Relational Objectivity — The Four-Category Ontology of Facts

Not all facts are intrinsic properties of isolated objects. Some facts live in relationships — between objects, between agents and reality, or between agents and their own states. Relational does not mean subjective.

Links: The Four Trunks, Epistemology, Morality, The Weighting Problem, Stance-Independence Retreat, Measurement, Causality, and Free Will, Logic and Mathematics, Wilson vs Objectivism, Newcomb’s Paradox, Performative Grounding

Trunk: This page is a core statement of Trunk 2 — Verification Epistemology. The working name for the trunk’s position is constitutive convergence.

The Problem

In the Wilson vs Objectivism debate, the central impasse was whether “stance-dependent” facts are really facts. Wilson assumed a two-category world: facts are either mind-independent (objective, real) or mind-dependent (subjective, opinion). If morality requires agents, it’s mind-dependent, therefore not factual.

ProphetDivinityy’s “taller than you” analogy was the right instinct: relational propositions can be factual. But he couldn’t name the category or articulate why relational ≠ arbitrary.

The problem: the two-category model is wrong. There are four.

The Four Categories

Category 1: Intrinsic Facts

Properties an object has regardless of anything else. Mass, charge, spatial extension. These are monadic — single-place predicates. “The sun has mass M” doesn’t reference any other object.

Not controversial. Everyone agrees these are objective.

Category 2: Relational Mind-Independent Facts

Properties that exist in the relationship between two or more things, with no agents required. This is the critical category that the two-category model misses.

The tree example: Tree A is taller than Tree B. This is:

• Relational — involves a comparison between two objects
• Mind-independent — no minds needed; the height difference exists whether anyone measures it
• Not arbitrary — there is a fact of the matter
• Causally consequential — the taller tree casts shade on the shorter one, winning the competition for sunlight. Reality “measures” the height differential through physical interaction, and the measurement has consequences.

More examples: Earth is closer to the Sun than Mars. Water dissolves salt. Acid corrodes iron. The speed of light exceeds the speed of sound. All relational, all mind-independent, all objective.

The measurement-causality connection: From the vault’s measurement-causality framework, measurement is physical interaction — not a mental act. The taller tree casting shade is a measurement. The photon doesn’t need to report to anyone. The universe is constantly “measuring” relational facts through causal interaction, producing consequences that track those facts.

This means relational facts are as old as the universe. “More massive than” existed before any minds did. The gravitational consequences of mass differentials were operating for billions of years before a single neuron fired.

Category 3: Agent-Relative Facts

Relational facts that require agents to exist but are not arbitrary given those agents. This is the category ProphetDivinityy needed and couldn’t name.

Defining features:

1. Wouldn’t exist without agents (remove all agents → these facts disappear)
2. Given agents of type T, determined by the nature of T and the structure of reality (not by any agent’s beliefs)
3. Agents can be wrong about them (there is a fact of the matter independent of what anyone thinks)
4. Independent agents of the same type in the same situation converge on the same answer

The cyanide test: “Cyanide is poisonous to humans” is agent-relative. Remove all humans, cyanide isn’t poisonous to anyone. But given humans, the toxicity is determined by biochemistry, not by belief. You can be wrong about whether something is poisonous. Independent scientists converge on the same LD50. This is a fact — a relational, agent-relative fact.

Moral facts work the same way: “Betraying every ally leads to isolation and vulnerability” is as much a relational fact as “cyanide disrupts cellular respiration.” Remove all agents, morality disappears. Given agents, the consequences of betrayal are determined by the structure of social reality and the nature of the agents — not by anyone’s opinion about betrayal.

More agent-relative facts:

• Nutrition: “Vitamin C prevents scurvy” — requires organisms, but is not a matter of opinion
• Danger: “Cliff edges are dangerous to bipedal organisms” — requires organisms, determined by physics + anatomy
• Economic value: “Division of labor produces surplus” — requires agents capable of trade, determined by comparative advantage (see Value and Profit)
• Moral norms: “Don’t murder, don’t steal, keep promises” — requires agents, converges across civilizations, you can be wrong about them, reality produces differential outcomes

Category 4: Mind-Dependent Facts

Properties that depend on an agent’s internal state and are not constrained by external reality. The genuinely subjective category.

