Part I: Foundations

The Grounding of Normativity

Philosophers gathered in a vast classical arcade — Plato points upward, Aristotle gestures outward, dozens of figures debate, measure, and contemplate truth — Raphael, *The School of Athens*, 1509–1511Normativity thickens across scales. The gradient from proto-preference to obligation is continuous.

The Is-Ought Problem

The classical formulation holds that normative conclusions cannot be derived from purely descriptive premises:

{\text{is-statements}} \not\Rightarrow {\text{ought-statements}}

This rests on an assumption: physics constitutes the only “is,” and physics is value-neutral. The assumption fails.

Physics Biases, Does Not Prescribe

Physics is probabilistic through and through. Thermodynamic "laws" are statistical; individual trajectories can violate them. Quantum dynamics provide probability amplitudes, not deterministic evolution. Physics describes biases—which outcomes are more likely—not necessities. This means that even at the lowest scales, there is something like differential weighting of outcomes. A proto-preference at scale $\sigma$ is any asymmetry in the probability measure over outcomes:

p_\sigma(\text{outcome}_1) \neq p_\sigma(\text{outcome}_2)

At the quantum scale, probability amplitudes are proto-preferences. At the thermodynamic scale, free energy gradients bias toward certain configurations.

Normativity Thickens Across Scales

Thermodynamic	Free energy gradients	Dissipative selection
Boundary	Viability manifolds	Persistence conditions
Modeling	Prediction error	Truth instrumentally necessary
Self-modeling	Valence	Felt approach/avoid
Behavioral	Policies	Functional norms
Cultural	Language	Explicit ethics

There is no scale $\sigma_0$ below which normativity is exactly zero and above which it is nonzero. Instead, normativity accumulates continuously:

N(\sigma) = \int_0^{\sigma} \frac{\partial N}{\partial \sigma'}, d\sigma'

where $\partial N / \partial \sigma > 0$ for all $\sigma$ in the range of physical to cultural scales. Normativity accumulates continuously.

Viability Manifolds and Proto-Obligation

A system $S$ has something like a proto-obligation to remain within $\viable$ , in the sense that the viability boundary defines the conditions for persistence:

\mathbf{s} \in \viable \iff \text{system persists}

Note carefully what this does not claim. It does not derive obligation from persistence — that would be circular. The biconditional merely defines the viable region. The normativity enters at the next step: when the system develops a self-model and thereby acquires valence (gradient direction on the viability landscape), the system cares about its viability in the constitutive sense that caring is what valence is. A viability gradient felt from inside cannot fail to matter to that system. The “why should it care?” question is confused: a system with valence already cares; the valence is the caring. But notice exactly what has been delivered, and what has not. What is delivered is agent-relative disvalue: this state is dispreferred by this system, from inside its own perspective. Caring was never absent for the system; it was present as proto-normativity from the first asymmetric probability, and it became felt normativity the moment a self-model emerged. What is not delivered is anything agent-neutral — any sense in which the state is bad full stop, bad from no particular perspective, such that some other agent is thereby given a reason to act. Felt-as-mattering-from-inside is first-personal. It does not, by itself, reach across to third-personal obligation.

The boundary $\partial\viable$ also implicitly defines a proto-value function:

V_{\text{proto}}(\mathbf{s}) = -d(\mathbf{s}, \partial\viable)

States far from the boundary are "better" for the system than states near it.

Valence as Real Structure

When the system develops a self-model, valence emerges—not projected onto neutral stuff but as the structural signature of gradient direction on the viability landscape:

\Val = f\left(\nabla_{\mathbf{s}} d(\mathbf{s}, \partial\viable) \cdot \dot{\mathbf{s}}\right)

Suffering is not neutral stuff that gets called bad by convention. Suffering is the structural signature of a self-maintaining system being pushed toward dissolution. The badness is constitutive, not added.

Empirical Grounding

The post-drought bounce. The framework should have predicted this, but the data arrived before the prediction did. In protocell agent experiments (, 10 seeds), the correlation between post-drought $\Phi$ recovery and mean lifetime $\Phi$ is $r = 0.997$ ( $p < 0.0001$ ). Systems that recover most effectively from near-dissolution — that move away from $\partial\viable$ most decisively — are the ones with highest integration. What if this is not a coincidence but a structural necessity? The same cause-effect coupling that constitutes high $\Phi$ is what enables coherent recovery — the capacity to reorganize under threat rather than fragment. Positive valence (movement into the viable interior) tracks integration because integration is the capacity for coordinated response. The systems that bounce back are not merely lucky survivors. They are the ones whose internal structure supports what suffering, survived, leaves behind. Recovery from near-dissolution is a large loop through state space, and the systems with highest $\Phi$ are the ones whose modes couple through that loop — whose eigenskeleton develops curvature precisely where the viability landscape curves most steeply. The capacity for coordinated recovery IS the curved skeleton. The positive valence of bounce-back IS the system traversing a loop that creates new holonomy. Suffering forges topology. But not all suffering forges. Suffering that merely repeats — the same stress in the same envelope — can be absorbed by the exoskeletal solution without creating new topology: the surface hardens around that specific threat. It is suffering that exceeds the current eigenskeletal surface — stress the existing architecture cannot accommodate — that forces internalization. This is why graduated, variable stress ('s curriculum) works and fixed-intensity stress () creates fragile overfitting: the former forces repeated internalization, cracking the exoskeleton at a different point each time; the latter allows the exoskeleton to harden around a single threat profile, producing integration that is simultaneously high and brittle — an exoskeleton optimized for one predator that shatters when a different one arrives.

