Chapter 3: Historical Precursors — Geometry Before Geometry

RUNNING EXAMPLE — Priya’s Model

Priya starts reading ethics, surprised to find her problem already described—in different language—by people who never imagined computers. Ross’s prima facie duties map directly: she has a duty of beneficence (match patients to life-saving trials), a duty of justice (match fairly), and a duty of fidelity (to her employer’s commission). These duties point in different directions—they are vectors, not scalars. She underlines a passage in Sen: a person’s advantage is not a matter of resources but of what they can do and be. Sen was describing capability as a multi-dimensional space. Priya’s TrialMatch score does exactly what Sen warned against: it collapses capabilities into a rank ordering.

Introduction: Structure Before Formalism

The mathematical apparatus of differential geometry was developed in the nineteenth and early twentieth centuries, reaching its canonical form in Einstein’s general relativity and the gauge theories of Yang–Mills. Moral philosophy predates this development by millennia. Yet the structural insights that geometric ethics formalizes — transformation behavior, multi-dimensional interdependence, coordinate invariance, the distinction between intrinsic and perspectival properties — have appeared throughout the history of ethics in various guises.

This chapter traces a genealogy of proto-geometric thinking in moral philosophy. The claim is not that Aristotle or Kant secretly knew differential geometry, but that they grappled with phenomena that resist scalar treatment and developed conceptual tools that, in retrospect, capture aspects of geometric structure. Reading these thinkers through a geometric lens both illuminates their insights and shows that the framework developed in this book has deep roots in the philosophical tradition.

We proceed roughly chronologically, though the ordering also reflects increasing mathematical sophistication in the proto-geometric concepts.

Aristotle: The Doctrine of the Mean as Context-Sensitive Calibration

Aristotle’s Nicomachean Ethics presents virtue as a mean (μεσότης) between extremes of excess and deficiency. Courage lies between recklessness and cowardice; generosity between prodigality and miserliness; proper pride between vanity and undue humility. The virtuous person hits the mean “at the right times, with reference to the right objects, towards the right people, with the right motive, and in the right way” (1106b21).

This is emphatically not a scalar doctrine. Aristotle explicitly rejects the idea that virtue is a single quantity to be maximized:

“It is no easy task to find the middle… anyone can get angry — that is easy — or give or spend money; but to do this to the right person, to the right extent, at the right time, with the right motive, and in the right way, that is not for everyone, nor is it easy.” (1109a26)

The mean is not a fixed point on a line but a context-dependent calibration across multiple dimensions. What counts as courage depends on the situation (battlefield vs. sickroom), the agent’s role (soldier vs. physician), the stakes involved, and the alternatives available. The mean for one person in one situation may be quite different from the mean for another person in a different situation.

Geometric Reading

In geometric terms, Aristotle’s mean is a section of a fiber bundle over the space of situations. At each point x in situation-space S, there is a fiber R_x of possible responses, and the virtuous response is determined by the local structure of the situation — not by a global, context-free rule.

More precisely, let S be the space of ethically relevant situations and let R be the space of possible responses. A character trait is a section σ:S→E of a fiber bundle π:E→S with fiber R, assigning a response to each situation. The virtuous character trait σ^* is the one that, at each point, selects the response appropriate to that situation’s specific configuration.

The “right time, right object, right person, right motive, right way” are coordinates on S. Virtue requires sensitivity to all of them. A scalar theory would say: “maximize courage” or “minimize cowardice.” Aristotle says: the courageous response is a function of the local coordinates, not a global maximum.

This is proto-geometric because it recognizes that ethical evaluation is:

Multi-dimensional (multiple “right X” conditions)

Context-dependent (the mean varies with situation)

Not reducible to optimization of a single quantity

What Aristotle lacks is the mathematical apparatus to describe how the mean transforms as we change coordinates — how the courageous response in one framing relates to the courageous response in another framing of the same situation. Differential geometry provides exactly this.

