What Is Universal Design for Learning (UDL)?

Universal Design for Learning (UDL) means designing a lesson from the start so that it works for every student, instead of designing it for an average student and patching it later for whoever does not fit. It is the same idea as a ramp on a building: it goes into the plans, not in at the end.

It was developed by CAST, an American education research organization, borrowing an idea from architecture. And it is probably the most cited and least well applied framework in inclusive education today, so this article also covers what the evidence says about whether it works, which is not what is usually said.

Key points

• UDL comes from architecture. The architect Ronald Mace coined "universal design" in 1985 to describe products and environments usable by everyone "without the need for adaptation or specialized design".
• There are three principles, set out by CAST: multiple means of engagement, of representation, and of action and expression. The current version of the guidelines is 3.0, released in July 2024.
• The difference from accommodations is timing, not content. An accommodation comes afterwards and targets one student; UDL is decided beforehand and targets the design of the lesson.
• UDL does not remove individual accommodations. The international definition of universal design itself says it does not exclude assistive devices where needed.
• The evidence is contested. One meta-analysis finds a moderate positive effect, and a published critical analysis concludes that the evidence CAST cites is weak. Both are true at once.
• The strongest criticism is not "it doesn't work" but "we don't know what it is". Studies that call themselves UDL do very different things, which makes the framework hard to measure as such.

Where it comes from: first there were ramps

UDL was not born in a school of education. It was born in a school of architecture.

The American architect Ronald L. Mace (1941–1998), who used a wheelchair after contracting polio at nine, coined the term universal design in an article published in November 1985 in Designers West (Mace, R. (1985). Universal Design: Barrier Free Environments for Everyone. Designers West, 147–152). The definition he gave is the one still circulating forty years later:

Universal design is the design of products and environments to be usable by all people, to the greatest extent possible, without the need for adaptation or specialized design.

That sentence is documented and dated in Eric Dolph's academic review The developing definition of universal design (Journal of Accessibility and Design for All, 11(2), 2021), which traces the history of the definition. The archive at North Carolina State University, which holds his papers, confirms it another way: Mace coined the term "universal design", and in 1989 he founded there the federally funded center that became the Center for Universal Design (Ronald L. Mace Papers, NC State University Libraries).

The example always used to explain it is the curb cut. It is put in for wheelchairs and ends up used by the delivery driver with a cart, the parent with a stroller and the traveler with a suitcase. It is not a concession to a minority: it is better design.

The same sentence ended up in an international treaty

Mace's definition did not stay within the profession. In 2006, the United Nations General Assembly adopted it almost word for word in Article 2 of the Convention on the Rights of Persons with Disabilities:

"Universal design" means the design of products, environments, programmes and services to be usable by all people, to the greatest extent possible, without the need for adaptation or specialized design. "Universal design" shall not exclude assistive devices for particular groups of persons with disabilities where this is needed.

That second sentence matters a great deal and almost nobody quotes it. We will come back to it. The full text is in the official PDF of the Convention.

From the building to the classroom: what CAST did

CAST (originally the Center for Applied Special Technology) was founded in 1984 in Massachusetts to use technology to improve learning for students with disabilities (CAST, "About CAST"). In the 1990s, Anne Meyer, David Rose and their colleagues set out Universal Design for Learning as a framework (CAST, "Universal Design for Learning: Theory & Practice"). The review by Ok, Rao, Bryant and McDougall dates it the same way: CAST has worked on UDL development and research since the 1990s, building on universal design, which originally dealt with access to the physical environment.

The conceptual leap is this: if in architecture the problem was physical access, in education it is access to learning. A book is a badly designed ramp if the only way to its content is reading it by sight at standard speed.

The three principles, from the official source

It is worth going to the source rather than to yet another secondhand infographic, because the guidelines changed recently and many materials still show the old version.

The current version is 3.0, released in July 2024 (CAST, "About the Guidelines 3.0 Update"). CAST's recommended citation is: CAST (2024). CAST Universal Design for Learning Guidelines version 3.0. The three principles are:

A detail that changes the meaning

"Design", not "Provide". In versions 2.0 and 2.2, the principles began with Provide multiple means of… (Guidelines 2.0, PDF). In 3.0 the verb is Design. It is not cosmetic: "provide" describes handing out ready-made options; "design" describes a decision made before the lesson. That is exactly the distinction this article is about, and CAST put it in the title of its own principles. If your training materials say "Provide", they are 2.2.

