Deep Self and Welfare

Working note. This proposes Deep Self as the welfare bearer and evaluative anchor required by welfare language. It distinguishes that framework-level claim from the stronger—and still unresolved—claim that a determinate Self/Mechanism boundary can be located.

Welfare must have a bearer.

$$\mathrm{Welfare}\Rightarrow\mathrm{WelfareBearer}$$

Welfare is not a free-floating property of a world-state. A condition is good or bad as welfare only because it is good or bad for someone. Pleasure, pain, preference satisfaction, money, life expectancy, behavior, and self-report may be extremely important evidence or proxies, but none by itself supplies the subject for whom the change matters.

I use Deep Self, or $DS_i$, for this person-level evaluative standpoint:

$$DS_i = \text{person-level evaluative structure} = \text{welfare bearer} = \text{agent-level standpoint}$$

The same person-level information can be described as selfhood structure, person-level ethics, utility-generating structure, or person-level meta-ethics. These phrases do not name several independent parts. They approach one high-dimensional evaluative structure from different directions: the structure through which a person cares, identifies, endorses, rejects, evaluates, is benefited, or is harmed.

The intended direction of explanation is:

$$\mathrm{DeepSelf} \to \mathrm{PersonLevelEvaluativeInformation} \to \mathrm{Welfare}$$

not:

$$\mathrm{Pleasure/Preference/Norm} \to \mathrm{WelfareBearer}$$

This does not make pleasure, preference, or norms irrelevant. It makes them mediated variables whose welfare significance depends on their relation to a bearer.

Why a mechanistic description is not enough.

The distinction between Self and Mechanism is not initially an empirical claim that psychology has already located two separate objects. It is introduced because a purely mechanistic explanatory space can describe transformation, regulation, stabilization, distortion, breakdown, compensation, and expression, yet still fail to supply an evaluative standpoint.

If the human explanatory space contains only variables and mechanisms, then it may contain behavior, institutions, coordination, discourse, and equilibrium. But these remain system states unless there is something for whom they matter.

$$\mathrm{NoSelf} \Rightarrow \mathrm{NoWelfareBearer}$$ $$\mathrm{NoWelfareBearer} \Rightarrow \text{welfare has no subject}$$ $$\mathrm{NoAgent} \Rightarrow \text{evaluation has no standpoint}$$

The recurring question is simply:

$$\text{For whom?}$$

Policy does not merely optimize system states; law does not merely regulate behavior; responsibility is not meaningful by itself; and welfare is not a floating number. Each ultimately refers to changes that matter to welfare-bearing agents.

The grounding direction is therefore:

$$\mathrm{Agent} \to \mathrm{Evaluation} \to \mathrm{Welfare} \to \mathrm{Norm/Law/Policy}$$

rather than the reverse. Rights, legal personhood, responsibility, punishment, contracts, citizenship, and institutional identity are social technologies for organizing relations among welfare-bearing selves. They may be indispensable, but they do not create the bearer whose welfare justifies them.

Meaning has the same indexed structure.

A world-state does not carry human meaning entirely by itself:

$$\mathrm{Meaning}\neq\mathrm{Meaning}(W)$$

Meaning is indexed to an agent:

$$\mathrm{Meaning}=\mathrm{Meaning}(W,S)$$

A world containing only one person could still contain meaning, because there would still be someone for whom things matter. This suggests that discourse and social equilibrium mediate and alter meaning, but are not required as its original source.

The anti-constructivist concern follows from the same structure. If discourse creates morality, morality defines the agent, and the agent participates in discourse, the proposed foundation becomes circular:

$$\mathrm{Discourse} \to \mathrm{Morality} \to \mathrm{Agent} \to \mathrm{Discourse}$$

The deeper problem is not only formal circularity. If the Self is entirely a product of current norms, welfare risks becoming merely a norm-indexed state-change rather than a change for a real bearer. On this view, murder is wrong because it destroys a Self and its welfare, not merely because it rearranges a system in a rule-disfavored way.

The strongest claim required by the framework is comparatively modest:

$$\mathrm{Welfare},\ \mathrm{Meaning},\ \mathrm{Evaluation},\ \mathrm{Policy} \Rightarrow \mathrm{EvaluativeSubject}$$

The existence of such a subject may be a framework-level or transcendental commitment rather than something proved from a more basic vocabulary.

Observed preference is not identical to the bearer.

