{ "version": 1, "schemaVersion": "1.0.0", "updatedAt": "2026-08-22", "title": "Reasoning Practice Scenarios", "description": "Realistic decision scenarios that teach the next reasoning move, not memory for cognitive-bias labels.", "difficultyLevels": { "starter": "One clear reasoning risk and one practical first move.", "intermediate": "Several plausible options and missing evidence to separate.", "advanced": "Competing lenses, structural causes or trade-offs that must remain visible." }, "boundary": "A scenario suggests a useful first lens and technique. It does not diagnose a person, prove a bias caused the decision or guarantee that the technique will improve the outcome.", "scenarios": [ { "slug": "hiring-first-impression", "title": "The confident first ten minutes", "difficulty": "starter", "situation": "hiring-interviews", "skill": "evidence-evaluation", "primaryLens": "cognitive-bias-anchoring-effect", "alternativeLenses": [ "cognitive-bias-confirmation-bias" ], "technique": "predefined-criteria", "prompt": "A candidate gives a confident opening answer. The panel spends the rest of the interview interpreting mixed evidence as proof that the candidate is strong.", "question": "What is the best first move before the panel gives a rating?", "options": [ { "id": "a", "text": "Ask the most senior interviewer for an immediate overall score." }, { "id": "b", "text": "Return to criteria fixed before the interview and score the same evidence against each criterion." }, { "id": "c", "text": "Add more informal conversation to see whether the first impression feels consistent." } ], "bestOption": "b", "explanation": "The opening answer may act as an anchor. Predefined criteria reduce the freedom to reinterpret later evidence around that first impression.", "missingEvidence": [ "The interview criteria and scoring rules agreed before the meeting.", "Independent notes from each interviewer.", "Evidence from comparable candidates assessed with the same questions." ], "evidenceNote": "Anchoring is a plausible first lens because an early judgment shaped later interpretation. The scenario does not prove that anchoring caused the final rating.", "nextAction": "Collect independent criterion-level scores before discussing an overall recommendation." }, { "slug": "hiring-evidence-shopping", "title": "The favourite candidate gets easier questions", "difficulty": "intermediate", "situation": "hiring-interviews", "skill": "evidence-evaluation", "primaryLens": "cognitive-bias-confirmation-bias", "alternativeLenses": [ "cognitive-bias-anchoring-effect" ], "technique": "consider-the-opposite", "prompt": "The team already prefers one candidate. Interviewers ask that candidate to explain strengths, while competing candidates are pressed mainly on weaknesses.", "question": "Which action would make the comparison more informative?", "options": [ { "id": "a", "text": "Ask what evidence would make the preferred candidate a poor choice, then collect the same evidence for every candidate." }, { "id": "b", "text": "Let each interviewer follow their intuition because structured questions can feel artificial." }, { "id": "c", "text": "Invite the preferred candidate to one more informal meeting and decide from team chemistry." } ], "bestOption": "a", "explanation": "The evidence search is asymmetric. Considering the opposite creates a real test of the preferred conclusion instead of collecting only supportive material.", "missingEvidence": [ "A shared question set.", "Evidence that would lower the preferred candidate’s rating.", "Comparable work samples or structured references." ], "evidenceNote": "Confirmation bias is a reasonable lens for the search pattern, but other explanations such as poor interview design may also matter.", "nextAction": "Write one disconfirming question and apply it to all serious candidates." }, { "slug": "hiring-past-success-story", "title": "The last star hire becomes the template", "difficulty": "advanced", "situation": "hiring-interviews", "skill": "evidence-evaluation", "primaryLens": "cognitive-bias-outcome-bias", "alternativeLenses": [], "technique": "predefined-criteria", "prompt": "A previous hire performed very well. The team now treats every feature of that hiring decision as a proven success formula, although several factors were uncertain at the time.", "question": "What should the team review before copying the old decision?", "options": [ { "id": "a", "text": "The original evidence, alternatives and uncertainty recorded before the successful outcome was known." }, { "id": "b", "text": "Only the qualities of the successful employee that are easiest to remember today." }, { "id": "c", "text": "Whether the old hiring manager still feels confident about the choice." } ], "bestOption": "a", "explanation": "A good outcome does not show that every part of the original process was good. Reapplying criteria fixed before the new hiring decision helps separate useful process evidence from the success story.", "missingEvidence": [ "The original scorecard and candidate pool.", "What risks were accepted at the time.", "Which later conditions helped the employee succeed." ], "evidenceNote": "Outcome bias is a useful review lens here, but one successful case is also too small to establish a reliable hiring rule.", "nextAction": "Reconstruct the old decision from records, then test each process element against criteria fixed for the new role." }, { "slug": "estimate-target-date-anchor", "title": "The date arrives before the estimate", "difficulty": "starter", "situation": "project-estimation", "skill": "forecasting", "primaryLens": "cognitive-bias-anchoring-effect", "alternativeLenses": [ "egocentric-bias-planning-fallacy" ], "technique": "independent-estimate", "prompt": "A sponsor says the migration should finish in twelve weeks. The delivery team then estimates each work package while that date remains visible.", "question": "What should the team do before negotiating the final schedule?", "options": [ { "id": "a", "text": "Build an independent range from work packages and assumptions before comparing it with the twelve-week target." }, { "id": "b", "text": "Split twelve weeks across the work packages so every activity fits the target." }, { "id": "c", "text": "Ask the sponsor to repeat why twelve weeks is important and use that as the baseline." } ], "bestOption": "a", "explanation": "The target can become the reference point for every later estimate. An independent estimate preserves a separate baseline and makes any movement toward the target inspectable.", "missingEvidence": [ "Independent estimates from the delivery team.", "Dependency and availability assumptions.", "Comparable completed migrations." ], "evidenceNote": "The visible target creates an anchoring risk, but the final schedule may still converge on twelve weeks if independent evidence supports it.", "nextAction": "Produce a range without the target in view and record why it changes after comparison." }, { "slug": "estimate-inside-plan-only", "title": "A detailed plan with no outside view", "difficulty": "intermediate", "situation": "project-estimation", "skill": "forecasting", "primaryLens": "egocentric-bias-planning-fallacy", "alternativeLenses": [ "heuristic-bias-availability-bias" ], "technique": "reference-class-forecasting", "prompt": "A team has a detailed task plan for a data migration. It predicts 40 days, but nobody has checked how long similar migrations actually took.", "question": "Which evidence should be added first?", "options": [ { "id": "a", "text": "More detail inside the current plan until every day is allocated." }, { "id": "b", "text": "A reference class of comparable completed migrations and their actual duration range." }, { "id": "c", "text": "A vote on whether 40 days feels optimistic or pessimistic." } ], "bestOption": "b", "explanation": "More inside detail can improve execution planning without correcting the baseline. Reference-class forecasting tests the plan against completed cases.", "missingEvidence": [ "Durations of comparable completed migrations.", "Reasons this project differs from the reference class.", "The distribution, not only the best or worst example." ], "evidenceNote": "Planning fallacy is a plausible lens when the estimate relies mainly on the inside plan. A poor reference class can also mislead, so comparability must be explained.", "nextAction": "Set an outside-view range first, then document any evidence-based adjustment." }, { "slug": "estimate-recent-disaster", "title": "One failed project resets every estimate", "difficulty": "advanced", "situation": "project-estimation", "skill": "forecasting", "primaryLens": "heuristic-bias-availability-bias", "alternativeLenses": [ "egocentric-bias-planning-fallacy" ], "technique": "reference-class-forecasting", "prompt": "The previous project missed its date by six months. Managers now double every estimate, even though most comparable projects had much smaller delays.", "question": "What is the best way to use the recent failure?", "options": [ { "id": "a", "text": "Ignore it because unusual failures contain no useful information." }, { "id": "b", "text": "Treat it as one observation inside a broader reference class and identify whether the same failure conditions apply." }, { "id": "c", "text": "Use it as the default scenario because recent events are the safest guide." } ], "bestOption": "b", "explanation": "The vivid failure deserves inspection, but not automatic dominance. A reference class shows how common that outcome is and whether the same conditions are present.", "missingEvidence": [ "Frequency of similar delays in comparable work.", "Specific causes of the previous failure.", "Whether those causes exist in the new project." ], "evidenceNote": "Availability bias may explain why the recent failure receives extra weight. The failure can still justify a larger estimate when relevant conditions repeat.", "nextAction": "Add the failed project to the reference class and state the conditions that warrant moving toward its outcome." }, { "slug": "incident-obvious-afterward", "title": "The root cause now looks obvious", "difficulty": "starter", "situation": "production-incident", "skill": "metacognition", "primaryLens": "cognitive-bias-hindsight-bias", "alternativeLenses": [ "cognitive-bias-outcome-bias" ], "technique": "decision-journal", "prompt": "After an outage, the failed component is easy to identify. Reviewers say the on-call engineer should have known immediately, although several alerts looked equally plausible during the incident.", "question": "What should the review reconstruct first?", "options": [ { "id": "a", "text": "The timeline of information available at each decision point, before the root cause was known." }, { "id": "b", "text": "The final root-cause diagram, because it contains the complete truth." }, { "id": "c", "text": "Which engineer now gives the most confident explanation." } ], "bestOption": "a", "explanation": "Hindsight compresses uncertainty after the outcome. A contemporaneous decision record restores what was knowable and which alternatives were reasonable at the time.", "missingEvidence": [ "The alerts visible at each time.", "Competing hypotheses considered