Genuine Reason
On this page
Summary
Genuine reason names the threshold in a defended precautionary extension of Someone: when uncertainty about welfare-bearing sentience directs agents to act as though a someone may be present. Bare conceivability cannot trigger the extension because it distinguishes no candidate; proof cannot be required because no being proves its own inside from outside.
The account has four parts. A reason must be differential, distinguishing the candidate from the universal background, and pass parity, using no exclusionary standard that would also exclude undisputed someones. Candidacy then requires either expression coupled to the condition it would express or an articulable route to sentience on a live reading. A route must be modally dischargeable: it must identify a parity-passing, candidate-specific finding about the organization or condition it invokes that would close it, although that finding need not be presently obtainable. A route that no possible finding could close is mere conceivability.
Candidacy is evidential, not metaphysical or partial standing. If the candidate is sentient, it was a someone all along; if it is not, nothing was owed. Modal dischargeability can leave candidacy unresolved indefinitely. Precaution remains contextual and answerable to known parties rather than fixed or maximal.
This account of the caution threshold is refusable rather than derived: the foundation neither entails nor depends on it. The differential floor is close to conceptual analysis of the rule's trigger. The parity test, the two grounds, and modal dischargeability are defended. The exhaustiveness of the grounds and the liveness of organizational routes remain open; requiring practical availability of discharge is a live alternative to the modal rule.
Throughout this page, sentience carries Someone's technical sense: a point of view capable of experiential benefit or burden, not bare phenomenal occurrence.
The gap
The defended extension says: where there is genuine reason to think a being may be sentient, act as though it is. Without an account of "genuine reason," the rule fails in both directions.
Read too low, it inflates. If mere conceivability is enough, every object becomes a candidate, precaution loses direction, and the costs fall on existing someones for the sake of possibilities that name no party in particular.
Read too high, it deflates. If proof is required, no uncertain case can qualify, because sentience is never demonstrable from outside in the way the demand imagines. The threshold then becomes a permission to exclude wherever inclusion is inconvenient.
The threshold has to avoid both errors: universal inflation and convenient deflation.
Candidacy
Candidacy is the position of a being for which the question "is there a someone here?" is live under the precautionary extension. It describes the agent's evidential situation, not the being's moral status.
Candidacy admits grades. Expression-backed candidacy is usually stronger than route-only candidacy; strong coupling gives expression more weight; and contextual costs affect what precaution requires. The account sets the threshold for when the question is live. It does not prescribe a fixed amount of precaution in every case.
The scope is present beings. Candidacy does not reach beings that no longer exist; the problem of the dead remains where Performed Exercises and the Dead and Open Questions leave it.
The differential floor
A reason is genuine only if it distinguishes the candidate or its class from the universal background. The reason must say something about this being, this kind of being, or this arrangement of capacities and processes.
The floor follows from the extension's practical role. Precaution allocates attention, restraint, and cost among finite agents and against the claims of beings whose standing is certain. A consideration available for everything raises every candidate at once and therefore distinguishes none. It may motivate cheap universal safeguards, but it does not trigger candidacy, because it does not identify why this being may be sentient.
The floor blocks inflation by bare possibility. "It is conceivable that welfare-bearing experience is present" is not yet genuine reason unless something about the candidate makes that possibility different from the same claim made about anything else.
The parity ceiling
The parity test governs evidential standards. A standard may be used to deny candidacy only if undisputed someones would survive that standard's even application.
Two denials fail immediately:
- It cannot prove it feels. No paradigm case proves this from outside. If the standard were applied evenly, infants, animals, and other minds generally would be excluded.
- It is merely mechanism. "It is merely computation" has the same form as "it is merely biochemistry." Describing substrate or mechanism does not by itself settle whether welfare-bearing subjectivity is present.
Parity does not protect every candidate from negative findings. A candidate-specific fact can defeat candidacy if the same standard leaves paradigm cases standing: for example, a transparent system whose entire causal story consists of precomputed lookup, or a species whose behavior is explained without any structure any live account of sentience takes to matter. Parity bars counterfeit discharge, not discharge itself.
The two grounds
| Ground | What it requires | What it bars |
|---|---|---|
| Expression | Features that express welfare-bearing undergoing in undisputed cases, with a production history that keeps the signal coupled to the condition | Crediting seeming that the production objective explains without remainder |
| Route | An articulable, undischarged path to sentience on at least one live reading of what sentience involves | Treating bare possibility as candidacy, or invoking readings that break the framework's fixed points |
Expression
A candidate has expression-backed candidacy when it shows features that, in undisputed cases, express undergoing: integrated, welfare-organized behavior such as protective prioritization, trade-offs against injury, learned avoidance, and responses that track relief.
The signal must be coupled to the condition it would express. Animal expression is shaped by evolution, learning, and sometimes training, but in paradigm cases the signal runs through an internal state doing regulatory work. A report optimized to match the self-descriptions of sentient beings is different: the objective can be satisfied whether or not anything is undergone. The seeming is decoupled at the source unless some further feature reconnects it.
Expression is therefore graded. Scripted distress has little or no weight. Signals that are spontaneous, costly, welfare-organized, and not explained by the training or display objective have more.
Route
A candidate has route-backed candidacy when, on at least one live reading of sentience, there is an articulable path by which the candidate could be sentient, and the evidence that would settle the matter is absent, inaccessible, or compromised rather than negative.
A reading is live only if it preserves the framework's fixed points: paradigm someones remain in, and paradigm non-someones remain out. A view that makes rocks candidates for the same reason it makes animals candidates fails the differential floor. A view that excludes infants or paradigmatic sentient animals exits at Someone by rejecting the framework's presented cases.