What CAST says the guidelines are not

In its own words, on the guidelines' FAQ page: "The UDL Guidelines are not meant to be a prescription or a checklist, but a tool that offers a set of suggestions" (Guidelines FAQ). It is the warning most often ignored: UDL is very often applied as a form to fill in when writing lesson plans.

UDL versus accommodations

This is the core, and it is explained by the phrase from Mace we have quoted twice: without the need for adaptation.

An accommodation is an after-the-fact, individual response: the lesson is already designed, there is a student it does not work for, and something is changed for them. An exam in braille, extra time, enlarged notes. It works, it is often legally required and in many cases it is the only option. But it has three costs: it comes late, it singles the student out, and it depends on someone remembering.

UDL is a before-the-fact response aimed at the design: the lesson is planned from the start with more than one way in and more than one way out, so most of the patches are no longer needed. The review by Ok and colleagues puts it this way: instead of adapting or modifying lessons while they are being taught, UDL builds supports proactively into goals, materials, methods and assessment (Ok, Rao, Bryant and McDougall (2017), Exceptionality, 25(2), 116–138).

The honest nuance: UDL does not replace accommodations

This is often told wrong, so it is worth saying plainly: UDL does not remove the need for individual supports, and no serious text claims it does. The UN Convention, in the same sentence that defines universal design, makes clear that it "shall not exclude assistive devices for particular groups of persons with disabilities where this is needed", and it keeps "reasonable accommodation" as a separate, additional legal obligation.

A blind student will still need their braille display, their transcribed textbook and their extra time, however well the classroom is designed. What UDL reduces is the number of patches and how exceptional they feel. The difference between asking for five accommodations and asking for one is huge for everyone involved, starting with the student.

What it means in a class with a blind student

This is where the framework becomes concrete: which of the usual measures are patches and which are design.

Look at the right-hand column: almost nothing costs more than the left, and almost everything helps more people. Naming who is speaking helps the blind student, the student with hearing loss, the new student who does not know names yet and the substitute teacher. That is the curb cut, literally.

The last row changes the year most, and almost no guide includes it. If only the blind student learns braille, it stays one person's private language. When the class learns it, it stops being an accommodation and becomes shared content: there is someone to write a note to. For doing it on Monday morning with egg cartons and card stock, see braille games and activities for the classroom.

The honest criticism: what the evidence says

UDL is cited everywhere. In the United States, the Higher Education Opportunity Act of 2008 and the Race to the Top program pushed it into teacher preparation, as the review by Ok and colleagues notes. It spread largely through education policy rather than through accumulated evidence.

So what does the evidence say? It depends on who you ask, and that is exactly the honest answer.

Three readings of that table, all defensible:

The favorable one. There is a methodologically serious meta-analysis with a moderate positive effect (g = 0.43) on academic achievement. For a general teaching framework, that is not nothing.

The skeptical one. That meta-analysis rests on twenty studies. The 2017 review found thirteen; the 2023 one, thirteen. It is a small base for a framework taught in teacher training around the world.

The uncomfortable one. Boysen does not test whether UDL works. He tests the evidence CAST itself presents as the basis of its guidelines, and concludes that it does not support them. His most cited finding is that most of the studies offered to support the guidelines on giving students choice were not about that, and that the framework's neuroscience basis, the well-known "three brain networks", is not backed by neuroscience studies of UDL.

The underlying problem: what counts as UDL

The most useful criticism is not "it doesn't work". It is that it is not clear what is being measured. Ok and colleagues noticed it back in 2017: although authors described their interventions as UDL, there was considerable variation in how they linked specific guidelines to the parts of their intervention. In other words, two "UDL studies" may be doing things that look nothing alike.

That has an awkward practical consequence. When a UDL study succeeds, it is hard to tell whether the framework worked or a specific practice did (sharing materials early, allowing several ways to answer, using digital text) that already had its own evidence. Boysen points to exactly that: the cited studies support individual effective teaching techniques more than the multiple-means thesis.

