Decision-Making and Mental Models: How to Think More Clearly

Sources checked September 9, 2026 · Health and Psychology

Advertising disclosure: this article contains labeled advertisement placements. No affiliate links are included.

A mental model earns its place when it changes a question you ask, an assumption you test or an action you take. Knowing the name of a bias is less useful than noticing what could make your preferred option wrong.

AI-generated conceptual editorial still life for A Wandering Mind.

Better decision-making is not about eliminating uncertainty. It is about making the uncertainty visible enough to work with. That can mean comparing two realistic options, testing one important assumption, or admitting that a decision needs expertise you do not have.

This guide uses a fictional workshop-planning example to connect the tools. The numbers are invented for explanation, not market prices or a business forecast. The framework and exercises are this article’s educational synthesis, not a validated psychological intervention or a substitute for professional advice on high-stakes matters.

First, define the decision you can actually make

“Should I organize a successful event?” is an aspiration, not a useful choice. “Which of these two available rooms should I book by Friday for a workshop next month?” identifies alternatives, a deadline and an action. It also reveals questions that a vague goal can hide: who has authority to book, what the room must provide, and whether postponing is still possible.

Separate three things before assigning scores:

• Requirements: conditions an option must meet, such as verified accessibility, the necessary date and permission to use the space for the planned activity.
• Preferences: things you would like to improve, such as convenience, atmosphere, capacity or money remaining after the event.
• Uncertainties: things you do not yet know, such as attendance, equipment availability or the cancellation terms.

A required feature should not disappear inside an average score. A beautiful room is not an acceptable substitute for an accessible room if accessibility is a requirement. Likewise, an unknown feature is not automatically a failed requirement: it may be a question to verify before deciding.

Include a realistic alternative to your favorite proposal. Sometimes that is the smaller version, a different date or not proceeding. “Do nothing” also has consequences, so describe them rather than treating the status quo as costless.

Five mental models, each with a job and a limit

Choose a tool for the question, not a collection of impressive names

Model	Useful question	Boundary to remember
Opportunity cost	What worthwhile alternative becomes unavailable if I commit this money, time or attention?	Compare feasible alternatives. An imaginary perfect option is not an opportunity you can actually take.
Expected value	What does a probability-weighted comparison imply under stated assumptions?	The average is not a promised outcome. It can conceal losses you cannot absorb and values the calculation leaves out.
Reversibility	Can I try a smaller version, change course or leave without disproportionate cost?	Read the actual commitments. A reversible booking may still consume time, trust or an opportunity that cannot be recovered.
Second-order effects	What might happen after the immediate benefit or cost?	Distinguish a plausible consequence from an elaborate story. Add a checkable assumption, not an endless chain of guesses.
Reference classes	What happened in genuinely comparable cases, including disappointing ones?	Similarity must be justified. A small, selectively remembered set of successes is not a reliable base rate.

For the workshop, opportunity cost might be the alternative use of the organizing team’s weekend. Reversibility might depend on whether a reservation is refundable. A second-order effect might be whether choosing a larger room creates extra staffing work. These are different questions; a financial calculation does not answer all of them.

Advertisement

A worked example: the larger upside is not always the better average

Suppose both rooms meet the event’s requirements. The large room costs $400 and holds 40 attendees. The small room costs $200 and holds 20. Each attendee contributes $20 after the assumed per-person costs. For this simplified comparison, subtract only the room cost from that contribution; taxes, administration and other possible costs are not modeled.

The organizer considers only three demand scenarios: 10, 20 or 40 people wanting to attend. For illustration, they assign probabilities of 25%, 50% and 25%. These probabilities are assumptions, not measured attendance forecasts. The small room admits no more than 20 people even in the high-demand scenario.

Fictional net contribution after room cost, under three assumed demand scenarios

Demand	Assumed probability	Large room	Small room
10 people	25%	10 × $20 − $400 = −$200	10 × $20 − $200 = $0
20 people	50%	20 × $20 − $400 = $0	20 × $20 − $200 = $200
40 people	25%	40 × $20 − $400 = $400	20 × $20 − $200 = $200

The large room’s expected value is (0.25 × −$200) + (0.50 × $0) + (0.25 × $400), or $50. The small room’s is (0.25 × $0) + (0.50 × $200) + (0.25 × $200), or $150. Under these assumptions and this financial objective, the small room is preferred, even though the large room has the bigger best-case result.

Neither room will produce its calculated average in any of the three listed scenarios. The calculation compares the assumed distribution of outcomes; it does not tell the organizer what will happen on the day. Nor are the three scenarios a complete risk inventory. Zero attendance, cancellation, additional costs or a different level of demand could change the picture substantially.

