Mental representations are the internal structures you use to perceive, understand, and act on the world. They are not the world itself; they are your compressed, structured model of it — patterns, relationships, and procedures stored in memory. A chess master doesn't see 32 pieces; she sees threats, pawn structures, and typical plans. A surgeon doesn't see tissue; he sees anatomy, landmarks, and sequences of moves. The quality of your representations determines how fast you make sense of situations and how accurately you predict outcomes.
Expertise research (e.g. Ericsson, Chase & Simon) shows that experts have richer, more differentiated representations than novices. Novices rely on surface features and step-by-step rules. Experts recognise deep structure and retrieve whole chunks — "this is a Sicilian," "this is septic shock." Better representations reduce cognitive load: you see the pattern instead of computing it. They also support anticipation and simulation: you can run scenarios in your head because the representation encodes how the system behaves. The goal of deliberate practice is partly to build these representations — to rewire how you see the domain.
In understanding and analysing, the model applies directly. When you build a mental model of a market, a product, or a team, you're building a representation. When you decompose a problem into first principles, you're making your representation explicit and testable. When you pattern-match to past cases, you're using stored representations to classify the current situation. The discipline is to notice when your representation is shallow (you're reacting to surface cues) and to invest in deepening it — through study, feedback, and repeated exposure to varied cases.
Section 2
How to See It
Mental representations reveal themselves when someone with experience sees structure that others miss, anticipates outcomes before they happen, or explains a situation in terms of principles and patterns rather than surface details. Look for compression: the expert's description is shorter and more structural than the novice's.
Analysis
You're seeing Mental Representations when an experienced analyst looks at a messy dataset and immediately points to the two drivers that matter, while a junior analyst lists every variable. The senior has a representation that compresses the domain into cause-effect structure; the junior is still at the level of raw features.
Strategy
You're seeing Mental Representations when a founder describes a market in terms of forces, incentives, and feedback loops rather than a list of competitors and features. The representation encodes how the system works, not just what's in it — and supports prediction and intervention.
Judgment
You're seeing Mental Representations when a doctor hears three symptoms and narrows to a differential diagnosis in seconds. The representation links symptom clusters to disease categories and typical trajectories; the novice would still be gathering information without a framework.
Execution
You're seeing Mental Representations when a programmer reads a codebase and quickly identifies the bottleneck or the design flaw. The representation includes typical patterns, failure modes, and where complexity hides; the novice sees only lines of code.
Section 3
How to Use It
Decision filter
"Before analysing a situation or making a call, ask: what representation am I using? Is it surface-level (lists, features) or structural (causes, patterns, dynamics)? If it's shallow, invest in building a better one — study, decompose, and get feedback."
As a founder
Build representations of your market, product, and organisation. That means moving from 'who are our competitors' to 'what are the forces that determine winner and loser' and from 'what features do we have' to 'what job does the product do and how does it create value.' The better your representation, the faster you see opportunities and threats and the less you rely on ad hoc reactions. Update representations when reality contradicts them.
As an investor
Your edge is often the quality of your representations — of industries, business models, and management. The investor who has a clear causal model of why a category will grow and how a company captures value can evaluate deals faster and with less noise. The one who only has a list of comparable companies is pattern-matching at the surface. Invest in deepening representations in your focus areas.
As a decision-maker
When facing a complex situation, make your representation explicit: what are the key variables, how do they relate, what would have to be true for X to happen? Test and refine the representation with evidence and feedback. Avoid the trap of acting on surface cues without a structural model — that's when you're surprised by second-order effects.
Common misapplication: Treating the representation as reality. Representations are always incomplete and sometimes wrong. When outcomes contradict your model, update the representation. The expert's trap is overconfidence in a representation that was built in a different regime.
Second misapplication: Building representations from anecdotes instead of structure. A list of war stories is not a representation. A representation encodes principles, relationships, and boundary conditions. Move from "this happened" to "this happens when these conditions hold because of this mechanism."
Feynman insisted on building representations from first principles and on explaining them in simple terms. His technique — explain as if teaching someone who doesn't know the domain — forces you to make your representation explicit and to find the structural level. When the representation is clear enough to teach, you've compressed the domain correctly. When you can't explain it simply, the representation is still fuzzy.
