970x125
A patient sits silent for 30 seconds, and then for a minute, and the silence could be any number of things—shame, trust, dissociation, fatigue, anger, or something just beginning to surface, or simply that nothing has arrived yet and won’t for a while—and reaching too quickly for one reading closes the others, while reaching too slowly lets something pass that may not come back.
This is the situation negative capability was named for, experiences of deep uncertainty, where trying to solve problems quickly is a failure mode. Keats’s Shakespeare’s-genius-inspired phrase—the capacity to remain in uncertainties, mysteries, doubts, without irritable reaching after fact and reason —is often treated as a stance, a kind of contemplative patience, sitting with the not-knowing and declining the urge to resolve it. That’s true, and it’s also incomplete, because uncertainty turns out not to be one thing, and tolerate ambiguity turns out not to be one discipline. The harder and more useful question is: What kind of not-knowing are you actually in?Doubtless, there are also times when asking, What kind of uncertainty is this? is problematic as well, a kind of “meta-uncertainty” to bear in mind.
Shades of Grey
Because, ironically, being particular about uncertainty is useful, here’s a working taxonomy, partial and revisable.
There’s the uncertainty of randomness, which is the world’s irreducible coin-flipping—the genuine dice, the stochastic, chaotic floor under all the rest.
There’s evidence uncertainty: You don’t yet know enough and more looking would help. It is distinct from model uncertainty, in which the frame through which you’re looking may itself be wrong or incomplete, and measurement uncertainty, in which the instruments are noisy or are capturing something other than what you mean to capture.
There’s causal uncertainty — you see the correlation, but you can’t see the counterfactuals, the worlds in which you didn’t act, didn’t say it, didn’t choose this.
There’s semantic and contextual uncertainty, whereby the words or signals don’t carry stable meaning and the frame keeps shifting under you.
here’s ethical and normative uncertainty, whereby even with the facts in hand aren’t settled, the question of what should matter most still in motion.
There’s human-state uncertainty—what is actually happening clinically, emotionally, relationally, right now in this room, in this body, between us,
And then there’s meta-uncertainty, which is the outlier among them: What kind of confidence is even available to you, here, about this? Given the power of metacognition, thinking about thinking, meta-uncertainty is a strong player.
The list goes on: provenance, trust, adversarial risk, authority, implementation, interpretation. Forms uncertainty takes are different from one another, asking different things from us. Evidence uncertainty asks for more looking. Model uncertainty asks for stepping back and questioning the question. Measurement uncertainty asks for triangulation across instruments. Causal uncertainty asks for humility about intervention. Ethical uncertainty asks for time more than for data. Human-state uncertainty asks for presence rather than analysis. Meta-uncertainty asks something higher-level—for marking what kind of confidence is even possible to have, before reaching for any of it.
In psychotherapy, uncertainty comes up on the regular and is easy to massify. The patient’s silence may not resolve into meaning for many sessions, and the clinician’s task is partly not to prematurely press meaning, not to fill the silence with an interpretation that’s, perhaps, the clinician’s projection; to resist giving advice. For example, genuinely reflective, present silence is different from silence in the face of dissociation and emotional numbing. Both are forms of staying with the not-knowing, but one feels very different from the other, and the ensuring conversations unfold. distinctively.
The Uncertainty of Everyday Life
Uncertainty is all around us, in the air we breathe, and premature collapse of the waveform leads to gross error. In science it is not taking hypothesis for conviction; in medicine, not turning a metric into a diagnosis; in leadership not turning a signal into a decision before knowing whether the signal is real; in personal life (the difficult conversation, the unsettled relationship, the choice that won’t crystallize), a willingness to recognize that the discomfort isn’t always a problem to solve, that the surface problem isn’t the real issue.
AI might be the most interesting case of all, or at least the most pressing: a system that produces confabulated confidence when the right output would be to question, leading to error, or even risk to life and limb (Brenner, Koyfman & Mehendale, 2026). The temptation is the same as the clinician’s: to impute or project certainty when the system, the situation, needs us to hold the center without grokking the maelstrom surround.
Cognition Essential Reads
A small practice that follows is to ask, before acting on a moment of not-knowing, which kind of not-knowing is this? Do I need more data or more time? Is my model wrong or is the situation genuinely ambiguous? Am I uncertain about what’s in front of me or about my own ability to read it? — and to notice that the asking itself slows the irritable reaching, often surfacing, quietly, which response the situation is actually asking for. And also to notice that frantic inquiry can itself be an effort to evade uncertainty. Between the asking and the answering falls the shadow.
Not knowing everything turns out to be fine and dandy. The harder and more interesting thing is knowing what kind of not-knowing you’re in, then deciding whether to wait, to look harder, to act provisionally on limited info, or to leave the mystery an unpopped soap bubble. How we approach our inner world, where we look and how we pay attention, the stances we adopt and the granular moves we make, directly affect emerging reality.
