Chapter 9 of 12
Truth, lies and uncertainty
Listen to this lesson
Synthetic voiceOn this page
Checking an account and detecting a lie are different tasks. Establish what is claimed, what the evidence supports and what remains uncertain about knowledge or intent. Someone can be inaccurate without deliberately misleading you; confidence and nervousness do not settle either question.
One habit to practise
When an answer matters to a decision, ask: “What would help me check that?”
What lie-detection research can tell us
Charles Bond and Bella DePaulo’s 2006 meta-analysis combined 206 research reports involving 24,483 judges. When people judged truth and lies without special aids, average accuracy was about 54%. Truths were classified correctly about 61% of the time and lies about 47%. These averages do not give the probability that a particular person is lying. Research record.
Aldert Vrij, Maria Hartwig and Pär Granhag’s 2019 review found known nonverbal deception cues faint and unreliable. Eye contact, fidgeting and confidence do not make a dependable lie test, even when several uncertain cues are combined. Research record.
Brief fictional example: A contractor looks away while discussing a delay. Check the claimed order date and delivery information rather than treating the movement as evidence of dishonesty.
Evaluate the account and its support
Definition
- Corroboration
-
Additional evidence that supports a claim, especially evidence from an independent source.
Example: A supplier’s order confirmation supports the contractor’s statement that a part was ordered.
Ask four questions:
- What event, timing or agreement is being claimed?
- What relevant independent information is available?
- Does it address that claim, or merely something nearby?
- What discrepancy or uncertainty remains?
A supplier’s quote can show an enquiry without showing that an order was completed. Records can also be incomplete or misleading, so check their origin and meaning. Multiple people repeating one source are not independent confirmation.
What about FBI approaches?
The FBI-managed High-Value Detainee Interrogation Group’s 2016 Interrogation: A Review of the Science brings together investigative research, including rapport-based, noncoercive information gathering. Its setting differs from an ordinary conversation. Official report.
An agency name or former investigator’s anecdote does not establish a method’s effectiveness. Ask whether the research measured more talk, more accurate information or more confessions: those are different outcomes. Do not import coercion, traps or a body-language checklist into family or business relationships.
Check your own stake
Wanting reassurance can make weak evidence feel enough. Earlier betrayal can make a new discrepancy feel decisive. Name what you hope or fear, then apply the same standard to welcome and unwelcome information.
A separate brief example: You want a purchase to succeed because you recommended it. That is a reason to inspect delivery evidence carefully, not a reason to discount the buyer’s concern or assume the seller lied.
Turn uncertainty into a practical decision
A date mismatch may reflect a mistake, a different meaning of “ordered” or deception. A plausible explanation is not proof either. Make the next arrangement depend on a relevant check, clear terms or a safeguard where appropriate.
You can change supplier, clarify a deadline or set a limit without obtaining a confession. Where threats, control or a formal dispute are involved, use appropriate support rather than conducting an amateur investigation.
Go deeper
Truth, lies and uncertainty · Denying an event · Comparing explanations · Difficulty trusting.
Key takeaways
- Accuracy, knowledge and intention are separate questions.
- Demeanour and agency-associated branding do not establish truth.
- Use relevant independent checks and make proportionate decisions under uncertainty.
Put it into practice
Recall
Why is a confident or nervous delivery weak grounds for deciding whether someone is lying?
Show the useful distinction
Known behavioural cues are unreliable and can have several causes. Ask “What would help me check that?” and focus on the particular claim.
Try it
A new fictional situation: A seller says an item was dispatched on Monday. Tracking shows the carrier received it on Wednesday.
What could the discrepancy establish, and what remains uncertain? Say or write your answer before opening the example.
Show an example answer
The dates need clarification: label creation, packing and carrier receipt may be different events. Ask what “dispatched” meant and compare the actual terms. The discrepancy matters but does not by itself establish what the seller knew or intended.
Use it this week
This week, choose one ordinary claim you strongly want to believe or disbelieve. State it precisely and make one legitimate, relevant check. Record what it established and how it changed your decision. Do not secretly monitor someone, set a trap or test a threatening situation.
My practice notes: Check a claim you want to believe
Notes save automatically in this browser when storage is available. Avoid names and sensitive details, especially on a shared browser. Download a text copy before clearing your browser data.
More practice (optional)
Another comparison
A separate fictional situation: Two friends repeat the same person’s account of an event. Do you now have three independent sources?
Show what to consider
No. The reports may share one origin. Establish who witnessed what before treating repeated accounts as independent corroboration.
Recall Chapter 2
This is a separate example. Before looking back, answer aloud or in your own notes: A delivery service’s warning system flags 12 parcels. Only 3 are actually late. Does a warning usually mean a parcel is late in this invented example?
Compare your answer
No. Only 3 of 12 flagged parcels are late; 9 are false alarms. Look at how often a warning is wrong before treating it as proof. Revisit the explanation.
Go deeper in the reference library
- Messages, tone & delayed replies — Interpret short or unanswered messages with attention to wording, medium, expectations and missing information.