during response.", "Actions available under the actual time pressure." ], "evidenceNote": "Hindsight bias is a useful lens for the review language, but the reconstruction may still reveal preventable process failures.", "nextAction": "Build an evidence timeline before judging the quality of incident decisions." }, { "slug": "incident-one-outage-policy", "title": "One outage creates a permanent rule", "difficulty": "intermediate", "situation": "production-incident", "skill": "metacognition", "primaryLens": "heuristic-bias-availability-bias", "alternativeLenses": [ "cognitive-bias-outcome-bias" ], "technique": "predefined-update-criteria", "prompt": "A rare database failure causes a severe outage. The team proposes an expensive control for every system without checking how often the failure occurs or whether the control addresses other risks.", "question": "What should happen before the control becomes policy?", "options": [ { "id": "a", "text": "Define the evidence and recurrence threshold that would justify the control across systems." }, { "id": "b", "text": "Approve it quickly because severe incidents should always produce broad rules." }, { "id": "c", "text": "Wait until the incident feels less memorable, then use intuition again." } ], "bestOption": "a", "explanation": "The vivid outage can dominate risk judgment. Predefined criteria make the policy depend on recurrence, exposure and control effectiveness rather than memory alone.", "missingEvidence": [ "Base rate of the failure mode.", "Systems with the same exposure.", "Expected risk reduction and operating cost of the control." ], "evidenceNote": "Availability is a plausible risk because one severe event dominates the proposal. A rare event can still justify a broad control when expected harm is high.", "nextAction": "Set a measurable adoption rule and test it against the system portfolio." }, { "slug": "roadmap-loud-customer", "title": "The loudest customer becomes the market", "difficulty": "starter", "situation": "product-prioritization", "skill": "decision-making-under-uncertainty", "primaryLens": "heuristic-bias-availability-bias", "alternativeLenses": [ "cognitive-bias-confirmation-bias" ], "technique": "base-rate-check", "prompt": "One large customer describes a painful workflow problem. The roadmap discussion treats that story as proof that most users need the same feature.", "question": "What evidence should be checked before reprioritizing the roadmap?", "options": [ { "id": "a", "text": "How common the workflow and pain are across the relevant customer population." }, { "id": "b", "text": "How strongly the customer repeats the request in the next meeting." }, { "id": "c", "text": "Whether the story is memorable enough to explain to executives." } ], "bestOption": "a", "explanation": "A vivid case can be important without being representative. A base-rate check asks how often the problem occurs and for whom.", "missingEvidence": [ "Number of affected customers and users.", "Severity and frequency of the problem.", "Evidence from customers who do not request the feature." ], "evidenceNote": "Availability bias may increase the story’s weight, but a single strategic customer can still justify action for commercial reasons. Those reasons should be explicit.", "nextAction": "Estimate prevalence and value before comparing the feature with other roadmap options." }, { "slug": "roadmap-already-built", "title": "Too much code to stop now", "difficulty": "intermediate", "situation": "product-prioritization", "skill": "decision-making-under-uncertainty", "primaryLens": "logical-fallacy-escalation-of-commitment", "alternativeLenses": [ "cognitive-bias-confirmation-bias" ], "technique": "incremental-value-check", "prompt": "A feature has consumed eight months of work. New research shows limited demand, but the team argues that stopping would waste the investment.", "question": "Which question should drive the next decision?", "options": [ { "id": "a", "text": "How much effort has already been spent compared with the original budget?" }, { "id": "b", "text": "Would we fund the remaining work today, based on future cost, value and uncertainty?" }, { "id": "c", "text": "How embarrassing would cancellation look to the steering committee?" } ], "bestOption": "b", "explanation": "Past work cannot be recovered. The incremental-value check evaluates only the future consequences of continuing, stopping or changing direction.", "missingEvidence": [ "Remaining cost and delivery risk.", "Current evidence of future value.", "Switching and cancellation consequences that still lie in the future." ], "evidenceNote": "Escalation of commitment is a plausible lens because past investment is used as a reason for further investment. Relevant future switching costs should not be ignored.", "nextAction": "Write a future-only business case for the next release step." }, { "slug": "vendor-demo-glow", "title": "The polished demo proves too much", "difficulty": "starter", "situation": "vendor-tool-selection", "skill": "evidence-evaluation", "primaryLens": "availability-heuristic-anthropomorphism", "alternativeLenses": [ "cognitive-bias-confirmation-bias" ], "technique": "predefined-criteria", "prompt": "A vendor presents a fluent AI assistant in a polished demo. The team begins describing the product as reliable and intelligent, although the critical workflow was not tested with company data.", "question": "What should the team require next?", "options": [ { "id": "a", "text": "A scripted test against predefined capabilities, failure cases and company data." }, { "id": "b", "text": "A longer