling, K. T. (2018). Reach everyone, teach everyone: Universal Design for Learning in higher education. West Virginia University Press.