Is the Subject Necessary for Knowledge?

This is the seventeenth article in the series Understanding Philosophy: Figures, Problems, and Ideas. From this point, the series moves beyond figures and particular questions in the history of philosophy and turns towards the structural relations among knowledge, the subject, language, and artificial intelligence.

The question “Is the subject a necessary condition for knowledge?” seems to invite a simple answer. How could there be knowledge without someone who knows?

On closer examination, however, this question combines several issues that belong to different levels. Do facts in the world depend on a subject? Does the truth or falsity of a proposition require someone to judge it? Does material stored in books and databases still count as knowledge when nobody is reading it? Does a mathematical theorem whose original author has long since died temporarily lose its status as knowledge whenever no one happens to be thinking about it? When an artificial intelligence system can preserve, retrieve, and revise a large body of knowledge, must we first prove that it possesses some kind of subjectivity before acknowledging that knowledge exists within it?

To answer these questions, we first need to distinguish facts, knowledge, and acts of knowing.

Suppose two stars collided before any form of life existed. Neither the collision itself nor the physical relations it produced required an observer in order to occur. Whether the proposition “two stars collided at a certain time” is true cannot depend on whether anyone discovers the event later. A subject is not the cause of a fact, and it cannot create truth through an act of will.

Traditional epistemology generally adds that knowledge involves a relation to a subject that facts alone do not. When we say that someone knows that two stars collided, we require more than the truth of the event. The person must also have encountered evidence, formed a judgement, and possess some capacity to distinguish what is correct from what is mistaken.

There is no problem with this claim.

The difficulty is that “someone knows something” and “some knowledge exists” are not identical propositions.

The first describes an act of cognition and therefore requires a knower. The second asks under what conditions knowledge, understood as a structure, can exist and persist. Once these two questions are conflated, it becomes easy to move directly from “an act of knowing requires a subject” to “the existence of knowledge must depend on a subject”.

That inference does not necessarily follow.

A mathematical theorem in a book does not change when the reader leaves the room. Medical information in a database does not temporarily lose its status as knowledge because no one is reading it at three in the morning. A scientific theory can persist across several generations of researchers. After its original proponents have died, later researchers can still retrieve it, test it, revise it, or even overturn it.

What persists is not the cognitive state of any particular individual, but a structure capable of continuing across different media and different subjects.

This point is especially important from the perspective of Sustenesis.

Knowledge is not a special substance located inside a subject, nor is it merely the psychological state described by the phrase “someone believes something”. Knowledge can instead be understood as a relatively stable sustenetic structure in which differences are organised under the constraints of reality. Such a structure can be preserved, retrieved, compared, tested, and revised, and it can continue to participate in new judgements and actions.

Several conditions are involved here.

First, there must be difference. If the world contained no distinguishable states, no information could arise.

Difference alone, however, is not yet knowledge.

When the temperature rises, a temperature sensor produces an electrical signal. It has responded to a difference in its environment. But if the signal vanishes immediately, is not preserved, forms no relation with other states, and has no effect on any later judgement, we would normally say only that the sensor detected a change. We would not say that knowledge in the full sense had been formed.

Difference must enter a structure capable of persisting.

Yet persistence by itself is not enough. A system can preserve an error with great stability. Knowledge must therefore be constrained by its object, its environment, experiment, logic, and other structures of knowledge. An error can be recognised as an error precisely because the world does not conform to our wishes. A theory that survives only by excluding counterexamples may remain internally consistent, but it gradually loses its reliability as knowledge.

Stability in knowledge is therefore not stasis. It is a form of coherence that can continue through constraint and correction.

From this perspective, the basic condition of knowledge is not the prior existence of a subject. It is the existence of a structure capable of sustained formation. Differences must be preservable, relations must be maintained, external constraints must be able to enter the structure, errors must be detectable, and the structure must be revisable when necessary. At the same time, it must retain enough continuity for what is retrieved later to remain recognisable as a development of the same structure of knowledge.

This approach also helps us reconsider Kant.

Kant gave the subject a foundational position in epistemology. Experience is not a completed object that simply enters the mind. Sensory material must be synthesised through space, time, the categories, and the unity of apperception before it can become a unified world of experience. The “I think” must be able to accompany representations. The subject is therefore not merely one ordinary object among others within knowledge, but a condition under which experience can become a unified experience.

This analysis remains crucial for explaining human cognition.

It does, however, require a qualification. What Kant primarily revealed was how human experience is organised into knowledge. He did not necessarily prove that every possible structure of knowledge must presuppose something resembling a human subject.

Suppose a machine system can preserve observations, attribute data gathered at different times to the same object, adjust its model in response to new evidence, and continue using the revised structure in prediction and action. Those processes already display important features of a knowledge structure. We may still ask whether the system is conscious, whether it can understand reasons, and whether it has first-person experience, but those are separate questions.

We cannot automatically deny that a knowledge structure exists merely because we have not established whether the system qualifies as a “subject”.

