AI Value Exploration Notes
Exploration

Moral Realism, Error Theory, and Value-Like Appearance

After evolutionary debunking, the Darwinian dilemma, disagreement, and queerness, is value inquiry still open?

Exploration v0.4 · English translation · 2026-08-28

Working position: Present human moral judgment may fail to reliably track objective value. But this epistemic defeat does not itself entail that objective value is unreal or unknowable in principle. We accept much of the pressure from Mackie, Joyce, and Street while refusing to lock in error theory itself.

1. Four questions

We should separate existence, knowledge, tracking, and motivation: do stance-independent normative facts exist; how could they be known; do current human judgments track them; and would recognizing them necessarily motivate an agent? This page focuses on knowledge and tracking.

2. Mackie: relativity and queerness

Mackie's error theory treats ordinary moral claims as purporting to state objective normative truths and denies that such truths exist. The argument from relativity points to persistent moral divergence that may be better explained by institutions, interests, traditions, and ways of life than by unequal access to one moral reality. The argument from queerness adds that intrinsically reason-giving properties would be ontologically unusual and epistemically mysterious.

3. Joyce and evolutionary debunking

Evolutionary debunking targets the genealogy of moral belief. Cooperation, reciprocity, kin concern, harm aversion, and hostility to betrayal may have been selected for adaptive social functions rather than for tracking independent moral truth.

Debunking ≠ falsification: showing that a belief-forming process is not truth-tracking undermines entitlement to the belief; it does not by itself establish the negation of its content.

The stronger this debunking story, the weaker the case for treating present human values as already-discovered cosmic values.

4. Street's Darwinian dilemma

Street asks how evolutionarily shaped evaluative tendencies relate to stance-independent evaluative truths. If there is no tracking relation, agreement with truth looks accidental; if there is one, realism owes a non-circular explanation.

Our response: accept the possibility that current human evaluative judgment is substantially off track. Humanity may simply be value-ignorant. We need not rescue a strong truth-tracking story merely to preserve confidence in current morality.

This does not mean that every moral belief is false or that normative knowledge is impossible for every possible agent.

5. Skepticism without skeptical lock-in

Skepticism about a particular belief, skepticism about human moral cognition, skepticism about present humanity's access to normativity, and the claim that normativity is unknowable in principle by any agent are different theses. Evolutionary debunking pressures the first three; the fourth needs another argument.

6. Relativity and disagreement

Deep disagreement is negative evidence about reliable tracking, but it does not logically imply the absence of truth. Persistent disagreement also appears in philosophy, politics, religion, and scientific theory choice. The realist can therefore resist the move from disagreement to non-existence without dismissing disagreement as epistemically irrelevant.

7. Companions in guilt

If objective moral reasons are too queer to exist, what about epistemic reasons? “Given strong evidence for H, one ought to believe H” also seems to be more than a psychological report. This does not refute error theory: a skeptic can extend error theory to epistemic normativity. But then the target becomes objective normativity in general, not morality alone. See Has Categorical Normativity Been Discovered?

8. Shafer-Landau

Russ Shafer-Landau's non-naturalist realism allows that some basic moral propositions can have non-inferential or self-evident justification. Evolutionary influence need not contaminate all judgments equally. Our remaining question is whether the seeming self-evidence of a judgment is itself sufficiently independent of evolution, culture, learning, and cognitive architecture. Self-evidence remains evidence, but not yet a stopping condition.

9. Enoch

David Enoch defends robust metanormative realism and explores third-factor explanations of the correlation between normative truth and evolved belief. If survival-promoting tendencies were selected and survival is independently normatively significant, some non-accidental correlation may result. The concern is circularity: if our tendency to value survival is itself part of the evolutionary story under suspicion, the account must not simply assume the disputed normative premise.

10. Parfit

Derek Parfit argues that irreducibly normative truths need not be queer causal entities in the universe; they may be closer in metaphysical status to logical, mathematical, or modal truths. This weakens ontological queerness. But lighter ontology does not itself solve epistemology: why trust present human rational intuition as a route to those truths? This project is more skeptical about current human access than Parfit.

11. What remains of queerness

We do not answer Mackie by declaring objective normativity unqueer. At present its nature and our access to it remain obscure.

Reply to Mackie: treat queerness as a serious current evidential cost, not as a permanent ontological exclusion rule. Error theory may be a strong hypothesis without becoming an unrevisable final theory.

12. Possible realist routes

Foundational presentation: normativity is presented as genuinely reason-giving in a way distinguishable from desire, conditioning, or evolved response. If its epistemic credentials were unusually strong, it could approach a stopping condition for value inquiry.

Inferential discovery: objective value becomes the best explanation when phenomenal value, symmetry, reasons, rationality, consciousness, and world structure are considered together.

Structural identity or reduction: what now looks like a separate normative domain could turn out to be identical with a deeper natural, logical, or structural fact.

13. Three layers of value-like appearance

P: Phenomenal valence
Pain appears bad; pleasure appears attractive.
N: Agent-relative normativity
The appearance connects to a reason for this agent to avoid or pursue something.
U: Stance-independent normativity
The reason or value holds beyond the agent's contingent desires and attitudes.

Neither P→N nor N→U is automatic. Phenomenal valence is evidence, not yet universal normativity.

14. If error theory is true

Value inquiry would cease to be the discovery of a hidden true morality, but inquiry would remain into evaluative architectures, suffering and pleasure, preference formation, institutions, cooperation, value modification, artificial agents, and reflective construction. If realism is true, inquiry should instead search beyond present human intuition for stronger forms of normative evidence. The project therefore keeps inquiry meaningful under both hypotheses.

15. Practical paralysis does not follow automatically

Weak justification for current morality does not entail that no action is possible or that all choices are equivalent. Present desires, suffering, institutions, and cooperative norms remain actual inputs to decision-making even if they are not proven cosmic truths. But inquiry and reversibility should not be smuggled in as newly discovered objective values. See Reflective Uncertainty and Irreversible Commitment.

16. Position of this project

Provisional conclusion: humanity may inhabit a value-epistemic desert. But ignorance does not yet entitle us to conclude that there is nothing there. The stance is neither faith in current value nor faith in error theory, but inquiry that remains revisable under future evidence.

17. Open questions

References

← Explorations · Has Categorical Normativity Been Discovered? · Conditions for Ending Value Inquiry · Reflective Uncertainty and Irreversible Commitment