presentation so the team can judge the product’s personality." }, { "id": "c", "text": "A list of well-known customers as a substitute for task testing." } ], "bestOption": "a", "explanation": "Humanlike presentation can inflate perceived capability. Predefined criteria force the product to demonstrate the task that matters in the real environment.", "missingEvidence": [ "Performance on the critical workflow.", "Failure behaviour and escalation paths.", "Evidence under the organisation’s data and controls." ], "evidenceNote": "Anthropomorphism is a relevant lens for capability inference from presentation. The demo may still represent real capability, which must be tested directly.", "nextAction": "Run the same evidence-based acceptance test for every serious vendor." }, { "slug": "vendor-first-price", "title": "The first price defines the negotiation", "difficulty": "intermediate", "situation": "vendor-tool-selection", "skill": "evidence-evaluation", "primaryLens": "cognitive-bias-anchoring-effect", "alternativeLenses": [ "cognitive-bias-confirmation-bias" ], "technique": "independent-estimate", "prompt": "A vendor opens with a three-year price of €900,000. Internal discussions now focus on whether a 15% discount is enough, although the team never estimated value or alternatives independently.", "question": "What should be created before the next commercial round?", "options": [ { "id": "a", "text": "An independent value range, reservation point and comparison with alternatives." }, { "id": "b", "text": "A more detailed explanation of why €900,000 is the vendor’s normal price." }, { "id": "c", "text": "A discount target expressed only as a percentage of the opening price." } ], "bestOption": "a", "explanation": "The opening number defines the frame. An independent estimate gives the team a separate basis for price and scope decisions.", "missingEvidence": [ "Expected value and total cost of ownership.", "Comparable offers or internal alternatives.", "A reservation point defined before the next offer." ], "evidenceNote": "Anchoring is plausible because all later numbers are expressed around the opening price. The opening price can still be reasonable if independent evidence supports it.", "nextAction": "Write an independent commercial range and the assumptions behind it." }, { "slug": "forecast-last-quarter", "title": "One quarter becomes the trend", "difficulty": "starter", "situation": "forecasting", "skill": "forecasting", "primaryLens": "heuristic-bias-availability-bias", "alternativeLenses": [ "probability-bias-subadditivity-effect" ], "technique": "base-rate-check", "prompt": "Sales rose sharply last quarter. The new annual forecast assumes the same growth will continue, although the wider history is more variable.", "question": "What should the forecast use as its starting point?", "options": [ { "id": "a", "text": "The most exciting recent quarter because it contains the newest information." }, { "id": "b", "text": "A broader distribution of comparable periods, then an explicit adjustment for current conditions." }, { "id": "c", "text": "The target required by the annual plan." } ], "bestOption": "b", "explanation": "Recent evidence matters, but it should be compared with a broader base rate. The adjustment should explain why this quarter is more informative than the rest.", "missingEvidence": [ "Historical distribution of comparable periods.", "Specific drivers of the recent change.", "Evidence that those drivers will continue." ], "evidenceNote": "Availability may give the latest quarter extra weight. A genuine structural change can justify a large adjustment when supported by independent evidence.", "nextAction": "Set the historical baseline and document each reason for moving away from it." }, { "slug": "forecast-risk-parts", "title": "The risks do not add up", "difficulty": "intermediate", "situation": "forecasting", "skill": "forecasting", "primaryLens": "probability-bias-subadditivity-effect", "alternativeLenses": [ "heuristic-bias-availability-bias" ], "technique": "base-rate-check", "prompt": "A team estimates a 25% chance that the project will be late. When asked separately, it gives 20% to supplier delay, 20% to testing delay and 15% to staffing delay, with overlap left undefined.", "question": "What should the team do next?", "options": [ { "id": "a", "text": "Keep every number because separate estimates are always more accurate." }, { "id": "b", "text": "Define overlap and reconcile the component risks with the total probability using comparable project data." }, { "id": "c", "text": "Choose the highest component probability as the total risk." } ], "bestOption": "b", "explanation": "Decomposed risks can receive more probability than the undivided event. Reconciliation and a reference distribution make the forecast internally checkable.", "missingEvidence": [ "How risk categories overlap.", "The frequency of each cause in comparable projects.", "A rule for combining component estimates." ], "evidenceNote": "Subadditivity is a useful lens for inconsistent whole-versus-part estimates. The numbers may also reflect unclear definitions rather than a stable cognitive effect.", "nextAction": "Create mutually clear risk definitions and reconcile the total before publishing the forecast." }, { "slug": "forecast-memory-rewrite", "title": "The forecast was “basically right”", "difficulty": "advanced", "situation": "forecasting", "skill": "forecasting", "primaryLens": "cognitive-bias-hindsight-bias", "alternativeLenses": [ "heuristic-bias-availability-bias" ], "technique": "predefined-update-criteria", "prompt": "A