Fewer visual elements does not mean less cognitive load — the two need to be verified separately

What it is

How many icons, lines of text, and colors sit on a screen is something you can just count. How much mental effort a user actually has to spend understanding, judging, and remembering things while completing a task is cognitive load — a different quantity. Reducing the first is routinely assumed to lower the second, and that assumption does not hold. Strip a label off an icon and the element count genuinely drops, but the user now needs an extra step to guess or recall what that icon stands for — and the mental effort that inference costs can exceed what it would have taken to just read a word. Looking lighter and feeling lighter to use are two curves that need to be measured separately, not two descriptions of the same curve.

Why it happens

The size of cognitive load depends on how many reasoning steps a task requires, how much information has to be held in mind, and how much effort it takes to retrieve an answer from memory. How many visible elements sit on the screen is only one factor among these, and often not the dominant one. Drop a text label in favor of a bare icon and the "looking" step does get lighter — less text to read — but the "thinking" step gets heavier: the user has to search memory for what that symbol means, and if the icon's meaning is not universal enough, that search can fail outright and require trial and error to resolve. The load has not shrunk along with the element count; it has simply moved from the looking stage to the thinking stage, and the total can even go up. Conversely, laying out several clearly-labeled text options sometimes carries a lower load than one ambiguous cluster of icons, precisely because it skips that extra decoding step.

Studying it

Separating "looks simple" from "feels light to use" cannot rest on counting visual elements or an expert's intuition about which version looks cleaner — it needs measures that actually track mental effort. A dual-task method has users perform a secondary task alongside the main one; if the main task's load is higher, performance on the secondary task drops accordingly. A subjective workload scale filled out after the task, combined with time-on-task and error rate, adds another angle. Applying this to a labeled version and an icon-only version side by side is what actually tells you whether the visual simplification bought a reduction in load, or did the opposite.

Where it stops holding

If an icon's meaning is already highly conventional for the target user population and needs essentially no decoding — a printer icon, a trash-can icon — dropping the text label really does add little load, and fewer elements and lower load move together. But for symbols whose meaning is not universal, or that require specialized background to interpret, removing the text label directly creates extra decoding load. Where this boundary falls depends on whether the icon's meaning is already familiar to the target users, not on how clean the icon itself looks.

Applying it

• Don't accept "fewer visual elements" as evidence of lower cognitive load when signing off on a design, especially after replacing text labels with icons or merging several options into one vague entry point.
• Before deciding a text label can be dropped, confirm the symbol is already highly conventional for the target users. When unsure, keep the label or add a lightweight hover explanation rather than assuming "more minimal, therefore better."
• To verify: measure secondary-task performance, time on task, and error rate for the pre- and post-simplification versions, rather than voting on visual impression in a review meeting. If the simplified version lowers secondary-task performance or raises error rate, that "simpler" design pushed load onto the user instead of reducing it for them.

Related

• Same group: A2.12.1 The visual system reads ambiguous figures as the simplest possible structure — the perceptual basis of the pull toward simplicity · A2.12.2 Visual minimalism as a style is not the same as perceptual-processing simplicity · A2.12.3 Stripping visual cues for a cleaner look can raise the cost of discovering an interaction · A2.12.4 Forcibly simplifying the appearance of inherently complex content shifts that complexity onto the user's own mental processing · A2.12.5 The pull toward simplicity works on figure recognition and does not automatically extend to information-architecture simplification
• Nearby: A9 Cognitive load · A6 Memory
• Search terms: cognitive load · dual-task method · NASA-TLX