Modality is usually introduced as a fairly technical topic: the possible, the necessary, the impossible, the actual. In logic, there are operators of possibility and necessity; in language, there are words such as “can,” “must,” “probably,” “necessarily,” and “impossible.” This presentation is convenient, but it almost immediately conceals the main question. Before asking whether a certain state of affairs is possible or necessary, we should ask: where did the space within which we determine what is possible and impossible come from in the first place?
When someone says “this is possible,” the statement appears to refer directly to reality. But possibility itself is never given unconditionally. Possibility exists relative to certain constraints, distinctions, and assumptions. One thing is physically possible, another logically possible, a third legally possible, and a fourth technologically possible. What is impossible within one theory may turn out to be admissible after its initial assumptions change. Thus, saying “this is impossible” without specifying a framework is a much stronger claim than it usually appears. It imperceptibly transfers the boundary of some system of description onto reality itself.
It is precisely here that modality ceases to be merely a branch of formal logic and becomes a problem of knowledge.
Possibility Is Not Located Within the Thing Itself
Suppose we say that some event is possible. This can mean at least several entirely different things. It may not contradict logical definitions. It may be permitted by a physical theory. It may be technically feasible with existing technology. It may be compatible with the data available to us. Finally, we may simply know of nothing that would allow us to rule it out.
In language, all these cases are easily collapsed into one word: “possible.”
But they are not identical. If I do not know whether someone is at home right now, the phrase “perhaps they are at home” speaks primarily about the state of my knowledge. Meanwhile, the person is either at home or not. Their location does not become ontologically indeterminate because I do not know it. I am dealing not with two states of reality, but with two states that my present information does not yet allow me to distinguish.
This simple distinction is remarkably easy to lose. The subject’s lack of knowledge begins to be perceived as a property of the object. From the fact that alternatives A and B are available to me, it is imperceptibly concluded that reality itself consists of alternatives A and B.
But this is an additional assertion, and it requires justification.
Possibility, therefore, cannot be treated as an independent substance hanging somewhere between existence and nonexistence. For something to become “a possibility,” it must first be distinguished as an alternative. The subject must distinguish a particular state of affairs, draw a boundary between it and its alternatives, and place it within some system of constraints. Only then does it become meaningful to ask whether this system admits that alternative.
Thus, modality is always secondary to distinction. Before one can say “A is possible,” one must already have A as something distinguished.
Possibility and Probability
Possibility is particularly often confused with probability. In everyday speech, “possible” almost imperceptibly turns into “has some probability.” It seems as though the difference is merely quantitative: the impossible has probability zero, the possible is somewhere between zero and one, and the necessary is at one.
But this picture presupposes far more than it reveals.
For probability, it is not enough to have possible events. One must already possess a space of outcomes, determine which events count as distinguishable, specify a measure, and decide how that measure is distributed among alternatives. A probability model is not a neutral photograph of the unknown, but a fairly complex construction.
If I say “the existence of life on this planet is possible,” no number whatsoever follows from that. I have not obtained 10%, 30%, or 80%. To obtain a probability, I need a model. And the model must include assumptions about which factors are relevant, which cases count as analogous, what sample I use, and how I connect observed data to an unknown case.
Consequently, the transition from possibility to probability is itself a cognitive operation.
This is especially clear where genuine unknownness appears. When an analyst says there are three scenarios and assigns them probabilities of 20%, 30%, and 50%, it creates the impression that an uncertain future has simply been measured. In reality, the future was first divided into three categories. It was then assumed that this division was sufficient. After that, the categories were assigned weights. And only at the last stage did the result begin to look like a property of the future itself.
But reality is under no obligation to know that an analyst has allowed it three scenarios.
It may produce a fourth.
And even the phrase “fourth scenario” is somewhat misleading here, because a new outcome may not be just another element of the same set. It may change the very mode of classification relative to which the first three scenarios were formulated at all.
Closed and Open Unknownness
This allows us to draw an essential distinction between two conditions of not knowing.
The first can be called closed unknownness. The space of alternatives is known; what is unknown is only which alternative has been realized. For example, a coin has been tossed and covered with a hand. We cannot see the result, but we know the distinction “heads / tails.” The lack of knowledge lies within an already specified form.
The second case is fundamentally different. The subject may not know not only which alternative has been realized, but what the possible alternatives are in the first place. Here, the unknown has not yet been enclosed within a ready-made system of coordinates.
In this condition, saying “A or B” is already quite a consequential action: it limits the unknown to what we have managed to distinguish beforehand.