The Doctrine of the Mean as a Metric Condition

There is a deeper geometric reading. The mean is defined relative to us — not the arithmetic mean of the extremes, but the mean “relative to us” (πρὸς ἡμᾶς). This suggests that the moral space has a metric structure that varies with the agent.

If we represent character traits as vectors in a space of dispositions, then “excess” and “deficiency” are directions away from the virtuous center. But what counts as excess depends on the metric: a step that is “too far” for one agent may be “not far enough” for another, because their metrics differ.

Formally, let g_μν(a) be the metric tensor on disposition-space, parameterized by agent a. The mean for agent a is the point equidistant (under g(a)) from the extremes. Different agents, with different metrics, locate the mean at different points.

This reading explains Aristotle’s insistence that virtue cannot be taught by rule. Rules are coordinate-dependent; the mean is metric-dependent. Without knowing the agent’s metric — their capacities, circumstances, history — one cannot specify the mean in advance.

In the language of Chapter 5, the agent-dependence of the mean corresponds to the fiber bundle structure of the moral manifold: the agent’s perspective lives in a fiber over the base situation, and the metric may vary along the fiber direction.

Kant: The Categorical Imperative as an Invariance Condition

Kant’s moral philosophy appears, at first glance, maximally anti-geometric. The categorical imperative demands universal laws, applicable to all rational beings regardless of circumstance. “Act only according to that maxim whereby you can at the same time will that it should become a universal law” (Groundwork, 4:421). What could be more scalar than a single test applied uniformly to all actions?

But look again. The categorical imperative is not a command to maximize a quantity. It is a constraint on the form of permissible maxims: only those maxims that can be universalized without contradiction are morally permissible.

Geometric Reading

In geometric terms, the categorical imperative is an invariance condition. It asks: which maxims remain valid under a specific transformation — the transformation from “I, in my particular circumstances” to “any rational being in relevantly similar circumstances”?

Let T be the transformation that generalizes a maxim from first-personal to universal form. A maxim m is permissible if and only if:

T(m)=m

That is, the maxim is a fixed point of the universalization transformation. Maxims that change under T — that work for me but fail when universalized — are impermissible.

This is structurally identical to how physicists identify scalars (quantities invariant under coordinate transformations) and more generally tensors (quantities that transform in specific lawful ways). Kant is asking: which moral claims are invariant under the transformation from particular to universal perspective?

The parallel is not superficial. In physics, the laws of nature must be the same in all reference frames — this is the principle of general covariance. In Kantian ethics, the laws of morality must be the same for all rational agents — this is the categorical imperative. Both are invariance conditions that constrain the form of legitimate laws.

This is also the structural ancestor of what Chapter 5 calls the Bond Invariance Principle (BIP): moral evaluations must be invariant under admissible re-descriptions of the situation. Kant discovered the principle; the geometric framework provides the transformation group.

The Kingdom of Ends as a Symmetry Group

Kant’s “kingdom of ends” deepens the geometric reading. In the kingdom of ends, every rational being is both legislator (author of universal laws) and subject (bound by those laws). The moral community is defined by the symmetry between these roles.

In mathematical terms, the kingdom of ends is closed under the transformation that swaps legislator and subject. If a law L is valid, then the transformed law T(L) — where agent and patient are exchanged — must also be valid. This is a symmetry condition on the structure of moral laws.

Symmetry conditions of this form are the hallmark of geometric equations. Maxwell’s equations are invariant under Lorentz transformations; Einstein’s field equations are invariant under general coordinate transformations. Kant’s moral laws are invariant under the permutation of rational agents.

What Kant identifies, without the mathematical language, is that moral objectivity is transformation invariance. A moral claim is objective not because it corresponds to some moral fact “out there,” but because it remains valid under all admissible transformations of perspective. This is the same structural insight that animates gauge theory in physics — and that the Bond Invariance Principle makes explicit in the ethical domain.

