Three numbers appear, and almost immediately a rule suggests itself. The feeling is satisfying: the pattern seems clear, the answer feels close, and the next examples come easily. Yet the moment we feel we understand may be precisely the moment to ask a different question. What would help us discover that our first explanation is incomplete?
The original Veritasium video above turns that question into a short encounter. Derek Muller offers a sequence and invites people to propose their own examples. A yes or no tells them whether each example fits his rule. Watching the exercise can be entertaining. Trying to notice your own strategy while watching makes it more revealing.
A yes can fit more than one explanation
The video begins with 2, 4, 8. Many viewers will recognise doubling and immediately try another doubling sequence. If the answer is yes, that can feel like progress. But two different rules can accept the same example. A sequence might double and also increase; agreement with one example does not tell us which description the person holding the rule intended.
Consider a fresh example of the same problem. Suppose you think the rule concerns multiples of three. Offering 3, 6, 9 may produce a yes, but that response does not distinguish your proposal from several alternatives. You have learned that the example is accepted. You have not necessarily learned why.
This is the hinge of the demonstration. Instead of producing another comfortable example, choose one whose possible answers separate your leading explanations. If you believe doubling is essential, an increasing sequence that does not double can be informative. An example that fails to fit can be useful too, provided you understand what it tests.
What the research actually investigated
A related laboratory task appears in psychologist Peter Wason’s 1960 paper, On the Failure to Eliminate Hypotheses in a Conceptual Task. His participants began with 2, 4, 6 and tried to discover a rule by suggesting further triples and receiving feedback. That starting sequence differs from the one in Veritasium’s demonstration.
Wason reported that some participants repeatedly reached incorrect conclusions and struggled to test their hypotheses effectively. The study examined a particular reasoning task with a defined procedure. It did not establish that a street interview can measure everyone’s intelligence, nor that every agreement-seeking question is irrational. Those larger conclusions would go beyond the evidence.
Confirmation bias is a useful name for a tendency to favour information that supports an existing view. But the name should help us investigate a process, not dismiss a person. Spotting a possible bias in someone else does not prove that their conclusion is false. It also does not show that our own way of arriving at a different conclusion is sound.
Change the question before collecting more answers
Everyday uncertainty is usually less tidy than a number puzzle. There may be no single person who knows the answer, and evidence may be incomplete or noisy. Even so, the demonstration suggests a practical discipline: make your explanation explicit before deciding what information to seek next.
Imagine that a colleague seems slow to respond, and you decide they are uninterested in your project. You notice another delay and treat it as confirmation. Before building a larger story, list other possibilities: an unclear request, competing work or a missed notification. None is automatically correct. The point is to create more than one explanation that evidence could distinguish.
A useful next step might be asking whether the request arrived and when a response is realistic. That is more informative than collecting memories of every delayed message. The example is ordinary because the skill is ordinary. You are trying to replace a repeating interpretation with a question that can produce a meaningful answer.
Keep the stakes proportionate. A small misunderstanding may need a conversation, not an investigation. Seeking disconfirming information should not become endless suspicion or a demand that every person prove their intentions. Thoughtful inquiry includes choosing a fair method and recognising when enough information is available to act.
Separate what happened from the story it supports
Try dividing a notebook page into three short entries. First record an observation in plain language. Then write the explanation that occurred to you. Finally describe another explanation that could fit the same observation. The distinction is useful precisely when the first story feels obvious.
For example, the observation might be that you completed less work on a day when you checked your phone frequently. Your first explanation may be that the phone caused the slower progress. Other factors could include a harder assignment or poor planning. Noticing alternatives does not clear the phone of responsibility. It prevents one day’s experience from carrying more certainty than it can support.
You might then try a limited comparison: define a similar work period, put the phone away and record what happens. This is a personal observation, not a controlled scientific study. Treat it as a way to improve your next choice rather than a final discovery about how every mind works.
Disagreement is useful when it can change something
Inviting another viewpoint can help, but only if the invitation is real. Ask a trusted person which part of your reasoning seems weakest. Listen for a specific objection rather than waiting for your turn to restate the conclusion. You do not have to accept the objection to understand what would answer it.
At the same time, disagreement alone is not evidence. A well-supported claim does not become doubtful merely because someone objects, and an unsupported claim does not become stronger through repetition. Compare the reasons, the quality of the sources and the possibility of checking them. Open-mindedness works best alongside standards for what counts as support.
It can help to name a condition for revision in advance. Perhaps a missing document, a clearer measurement or a direct answer would change your view. If nothing could possibly alter the conclusion, pause and ask whether you are still investigating it. Sometimes you are expressing a value or commitment instead; recognising that difference can make a conversation more honest.
Also notice the temptation to reward yourself for recognising the bias in the video. Understanding an example while an explanation is supplied is easier than choosing a useful test in the middle of an uncertain situation. Recognition is a beginning. The harder practice is remembering the question when you care about the answer.
You can make that practice less dramatic by starting with reversible decisions: the order of a work session, a way to organise a shelf or an assumption about a familiar route. Small corrections help separate learning from embarrassment. Changing your mind need not become a verdict on your character.
Practise being correctable
The point of the number puzzle is not to remain uncertain forever. It is to improve the route from a plausible idea to a justified conclusion. A good question can reveal that you were partly right, completely wrong or asking about the wrong feature. Each outcome gives you something more useful than another automatic yes.
Choose one small belief to examine today. State it clearly, identify an alternative and look for information that could distinguish them. Then decide what the result supports, including its limits. Awareness here is an action: making room for reality to answer differently from the way you expected.