Most accessible evidence is mediated output:

pleasure
preference
behavior
self-report
reflective judgment
pain
suffering
motivation

A reported preference may be unstable, compulsive, socially learned, strategically expressed, or distorted by a mechanism. Suffering may arise because a mechanism overrides the Self, an environment blocks self-expression, a conflict within the Self is painful, or an ordinary loss genuinely matters. The reality of the report or suffering does not by itself settle its provenance.

Psychosis, compulsion, ego-syntonic conditions, and personality-related cases make the difficulty especially visible. Treating present self-report as the final authority can collapse the bearer into one mediated output; treating clinical judgment as the final authority merely replaces one proxy with another. First-person evidence remains indispensable without becoming automatically sovereign.

self-report = indispensable evidence
self-report ≠ the Self
self-report ≠ final sovereignty

Individual welfare functions can therefore be rich. They may contain autonomy, fairness, trust, recognition, identity, security, agency, non-humiliation, non-coercion, a sense of justice, and institutional legitimacy. To ground welfare in agents is not to reduce it to pleasure or a tally of stated preferences.

Physical realization and the boundary problem.

Nothing here requires a non-physical substance. Even if all biological and neurological facts were known, Self could be physically realized. The remaining question would be how the total known structure is partitioned for explanatory purposes.

Let:

$$P_{\mathrm{total}}$$

be the fully known biological or mechanistic structure, and let:

$$P_{\mathrm{total}}=P_S\cup P_M$$

where $P_S$ realizes Self and $P_M$ is non-self Mechanism. The crucial condition is that, in the relevant explanatory geometry,

$$P_S\cap P_M=\varnothing.$$

A point cannot be both the agent and merely one of the mechanisms acting on or through the agent. The boundary may be complex, curved, fuzzy in observation, or practically inaccessible without being ontologically absent.

If no such distinction exists, it becomes unclear what answers the questions:

What is me?
What is my function, mechanism, state, or output?
Where is the agent for whom welfare matters?

The boundary cannot simply be fixed by law, responsibility, social recognition, or self-report, since those already presuppose some account of the agent. That would reproduce the same self-reference:

$$\mathrm{Norm} \to \mathrm{AgentBoundary} \to \mathrm{Agent} \to \mathrm{Norm}$$

Complete empirical knowledge may supply the full physical facts:

$$\mathrm{EmpiricalTruth}\to P_{\mathrm{total}}.$$

An additional non-normative boundary principle may still be required:

$$\mathrm{BoundaryPrinciple}(P_{\mathrm{total}}) \to (P_S,P_M).$$

That principle might be axiomatic, metaphysical, or bridge-like. This is deliberately left open. The note does not claim that the principle has already been found. It records the unresolved pressure: empirical description of every process may still not tell us which part of the structure is the welfare-bearing agent.

Pain, suffering, and moral status.

Pain is morally important, but it is not the primitive foundation of welfare. The deeper relation is:

$$\mathrm{Self} \to \mathrm{WelfareSignificance} \to \mathrm{Pain/Suffering\ as\ powerful\ signals}$$

rather than pain generating a welfare bearer by itself. This matters for non-human moral status. Species membership need not be the final criterion; a non-human being with the relevant selfhood or evaluative structure could be a welfare bearer. Pain would remain important evidence, but the underlying question would be whether there is a subject for whom the state matters.

Human-bounded objectivity.

The proposed evaluative objectivity is human-bounded. Human moral concepts arise within human biological, cognitive, social, and selfhood structure. Pain, dignity, agency, personhood, and welfare are not read directly from the universe; they become intelligible through the structure of human welfare-bearing selves.

$$\mathrm{HumanOntologicalStructure} \to \mathrm{HumanMetaEthics}$$

This is not simple universal realism, because beings with radically different self-structures could have a different evaluative structure. It is not simple relativism, because human meta-ethics would still be constrained by human ontological structure rather than created by a vote, a discourse, a historical convention, or a current norm.

The guiding distinction throughout is:

$$\mathrm{Reality}\neq\mathrm{Representation}$$ $$\mathrm{Self}_{real}\neq\mathrm{Self}_{reported}$$

The open task is to explain how a real evaluative bearer is represented, mediated, and imperfectly accessed without either reducing it to its proxies or treating it as already empirically located.

notes: Frankfurt's discussion of desire and caring, and Sripada's account of the deep self, form part of the early background to the questions considered here. The phrase deep self, some care-based vocabulary, and several addiction scenarios follow Sripada's usage. Crisp's Well-Being is cited for the familiar formulation of well-being as what is good for a person.