Defining features:

1. Requires agents to exist
2. NOT determined by external reality — different agents in the same situation give different answers
3. You can’t be wrong (there is no fact of the matter beyond the agent’s state)
4. Independent agents do NOT converge

“Bell-bottoms are cool.” “Chocolate is better than vanilla.” “I prefer jazz.” These are real psychological facts about the agent, but they don’t track anything external. No amount of investigation will reveal the “objectively correct” ice cream flavor.

The Argument Chain

This four-category framework enables a clean argument that moral facts are objective. The chain:

Step 1 — Relational facts between non-agent objects are real.

The taller tree casts shade. No minds required. Undeniable.

This establishes that relational ≠ subjective. It kills the two-category assumption at the root.

Step 2 — Adding agents to the relation doesn’t degrade objectivity.

“Cyanide is poisonous to humans” is as factual as “tree A is taller than tree B.” The fact that one relation involves agents and the other doesn’t is irrelevant to whether it’s a fact. Both are determined by the nature of the relata and the structure of reality.

The burden of proof shifts: anyone claiming agent-involvement degrades objectivity must explain why. Why would adding a more complex relatum (an agent) to a relation make the relation less real than one involving simpler relata (trees)?

Step 3 — Moral facts are agent-relative facts.

They require agents to exist (remove all agents → no morality). They’re determined by the nature of agents and reality (not by belief). You can be wrong about them. Independent civilizations converge on the core. Reality produces differential outcomes (societies built on betrayal collapse; societies built on cooperation persist).

Step 4 — The boundary between 3 and 4 is where subjectivity enters.

Not all agent-involving claims are category 3. “Chocolate is better than vanilla” is category 4 — no convergence, no error, no differential outcomes. The diagnostic tests (below) distinguish them.

Diagnostic Tests

Three tests separate agent-relative (category 3) from mind-dependent (category 4):

Murder: Convergence across civilizations? Yes. Can you be wrong about whether it’s harmful? Yes. Differential outcomes for murderous vs. non-murderous societies? Yes. → Category 3.

Ice cream flavor: Convergence? No. Can you be wrong about what you prefer? No. Differential outcomes? No. → Category 4.

The fuzzy boundary: The diagnostic tests are clear at the poles but fuzzy at the margins. “How much personal risk is acceptable?” — some convergence (don’t walk off cliffs), but significant variation in degree. “Is moderate drinking harmful?” — convergence is still evolving as evidence accumulates. This is where the weighting problem lives: the core is agent-relative (category 3), the margins shade into mind-dependent (category 4), and the boundary is a gradient, not a wall.

The Measurement Backbone

The four categories map cleanly to the vault’s measurement-causality framework:

Each category is “measured” at a different level, and the objectivity question at each level reduces to: does reality constrain the answer, or only the agent’s internal state?

For categories 1-3, reality constrains. The tree’s shadow, the cyanide’s toxicity, and the social consequences of betrayal are all determined by how reality works, not by what anyone thinks. For category 4, reality doesn’t constrain — the answer lives entirely in the agent’s internal state.

How This Unifies the Vault

Morality Framework

The morality page already holds: “Change the creature, change the Tao. The moral law reflects what the agents are, not some law independent of them.”

This is the agent-relative thesis (category 3). Moral facts are real but relational — they exist in the interaction between agents and reality. Remove agents, morality disappears. Given agents, the facts are constrained. The four-category framework gives this position a name, a formal structure, and a philosophical lineage.

The morality page’s scope model maps directly:

The Weighting Problem

The weighting problem identifies exactly where category 3 transitions to category 4:

• Measurements (step 1) → Categories 1-2 (mind-independent)
• Core weightings that converge across civilizations → Category 3 (agent-relative)
• Marginal weightings that diverge → Category 4 (mind-dependent)

The weighting problem’s “outcome 3” — convergence of weightings is evidence of objective ranking structure — is the claim that core moral weights live in category 3. The four-category framework explains why: core moral weights are relational facts about agents and reality, measured by the causal consequences of different choices across civilizations of independent observers.