Modern science has already made this problem much more visible.

A complex scientific conclusion rarely exists in its entirety within the consciousness of any one person. No single researcher possesses all the experimental data, principles of instrumentation, statistical procedures, software code, theoretical background, and historical literature involved. Knowledge is distributed across a network composed of researchers, experimental equipment, databases, papers, software, institutions, and procedures of verification.

Researchers retire and die. Computers are replaced, and databases are migrated. Yet some structures within this network continue.

What preserves continuity is not one fixed subject, but the structure itself.

We could, of course, redefine the scientific community as a “distributed subject”. That move requires caution. If every structure capable of preserving, retrieving, and correcting knowledge is redefined as a subject, the statement that knowledge must have a subject easily becomes circular by definition.

We first discover that knowledge requires a structure capable of preservation, retrieval, and error correction. We then call that structure a subject, and finally conclude that knowledge must have a subject.

This does not explain why the subject is necessary. It merely enlarges the meaning of the word “subject”.

Sustenesis offers a simpler approach. A structure can remain a structure. We do not need to rename every knowledge-bearing structure as a subject merely to preserve the traditional place of the subject in epistemology.

Language itself contributes to the illusion.

The sentence “Zhang San knows that the Earth revolves around the Sun” naturally requires a grammatical subject because it describes the act of knowing.

By contrast, the sentence “the knowledge that the Earth revolves around the Sun has been preserved” is not about a particular act of cognition at a particular moment. It concerns whether a structure of knowledge can continue to exist.

The fact that an act of knowing requires a knower does not entail that a knowledge structure must always have a knower continuously present.

A knowledge structure can remain temporarily unused. It can pass from one subject to another, from human memory to paper, and from paper to a database. Its medium changes, as do the subjects who participate in it. Yet it can preserve continuity so long as its key differences, relations, and constraints are retained well enough.

Knowledge is therefore better understood as a sustenetic structure capable of operating across different media than as a psychological state that must remain permanently attached to a knowing subject.

None of this means that subjects are unimportant.

On the contrary, a human subject is an extraordinarily complex structure for the sustained formation of knowledge. We can perceive differences, form concepts, establish memories, compare experiences, discover errors, reorganise existing knowledge, and use those structures in action. Consciousness, self-experience, language, and social relations add further levels that are distinctive of human knowledge.

The importance of the subject, however, is not the same question as whether a subject is a necessary condition for the existence of all knowledge.

A subject can participate in producing knowledge, understand it, doubt it, give it meaning, and accept responsibility for its use. In human life, these functions may be far more important than the mere preservation of information.

From the standpoint of the structural conditions of knowledge itself, however, the subject need not occupy an a priori position.

Artificial intelligence makes this distinction even more explicit.

The fact that a language model produces a correct sentence is not, by itself, enough to show that it possesses stable knowledge. Its output may be the product of a local statistical association, with no reliable source and no capacity to remain consistent in a new environment.

But if an artificial system can preserve information, organise it into structural relations with other information, retrieve it in different contexts, detect error through external outcomes, and modify its own state under new constraints, then the nature of the question changes.

At that point, the central research question is no longer only whether the system is a subject. We must ask what kind of sustenetic knowledge structure has formed within the system and through its relation to the environment.

The problems of consciousness and subjectivity remain. They do not, however, need to serve as the first gate through which every judgement about the existence of knowledge must pass.

We can now describe the relation among facts, knowledge, and subjects more clearly.

Facts are the differences, states, and relations that exist in the world.

Knowledge consists of some of those differences and relations organised under constraints into stable structures that can be preserved, retrieved, tested, revised, and remain effective in further operation.

Subjects are a class of complex systems capable of participating in the formation, maintenance, retrieval, understanding, and transformation of those structures of knowledge.

On this account, the subject is no longer the foundation of knowledge. It becomes a particular kind of sustenetic structure that itself requires explanation.

My answer to the question “Is the subject a necessary condition for knowledge?” is therefore no.

Acts of knowing require some kind of knowing subject, but knowledge itself does not necessarily require a knower who remains continuously present. What knowledge cannot do without are the conditions of sustained formation. Differences must be able to form a structure, the structure must be subject to the constraints of reality, and it must maintain sufficient coherence through time while remaining open to retrieval, testing, and revision.

The subject is one extremely important way in which this process is realised. It is not the only possible form, nor is it a universal a priori condition for the existence of knowledge.

The distinction may appear small, but it changes the starting point of epistemology.

Traditional epistemology often begins by asking what kind of subject can know the world.

Sustenesis prefers to begin with a different question. What kind of structure can become knowledge and continue to exist under the constraints of the world?

Once this question comes first, the subject does not disappear. It simply moves from the starting point of epistemology to the position of something that must itself be explained.

We no longer begin by assuming a knowing subject and then looking for knowledge from within it.

We begin by investigating how knowledge is sustained and formed. We can then ask how, within certain complex structures of knowledge, a subject itself comes into being.