forecast missed badly. After the outcome, the owner says the result was always possible and that the original forecast should be read more generously.", "question": "What would make the review useful?", "options": [ { "id": "a", "text": "Compare the outcome with the frozen forecast, range and update rules written before the result." }, { "id": "b", "text": "Allow the owner to restate what they meant now that the result is known." }, { "id": "c", "text": "Judge only whether the final narrative sounds reasonable." } ], "bestOption": "a", "explanation": "A frozen record prevents the forecast from changing in memory. Predefined update criteria also show whether new evidence should have triggered a revision before the outcome.", "missingEvidence": [ "The original point estimate or range.", "Assumptions and update triggers.", "Evidence available before each review date." ], "evidenceNote": "Hindsight is a plausible review risk because the original forecast is being reinterpreted after the result. Ambiguous documentation can be a separate process problem.", "nextAction": "Score the forecast against the original record, then improve the future recording standard." }, { "slug": "review-latest-quarter", "title": "The last quarter becomes the whole year", "difficulty": "starter", "situation": "performance-review", "skill": "metacognition", "primaryLens": "heuristic-bias-availability-bias", "alternativeLenses": [ "cognitive-bias-confirmation-bias" ], "technique": "predefined-criteria", "prompt": "An employee had one difficult final quarter. The annual review now focuses almost entirely on that period and gives little weight to earlier results.", "question": "What is the best first correction?", "options": [ { "id": "a", "text": "Use the same predefined criteria across the full review period and inspect evidence by period." }, { "id": "b", "text": "Trust the manager’s memory because recent work is usually the most relevant." }, { "id": "c", "text": "Ask the employee to provide only their three strongest examples." } ], "bestOption": "a", "explanation": "Recent events are easier to recall. Predefined criteria and period coverage reduce the chance that one vivid interval becomes the entire evaluation.", "missingEvidence": [ "Evidence across the full review period.", "Criteria and expectations set before the review.", "Comparable treatment of positive and negative examples." ], "evidenceNote": "Availability may explain the heavy focus on recent events. The latest quarter can still deserve more weight when the role or expectations changed.", "nextAction": "Build a criterion-by-period evidence table before setting the rating." }, { "slug": "review-lucky-result", "title": "A good result proves a good process", "difficulty": "intermediate", "situation": "performance-review", "skill": "metacognition", "primaryLens": "cognitive-bias-outcome-bias", "alternativeLenses": [ "heuristic-bias-availability-bias" ], "technique": "decision-journal", "prompt": "A manager skipped required checks but the project succeeded. The review treats the success as proof that the manager’s approach was excellent.", "question": "What should be evaluated separately?", "options": [ { "id": "a", "text": "The quality of the decision process using what was known before the outcome." }, { "id": "b", "text": "Only the final business result because process is secondary." }, { "id": "c", "text": "Whether the manager can tell a convincing success story." } ], "bestOption": "a", "explanation": "Good outcomes can follow weak processes. A decision record helps judge whether the risk was understood and managed before luck was known.", "missingEvidence": [ "Required checks and why they existed.", "Risks known before execution.", "Whether similar choices would perform reliably across repeated cases." ], "evidenceNote": "Outcome bias is a relevant lens, but process rules may also need review if they add cost without reducing meaningful risk.", "nextAction": "Rate process quality and outcome quality in separate fields." }, { "slug": "negotiation-first-offer", "title": "The opening offer owns the range", "difficulty": "starter", "situation": "negotiation", "skill": "decision-making-under-uncertainty", "primaryLens": "cognitive-bias-anchoring-effect", "alternativeLenses": [ "cognitive-bias-confirmation-bias" ], "technique": "independent-estimate", "prompt": "A supplier opens at €120 per unit. The buying team immediately discusses whether €105 would be a good deal, although no independent range was prepared.", "question": "What should the team establish before responding?", "options": [ { "id": "a", "text": "An evidence-based range, reservation point and alternatives independent of the opening offer." }, { "id": "b", "text": "A percentage discount that looks substantial against €120." }, { "id": "c", "text": "How confident the supplier sounded when naming the price." } ], "bestOption": "a", "explanation": "The opening number frames the discussion. An independent estimate moves the negotiation back to value, alternatives and evidence.", "missingEvidence": [ "Comparable prices or cost drivers.", "Value of alternatives and switching costs.", "A reservation point agreed before the counteroffer." ], "evidenceNote": "Anchoring is a plausible risk because the counteroffer is derived from the opening number. The opening number may still contain useful market information.", "nextAction": "Freeze the independent range before reviewing the supplier’s justification." }, { "slug": "negotiation-public-position", "title": "The team defends what it already