The financial objective is also a choice. If serving more people is central to the event’s purpose, capacity has value that this dollar comparison omits. State that value openly rather than quietly changing the probabilities to make the larger room look better. Cash available upfront and tolerance for loss matter separately from an average.

Find the assumption that could reverse the decision

Now keep the low-demand probability at 25% and vary the probability of high demand. Call that high-demand probability p; the middle scenario then has probability 75% minus p. This particular exercise allows p to range from 0% to 75%, so all three probabilities remain nonnegative and total 100%.

The large room’s expected value becomes −$50 + ($400 × p), with p written as a decimal. The small room stays at $150 because both its middle- and high-demand results are $200. They tie when p is 50%. At 60% high-demand probability, with 15% middle demand, the large room’s expected value is $190.

This does not establish that high demand is likely. It identifies what would have to change for the financial ranking to reverse. The organizer can now ask a focused question: what evidence would justify believing high demand is more likely than the original assumption?

The UK Treasury’s Green Book 2026, written for public-sector appraisal, discusses sensitivity analysis, switching values, uncertainty and optimism bias. The personal-scale example here is an original illustration of that general reasoning, not an application of government appraisal rules to an actual booking.

Improve the inputs before polishing the answer

A spreadsheet can multiply weak assumptions perfectly. Before adding decimal places, ask where the attendance scenarios came from. Similar past events may help, but define “similar”: audience, venue, promotion, ticket conditions, season and the meaning of attendance. Registrations, paid bookings and people who actually arrive are different measures.

Include events that were canceled or attracted little interest if they belong in the comparison. Looking only at successful events answers a narrower question than “What happens when people attempt this?” When only a few comparable cases exist, explain the limitation instead of presenting a precise percentage as established fact.

Ask a second person to challenge the input most favorable to your preferred option. They do not need to oppose the event. Their useful role is to ask what evidence would change the estimate, whether an omitted cost matters, and whether another plausible scenario would affect the choice.

Do not make yesterday’s spending decide tomorrow’s booking

Suppose $60 has already been spent on nonrefundable printing that cannot be reused or sold, whichever room is chosen. That money is gone under both options. Adding it to both totals changes the event’s overall accounting result, but it does not change the difference between the two room choices.

That is different from a refundable deposit, a future cancellation fee or materials that retain useful value. Those can change future consequences and belong in the comparison. “Ignore sunk costs” means separating irrecoverable past spending from what the current decision can still affect—not ignoring contracts or pretending the project has never lost money.

When should you trust intuition?

In their 2009 paper, Daniel Kahneman and Gary Klein identified conditions supporting skilled intuition: an environment with learnable regularities and an opportunity to learn them through experience. They also cautioned that confidence alone does not establish accuracy. Conditions for intuitive expertise.

For the organizer, familiarity with how long setup usually takes may be better grounded than a feeling about demand from an entirely new audience. Ask what kind of feedback produced the intuition. Did you see the results repeatedly and promptly? Or do you mostly remember the occasions when your impression turned out well?

Intuition can suggest a question without settling it. “This attendance estimate feels optimistic” is worth investigating. “I feel certain” is not evidence that an uncertainty has disappeared. Equally, adding a scorecard does not make a judgment objective if the scores simply restate the same feeling.

Advertisement

Use a small test to answer a specific question

Before committing, consider whether a limited trial could resolve the uncertainty that matters most. For the workshop, that might mean checking interest with the intended audience before the booking deadline. An interest form can tell you something about stated interest; it cannot by itself establish paid attendance or show-up rates.

Specify the question, evidence and decision rule before the results arrive. For example: “We will use this outreach to learn whether the proposed date is a common obstacle. If it is, we will reconsider the date before choosing a room.” That is more informative than collecting favorable comments and calling the event validated.

A trial also has costs. It takes time, may reach an unrepresentative group and can create expectations you need to manage. Do not advertise a confirmed event that is not confirmed. If a test cannot influence the choice before the deadline, its value for this particular decision may be limited.

Set a stopping point for research. Identify the next piece of information that could change the choice, what it will cost to obtain, and how long you can reasonably wait. More information is useful when it changes what you do, not simply when it postpones the discomfort of deciding.

A pre-mortem needs actions, not just worries

A pre-mortem asks you to imagine that the plan has failed and work backward to possible explanations. A 1989 study of prospective hindsight examined how temporal perspective and treating an outcome as certain affected explanations. Its abstract reports stronger effects from outcome certainty than from temporal perspective. It does not establish a universal percentage improvement in decision success. Mitchell, Russo and Pennington’s study.