This is precisely where probability becomes particularly deceptive. Probability gives the impression of respecting uncertainty because it does not assert a single outcome with complete confidence. But it may destroy unknownness long before an outcome is chosen. It destroys unknownness at the moment when it declares the space of alternatives to be known.
After that, all that remains is to calculate the weights.
But if the space itself is unknown, no sum of probabilities restores the openness that has been lost. One can distribute 100% with extraordinary care among the wrong categories.
Thus, unknownness is not a low degree of precision in a probability model. In some cases, it is a qualitatively different epistemic condition: the subject does not yet even know which distinction should be made here.
Necessity and the Framework
The same problem arises with necessity.
Formal logic allows us to work quite rigorously with the expression “it is necessary that P.” If P is true in all admissible states of some model, it is considered necessary. But the word “necessary” has a rhetorical force that far exceeds this formal construction. It easily creates the impression of an absolute property of reality.
Yet necessity almost always exists relative to a framework.
Once the rules of chess have been specified, a bishop necessarily moves diagonally. This is a strict necessity, but it is not a law of nature. It is a consequence of the adopted system of rules. If we change the game, the necessity disappears.
Likewise, a mathematical proof can establish necessity within a specified axiomatic and operational structure. It does not follow that the structure itself is the only possible form of reality.
Here it is useful to distinguish necessity within a form from the necessity of the form itself.
The first can be absolute under fixed assumptions. The second does not follow from the first.
It is precisely this distinction that gets lost whenever the restrictions of a model begin to be described as restrictions of being. The theory says: “Under these assumptions, X is impossible.” Then the first half of the sentence disappears, leaving: “X is impossible.”
The form ceases to be perceived as a form.
Actuality Is Not Maximum Probability
Another common error consists in placing actuality on the same scale as possibility and probability. The result is something like: impossible, improbable, probable, almost certain, actual.
But actuality is not an extreme degree of probability at all.
Before an event, its probability may be arbitrarily high. After the event, the question changes. If a coin has already landed heads, heads does not become “an outcome with 100% probability” in the same sense in which probability was discussed before the toss. Now we are dealing with the fact of a particular observed state.
And conversely, an event with an extremely low prior probability does not become any less actual after it occurs.
This boundary matters because it shows that modal categories cannot be reduced without remainder to a single numerical axis. Between “may happen” and “has happened” lies not a quantitative transition, but a change in the very relationship to the event.
The same applies to necessity. If X happened, it does not follow that X had to happen. Actuality does not prove necessity.
Nevertheless, human thinking constantly makes this substitution retrospectively. After an event occurs, the space of earlier alternatives begins to disappear from perception. The outcome that occurred becomes the center of a reconstruction of the past, and everything preceding it is arranged as a road leading specifically to that outcome.
This is how the phrase “it was inevitable” is born.
But the fact that X happened tells us only that X happened. To assert inevitability, one must show the impossibility of the relevant alternatives. That is a substantially stronger claim.
Modality and Reconstruction
Here modality comes into direct contact with the problem of distinguishing observation from reconstruction.
Suppose someone enters a room and sees a broken glass. What is immediately given to them is the broken glass. They then say: “The glass fell off the table.” This is already a reconstruction. It may be an extremely plausible reconstruction, but it is not identical to what was observed.
The next step: “Someone knocked the glass over.” Another transition.
Then: “The person probably did it accidentally.”
If all these transitions are collapsed, we get the ordinary sentence: “Someone accidentally broke the glass.”
Language has erased several modal layers and presented the final construction as a simple description of an event.
That is precisely why modality is directly connected to epistemic discipline. It makes it possible to preserve the status of every element of a statement: what is observed, what is inferred, what is supposed, what is considered probable, what is excluded, and on what grounds.
The problem does not begin when a person constructs a reconstruction. Thinking is practically impossible without reconstructions. The problem begins when the distinction between reconstruction and observation disappears.
Likewise, the problem with probability does not lie in probability itself. The problem begins when the distinction between a probabilistic model and the structure of reality itself disappears.
Possible Worlds and the Cost of Formalization
Classical modal logic addresses many of these problems through possible-world semantics. In rough form, the principle is familiar: P is possible if there exists an accessible world in which P is true; P is necessary if it is true in all accessible worlds.
This is an extraordinarily powerful formal tool. But its very power can create another substitution.
To speak of a set of possible worlds, one must already determine which worlds count as accessible. In other words, the formal system must establish criteria of admissibility in advance.