announced", "difficulty": "advanced", "situation": "negotiation", "skill": "decision-making-under-uncertainty", "primaryLens": "cognitive-bias-confirmation-bias", "alternativeLenses": [ "logical-fallacy-escalation-of-commitment" ], "technique": "predefined-update-criteria", "prompt": "A team publicly announced that it would never accept a contract longer than one year. New evidence shows a longer term could reduce total risk, but every analysis is now used to defend the public position.", "question": "What would make an update legitimate rather than convenient?", "options": [ { "id": "a", "text": "Apply evidence thresholds and review conditions defined before the new offer arrived." }, { "id": "b", "text": "Keep the position because changing it would always show weakness." }, { "id": "c", "text": "Change the position privately without recording why." } ], "bestOption": "a", "explanation": "A public commitment can shape later evidence use. Predefined update criteria allow change when relevant evidence arrives without silently rewriting the logic.", "missingEvidence": [ "Original reasons for the one-year limit.", "Evidence that changes those reasons.", "Future risks and benefits of each contract term." ], "evidenceNote": "Confirmation bias may influence the analysis, while reputation and bargaining strategy are also real considerations that should be stated separately.", "nextAction": "Record the evidence threshold for changing the position and test the new offer against it." }, { "slug": "ai-first-answer-frame", "title": "The model supplies the first hypothesis", "difficulty": "starter", "situation": "ai-assisted-research", "skill": "ai-assisted-reasoning", "primaryLens": "cognitive-bias-anchoring-effect", "alternativeLenses": [ "false-priors-automation-bias" ], "technique": "independent-estimate", "prompt": "Before reviewing the source material, a researcher asks an AI assistant for the likely explanation. The rest of the search follows the categories in that first answer.", "question": "What should be done before relying on the model’s frame?", "options": [ { "id": "a", "text": "Write an independent provisional question, alternatives and evidence needs before reading the generated explanation." }, { "id": "b", "text": "Ask the same model to make the original explanation more detailed." }, { "id": "c", "text": "Use the first answer as the literature-review structure because it is efficient." } ], "bestOption": "a", "explanation": "The first generated frame can become an anchor for search and interpretation. An independent provisional view preserves alternatives before the model shapes them.", "missingEvidence": [ "The researcher’s pre-AI question and candidate explanations.", "Primary sources that support or challenge each explanation.", "Evidence not represented in the model’s first frame." ], "evidenceNote": "Anchoring is a plausible interaction risk. The model’s first hypothesis can still be useful, but it should compete with independently stated alternatives.", "nextAction": "Write the research question and at least two plausible alternatives before reading the generated answer." }, { "slug": "ai-polished-unsupported", "title": "Fluent text without inspectable support", "difficulty": "intermediate", "situation": "ai-assisted-research", "skill": "ai-assisted-reasoning", "primaryLens": "false-priors-automation-bias", "alternativeLenses": [ "truth-judgment-illusory-truth-effect" ], "technique": "source-trace", "prompt": "An AI assistant gives a clear recommendation and names several studies, but the citations cannot be located. The team wants to use the recommendation because it sounds complete.", "question": "What is the best next move?", "options": [ { "id": "a", "text": "Trace every consequential claim to an inspectable independent source and mark unsupported claims as unverified." }, { "id": "b", "text": "Ask another AI assistant whether the first answer sounds correct." }, { "id": "c", "text": "Keep the recommendation but remove the citations from the final document." } ], "bestOption": "a", "explanation": "Automation bias concerns inappropriate reliance on automated output. Source tracing makes the recommendation depend on evidence that can be inspected outside the model.", "missingEvidence": [ "Existence and content of the cited studies.", "Whether sources support the exact claim.", "Alternative evidence or expert review for consequential decisions." ], "evidenceNote": "Automation bias is a possible lens because fluency is substituting for verification. The recommendation may still be correct, but correctness must be checked independently.", "nextAction": "Create a claim table with verified, disputed and unsupported status." }, { "slug": "ai-many-summaries-one-source", "title": "Five AI summaries, one origin", "difficulty": "advanced", "situation": "ai-assisted-research", "skill": "ai-assisted-reasoning", "primaryLens": "truth-judgment-illusory-truth-effect", "alternativeLenses": [ "false-priors-automation-bias" ], "technique": "source-trace", "prompt": "Five generated summaries repeat the same market statistic. The team treats the repetition as strong agreement, but every answer appears to depend on the same unverified report.", "question": "How should confidence be assessed?", "options": [ { "id": "a", "text": "Count each generated answer as an independent source because the wording differs." }, { "id": "b", "text": "Trace the statistic to its origin and count genuinely independent evidence, not repeated outputs." }, { "id": "c", "text": "Accept the statistic if it appears in at least three model responses." } ], "bestOption": "b", "explanation": "Repetition