At night, a bang rings out. A few streets away, blue lights are flashing. By morning, photographs, two sentences from a witness, and a short video with no beginning appear on social media. Within an hour there is already a perpetrator, a motive, a system failure, and someone who “knew it all along.”
In the afternoon, another piece of information arrives.
The story changes.
In the evening, another one.
It changes again.
What is remarkable is not only that we were wrong. What is remarkable is how quickly, beforehand, we stopped feeling that we did not know something.
Psychology uses the term need for cognitive closure for part of this problem. Kruglanski and Webster described it as a desire for a definite, unambiguous answer instead of confusion and ambiguity. It is not simply a “flaw in certain people”: we differ in it individually, and situations can intensify it as well.1
The more costly uncertainty is for us— in time, emotion, or practical terms—the more tempting it becomes to finally close something down.
“I Don’t Know” Is Not an Empty Answer
The words I don’t know often sound like failure in ordinary conversation.
You ask an expert and want an explanation. You ask a doctor and want a diagnosis. You ask an investigator and want a perpetrator. You ask an analyst and want a prediction.
“I don’t know yet” sounds weak because it offers no closure. It does not let us put the problem away in a mental drawer. It does not provide a clear model of what will happen next.
And yet it can be more accurate, informationally, than a long answer.
The sentence “I don’t know” may contain important information: the current evidence does not yet distinguish between several possibilities.
That is different from ignorance.
Ignorance may mean that we have not looked into the problem. Epistemic restraint means that we have looked into it enough to know where the evidence ends.
The difference: “I don’t know because I investigated nothing” and “I don’t know because the data still allow three equally plausible hypotheses” are two entirely different levels of knowledge.
Seize. Freeze. Grasp. Lock In.
The theory of the need for cognitive closure describes two important tendencies: the urgency to obtain an answer quickly and the desire to hold on to an answer once obtained. In the English literature, the image of seizing and freezing has become established—a person quickly “seizes” on something and then “freezes” there.1
It is an elegant description of a mechanism everyone knows.
First, we see one piece of information.
“Aha. So that’s what happened.”
Further data no longer enter an empty space. They arrive in a system that already has a story.
Information that confirms the story fits smoothly.
Information that contradicts it must first overcome something extra: our investment in the first explanation.
This does not mean that a quick decision is always wrong. In many situations speed is necessary, and complete information never arrives. The need for closure is not pathological in itself; it enables action.
The problem arises when the psychological need for an answer starts to be mistaken for evidence that we have the answer.
Uncertainty Is Not a Neutral Feeling
Why does ambiguity bother us so much?
Because uncertainty is not merely an abstract state of information. It has psychological and physiological consequences.
In an experiment published in Nature Communications, participants waited to see whether they would receive a mild electric shock. The researchers found that subjective estimates of uncertainty predicted the course of both subjective and physiological stress responses; under some conditions, uncertainty itself mattered more for stress than the simple probability of a negative outcome.3
In other words, a person may sometimes tolerate a bad certainty better than an open possibility.
That is important context for our desire for an explanation.
A story is not just information.
A story is also the end of waiting.
When Chance Gets a Plot
Imagine three events.
On Monday, your car breaks down.
On Wednesday, you lose your wallet.
On Friday, a strange email arrives.
Separately, they are three things.
Then someone convinces you that they are connected.
Suddenly every detail acquires a role.
That is the power of a causal story. Human cognition is exceptionally good at looking for causes; causal explanations are essential to learning, prediction, and understanding the world.4
But the same ability can create the illusion that a coherent story is automatically a true story.
It is not.
Some real events have terribly complicated causes. Some have no clear culprit. Some arise from the convergence of ten small factors. Some contain randomness that is hard to fit into a story.
And some we simply cannot explain yet.
We Think We Understand—Until We Have to Explain It
There is a second problem: it is not only that we want an answer. We often feel that the answer we have is much more complete than it really is.
Rozenblit and Keil called this the illusion of explanatory depth. In a series of experiments, people rated how well they understood the workings of ordinary mechanisms and phenomena. When they then had to explain them step by step, their assessment of their own understanding fell.2
A classic example is simple: do you know how a flush toilet works?
Of course.
Now try to explain it without a diagram, from pressing the button to the tank filling again, so that someone could build the mechanism from your description.
Suddenly gaps appear that had not been subjectively visible before.
This illusion is not evidence that people are stupid. On the contrary, it may be a by-product of a highly efficient cognitive architecture.
We do not need to carry the entire technical world in our heads. We need to know enough to use it; the rest is distributed among the people, documentation, tools, and institutions around us.
The problem begins when we confuse access to knowledge with knowledge in our heads.
The Internet Eliminated the Cost of an Answer. Not the Cost of Truth.
Thirty years ago, getting an answer was harder.
Today we get one in a second.
A search engine offers ten results. A social network offers a thousand opinions. Generative AI can formulate a coherent explanation almost instantly.
Technologically, this is fantastic.
Epistemically, it creates a new problem: the fluency of an answer keeps getting cheaper, but the truth of an answer has not become cheaper at the same rate.
The fact that text sounds convincing does not change the basic questions:
What is the source?
How strong is the evidence?
Are there alternative explanations?
What would disprove this conclusion?
Which parts are facts and which are inferences?
“I don’t know” becomes even more valuable in such an environment because it competes with a practically endless supply of ready-made explanations.
The Greater the Pressure, the More Expensive Uncertainty Becomes
It is no coincidence that the fastest stories emerge after disasters, attacks, outages, political crises, or people going missing.
At such moments, ambiguity is not an academic problem. It is emotional.
