Every decision you make passes through mental models — simplified representations of how the world works. Some you chose deliberately; most you absorbed from culture, education, family, and media without examination.

Useful models: supply and demand, compound interest, feedback loops, opportunity cost, second-order effects, inversion (solving problems by identifying what to avoid), first principles thinking (reasoning from fundamentals rather than analogy).

The map-territory distinction (Alfred Korzybski): your model of reality is not reality itself. The model is always simpler than the system it represents. This simplification is both its power (makes complex systems navigable) and its limitation (misses features the model doesn't include). Holding multiple models simultaneously — and knowing which one applies in which context — is the practice of cognitive mastery.

Charlie Munger's "latticework of mental models": instead of relying on one framework, build a diverse toolkit. Economics, psychology, biology, physics, statistics, history, game theory — each provides models that illuminate different aspects of the same situation. The person with one model tries to fit every problem into it. The person with many models selects the most appropriate one.

Start with the highest-utility models across domains:

From psychology: confirmation bias, loss aversion, social proof, narrative bias, Dunning-Kruger. From economics: opportunity cost, comparative advantage, incentive structures, sunk cost. From systems: feedback loops, leverage points, emergence, unintended consequences. From statistics: base rates, regression to the mean, sample size, survivorship bias. From philosophy: falsifiability, steel-manning, epistemic humility, Occam's razor. From biology: evolution, adaptation, niches, antifragility.

The practice: when facing a decision, run it through multiple models. "What does economics say about the incentives here?" "What does psychology say about my biases?" "What does systems thinking say about unintended consequences?" The convergence of multiple models produces more robust decisions than any single framework.

Maintenance: models need updating. Test them against reality. When a model consistently fails to predict or explain, revise or replace it. The goal is a living toolkit, not a fixed dogma.