What this trains: the working judgment an investigator, auditor, recruiter or journalist needs. Each case gives you a short file, then the subject's own words. You decide whether the account holds, and then you say what the decision rests on.
How a case works:
- Read the file lines first. They carry the records, the logs, the baselines and the incentives.
- Then read the statement, which is the subject speaking in their own words.
- Step 1: give a verdict, Truthful or Deceptive, on the specific claim under review. The verdict is about that claim, not guilt or innocence of any broader allegation.
- Step 2: choose the observation that carries the weight. The alternatives are plausible, but they may misread, overinterpret, or give too much weight to part of the file. Only one supports the defensible conclusion.
- Training Focus: use Mixed Skills for 8, 16, or 24 cases, or isolate one skill for a six-case drill. A focused drill uses all six unique authored cases in that skill. The case bank is balanced across truthful and deceptive accounts, but the session quota is not disclosed while you play.
- In Training Mode the answer, the reasoning and the key principle appear at once, with the decisive record and the decisive phrases marked. In Test Mode nothing is revealed until the end, and every explanation is waiting in the case review.
The eight skills scored separately:
- Contradictions: two claims, or a claim and a record, that cannot both be true as stated.
- Timeline and Sequence: whether clock times, travel, order of events and claimed actions can all fit together.
- Changing Stories: separating a material revision from ordinary added detail, and asking what the speaker learned between versions.
- Motive and Incentives: the rewards, pressures and loyalties that raise the risk of distortion without proving it happened.
- Context and Corroboration: counting independent origins rather than repetitions, and reading a record for what it actually captures.
- Behavior and Baseline: change from a person's own normal pattern, weighed against everything else in the file.
- Statement Language: the verbal credibility markers, including detail, distancing, oaths, timeline bridges, hedging and self correction.
- Diagnostic Questioning: choosing the follow-up that separates competing explanations instead of pressing the same denial again.
The habit these cases keep pushing: keep the conclusion the same size as the evidence. Roughly a quarter of the wrong options are overreaches, statements that are perfectly sensible one step further along than the file can carry. A record that shows a badge opened a door does not show who carried the badge, and a document that arrives late does not show anyone instructed a backdate. Learning to stop at the right place is most of the skill.
An honest limit worth knowing: across decades of research, people asked to judge truth from behavior alone perform close to chance. Working from records, sequence, incentives and language does better than that, which is why this training is built the way it is, but every one of these signals describes a tendency across many cases rather than proof about any single one. These files are written so that a genuine deciding factor is present. Real ones are messier, and a truthful person under pressure can produce several warning signs at once. Treat what you learn here as a way to ask better questions, never as grounds to conclude that a particular person lied.
Scoring: the verdict and the observation are worth one point each, so each case is worth two points. Guessing throughout would land at about 38 percent, since the verdict is a coin flip and the observation is one option in four. The Called Deceptive figure shows how often you reached for a deceptive verdict. The case bank is balanced across truthful and deceptive accounts, so the Called Deceptive figure can help reveal a response bias across repeated sessions. Turning on Confidence Rating adds a calibration figure showing how often you were right when you were sure.