increases familiarity without creating independent evidence. Source tracing collapses dependent summaries back to their shared origin.", "missingEvidence": [ "The earliest inspectable source.", "Independent datasets or analyses.", "Definitions, sample and date behind the statistic." ], "evidenceNote": "Illusory truth is a relevant lens because repeated wording can increase acceptance. Dependence between model outputs is also a data-provenance problem.", "nextAction": "Replace the answer count with an independent-source count." }, { "slug": "claim-ten-articles-one-source", "title": "Ten articles repeat one report", "difficulty": "starter", "situation": "checking-repeated-claims", "skill": "information-verification", "primaryLens": "truth-judgment-illusory-truth-effect", "alternativeLenses": [ "memory-bias-continued-influence-effect" ], "technique": "source-trace", "prompt": "A claim appears in ten articles. Following the links shows that nine articles cite the tenth, which cites one small report.", "question": "What should determine confidence?", "options": [ { "id": "a", "text": "The number of pages containing the sentence." }, { "id": "b", "text": "The quality of the original report and any genuinely independent evidence." }, { "id": "c", "text": "How high the articles rank in search results." } ], "bestOption": "b", "explanation": "Ten repetitions do not equal ten independent observations. Source tracing reveals the evidence structure behind the apparent agreement.", "missingEvidence": [ "Method and sample of the original report.", "Independent replications or datasets.", "Whether later articles add new evidence or only repeat the claim." ], "evidenceNote": "Illusory truth may make the repeated claim feel more credible. The main operational problem is source dependence, which can be checked directly.", "nextAction": "Draw a simple source tree and assess each independent root." }, { "slug": "claim-correction-without-replacement", "title": "The correction removes the story but leaves a gap", "difficulty": "intermediate", "situation": "checking-repeated-claims", "skill": "information-verification", "primaryLens": "memory-bias-continued-influence-effect", "alternativeLenses": [ "truth-judgment-illusory-truth-effect" ], "technique": "source-trace", "prompt": "A report corrects a false explanation for an event but gives no supported alternative. Later discussions still use the old explanation because it is the only coherent story people remember.", "question": "What would make the correction more usable?", "options": [ { "id": "a", "text": "Repeat only that the old claim was false." }, { "id": "b", "text": "State the supported replacement explanation, its evidence and what remains unknown." }, { "id": "c", "text": "Avoid discussing the event again so the old story fades naturally." } ], "bestOption": "b", "explanation": "Corrections are easier to use when they replace the old causal story with a supported account. Source tracing also shows what evidence supports the replacement.", "missingEvidence": [ "Evidence against the old explanation.", "Evidence for the replacement explanation.", "Uncertainty that remains after correction." ], "evidenceNote": "Continued influence is a relevant lens because the rejected explanation still guides reasoning. The effect can also reflect an unresolved information gap.", "nextAction": "Write a replacement account with source links and explicit unknowns." }, { "slug": "career-salary-anchor", "title": "One salary number defines the career choice", "difficulty": "starter", "situation": "career-change", "skill": "decision-making-under-uncertainty", "primaryLens": "cognitive-bias-anchoring-effect", "alternativeLenses": [ "logical-fallacy-escalation-of-commitment" ], "technique": "predefined-criteria", "prompt": "A recruiter names a salary 20% above the current role. The career discussion now revolves around that number, while learning, stability, workload and future options remain vague.", "question": "What should happen before comparing the offers?", "options": [ { "id": "a", "text": "Define the small set of career criteria and evidence for each criterion before weighting the salary difference." }, { "id": "b", "text": "Convert every non-salary factor into a guess based on the new salary." }, { "id": "c", "text": "Accept that compensation is the only objective criterion." } ], "bestOption": "a", "explanation": "The first salient number can dominate the frame. Predefined criteria protect the decision from collapsing into one dimension.", "missingEvidence": [ "Evidence about role scope, manager, workload and learning.", "The relative importance of criteria before the offer.", "Total compensation and downside conditions, not only headline salary." ], "evidenceNote": "Anchoring is a plausible risk because one number organizes the decision. Salary can still be the most important criterion when that priority is explicit.", "nextAction": "Create and weight the criteria before scoring either role." }, { "slug": "career-years-invested", "title": "Twelve years means another year", "difficulty": "intermediate", "situation": "career-change", "skill": "decision-making-under-uncertainty", "primaryLens": "logical-fallacy-escalation-of-commitment", "alternativeLenses": [ "cognitive-bias-anchoring-effect" ], "technique": "incremental-value-check", "prompt": "A specialist dislikes the current path but says changing direction would waste twelve years of experience. The next year is evaluated mainly through the past investment.", "question": "Which comparison is most useful?", "options": [ { "id": "a", "text": "Compare