The idea for this text came to me while listening to a conversation about nutrition. The expert repeatedly answered “it depends” or “I don’t know.” I did not take this as a lack of competence. On the contrary — it sounded like the responsible conduct of someone who understands the limits of her knowledge.

Complete knowledge of reality would mean knowing every aspect of it. No single person can achieve that. Someone can spend thirty years exploring one part of the world and know more about it than almost everyone else, but it will still be a part. The problem begins when the depth of local knowledge is confused with the right to speak with the same authority about everything else.

Specialization deepens a part; it does not extend it to everything

Specialization is one of the mechanisms through which we develop science, technology and society. I do not need to know how an engine is built to use a car. I do not need to study medicine to receive treatment. The division of knowledge allows us to benefit from the work of people who have explored a particular problem much more deeply.

Yet this same mechanism has a boundary. An academic title, popularity, sporting success or professional achievement does not automatically transfer competence to another field. An expert can, of course, speak outside their specialty. They can learn more, gather sound arguments and be right. Their previous status, however, is not evidence that they are right.

A study by Stav Atir, Emily Rosenzweig and David Dunning shows an interesting problem even within a single domain. People who rated their own knowledge more highly were more likely to claim familiarity with concepts that did not actually exist. The authors did not demonstrate that every expert is overconfident. They showed something more modest but important: a sense of competence can make it harder to notice one’s own ignorance (Atir, Rosenzweig and Dunning, 2015).

There is also the halo effect. A generally positive or negative impression of a person can color judgments of their individual characteristics, even when the observer does not recognize this influence (Nisbett and Wilson, 1977). Success and likability attract attention, but they still do not replace justification. An argument does not become true because it was voiced by someone high in the social hierarchy.

Credibility does not reside in a name or a position

I do not want to reverse this mistake and claim that a formal expert is always wrong while someone without a title always sees more. That would merely be another form of wishful thinking. The boundary does not run between expert and layperson. It runs between a claim that can be tested against reality and a claim sustained solely by status, opinion or emotion.

I apply the same principle to myself. I work in IT, but that does not mean everyone should use a computer or understand the binary system. My specialization can help me explain a particular mechanism. It does not give me the right to organize other people’s entire lives simply because I know technology better than they do.

Likewise, an expert’s personal inconsistency does not automatically invalidate their knowledge. A doctor can communicate true information about metabolic diseases even if they themselves are obese. Their conduct may affect credibility, trust or judgments of their ability to apply knowledge, but it does not determine the truth of a particular claim. Criticizing a person cannot replace examining an argument.

Emotion can prompt a question, but it does not determine reality

I am troubled by the moment when someone confuses opinion with fact or tries to evoke an emotion instead of providing justification. Emotion conveys information about a person. It may indicate that a problem matters to them, that they perceive an injustice or that they are afraid of something. It is not, however, independent evidence of how the world works.

Ziva Kunda described the mechanism of motivated reasoning: we can select, construct and evaluate arguments so as to reach a desired conclusion rather than the conclusion most consistent with the available information (Kunda, 1990). This does not mean that everyone who is emotionally invested is wrong. It means that strength of conviction cannot be the measure of truth.

That is why I ask not only “who said it?” and “how confidently did they say it?” but also “what exactly was said?”, “on what basis?” and “what could show that it is untrue?”. If a position allows no means of correction, it begins to resemble the protection of an identity rather than an attempt to understand reality.

“I don’t know” and “it depends” need to open something up

Simply saying “I don’t know” does not solve a problem. It can honestly mark a boundary or conveniently end a conversation. The difference becomes apparent in the next step.

An honest “I don’t know” should help name what is missing: what information we need, where we can look for it and what we will do if a decision must be made before certainty is available. An honest “it depends” should identify the dependencies: which conditions change the answer and why.

Research on communicating uncertainty offers no simple recipe claiming that an explicit caveat always increases trust. Responses depend on the presentation, the audience and the situation (van der Bles et al., 2019). Epistemic humility is therefore not an image-management technique. It is an acceptance that one’s own belief may need correction. Research on intellectual humility describes it, among other things, through recognition of one’s own fallibility and willingness to revise a position in light of appropriate evidence (Krumrei-Mancuso and Rouse, 2016).

From the general to the particular and back

To examine my own reasoning, I try to move in both directions. A general model should explain specific cases. Specific cases should, in turn, be able to correct the model. If a detail does not fit, that does not always mean the entire model is false. It may, however, reveal a missing condition, an overgeneralization or a poorly framed question.

The first practical step is sometimes simply to read more. People are not as original as they sometimes like to think. A similar problem may very well have occurred before; someone has described it, made mistakes and left results behind. A search engine can help find these traces, but it is not an arbiter of truth. We still need to examine the source, context, date and method, and whether the material actually answers our question.

Not everyone has an obligation to explore every field. Not everyone needs to know how an internal combustion engine, the binary system or a particular research method works. An honest boundary, however, sounds different from a pretense of omniscience: I may not know something, but I should not present a lack of intellectual effort as an ever-better understanding of the world.

Today, a huge share of accumulated knowledge is accessible from a device that fits in a pocket. Access to knowledge is not the same as possessing it. Intellectual maturity begins not when a person has an answer to everything, but when they can recognize what they truly know, what they are merely guessing and where they should say: “I don’t know that yet.”

Bibliography

✦Building systems for greater autonomy.