Mental Models

Function

A model's job is to let you act before you have all the information — which is to say, always. Without explicit models you still use models; you just don't know what they are. That is the worst situation, because you cannot correct what you cannot see. Naming your models makes them debuggable.

Mechanism

Each model is a trade between accuracy and usability. A perfect map of the territory is the territory; useless. A good model throws away enough detail to be carried in the head, kept close enough to truth that it predicts more often than it misleads. You upgrade models by noticing when reality refuses to behave as predicted — and treating that refusal as data, not insult.

Breakdown

1. Naming the model you are about to use before acting on it.
2. Treating failed predictions as data, not noise.
3. Holding multiple models for the same domain, with rules for when each applies.
4. Retiring models that no longer match the territory.

Example

Two managers face the same disappointing quarter. The first uses the model 'people are lazy', and responds with surveillance. The second uses the model 'systems produce the behavior they reward', and audits the incentives. Six months later one has a sullen team and the same numbers. The other has different numbers. They saw the same facts. They were operating two different worlds.

Connected concepts

• depends on Self-Model Updating
• reinforces Cognitive Clarity
• causes Reframing

Questions this answers

What are mental models?

Mental models are the compressed maps you use to predict and act in the world — internal representations of how things work, simpler than reality and far more useful than no map at all.

How do I upgrade my mental models?

Name the model you are using before acting. When reality refuses to behave as predicted, treat the refusal as data, and replace the model with one that handles the new case.