Mental Models & Cognitive Load
Mental Models & Cognitive Load
Every user arrives at an interface with assumptions. They assume the app works like other apps they know. They assume the icons mean what they've seen elsewhere. They assume the navigation follows patterns they've learned. When those assumptions are wrong — when the interface violates their mental model — they make errors, feel frustrated, and often abandon the task.
HCI's fundamental insight about cognition is that designers cannot change how the human brain works. They can only design interfaces that align with it.
How Mental Models Form
A mental model is a user's internal representation of how a system works. It is built through:
- Experience with the current system — using it, making errors, learning from them
- Prior experience with similar systems — the mental model of email applies to every new email client
- Analogy and metaphor — the desktop metaphor (files in folders, a trash can for deletions) imports understanding from the physical world
Mental models are almost never complete or accurate. Users operate on partial, often wrong models — and navigate successfully anyway, by recognising patterns that worked before.
Design implication: Match your system's actual model (how it really works) to users' expected mental model. Where they diverge, users will make systematic, predictable errors. These errors are not user failures — they are design failures.
The classic example: a user who types a URL into the Google search bar (rather than the address bar) is not wrong about how to access a website. They have learned that "typing something there takes you somewhere." The distinction between search and direct navigation is not obvious to non-technical users. Google's search handles both gracefully because it understands this mental model.
Cognitive Architecture: What Designers Must Respect
Working Memory holds active information. It is limited — roughly 7 ± 2 items, though modern research suggests closer to 4 chunks for complex information. Interfaces that demand users hold many things in mind simultaneously overwhelm working memory, causing errors and frustration.
Long-Term Memory stores learned patterns, skills, and concepts. Good interfaces exploit long-term memory by following conventions users already know. Deviating from conventions (even to "innovate") forces users to learn new patterns — a working memory cost that only pays off if the new pattern is substantially better.
Attention is selective and limited. Users do not read interfaces — they scan. Eye-tracking studies show F-pattern or Z-pattern reading, with attention concentrated at the top-left and rapidly declining. Important information buried below the fold or in right-side columns is frequently missed.
Cognitive Load in Practice
Cognitive load theory (Sweller, 1988) distinguishes three types:
- Intrinsic load — the inherent complexity of the task itself
- Extraneous load — complexity added by poor design (unclear labels, excessive options, inconsistent layouts)
- Germane load — cognitive effort that builds useful mental models (good complexity that teaches)
Interface design's job is to minimise extraneous load. Intrinsic load is unavoidable — a tax filing form is genuinely complex. But a confusing layout, ambiguous labels, and unclear error messages add extraneous load on top of an already difficult task.
Hick's Law formalises one dimension: decision time grows logarithmically with the number of choices. A menu with 12 items takes measurably longer to navigate than one with 4 — even for experienced users. This is why great navigation systems (Google's homepage, Apple's product pages) are ruthlessly minimal.
Gulf of Execution and Gulf of Evaluation
Don Norman introduced two critical gaps:
Gulf of Execution: The gap between what a user wants to do and the actions available. "I want to delete this item — how do I do that?" If the answer is not obvious, the gulf is large. Design closes this gap through clear affordances, visible controls, and progressive disclosure.
Gulf of Evaluation: The gap between what happened and what the user expected. "Did my action work? Was that a success or failure?" Design closes this through immediate, clear feedback — progress indicators, confirmation messages, error states that explain what went wrong.
Large gulfs in either direction produce the "I don't know what I'm doing and I don't know if it worked" experience that drives user abandonment.
Designing for Cognitive Alignment
Practical principles that follow directly from cognitive science:
- Reduce choices at every decision point — Hick's Law is always operating
- Follow conventions — deviate only when the benefit dramatically outweighs the relearning cost
- Make system state visible — users need to always know where they are and what happened
- Support recognition over recall — users can recognise options more reliably than recall them from memory; keep commands visible rather than requiring memorisation
- Design for error recovery — users will make errors; interfaces must make correction obvious and easy