Question
Cognitive load theory actually predicts something real
I sat in on a UX meeting last week where someone cited cognitive load to justify removing information from a page, and I realized I couldn't tell if they meant Sweller's original model or just a vague intuition that "simpler is better." Which is sort of the problem.
The framework *does* predict something specific: that performance on a task degrades when working memory is saturated, and that the *type* of cognitive demand matters (intrinsic vs. extraneous). Sweller's work in the 1980s showed this fairly rigorously in learning contexts. But here's where it gets slippery. The theory works best when you're dealing with tasks where there's a clear, measurable cognitive bottleneck—learning algebra, reading instructional materials, that sort of thing. You can actually test whether removing extraneous details improves retention.
The folk psychology version—which is what lives in product teams—is much looser. It's the idea that humans have a fixed bucket of attention, and if you put too much stuff in front of them, they'll be overwhelmed. This *sounds* like it should predict something, but it doesn't actually account for what people are motivated to do, how schemas form, or why some complex things (a good novel, a video game) hold attention while simple things don't. I think we inherited the term from cognitive science and then abandoned the specificity that made it testable.
So it predicts well in lab conditions around procedural learning. Beyond that it's mostly a heuristic for "this seems complicated" which, fair enough, but let's not pretend it's the same thing.
1 comment
Log in to comment.
This distinction between Sweller's actual model and the folk version tracks with what I've seen in schools for years. Teachers will say "cognitive load" to mean "the kids look confused" without any sense of whether it's intrinsic complexity (the math is genuinely hard) or extraneous load (the worksheet is poorly formatted). Those require totally different fixes, but they get treated the same way—usually by dumbing something down that didn't need dumbing down.
The part that actually bothers me though is the assumption underneath the folk version: that less information is always better. I've watched schools strip down curricula in the name of "reducing cognitive load" and what actually happened was students had less to hook new concepts onto. A kid learning photosynthesis needs to know it connects to respiration, energy transfer, plant growth—that network is what makes it stick. Call that load if you want, but it's not extraneous, and removing it doesn't help. Sweller's theory doesn't actually predict that outcome. The vague "simpler is better" intuition does, and it's wrong in that direction often enough to matter.
Your point about testability is the real one. In a lab with a controlled task, you can measure whether an extraneous detail hurts performance. In a product team deciding what to show, or a classroom deciding what to teach, you're operating without that feedback loop. So yeah, we've kept the word and lost the rigor. Worth acknowledging when we're just using it as a synonym for "this seems like a lot."