For an original workshop exercise, imagine that attendance disappointed and setup went badly. Three possible explanations might be a date conflict, interest that did not translate into attendance, and equipment nobody had explicitly agreed to bring. Each becomes useful only when connected to an action:

• Date conflict: check relevant audience calendars before confirming the date; assign a named organizer to resolve the question.
• Weak conversion from interest: keep the attendance estimate provisional and record what evidence would justify revising it.
• Missing equipment: confirm the required items, responsible person and check date, rather than assuming “the venue handles it.”

The exercise is not a prediction that these failures will occur. It is a way to make overlooked assumptions discussable. Keep plausible concerns separate from remote possibilities so the discussion does not become an unbounded search for reasons nothing can work.

Why a checklist is support, not a guarantee

Evidence from a different, high-stakes setting illustrates the need for care with checklist claims. A 2009 before-and-after study across eight hospitals associated a surgical-checklist program with lower complications and mortality. A 2014 study across 101 Ontario hospitals did not find statistically significant reductions in the measured outcomes after checklist adoption. Haynes and colleagues; Urbach and colleagues.

These historical studies do not test the personal decision worksheet below, prove that checklists always work, or prove they never work. They also do not support abandoning clinical protocols. The narrower takeaway for this article is to avoid treating possession of a checklist as proof of an improved outcome. Check whether its questions are understood, answered and connected to action.

Keep a decision record you can actually review

A short record made before the outcome is known gives you something more useful than a reconstructed memory. Record enough detail to understand the choice later, without unnecessarily storing other people’s personal information. For the fictional event, that might look like this:

1. Decision and deadline: choose a room by Friday, after confirming the required features and terms.
2. Options: small room, large room or postponement; describe the consequences of each feasible option.
3. Reason: the small room has the higher modeled net contribution under the current assumptions; note separately the trade-off in capacity.
4. Key uncertainty: demand from the intended audience, with the three scenario probabilities clearly labeled as assumptions.
5. Evidence that could change the choice: credible new demand information, changed room costs, or a requirement an option cannot meet.
6. Action and owner: who verifies the terms, who makes the booking and what remains unresolved.
7. Review point: after the event, compare actual demand, costs and execution with what was recorded beforehand.

Do not rewrite the original entry to make it fit the result. Add a dated update. If the plan changes, preserve why it changed. A useful record distinguishes what you knew at the time from what became obvious only afterward.

Review the process without giving luck all the credit

Baron and Hershey’s 1988 experiments found that outcomes influenced evaluations of decisions even when the information available to the decision-maker was held constant. A preregistered 2023 replication also found outcome bias in a medical-decision scenario. These were judgments of scenarios, not tests establishing the quality of real clinical care or every kind of everyday choice. Original outcome-bias study; replication by Aiyer and colleagues.

If the workshop fills the large room, that result alone does not prove the organizer’s earlier demand estimate was well supported. If attendance is disappointing, that alone does not prove that every step was careless. Ask two separate questions: was the process reasonable with the information available, and what does the result teach us now?

That distinction should not become an excuse to ignore repeated surprises. If forecasts repeatedly exceed actual demand, revisit the reference cases, definitions and assumptions. Separate problems with the choice from problems carrying it out: an appropriate room choice and a missed reminder can coexist. Look for the change you can make on the next attempt.

The aim is not a decision you can defend forever. It is one whose reasons are clear enough to examine, whose uncertainties are honest enough to update, and whose next step is practical enough to take.

AI-generated conceptual editorial illustration for A Wandering Mind; not a measured decision diagram or documentary photograph.

Sources and further reading

Primary sources checked September 9, 2026. Research summaries are limited to the original-paper abstracts retrieved and the Green Book sections on uncertainty, optimism and sensitivity analysis; this is not a comprehensive literature review. The workshop, calculations and worksheet are original educational examples. No independent professional assessment or real-world test of this framework is claimed.

• Kahneman and Klein (2009) — Conditions for intuitive expertise
• Mitchell, Russo and Pennington (1989) — Back to the future: Temporal perspective in the explanation of events
• Haynes and colleagues (2009) — A surgical safety checklist to reduce morbidity and mortality in a global population
• Urbach and colleagues (2014) — Introduction of surgical safety checklists in Ontario, Canada
• HM Treasury — The Green Book 2026
• Baron and Hershey (1988) — Outcome bias in decision evaluation
• Aiyer and colleagues (2023) — A preregistered replication of Baron and Hershey’s outcome-bias study