Learn it → Apply it → Save as Skill
The best possible decision and the best decision you can actually make are not the same thing. Classical rationality assumes complete information, unlimited time and unlimited computing power to weigh every option before choosing. Real people, and real organizations, have none of these in unlimited supply. Bounded rationality describes how decisions actually get made once information, time and attention are treated as the scarce resources they really are.
The Dr. Wallace’s Favorite Mental Models is designed to turn useful ways of thinking into reusable tools. Each model gives you a fast visual overview, deeper interactive learning, practical application exercises and an AI-ready Markdown skill you can save for future use.
Instead of merely reading about mental models, you can gradually build your own AI skills library: a collection of portable reasoning skills that can help ChatGPT, Claude, Grok and other AI assistants apply the mental models you value.
Herbert Simon argued that "economic man," the textbook actor who surveys every option and picks the provably optimal one, does not describe how real decisions get made. Real decision-makers face limited information, limited time and limited processing capacity, so instead of optimizing they satisfice: they search until they find an option that clears a "good enough" bar, then they stop. This is not a failure of rationality. It is rationality adapted to the constraints that actually exist.
Important: options 4 through 6 might contain something better than option 3. Bounded rationality says the cost of finding out is not always worth paying.
The model is most useful right when a search for the right hire, vendor, apartment, feature or partner keeps extending past the point where you can name what more information would actually change, or when a decision feels stuck because no option can be proven optimal.
Searching for a decision without a stopping rule happens automatically. Slowing it down to these four steps turns an open-ended search into a deliberate, defensible choice.
Name the real limits on this decision: how much time, information and attention you actually have, not how much you wish you had.
Decide what "good enough" looks like before you start comparing options, so the bar cannot quietly shift to match whatever you happen to find.
Review options in a sensible order, giving each plausible candidate a serious look as it appears rather than trying to survey the whole field first.
Take the first option that clearly clears the bar. Treat any further search as a new decision with its own cost, not something you automatically owe yourself.
Open a use case to see what the model changes.
Reviewing two hundred resumes for one open role costs real time a growing team often cannot spare. Setting a clear, good-enough bar before reading resume one, and stopping once a candidate clearly clears it, usually beats an exhaustive search that delays a needed hire for the chance of a marginally better one.
Comparing every vendor, tool or plan on the market has a real cost in time, and that cost grows the longer the search runs. A threshold set in advance, "does this meet the actual requirements," ends the search at a defensible point instead of an arbitrary one.
Product teams rarely have complete information about how a feature will perform before it ships. Bounded rationality supports shipping the version that clears a reasonable confidence bar and learning from real usage, rather than waiting for data that would only arrive after the decision window has already closed.
A negotiator who knows their BATNA can satisfice: accept a deal that clearly beats the walk-away option rather than continuing to hold out for an unverifiable "best possible" deal that may never arrive before the opportunity closes.
Simon developed the model while studying how administrators actually make decisions inside real organizations, where nobody has complete information about every downstream effect of a policy. Administrative decisions get made using standard procedures and reasonable-enough evidence, not exhaustive analysis of every consequence.
Choosing among an effectively unlimited set of apartments, restaurants or laptops is exactly the situation the model describes: pick the few criteria that actually matter, then take the first option that satisfies them rather than surveying the entire market first.
This worked example highlights the manifestation doing the most work in the scenario below.
Expand for the research behind the model and where it does and does not apply.
Simon introduced the term in his 1947 book Administrative Behavior and developed it further in Models of Man (1957), contrasting "economic man," assumed to have complete information and unlimited computational power to find the single optimal choice, with what he called "administrative man," who satisfices: choosing the first alternative that meets a minimum standard of acceptability given real limits on information, time and cognitive capacity. Simon received the 1978 Nobel Memorial Prize in Economic Sciences "for his pioneering research into the decision-making process within economic organizations."
No. Simon's framing treats bounded rationality as a description of genuine rationality operating under real constraints, not as a bias or an error mode. A decision-maker who satisfices is not failing to be rational; they are being rational about the cost of continued search itself, a cost the unlimited-information model ignores by assumption.
Simon's work set the foundation that Daniel Kahneman and Amos Tversky built on starting in the 1970s, studying specific mental shortcuts, or heuristics, and the systematic errors they can produce. Later, Gerd Gigerenzer's research on "ecological rationality" pushed back on framing these shortcuts as mostly error-prone, arguing that many heuristics perform well precisely because they are matched to the structure of real environments rather than an idealized, unlimited-information one.
Satisficing is a deliberate strategy with a threshold set in advance, not a license to decide carelessly or skip research the actual stakes justify.
Deciding what "good enough" means only after seeing the options lets whichever option you already like redefine the bar in its own favor.
Continued search has a real cost in time and attention. Past a point, that cost outweighs the expected improvement in the outcome.
The model describes constrained rationality, not a bias to be corrected. Treating it as a flaw to eliminate misses the point Simon was making.
Comparing every option on the market for a low-stakes, reversible choice usually costs more time than it returns in a better outcome.
Standard operating procedures and routines exist because organizations face the same information and time limits Simon described, not only individuals.
the decision is high-stakes, largely irreversible, has a small, well-defined set of options, and the time and data to properly evaluate each one are actually available.
probabilities and payoffs are known or estimable well enough to calculate directly, rather than approximated through a threshold.
the option space is large but well-structured enough for software to search it exhaustively, such as route planning or portfolio allocation, where the unbounded search a person cannot do by hand is now genuinely available.
This browser tool structures your thinking. It does not send your information anywhere or perform AI analysis.
Save as Skill converts the model into portable Markdown instructions that teach an AI when and how to apply Bounded Rationality, what questions to ask, what evidence to seek, what mistakes to avoid and how to structure its analysis.
Save the skills you find useful and organize them into your own reusable library. The Markdown is intentionally model-neutral, so it can be used wherever an AI system accepts uploaded files, project knowledge, custom instructions or reusable text context. Platform capabilities differ, so this is a portable skill file rather than a claim of one-click installation.
Not in practice. Optimizing requires knowing the full option set in advance and being able to score each one accurately, conditions that rarely hold outside clean, well-structured problems. Once the real cost of search is counted, satisficing is often the better strategy actually available, not a fallback for the strategy you could not afford to run.
Simon's original interest was organizational: firms and agencies bound decisions through standard operating procedures, division of labor and routines specifically because no single person inside them has complete information either. An organization's routines are, in effect, a pre-set satisficing threshold applied at scale.
Bounded rationality concerns limits on information, time and computational capacity available for a decision. A separate, related but distinct problem is bounded willpower: knowing the better option and still failing to choose it. The two frequently overlap in practice, for example when someone stops searching for a good option not because the bar was cleared but because searching is unpleasant, but they describe different constraints and call for different fixes.