Munger advocates a "latticework of mental models" — multiple representations from different disciplines that you use to analyse the same situation. Each model is a representation of how one slice of reality works; combining them reduces the chance that a single shallow representation drives your decision. The discipline is building and maintaining these representations and knowing when to apply which.
Section 6
Visual Explanation
Mental representations compress experience into structure. Novices hold surface features (many pieces); experts hold patterns and relationships (fewer, richer chunks). Better representations reduce load and support prediction.
Section 7
Connected Models
Mental representations connect to how we learn, reason, and simplify. The models below either reinforce representation-building, create tension with it, or follow from it.
Reinforces
Chunking
Chunking is the compression of information into units. Mental representations are built from chunks — patterns and relationships that group details. The more you chunk, the richer your representation can be without overloading working memory.
Reinforces
First Principles Thinking
First principles thinking is breaking a domain down to fundamental truths and rebuilding. That process makes your representation explicit and testable. The two reinforce each other: first principles help you build better representations; good representations help you see which principles apply.
Tension
Curse of Knowledge
The curse of knowledge is when your expertise makes you forget how others see the world. Your representation is so automatic that you can't simulate the novice's. The tension: deep representations are an asset, but they can blind you to what others don't yet see. Mitigate by teaching and by seeking disconfirming evidence.
Tension
Map vs Territory
The map is not the territory — your representation is not reality. The better your representation, the more you might confuse it with the full picture. The discipline is to hold the representation as a model to be updated when the territory doesn't match.
Section 8
One Key Quote
"The capacity of the human mind for forming and manipulating images is one of the chief means by which we transcend the bounds of immediate experience and bring the distant past and the uncertain future into the same cognitive space as the present."
— Herbert Simon, cognitive psychologist and Nobel laureate
Representations let you simulate, anticipate, and reason about what you can't directly observe. They compress experience into structure so you can run scenarios and make decisions. The practitioner's job is to build them deliberately and to update them when the world doesn't match.
Section 9
Analyst's Take
Faster Than Normal — Editorial View
The quality of your representations is the ceiling on your judgment. If you're still at the level of lists and surface features, you'll be slow and brittle. If you've built structural models — causes, patterns, dynamics — you can analyse faster and anticipate better. Invest in representation-building in the domains that matter for your role.
Make representations explicit. The Feynman technique (explain simply) and first-principles decomposition both force you to make your representation visible. When you can state it, you can test it and refine it. When it stays implicit, you can't tell if it's wrong.
Update when reality contradicts. The biggest failure mode is clinging to a representation that no longer fits — because the market shifted, the product changed, or you were wrong about a mechanism. Treat representations as hypotheses: use them until evidence says otherwise, then revise.
Borrow representations from other domains. Munger's latticework is exactly this: use mental models from physics, psychology, and economics to analyse business. Cross-domain representations reduce the chance that one narrow model drives everything.
Don't confuse data with representation. Having more data doesn't automatically improve your representation. You need structure — relationships, principles, boundaries. Build the representation; use data to test and refine it.
Section 10
Test Yourself
Is this mental model at work here?
Scenario 1
A senior analyst explains a market in three sentences: forces, feedback loops, and the one lever that matters. A junior presents 15 data points with no clear thesis.
Scenario 2
A founder insists the market will grow because 'everyone says so' and lists three anecdotes. She has no model of why growth would occur or what could stop it.
Section 11
Summary & Further Reading
Summary: Mental representations are the internal structures you use to perceive, understand, and act — compressed, structured models of a domain. Experts have richer, more differentiated representations than novices; that's why they see structure others miss and anticipate better. Build representations by study, feedback, and first-principles decomposition. Make them explicit (e.g. by teaching), test them against reality, and update when they're wrong. Use them in understanding and analysing to move from surface features to causal structure.
Ericsson's popular treatment of deliberate practice and the role of mental representations in expertise. Clear account of how experts build and use representations.
A method for making your representation explicit by explaining in simple terms. If you can't explain it, the representation isn't clear enough.
Leads-to
Pattern Matching
Pattern matching is applying stored patterns to new situations. It depends on having representations that encode those patterns. Better representations improve the accuracy and speed of pattern matching; poor representations lead to false positives and missed signals.
Leads-to
Cognitive Load Theory
Cognitive load theory says working memory is limited. Good representations reduce load by compressing information into meaningful units. Building representations is an investment that pays off in lower load during analysis and decision-making.