Ross: Prima Facie Duties as Vector Components

W.D. Ross’s The Right and the Good (1930) introduced the concept of prima facie duties: duties that are binding unless overridden by stronger duties. We have prima facie duties of fidelity (keeping promises), reparation (making amends), gratitude, justice, beneficence, self-improvement, and non-maleficence.

These duties can conflict. A promise to meet a friend may conflict with an opportunity to prevent harm to a stranger. When they conflict, we must judge which duty is stronger in this particular situation — a judgment Ross calls the determination of our actual duty.

Geometric Reading

Ross’s prima facie duties are components of a moral vector. Each duty type corresponds to a dimension of the obligation space:

O=(O_fidelity,O_reparation,O_gratitude,O_justice,O_beneficence,O_improvement,O_{non-maleficence})

In any given situation, each component has some magnitude (possibly zero). The actual duty is some function of these components — but, crucially, not a simple sum.

Ross explicitly rejects the utilitarian move of reducing all duties to a single scalar (utility). He also rejects the idea that there is a fixed priority ordering among duty types. Instead, the determination of actual duty is a matter of judgment that weighs the components contextually.

In geometric terms, Ross is grappling with the problem of contraction: how do we go from a multi-component vector to a single action-guiding prescription? His answer — that there is no mechanical rule, only trained judgment — reflects the fact that different situations call for different contraction operations. This is precisely the structure developed in Chapter 6: the scalar S=I_μO^μ depends on the interest covector I_μ, which varies with the agent and context.

The Interaction Problem

Ross’s framework faces a difficulty that has occupied his commentators for nearly a century: how do prima facie duties interact? If I have a strong duty of fidelity and a weak duty of beneficence, does the fidelity duty simply win? Or do they combine in some more complex way?

The geometric framework suggests an answer. Duties are not merely magnitudes to be compared; they have directions in moral space. Two duties may be:

Aligned: both point in the same direction (keeping a promise that also helps someone)

Orthogonal: independent, neither reinforcing nor conflicting (a promise to one person, a beneficence opportunity involving another)

Opposed: pointing in opposite directions (a promise that requires harming someone)

The combination of duties is then a vector sum, with the geometry determining how they add:

O_actual=∑_i^​ O_i

When duties are aligned, their magnitudes add. When orthogonal, they combine by the Pythagorean theorem (their resultant is √(O₁²+O₂²), not O₁+O₂). When opposed, they partially cancel.

This explains a phenomenon that has puzzled Rossian commentators: why strong orthogonal duties can coexist without conflict (I can keep my promise and help the stranger, if the actions lie in independent dimensions), while even weak opposed duties create tension (breaking even a minor promise to prevent trivial harm still feels like a genuine moral loss — the cancelled component leaves what Chapter 15 will call moral residue).

Ross’s “judgment” can now be understood as sensitivity to the geometry of the duty configuration — something that resists codification in scalar terms but has definite mathematical structure.

Rawls: The Original Position as a Symmetry Constraint

John Rawls’s A Theory of Justice (1971) proposes that principles of justice are those that would be chosen by rational agents behind a “veil of ignorance” — not knowing their place in society, their natural talents, or their conception of the good. The original position is a thought experiment designed to identify principles that are fair because they are chosen without knowledge of how they will affect the chooser.

Geometric Reading

The original position is a symmetry condition. By removing knowledge of particular position, it forces the choice of principles that are invariant under permutation of agents. If a principle benefits position A at the expense of position B, it cannot be chosen behind the veil, because the chooser might turn out to occupy position B.

Formally, let π be a permutation of social positions. A principle P is admissible in the original position if and only if:

π(P)=P for all permutations π

This is precisely the condition for a symmetric tensor — a tensor unchanged under index permutation.

Rawls’s two principles of justice can be understood as the unique symmetric solution (up to specification) to the problem of social cooperation. The first principle (equal basic liberties) is symmetric by construction: everyone gets the same liberties. The second principle (difference principle) permits inequalities only if they benefit the worst-off position — a symmetric condition because the worst-off position is defined relative to the structure of positions, not to any particular occupant.