And the conclusion Ok and colleagues wrote in 2017 is still the most accurate description of where things stand: UDL's promise as an equalizer will remain unfulfilled until an empirical evidence base validates its benefits and identifies which specific practices improve outcomes.

So, should we use it?

Our reading, and it is an opinion: UDL is a good design principle with weak empirical proof. Both statements fit together.

That a classroom planned with several ways in and out works better than one planned for an imaginary student is sensible, consistent with what we know about accessibility and consistent with a moderate positive effect. That this justifies presenting it as a "neuroscience-based" framework, or filling in lesson plan forms with checkpoints, does not hold up, and CAST itself says its guidelines are not a checklist.

The practical takeaway is the design question, which does not depend on any meta-analysis: what part of what I am about to do for this one student can I do, at no extra cost, for the whole class?

Braillito as an example

An example of what this looks like in a tool, not an advertisement.

Braillito does not have one version for blind students and another for everyone else. It is the same app, with the same exercises, working the same with a screen reader as with sight. That is not a marketing decision: it is the difference between designing for two ways of access from the start and building a visual app with an "accessible mode" bolted on later. When an adapted version exists, someone has to maintain it, and it is always behind.

The consequence in a classroom is what matters: twenty-five students and their blind classmate can do the same lesson at the same time, without anyone getting separate material.

We give it as an example because it is the case we know from the inside, not because it is the only one. Any material that works through two channels at once does the same job.

What we could not confirm

• The exact date and outlet of the term's first use. Dolph's review (2021) dates Mace's article to Designers West, November 1985, and that is the reference we followed. We could not consult the original article, which is not openly available.
• Whether King-Sears et al. describe their own effect as "moderate to large". CAST's evidence page summarizes that meta-analysis as showing moderate to large effects. The published pooled value is g = 0.43, which by convention reads as moderate. We could not read the full paywalled article, so we do not know whether CAST's range refers to the fifty individual effects or is a generous reading of the overall result.
• The number of "considerations" in version 3.0. Version 2.2 had 31 checkpoints; for 3.0 we have seen different figures in secondary sources and found no official count, so we give none.
• Whether CAST has formally replied to Boysen's analysis. There are response articles in the same journal, but we did not find a point-by-point institutional reply from CAST.
• Any legal obligation to apply UDL in a given country. The framework appears in education law and policy in several countries, but its scope and enforceability vary widely and we will not summarize them here.

Frequently asked questions

What does UDL stand for?

Universal Design for Learning, the framework developed by CAST in the 1990s. It builds on universal design, the broader architectural concept coined by Ronald Mace in 1985.

What are the three principles of UDL?

In version 3.0 of CAST's guidelines: Design Multiple Means of Engagement (the why of learning), Design Multiple Means of Representation (the what) and Design Multiple Means of Action & Expression (the how). Each principle has three guidelines, nine in total.

What is the current version of the UDL Guidelines?

Version 3.0, released by CAST in July 2024. It replaces version 2.2 from 2018. The most visible change for everyday use is that the principles went from "Provide multiple means of…" to "Design multiple means of…". Many training materials still use 2.2.

Does UDL replace accommodations?

No. It reduces them. A well-designed classroom removes many of the patches, but students with disabilities will still need individual supports (braille materials, assistive technology, extra time), and those supports are a separate, additional obligation. The international definition of universal design says explicitly that it does not exclude assistive devices where needed.

Is there evidence that UDL works?

Partly, and it is disputed. A 2023 meta-analysis of twenty studies found a moderate positive effect on achievement (g = 0.43). A 2024 critical analysis in Policy Futures in Education concluded that the evidence CAST cites for its own guidelines is weak. The evidence base is small and very varied, and part of the problem is that studies do not agree on what counts as UDL. It is reasonable to use it as a design principle; it is not reasonable to present it as an intervention with proven effectiveness.

Where do I start if I have a blind student in my class next week?

With what is urgent, not with the framework. Order the materials that take months now, walk the classroom with the student before it fills up, and start saying out loud what you write on the board. UDL is the question you ask afterwards, when you plan the next unit: of all this, what can I do for the thirty instead of for the one?