Knowledge in a Domain: A Domain-Relative Response to the Gettier Problem

Let $D(a)$ be the epistemic domain of an agent $a$: the set of evidence, concepts, rules, background assumptions, and informational access available to that agent.

Let $U_D$ be the universal domain: the complete set of relevant facts, evidence, and epistemic standards.

We can then distinguish:

$$K_{D(a)}(a,p)$$

from:

$$K_{U_D}(a,p)$$

The first means: “$a$ knows $p$ relative to $a$’s domain.”

The second means: “$a$ knows $p$ relative to the universal domain.”

A domain-relative JTB account could be stated as:

$$K_{D(a)}(a,p) \iff B(a,p) \land T(p) \land J_{D(a)}(a,p)$$

In plain language: $a$ has domain-relative knowledge that $p$ when $a$ believes $p$, $p$ is true, and $a$’s belief is justified according to the standards of $a$’s domain.

In a standard Gettier case, an agent forms a belief through reasoning that is justified by the evidence available to them. The resulting belief happens to be true, but its truth depends on some fact outside the agent’s awareness. From the agent’s perspective, the inference seems sound. From a God’s-eye or universal perspective, however, the connection between justification and truth is accidental.[1]

The domain-relative proposal says:

• Within $D(a)$: the agent has a justified true belief, and it may properly count as knowledge-for-that-domain.
• Within $U_D$: the hidden false premise, missing evidence, or accidental truth-making connection becomes visible; the belief may fail to be knowledge.

So the claim is not:

$$K(a,p) \land \neg K(a,p)$$

That would be a contradiction.

Instead, the claim is:

$$K_{D(a)}(a,p) \land \neg K_{U_D}(a,p)$$

There is no contradiction because the knowledge predicate is indexed to different domains.

This approach preserves an important intuition: agents often do know things relative to the evidence and frameworks available to them. A doctor may know a diagnosis relative to the medical evidence available at the time. A scientist may know a result relative to the methods and instruments of their field. An AI system may know a fact relative to its knowledge base and inference rules.

In each case, the agent may be epistemically responsible within its domain, even if later information reveals that the belief was incomplete, defeated, or only accidentally true.

This also fits a broader relativist thought: epistemic justification may be assessed relative to an epistemic system or framework, rather than “simpliciter.” The proposal here is more limited than full relativism, however. It does not necessarily say that all truth is relative; it primarily says that knowledge-ascription and justification are domain-indexed.[2]

The account works best if we distinguish two notions of knowledge:

Notion	Meaning	Gettier status
Domain-relative knowledge	A belief that is true and justified by the standards of $D(a)$	Can count as knowledge
Universal knowledge	A belief that is true, justified, and non-accidentally connected to the fact across $U_D$	May fail in Gettier cases

On this view, Gettier cases show that domain-relative knowledge does not automatically entail universal knowledge. They do not necessarily show that domain-relative knowledge is impossible.

The main objection is simple: if a belief is only accidentally true from the broader standpoint, should we still call it knowledge?

A critic might say that the agent has merely a locally warranted belief, not knowledge. The domain-relative theorist can reply that “knowledge” is not one single, context-free notion. There is practical or operational knowledge—what an agent is entitled to rely on within its working domain—and there is ideal or absolute knowledge—what would survive evaluation from the complete epistemic standpoint.

That reply is philosophically significant, but it comes at a cost. It changes what we mean by “knowledge.” If knowledge is always indexed to a domain, then the traditional debate about whether JTB is sufficient for knowledge may need to be reframed: JTB may be sufficient for domain-relative knowledge, but not for universal knowledge.

If you want to develop this into a fuller argument, several issues need clarification.

• Define $D(a)$ precisely. Is it the agent’s actual evidence, their accessible evidence, their conceptual framework, their social practices, or all of these? The strength of the theory depends heavily on this definition.

• Decide whether truth is domain-relative. The safest version keeps truth universal: $p$ is either true or false independently of $D(a)$. Only justification and knowledge-ascription are domain-relative. If truth is also relative, the view becomes much more controversial.

