Working memory is a short shelf.
A few chunks, not a paragraph. The shelf on the neuroinclusive page is a demonstration of that limit.
Nelson Cowan's estimate is that people hold a few chunks at once, not a wall of clauses. John Sweller's cognitive-load work says the same thing from the design side: extra pieces that are not the idea itself use up the shelf. Richard Mayer's segmenting principle is the practical version. Break the idea. Do not stack it.
The shelf on the neuroinclusive page is a demonstration of that limit, not a study we ran. Four slots. Seven pieces do not fit. Three chunks do. The point for a school or a training team is the design choice: chunk the lesson, or let the learner decide first so the sentence has something to attach to. CAST's Universal Design for Learning is the broader frame. We are not making a clinical claim.
There is a second pass in that offer. The people who teach try the overloaded version themselves, so the barrier is felt rather than described in a workshop slide.
Sources: Cowan, N. (2001). The magical number 4 in short-term memory. Behavioral and Brain Sciences, 24(1), 87–114. doi:10.1017/S0140525X01003922 · Sweller, J. (1988). Cognitive load during problem solving. Cognitive Science, 12(2), 257–285. doi:10.1207/s15516709cog1202_4 · Mayer, R. E. (2009). Multimedia Learning (2nd ed.). Cambridge University Press. · CAST (2024). Universal Design for Learning Guidelines version 3.0. udlguidelines.cast.org