Structural Realism (Logic and Math)

The logic-and-math page holds: “The patterns they describe are real. The formalisms themselves are constructions.”

Same structure. The patterns in agent-reality interaction are real (category 3 facts). The descriptions we build of those patterns are constructed (but constrained). Structural realism and relational objectivity are the same epistemological move applied to different domains:

Performative Grounding

Performative grounding works because moral facts are agent-relative (category 3).

• If moral facts were intrinsic (category 1) — you wouldn’t need performative grounding; just point at them
• If moral facts were mind-dependent (category 4) — performative grounding wouldn’t work; there’d be nothing for the agent’s actions to contradict
• Moral facts being agent-relative (category 3) — performative grounding works precisely here: the agent’s own activity constitutes one side of the relation. The fact is real but relational, and the agent can’t escape the relation because existing and acting is being in the relation

Newcomb’s Paradox

Newcomb’s is an agent-relative problem. The “right” answer depends on what kind of agent you are — but it’s not arbitrary. Given your agent-type, the supercomputer’s prediction is a fact about the relationship between your type and the box contents. The one-boxer’s advantage is an agent-relative fact: agents who honor pre-commitments are the type the computer rewards. Not mind-independent (no agents, no paradox). Not mind-dependent (you can be wrong about what to do — two-boxers demonstrably are). Agent-relative.

The Philosophical Lineage

The agent-relative category has serious pedigree:

Foot and Korsgaard are the most directly relevant:

• Foot gives the cleanest version: moral facts are natural relational facts, the same kind of thing as “good roots for an oak.” Biological, not Platonic.
• Korsgaard gives the binding mechanism: you can’t opt out of agent-relative moral facts because agency itself generates them. This is structurally identical to the vault’s performative grounding.

What’s Established vs. What’s Synthesis

The individual pieces of this framework are all well-established in analytic philosophy. The specific four-category packaging is a synthesis — drawing from multiple traditions that each carved the landscape differently. Honesty about what’s inherited and what’s new:

What’s established

Category 2 (relational mind-independent) is standard analytic metaphysics. The intrinsic/extrinsic distinction — David Lewis, D.M. Armstrong, Kit Fine — treats relational properties as fully real. “Tree A is taller than Tree B” is a textbook extrinsic relational property. No one in the field disputes its objectivity. But the tradition doesn’t treat it as a separate ontological category requiring its own label — it’s just filed under “relational property” within the broader class of “real.”

Category 3 (agent-relative) is well-defended across multiple traditions. Foot’s natural goodness, Korsgaard’s normativity-from-agency, Railton’s moral realism, Johnston’s response-dependence — all argue that agent-involving relational properties can be objective. The philosophical heavy lifting has been done.

The 3/4 boundary is an active research area. The distinction between agent-relative facts (objective despite involving agents) and mind-dependent facts (genuinely subjective) is where most of the contemporary metaethics debate lives.

How existing traditions carve the landscape

Each tradition draws lines in different places, and none produces all four categories cleanly:

What’s new in this synthesis

Two things:

1. The explicit four-category taxonomy with category 2 as the argumentative bridge. Each tradition above recognizes some of the categories but lumps others. Lewis/Armstrong split 1 from 2-4. Johnston splits 1-2 from 3-4. Nobody (as far as we can find) splits all four cleanly in a single framework and uses category 2 as the stepping stone that forces the moral realism conclusion.

The argumentative move — “you grant tree shade is a relational fact; now explain why adding agents degrades the relation” — is where the specific value lives. Foot comes closest: she uses “good roots for an oak” as the bridge from biological relational facts to moral facts. But her oak is already an organism, so it’s arguable whether it’s category 2 or 3. The tree-shade example is cleaner because it’s purely physical — no organisms evaluating anything, just photons and chloroplasts. The height differential has causal consequences (competitive advantage for sunlight) with no agent in the loop at all.