• Explain the relation between $D(a)$ and $U_D$. Is $U_D$ the union of all individual domains, an idealized complete perspective, or simply the domain containing all relevant facts? These are not obviously the same.

• Handle conflicting domains. If two agents have different domains and reach incompatible conclusions, can both have domain-relative knowledge? If so, explain why this does not collapse into “anything goes” relativism.

• Address defeaters. A belief may be justified within $D(a)$, but there may be evidence outside $D(a)$ that undermines it. The theory should say when such external defeaters remove domain-relative knowledge and when they only remove universal knowledge.

• Clarify the role of luck. Gettier cases are usually diagnosed as involving epistemic luck. The proposal should explain whether luck is absent within $D(a)$, present only from the perspective of $U_D$, or somehow domain-dependent.

• Give a concrete Gettier example. Take one standard case and show exactly what belongs to $D(a)$, what belongs to $U_D$, why the belief is $K_{D(a)}$, and why it is not $K_{U_D}$.

• Consider practical applications. This framework may be especially useful for AI systems, scientific models, legal reasoning, and organizational decision-making, where agents must act on domain-bound information even when absolute certainty is unavailable.

The domain-relative proposal does not deny the force of Gettier cases. It reframes them. Gettier cases may show that JTB is insufficient for universal knowledge, while still allowing JTB to be sufficient for knowledge within a particular epistemic domain.

That makes the view neither a simple defense of traditional JTB nor a complete rejection of it. Instead, it treats JTB as domain-sensitive: true, justified belief can be knowledge where the agent stands—but it may not remain knowledge once the domain expands.

Sources
[1] The Analysis of Knowledge (Stanford Encyclopedia of Philosophy ... https://plato.stanford.edu/archives/win2024/entries/knowledge-analysis/index.html
[2] Relativism - Stanford Encyclopedia of Philosophy https://plato.stanford.edu/entries/relativism/
[3] The Analysis of Knowledge - Stanford Encyclopedia of Philosophy https://plato.stanford.edu/archives/sum2006/entries/knowledge-analysis/
[4] Relativism (Stanford Encyclopedia of Philosophy/Spring 2024 Edition) https://plato.stanford.edu/archives/spr2024/entries/relativism/
[5] Relativism (Stanford Encyclopedia of Philosophy/Spring 2025 ... https://plato.stanford.edu/archives/spr2025/entries/relativism/
[6] Relativism (Stanford Encyclopedia of Philosophy/Fall 2025 Edition) https://plato.stanford.edu/archives/fall2025/entries/relativism/
[7] The Analysis of Knowledge - Stanford Encyclopedia of Philosophy https://plato.stanford.edu/archives/spr2010/entries/knowledge-analysis/
[8] Relativism (Stanford Encyclopedia of Philosophy/Spring 2023 Edition) https://plato.stanford.edu/archives/spr2023/entries/relativism/
[9] Relativism (Stanford Encyclopedia of Philosophy/Winter 2019 ... https://plato.stanford.edu/archives/win2019/entries/relativism/
[10] Relativism (Stanford Encyclopedia of Philosophy/Fall 2016 Edition) https://plato.stanford.edu/archives/fall2016/entries/relativism/
[11] Relativism (Stanford Encyclopedia of Philosophy/Winter 2023 Edition) https://plato.stanford.edu/archives/win2023/entries/relativism/
[12] Relativism (Stanford Encyclopedia of Philosophy/Winter 2020 Edition) https://plato.stanford.edu/archives/win2020/entries/relativism/
[13] Relativism - Stanford Encyclopedia of Philosophy https://plato.stanford.edu/archives/sum2017/entries/relativism/
[14] Relativism (Stanford Encyclopedia of Philosophy/Winter 2015 Edition) https://plato.stanford.edu/archives/win2015/entries/relativism/
[15] Relativism (Stanford Encyclopedia of Philosophy/Winter 2018 Edition) https://plato.stanford.edu/archives/win2018/entries/relativism/