The core function of a mental model is predicting system behavior for a given action, not memorizing steps

What it is

The core value of a mental model isn't to serve as a stored operating manual for the user — it's to let them roughly predict how the system will respond to a given action, even one they've never performed before. This is not the same thing as step memory, which answers "which button do I click at step three"; a mental model answers "what happens if I do this."

Why it happens

A model with genuine predictive power contains a causal structure that can be "run": what states the system could be in, what actions are available, and how each action transitions the state. Faced with a new action, the user runs this structure mentally and derives an expected outcome, rather than retrieving a pre-stored step sequence from memory. Pure step memory breaks the moment a scenario changes even slightly — move an entry point and it can no longer be found — while a predictive model can absorb such changes, because what it produces is "what the outcome will be," independent of whether some fixed operational path is still sitting exactly where it was.

Studying it

The classic way to distinguish the two is to give users a novel action they've never performed but whose outcome can be derived from rules they already know: a reasonable prediction indicates a runnable causal model, while guessing or admitting ignorance suggests their prior fluency relied more on a fixed step sequence than on transferable causal understanding. The key to this diagnostic is that the action genuinely has to be untaught — otherwise what gets measured is still step memory, not predictive capacity.

Where it stops holding

Predictive capacity isn't all-or-nothing: most users' models only cover a limited depth of inference — able to predict the immediate next step but not the chain of consequences two or three steps further out — beyond that depth the model degrades into guessing. Step memory and predictive modeling also typically coexist within the same user; it's hard to find anyone who relies purely on one kind of knowledge, so what gets measured is usually a mixture of the two rather than a clean either-or category.

Related

• Same group: A7.01.1 A mental model is a user's internal explanation of how a system works · A7.01.2 A model can be incorrect and still support successful operation · A7.01.3 The model determines a user's expectations and response when something goes wrong · A7.01.5 A sufficiently complete mental model lets a user derive operations they were never taught · A7.01.6 A model's internal consistency and its accuracy are two separate things — a consistent but wrong model still works · A7.01.7 Users lack insight into the accuracy of their own model and rarely question it
• Nearby: A6.09 Procedural memory and automaticity · A7.15 Model layers: functional, structural, and implementation
• Search terms: predictive mental model · procedural recall · runnable model