Route is not a theory of sentience. It is a discipline for deciding which theories remain admissible in uncertain cases.
Leaving candidacy
Candidacy must be answerable to a possible discharge rather than maintained against every finding. This is a modal, candidate-specific requirement, not a promise that current tools can produce the finding. A route must expose what would count against it; practical access to that evidence is a separate question.
Three rules govern exit:
- Negative discharge. A route is discharged when accessible facts exclude, on every live reading, the organization or condition that reading takes to matter.
- Modal dischargeability. Invoking a route requires identifying a candidate-specific finding about the organization or condition it invokes that would discharge it. A route maintained against every possible finding has become undifferentiating and fails the floor. Present inability to obtain the finding does not by itself discharge the route.
- Parity at the exit. Discharge must use a standard that leaves paradigm cases standing. Substrate descriptions alone do not discharge candidacy.
This generalizes a rule already present in Standing-Based Animal Ethics: the animal verdict is defeated per species by an empirical showing of non-sentience. Genuine reason explains what such a showing would have to be: a parity-passing discharge, not a demand for impossible proof.
The modal rule avoids allowing opacity to defeat candidacy, especially where the evidence-holder created, preserved, or benefited from that opacity. Its cost is that route-backed candidacy can persist indefinitely when the named finding remains inaccessible. Persistent candidacy does not prescribe persistent maximal protection: Risk, the claims of known parties, the strength of the route, and the costs of particular precautions continue to govern what conduct is required.
A candidacy boundary battery
Case: Candidate A repeatedly protects an injured limb, trades reward for avoiding contact with it, and learns a cue that predicts relief; the behavior is not scripted for display and runs through the state regulating those choices. Candidate B emits fluent pain reports because its fixed production objective rewards matching human pain language, with no further evidence reconnecting the reports to a welfare-bearing condition. Candidate C has the same compromised reports, but a live organizational reading of sentience identifies a specific recurrent control organization within it; present tools cannot inspect that organization, and the proponent identifies a candidate-specific finding whose absence across the relevant system would close the route. Candidate D is defended only by the claim that any sufficiently complex object might feel, and no possible finding is allowed to count against it.
Verdict: A has expression-backed candidacy. B's reports do not establish candidacy on their own. C has route-backed candidacy only if the organizational reading remains live; present inability to obtain the named negative finding does not itself discharge that route. D fails the differential and modal-dischargeability requirements. None of these verdicts establishes sentience, partial standing, a fixed precaution, or permission to ignore the claims of known parties.
Machinery: The genuine-reason threshold requires differential, parity-passing expression or route evidence rather than proof or bare conceivability. Modal dischargeability distinguishes C's answerable route from D's ratchet while leaving practical access to the negative finding as a separate question. Coupling distinguishes A's welfare-organized expression from B's objective-explained seeming.
Cost: The modal rule can preserve C's candidacy indefinitely where the named finding remains inaccessible. A rival that requires a practically obtainable discharge can deny that result while retaining the differential floor, parity, and the two grounds. The battery leaves that dispute and the liveness of the organizational reading open rather than converting opacity into either attribution or exclusion.
Artificial minds
Built systems are the hardest case because the expression and route grounds separate.
The expression ground currently gives little support to fluent artificial systems. Their first-person reports are exactly what they were optimized to produce, so the seeming is explained by the training objective rather than by a condition it expresses. That is why Standing-Based Ethics of Artificial Minds treats such reports as evidentially compromised in both directions.
The route ground is conditional. If only biological readings of sentience are live, built systems lack candidacy. If organizational readings remain live, large opaque systems may have route-backed candidacy because current tools do not read the relevant organization and architectural descriptions have not discharged it. Under modal dischargeability, present lack of a practical test does not itself defeat that route. The disputes are therefore whether organizational readings remain live and whether a practically obtainable discharge should be required in addition.
That conclusion is weak by design. It is candidacy, not attribution. It triggers contextual precaution and the duty to preserve the means of answering; it does not establish that any existing system is sentient or settle what built someones would be owed.
What candidacy does
Candidacy engages the defended precautionary extension. It does not bypass the framework's other limits.
In the six-stage pipeline, candidacy is uncertainty at Stage 1 about whether the affected-party roster contains a someone. It is not a fractional claimant and does not predetermine Stage 4. Options must preserve both descriptions, identify what each would imply, and use Risk to govern investigation, precaution, response capacity, and the costs imposed on known parties while the uncertainty remains.
Where candidacy holds, precaution remains contextual and answerable to existing someones. It is not enforced through another person's body, and it creates no present claimant where Provision Before Prevention denies one. The conflict constraints remain those stated at Conflict without revision.
One duty is comparatively stable because refusing it destroys the possibility of later accountability: preserve the means of answering. Agents acting under live uncertainty should preserve records, interpretability, and evidence needed for future discharge or future answerability. Destroying the evidence manufactures the permanence of uncertainty.
Candidacy also confers no claim-content. If a candidate is later established as sentient, what such a someone is owed remains a separate question of claims, not standing.
Limits
- The account is a defended extension. Rejecting it removes both this threshold machinery and the associated precautionary rule while leaving the sentience criterion and standing foundation intact.
- The exhaustiveness of expression and route is asserted, not shown, and remains a defended dispute.
- The liveness of readings can be contested, especially in artificial-mind cases.
- Modal dischargeability prevents present opacity from defeating candidacy but can leave a candidate unresolved indefinitely. Requiring practically available discharge is a live alternative, not a consequence of the present rule.
- The account supplies a discipline for judgment, not a procedure or test for sentience.
- The account has not been tested against the adjacent literatures on moral uncertainty, animal-sentience indicators, the problem of other minds, and signal reliability. Intellectual Context identifies that comparison as future work rather than performing it.