the future value, cost and options of staying versus changing from today onward." }, { "id": "b", "text": "Count the years already invested and continue whenever the number is large." }, { "id": "c", "text": "Choose the path that makes the past story look most consistent." } ], "bestOption": "a", "explanation": "Past experience may create transferable value, but it is not itself a reason to invest another year. The decision should compare future consequences.", "missingEvidence": [ "Transferable skills from the existing path.", "Future learning and income under each option.", "Real switching costs that still occur in the future." ], "evidenceNote": "Escalation of commitment is a plausible lens because past time drives the next investment. Past experience can remain relevant through future benefits and switching costs.", "nextAction": "Write a future-only comparison for the next twelve months." }, { "slug": "career-all-or-nothing", "title": "The career change is treated as irreversible", "difficulty": "advanced", "situation": "career-change", "skill": "decision-making-under-uncertainty", "primaryLens": "cognitive-bias-projection-bias", "alternativeLenses": [ "logical-fallacy-escalation-of-commitment" ], "technique": "reversible-test", "prompt": "A person feels exhausted today and considers leaving the field completely. The choice is framed as staying forever or resigning immediately, although smaller tests are possible.", "question": "What is the best way to reduce uncertainty?", "options": [ { "id": "a", "text": "Make the full change now because current feelings reveal future preferences." }, { "id": "b", "text": "Design a small reversible test, such as a short project, course or internal rotation, with a result that would update the larger choice." }, { "id": "c", "text": "Delay every action until certainty appears." } ], "bestOption": "b", "explanation": "Current feelings can be projected too far into the future. A reversible test creates evidence about the alternative without requiring an immediate permanent move.", "missingEvidence": [ "Whether exhaustion is role-specific or field-wide.", "Experience of the alternative under realistic conditions.", "Recovery, workload and family constraints over time." ], "evidenceNote": "Projection bias is a possible lens because today’s state is used to predict long-term preference. The current state may still signal a real and urgent problem.", "nextAction": "Choose one reversible test and define what result would change the larger decision." }, { "slug": "kpi-score-becomes-proof", "title": "The benchmark score ends the discussion", "difficulty": "intermediate", "situation": "kpi-review", "skill": "evidence-evaluation", "primaryLens": "cognitive-bias-confirmation-bias", "alternativeLenses": [ "heuristic-bias-availability-bias" ], "technique": "red-team-review", "prompt": "A team prefers a new platform and highlights its strong benchmark score. It dismisses workload, data quality and operating constraints that are not measured by the benchmark.", "question": "What should a red-team review ask?", "options": [ { "id": "a", "text": "Which important failure modes and decision criteria are not represented by the score?" }, { "id": "b", "text": "How can the score be presented more prominently to the steering committee?" }, { "id": "c", "text": "Which competing platform has the weakest marketing material?" } ], "bestOption": "a", "explanation": "The preferred conclusion is selecting supportive evidence. A red-team review tests what the benchmark omits and whether the score transfers to the real operating context.", "missingEvidence": [ "Definition and construction of the benchmark.", "Performance under the organisation’s workload and data.", "Important outcomes not represented by the score." ], "evidenceNote": "Confirmation bias is a plausible lens because evidence outside the preferred score is discounted. The benchmark may still be highly informative when it matches the decision goal.", "nextAction": "List the benchmark’s blind spots and test the two most consequential ones." }, { "slug": "kpi-good-score-bad-process", "title": "A good KPI excuses a weak process", "difficulty": "advanced", "situation": "kpi-review", "skill": "evidence-evaluation", "primaryLens": "cognitive-bias-outcome-bias", "alternativeLenses": [ "cognitive-bias-confirmation-bias" ], "technique": "decision-journal", "prompt": "A support team meets its response-time KPI by closing tickets early and reopening them later. Leaders call the process successful because the dashboard is green.", "question": "What should be reviewed separately?", "options": [ { "id": "a", "text": "The measured score, the underlying customer outcome and the process used to produce the score." }, { "id": "b", "text": "Only the dashboard because formal KPIs define success." }, { "id": "c", "text": "Whether the team feels proud of the result." } ], "bestOption": "a", "explanation": "A favourable measured outcome can hide a poor decision process and a weak real-world outcome. A decision record preserves why the KPI was chosen and what it was meant to represent.", "missingEvidence": [ "Customer resolution and repeat-contact outcomes.", "How ticket handling affects the metric.", "The original purpose and known limits of the KPI." ], "evidenceNote": "Outcome bias is a useful lens for treating a green result as proof of process quality. Metric design and incentives are also structural causes that require direct analysis.", "nextAction": "Review the KPI, underlying outcome and process as three separate evidence columns." } ], "releaseVersion": "2026.08.18" }