For logical possibility, these will be one set of restrictions. For physical possibility, another. For historical possibility, a third.
Consequently, the accessibility relation is the hidden boundary of the form.
Modal logic does not directly discover every possible state of reality. It constructs a system within which certain states are considered admissible and others are not. Then it rigorously investigates relationships within that system.
This is not a defect of logic. It is the normal operation of a formal instrument.
The error begins only when the instrument forgets its own boundary.
A possible world in semantics need not be a “real parallel world.” It can simply be an element of a formal space created to analyze a modal statement. The transition from a semantic construction to ontology is a separate philosophical step, not part of the logical notation itself.
The Subject and the Space of Alternatives
Now we can return to a question that usually remains outside introductory texts on modality.
Who determines the space of alternatives?
When a person is given a problem with options A, B, and C, they can choose among them. That does not mean they are working directly with reality. They are working within a space of distinctions that has already been formed.
A more fundamental operation arises when the subject asks: why precisely A, B, and C? Why is the boundary drawn here? What was left out of the classification? Are the proposed alternatives mutually exclusive? Are they exhaustive? Might there be a distinction relative to which the entire original classification would disintegrate?
That is how a metamodal level emerges.
The subject is now working not only with possibilities within a form, but also with the possibility of the form itself being different.
This is a fundamentally important transition. As long as the subject selects an option within a given set, the system of distinctions remains invisible. As soon as the subject begins to distinguish the system itself, a new space appears.
And here it becomes clear why a new distinction cannot simply be presented as another preexisting option. Before that distinction emerges, it may not exist for the given space of description at all. It was not on the list with a zero or low probability. The list was structured differently.
Thus, genuine novelty sometimes changes not the answer but the question.
Modality and Freedom
The problem of freedom also has a modal structure.
The phrase “I could have acted differently” is a claim about the possibility of an alternative action. It is fundamentally different from the assertion “I acted differently.” The first concerns the space of available continuations; the second concerns actuality.
A deterministic construction often imperceptibly conflates these two levels. If X occurred, then with complete knowledge of the causes, only X could have occurred. But an additional assumption has already been introduced here: that the state of the world and its laws completely determine one unique outcome.
The mere fact that X occurred does not prove this.
To prove that X was necessary, one must show the impossibility of any other outcome under relevantly identical conditions. That is precisely a modal assertion.
Therefore, the dispute over freedom cannot be resolved merely by pointing to causation or to the fact of the event that occurred. It requires an answer to a deeper question: what is the space of admissible continuations, and how do we know its boundaries?
And here the subject appears again, because even the description of those boundaries is constructed through a particular system of distinctions.
Modality as Preserving the Status of an Assertion
Ultimately, the value of modality lies not in a set of terms, but in its preventing all assertions from lying on the same plane.
“I see X.”
“Y follows from X.”
“Y is possible.”
“Y is probable.”
“My model does not admit not-Y.”
“Y is necessary.”
“I do not know what lies beyond the boundaries of this model.”
These are not different degrees of the same assertion. They are different relationships between subject, form, and reality.
When they collapse into one another, thought becomes outwardly confident and inwardly undifferentiated. The model begins to speak in the voice of reality. Unknownness turns into a probability distribution. A system’s restriction becomes impossibility in general. A realized outcome retrospectively becomes necessary. Reconstruction becomes observation.
Modal thinking, by contrast, preserves the transitions.
It compels us to ask not only “is this true?” but also “what kind of assertion am I making right now?” Not only “is this possible?” but also “which system of restrictions makes it possible or impossible?” Not only “what is the probability?” but also “why has the space of outcomes been divided up in this particular way?” Not only “what happened?” but also “have I turned knowledge of the outcome into a claim about its necessity?”
In this sense, modality is not a decorative addition to a statement, but a way of preserving the boundaries of knowledge.
It shows the moment when form ceases to be transparent.
We see that “the possible” does not exist separately from the space within which something is regarded as possible; that “the necessary” requires a framework relative to which it is necessary; that probability requires an already formed set of outcomes; that unknownness may lie deeper than that set; that actuality does not prove necessity; and that the boundaries of a model have no right to imperceptibly turn into the boundaries of reality.
And then the central question of modality can be formulated much more precisely.
Not:
“What is possible?”
but:
“How did the subject establish the space within which something acquired the status of possible, impossible, or necessary in the first place, and where does the necessity of a form end and an assertion about reality itself begin?”
It is at precisely this point that modality ceases to be a table of operators and becomes an inquiry into the boundary between distinction and what has yet to be distinguished.