The Metric of the Original Position

The original position also implicitly specifies a metric on the space of social positions. The difference principle uses a maximin criterion: maximize the minimum position. This is equivalent to choosing a specific metric in which distance is measured by the worst-off coordinate.

In geometric terms, the Rawlsian metric is the supremum metric:

d(x,y)=max_i|x_i-y_i|

This gives decisive weight to the worst-off dimension. Alternative metrics yield different principles:

The utilitarian metric ( L¹ ): d(x,y)=∑_i^​ |x_i-y_i| , weighting all positions equally

The Euclidean metric ( L² ): d(x,y)=√(∑_i^​ (x_i-y_i)²) , weighting by squared deviations

The prioritarian metric: a weighted L^p metric giving greater weight to worse-off positions, interpolating between utilitarian and Rawlsian

Rawls’s argument against utilitarianism can be read as an argument about which metric is appropriate for justice. The utilitarian metric permits sacrificing some positions for aggregate gain; the Rawlsian metric does not. This is a substantive geometric claim about the structure of fair social evaluation.

The geometric framework (Chapters 6 and 9) makes this explicit: different ethical theories correspond to different metric tensors on the moral manifold. The metric is not given by the mathematics; it encodes a substantive commitment. What geometry provides is the vocabulary for stating the commitment precisely — and the tools for deriving its consequences.

Sen and Nussbaum: Capabilities as a Basis for Moral Space

Amartya Sen and Martha Nussbaum developed the capabilities approach as an alternative to both utilitarian welfare measures and Rawlsian primary goods. The core idea is that what matters morally is not subjective well-being (utility) or objective resources (income, rights) but capabilities: the real freedoms people have to achieve “functionings” they have reason to value.

Sen identifies a plurality of capabilities — life, bodily health, bodily integrity, senses/imagination/thought, emotions, practical reason, affiliation, relation to other species, play, control over one’s environment — that are irreducibly plural. They cannot be reduced to a single scalar measure without violence to their distinct characters.

Geometric Reading

The capabilities are basis vectors for a moral space. Each capability defines an independent dimension along which a person’s life can go well or badly. A person’s situation is a vector in capability space:

c=(c_life,c_health,c_integrity,c_senses,c_emotions,c_reason,c_affiliation,c_nature,c_play,c_control)

This is explicitly multi-dimensional. Sen insists that capabilities cannot be aggregated into a single index without loss of essential information — precisely the claim that moral evaluation requires richer structure than a scalar.

The connection to the nine-dimensional moral manifold of Chapter 5 is suggestive. The nine dimensions derived from the 3 × 3 matrix of ethical scopes and epistemic modes do not map one-to-one onto Sen’s capabilities, but the structural move is the same: identifying the basis of moral space by systematic analysis of what dimensions are needed to distinguish morally distinct situations. Sen proceeds by philosophical reflection on human flourishing; the geometric framework proceeds by combinatorial analysis of ethical scope and epistemic mode. That both arrive at a finite-dimensional space of roughly similar cardinality (∼10) is at least noteworthy.

The Incompleteness Thesis

Sen argues that comparative judgments of capability sets are incomplete: we can often say that one situation is better than another along some dimensions and worse along others, without being able to say which is better overall. This incompleteness is not a failure of the theory but a feature of moral reality.

In geometric terms, this is the claim that the moral metric is degenerate or partial. Not all vectors can be compared in length. Given two capability vectors c₁ and c₂, we may have:

c₁ⁱ>c₂ⁱ along some dimensions

c₁^j<c₂^j along other dimensions

No legitimate contraction to an overall comparison

This is the geometric signature of incommensurability. A scalar theory would force a comparison (by summing or by lexical priority); Sen’s theory preserves the genuine incompleteness. The geometric framework represents this as a structural feature of the moral manifold — a degenerate metric — rather than a failure of analysis.

