Epistemic uncertainty is uncertainty in the subject’s knowledge of reality, not uncertainty in reality itself. It exists where the subject cannot unambiguously establish a state, event, relationship, or boundary on the basis of the knowledge available to them.
Reality itself may nevertheless be determinate in relation to the distinction under consideration. A playing card already has a specific value although it lies face down; an event has already occurred although its circumstances have not been established; an object already possesses a particular property although it has not yet been measured. Uncertainty here concerns not these states as such but the subject’s access to their determinacy.
Epistemic uncertainty can arise from the absence of observation, incomplete data, measurement error, an inadequate model, conceptual ambiguity, or unknown causal relationships. It is therefore not reducible to a mere lack of knowledge about an individual fact. The subject may not know which value holds within a given set of options, may not know the complete set of options, or may lack the distinction within which the state could be established at all.
Further investigation can reduce this uncertainty: the subject observes, measures, compares evidence, clarifies concepts, and reconstructs a model. But the epistemic nature of uncertainty does not imply that it must be eliminable in practice. Traces of an event may have been lost, measurement may be inaccessible, and the possibilities for observation may be limited. The determinacy of reality does not in itself guarantee the subject complete knowledge of it.
Epistemic uncertainty is also not identical to error. Under uncertainty, the subject has not yet fixed a single answer or acknowledges that the grounds for doing so are insufficient. In the case of an error, the subject has already taken some state to be determinate, but their determination does not correspond to reality. Uncertainty can lead to error if the subject conceals it behind unjustified confidence, but uncertainty itself leaves the question open.
It cannot be defined solely in terms of probability. A probabilistic assessment already presupposes a model: outcomes have been distinguished, relationships between them have been established, and a method for distributing degrees of confidence has been selected. Epistemic uncertainty can concern the foundations of that model, the completeness of the set of outcomes, and the applicability of probabilistic description itself.
Epistemic and ontological uncertainty can be present simultaneously. A subject may not know which state will arise because the future has not yet become determinate; moreover, they may not know which possibilities are open, which causes are operating, or where the boundary of their available knowledge lies. In the first case, uncertainty concerns reality; in the second, the subject’s knowledge of it.
Therefore, distinguishing these two kinds of uncertainty does not always give the subject a ready-made classification. To call an uncertainty epistemic, it is insufficient to say “we do not know.” There must be grounds for believing that determinacy already exists in reality and that what remains open is our knowledge of it. If the ontological status of the state under consideration itself remains uncertain, that must be preserved in the description as a separate question.
Epistemic uncertainty denotes a boundary of knowledge, not a boundary of reality. By refining their knowledge, the subject establishes determinacy for themselves, but does not thereby create the determinacy of what has already occurred. The subject distinguishes and fixes what they were previously unable to establish.