2. The measurement-causality backbone connecting all four categories. The vault’s position that measurement = physical interaction (from the measurement-causality page) provides a unified account of what “measures” each category: physics measures category 1-2 facts, agent-reality interaction measures category 3 facts, and only the agent’s internal state “measures” category 4 facts. This connects the ontology of facts to the philosophy of physics in a way the existing traditions don’t, because they weren’t working from a measurement-as-interaction framework.

The MacIntyre connection

MacIntyre’s After Virtue (1981) is worth flagging specifically. His central thesis is that modernity lost the category-3 framework — the Aristotelian understanding that “good” is a relational fact about organisms and their telos — and defaulted to a two-category model (objective science vs. subjective values). The Enlightenment project of grounding morality in pure reason failed precisely because it was trying to make moral facts category-1 (intrinsic, mind-independent) when they’re actually category-3 (agent-relative). The four-category framework is, in a sense, the recovery of what MacIntyre says was lost — but grounded in measurement-causality and performative contradiction rather than in Aristotelian teleology.

The Wilson Counter and Its Answer

Wilson’s likely objection: “Who determines what counts as ‘the nature of agents’? Isn’t that just another stance-dependent fact?”

The answer: The agents determine it — by existing. You don’t choose your nature any more than you choose the gravitational constant. Human nature is an empirical fact about what humans are, discoverable by observation and confirmed by cross-cultural convergence. The fact that it requires agents to exist doesn’t make it optional or arbitrary. Water’s chemical composition requires water molecules to exist — that doesn’t make H₂O a matter of opinion.

Wilson’s deeper move: “But you’re using your nature to evaluate your nature. That’s circular.”

The answer (performative grounding): Yes — and there’s no alternative. Any agent evaluating anything is using their cognitive faculties, which are features of their nature. This isn’t a vicious circle; it’s the ground floor. The tree doesn’t need to “justify” casting shade. The agent doesn’t need to “justify” having a nature. Demanding justification for the preconditions of justification is a category error — and making the demand demonstrates the preconditions.

Open Questions

1. The boundary gradient. The diagnostic tests (convergence, error, causal consequence) are clear at the poles but fuzzy at the margins. Is there a principled way to locate where category 3 shades into category 4, or is the gradient itself the answer? The weighting problem suggests the gradient is the answer — and the enforceability gradient from scope confusion maps enforcement to the degree of convergence.

2. Does category 2 actually need to be separate from category 1? “Tree A is taller than Tree B” could be analyzed as an intrinsic fact about the pair {Tree A, Tree B}. Kit Fine’s formal ontology treats relational properties as multi-place predicates — real properties, just not monadic ones. The separation may be more pedagogical than ontological. But the pedagogical value is high: it makes the relational-but-objective point undeniable before agents enter the picture.

3. Cross-species agent-relative facts. “Cyanide is poisonous to humans” and “cyanide is poisonous to dogs” are different facts — same relation, different relata. Moral norms presumably work the same way: “cooperation is beneficial for social primates” might be a different (but related) fact from “cooperation is beneficial for eusocial insects.” How far does agent-relative morality generalize across agent types? Connects to the animal cognition thread from Wilson debate and epistemology — non-human concept formation.

4. Can category 4 facts become category 3? Fashion seems mind-dependent — but if wearing culturally appropriate clothing improves social outcomes (convergent, error-possible, causally consequential), does it shift categories? This would mean some apparently subjective facts are actually agent-relative facts that we haven’t recognized yet. The history of nutrition suggests this happens: “eat what tastes good” was category 4 until we discovered vitamins, at which point “eat vitamin C” became category 3.

5. The relationship between categories 2 and 3 under measurement-causality. If measurement is physical interaction, and agent-reality interaction is a kind of measurement, then category 3 might be a subset of category 2 where the relata happen to include agents. The shade cast by the taller tree and the death caused by cyanide are both “measurements” — reality registering relational facts through causal consequences. The only difference is the complexity of one relatum (agent vs. tree). Does this mean the ontological distinction between 2 and 3 is merely one of complexity, not of kind?

Suggested Reading

Tags

philosophy, epistemology, morality