According to the original theory, the need for cognitive closure can also be triggered situationally—for example, by pressure to decide or by the costs associated with continued information processing.1 Experimental research also shows that time pressure changes how people explore possibilities and respond to information.5
This creates a dangerous combination:
The worst moment for certainty is therefore often the very moment when we want it most.
Why a Simple Explanation Sometimes Beats a Probable One
This topic often appears in research on conspiracy beliefs. A review by Karen Douglas and colleagues lists epistemic needs among the psychological motives for conspiracy theories—the effort to understand significant and uncertain events, gain certainty, and create a causal explanation.6
But beware the shortcut.
The need for certainty does not mean that a person will automatically believe a conspiracy. And an “official” explanation is not true merely because it is official.
The underlying mechanism is more general: when an event is large, random, and unclear, an explanation with a clear actor, motive, and cause may feel psychologically more satisfying than the sentence:
“It is a combination of several factors, and we still do not know some of them.”
Reality has no obligation to be narratively elegant.
The Forensic Sentence That Can Save a Case
In criminal investigation, “I don’t know” can sometimes be a sign of discipline.
An analyst receives a digital trace and does not know who created it.
An expert sees damage, but two technical causes are possible.
A witness remembers a face, but does not know whether they actually saw the person that evening or only later in a photograph.
An investigator has a strong hypothesis, but one piece of evidence is still missing.
An unprofessional response is not only making a mistake.
It can also be rewriting a degree of uncertainty as certainty because certainty is easier to put in a conclusion.
Good analysis therefore separates:
What we know.
What we reasonably infer from it.
What is merely possible.
What we do not know yet.
How to Learn to Say “I Don’t Know” Better
The point is not to answer “I don’t know” to everything.
That is not humility. It is resignation.
A good “I don’t know” has structure.
What exactly do I not know?
Which possibilities remain open?
What evidence could distinguish between them?
How certain am I of what I am nevertheless claiming?
One of the most effective techniques comes directly from research on the illusion of explanation: try to explain the conclusion step by step for real. Do not repeat the name of the mechanism. Do not use a technical term. Describe the actual chain.
If, between two steps, you find:
“…and then somehow it happens…”
you have found the place where your certainty probably got ahead of your understanding.
Uncertainty Is Not Weakness. It Is Information.
The public sphere rewards the opposite behavior.
A person who says “it is complicated and we do not have enough data yet” sounds less impressive than someone who explains the whole world in thirty seconds.
Social status is often tied to having an answer.
An expert is expected to know.
A commander is expected to decide.
A politician is expected to have a plan.
An analyst is expected to find something.
And yet there are situations in which the highest form of competence is to correctly name the boundary of one’s information.
Science works precisely because it allows answers to remain provisional. A hypothesis can be the best one available and still remain open to revision. A result can be probable without being certain. A model can work and still not be the final explanation.
“I don’t know” is therefore not the opposite of knowledge.
Sometimes it is its boundary.
The Empty Space We Should Leave Empty
When something happens, the brain begins assembling a story almost immediately.
Who.
Why.
How.
What happens next.
That is one of its greatest abilities. Without fast models of the world, we could not decide, learn, or survive.
But every powerful ability has a point at which it becomes a weakness.
With explanations, that is the moment when a story stops being a working model and becomes a substitute for missing evidence.
Perhaps the point, then, is not to learn to doubt everything.
It is something more precise:
to recognize the moment when our need for an answer is growing faster than the amount of information that can support it.
And at such a moment, to be able to say a sentence that sounds neither clever, authoritative, nor definitive.
But it may be the most accurate of all:
Sources and further reading
- Kruglanski, A. W. & Webster, D. M. (1996). Motivated Closing of the Mind: “Seizing” and “Freezing”. Psychological Review, 103(2), 263–283. The original theoretical framework for the need for cognitive closure as a desire for a definite answer and an avoidance of confusion and ambiguity. PubMed.
- Rozenblit, L. & Keil, F. (2002). The misunderstood limits of folk science: an illusion of explanatory depth. Cognitive Science, 26(5), 521–562. A series of studies showing that people may overestimate how deeply they understand causally complex mechanisms and revise their assessment after attempting a detailed explanation. PMC.
- de Berker, A. O. et al. (2016). Computations of uncertainty mediate acute stress responses in humans. Nature Communications 7, 10996. The experiment showed a relationship between subjective uncertainty and the dynamics of subjective and physiological stress responses. Nature Communications.
- Keil, F. C. (2024). Causal Explanations and the Growth of Understanding. Annual Review of Developmental Psychology. A review of the importance of causal explanation for learning and cognitive development. Annual Reviews. See also Sloman & Lagnado, Causality in Thought, Annual Review of Psychology.
- Wu, C. M. et al. (2022). Time pressure changes how people explore and respond to uncertainty. Scientific Reports 12. A study of how time pressure changes search and decision-making in an uncertain environment. Scientific Reports.
- Douglas, K. M., Sutton, R. M. & Cichocka, A. (2017). The Psychology of Conspiracy Theories. Current Directions in Psychological Science, 26(6), 538–542. A review of epistemic, existential, and social motives for conspiracy beliefs; the authors emphasize that the relationships are neither simple nor deterministic. PMC.
- Growiec, K. et al. (2026). Need for cognitive closure predicts preference for similar over dissimilar social interactions. Recent research reminding us that the need for closure is a measurable individual characteristic associated with tolerance of uncertainty and potentially influencing social choices as well. PMC.
Comments
Have a view or an additional source? Add a comment.
Discussion is not active yet.