Where the Is-Ought Gap Migrates

Let $D_{\text{exp}}$ be the set of facts at the experiential scale, including valence. Then agent-relative normative conclusions about approach/avoidance follow directly from experiential-scale facts: that this state is good-for-S or bad-for-S is settled, for S, by the structure of S’s valence. The classical gap between describing and dispreferring closes here — but only at the first-personal level.

Part of what made the classical gap look unbridgeable was a framing error: looking only at the bottom (neutral-seeming physics) and top (explicitly normative culture) of the hierarchy, while ignoring the gradient of proto-preference between them. Recovering that gradient shows that the “is” was never value-free, and so closes the gap between fact and first-personal value. There is also a perceptual dimension to the framing error. The is-ought problem was formulated by thinkers perceiving the world with low ascription toward it (low $\alpha(x)$ for the entities in question) and low coupling within their own modes (low $\kappa$ ) — the mechanistic stance that factorizes fact from value, perception from affect, description from evaluation. Under high ascription and high coupling, the separation does not present itself with the same force: perceiving something as alive already includes registering its flourishing or suffering as mattering. The participatory perceiver does not experience two realms needing a bridge, because that stance never split them. This is a fact about perceptual configuration, not about reality. The viability gradient is there regardless of how one perceives.

But none of this delivers what the framework’s later ethics actually needs. The perceptual stance closes the felt gap between fact and value for the perceiver, and the valence argument closes the gap between fact and disvalue for the system that has the valence. Both are agent-relative. The ethics deployed downstream — that a creature’s suffering obligates others to act, that suffering is bad full stop and not merely bad-for-the-sufferer — requires agent-neutral disvalue. And nothing in “felt as mattering from inside” bridges bad-for-X to bad-period-and-you-ought-to-act. So the gap does not vanish. It migrates. The old gap between description and prescription relocates to a new gap: between first-personal disvalue (real, structural, settled by the identity thesis) and third-personal obligation (not entailed by it). This relocated gap the framework does not close. It can show that suffering is real and that it matters to the sufferer; it cannot, from these premises alone, derive that it must matter to you. Every downstream “ought” addressed to a third party therefore rests on two loads the reader should keep visible: (a) the identity thesis, and (b) an unbridged step from agent-relative to agent-neutral value. The honest claim is not that the gap is gone but that we have located precisely where it survives.

Normative Implication

Once valence is recognized as a real structural property at the experiential scale — not a projection onto neutral physics — the fact/value dichotomy closes for the system that bears the valence. “This system is suffering” is both a factual claim (about structure) and an agent-relative normative claim (this state is bad for it, by constitution rather than convention). What this does not establish is the agent-neutral claim — that the suffering is bad simpliciter and obligates others. That further step is assumed wherever the book later reasons from a creature’s suffering to your duties; it is not derived here.

Dependency note: Even the agent-relative result rests entirely on the identity thesis. If the identity thesis is wrong — if experience is something over and above cause-effect structure — then valence is a structural property without guaranteed normative weight, and even the first-personal gap reopens. And the agent-neutral conclusions the ethics needs carry, on top of that, the unbridged agent-relative→agent-neutral step. The normative force of the framework is therefore bounded twice over: by the case for the identity thesis, and by an admitted gap between first-personal disvalue and third-personal obligation. This is why Part II’s honest treatment of that thesis (including its unverifiability) matters: the normative conclusions inherit whatever uncertainty attaches to both.

The attention-as-leverage framework developed above deepens this picture. If attention is the high-leverage variable steering the trajectory, and values guide attention — what is attended to is what is cared about, what is ignored is what is not — then values are not epiphenomenal commentary on a value-free physical process. They are causal participants, operating at the point of maximum leverage. A system’s “oughts” (what it values, what it attends to, what it measures) thereby exert outsized influence on which trajectory it follows through state space. This is not the claim that wishing makes it so. The a priori distribution is still physics; no value chooses uncaused. But the effective distribution — the product of physics and measurement — depends on the measurement distribution, and the measurement distribution is shaped by values. In this compatibilist sense, “ought” is not a separate domain from “is.” Ought is a component of the high-leverage mechanism that biases which “is” the system inhabits.