Nussbaum’s Threshold and Stratum Structure

Nussbaum modifies the capabilities approach by introducing thresholds: minimum levels of each capability below which a life is not fully human. This introduces stratum structure into capability space.

Below the threshold, we are in a different moral regime — one where the imperative is to raise capabilities to the threshold level, admitting no trade-offs. Above the threshold, trade-offs and choices become permissible. The threshold is a stratum boundary separating regions with different moral rules.

This is proto-stratified geometry. Nussbaum’s capability space is not a smooth manifold but a stratified space with distinguished hypersurfaces (the thresholds) where the rules change discontinuously. In the vocabulary of Chapter 8, Nussbaum’s thresholds are codimension-1 stratum boundaries; the regime below the threshold (where the only permissible direction is “up”) is an absorbing stratum with restricted tangent directions.

The structural parallel to the empirical findings described in Chapter 17 is striking: analysis of the Dear Abby corpus reveals that qualifying conditions — abuse, danger, incapacity — function as nullifiers that override all other moral considerations. These are absorbing strata in exactly Nussbaum’s sense: once you cross the threshold (a child is being harmed, a spouse is abusive), the moral calculus collapses to a single imperative, admitting no trade-offs.

Hohfeld: Jural Relations as Discrete Structure

Wesley Newcomb Hohfeld’s Fundamental Legal Conceptions (1913, 1917) introduced a taxonomy of legal relations that has become foundational in jurisprudence. Hohfeld identified eight fundamental jural positions organized into two squares of opposition:

Within each square, the relations are connected

by correlatives

(Claim↔Duty, Liberty↔No-claim)

and opposites

(Claim↔No-claim, Duty↔Liberty).

Geometric Reading

Hohfeld’s taxonomy is remarkable because it is entirely discrete — there are no continuous parameters, no “degrees” of claim or liberty. The positions are qualitative states, and the transitions between them are all-or-nothing.

In geometric terms, the Hohfeldian positions define the vertices of a discrete symmetry structure. The correlative and opposition relations generate a group of transformations on these vertices — and that group turns out to be D₄, the dihedral group of order 8, which is also the symmetry group of the square.

This identification is not metaphorical. The eight Hohfeldian positions map onto the eight elements of D₄ (four rotations, four reflections), and the correlative and opposition operations correspond to specific group elements. The transitions between positions — obligation flipping to liberty, claim dissolving to no-claim — are group actions.

The significance for geometric ethics is twofold:

Discrete symmetry in moral space. The Hohfeldian relations show that moral space contains irreducibly discrete structure alongside its continuous dimensions. The moral manifold is not purely smooth; it has a discrete component that no amount of refinement can smooth away. This is the mathematical content of stratification (Chapter 8).

Empirical confirmation. Analysis of natural moral reasoning (Chapter 17) reveals that these transitions behave as step functions: a qualifying phrase (“you promised”) flips the moral state from liberty to obligation with no intermediate values. The correlative symmetry — Claim↔Duty at 87%, Liberty↔No-claim at 82% — is a structural invariant of the moral manifold, detectable in data.

Hohfeld, working in the formalist tradition of legal analysis, discovered discrete geometric structure in the moral domain nearly a century before the mathematical framework existed to describe it.

Moral Uncertainty: Superposition and Structured Credence

Recent work in moral philosophy has focused on moral uncertainty: what should we do when we are uncertain which moral theory is correct? If I am 60% confident in utilitarianism and 40% confident in deontology, and they recommend different actions, what should I choose?

Various approaches have been proposed: “my favorite theory” (act on whichever theory you find most plausible), “maximize expected moral value” (weight each theory’s recommendation by credence), and more sophisticated methods that account for intertheoretic comparisons.

Geometric Reading

The structure of moral uncertainty is naturally vectorial. An agent under moral uncertainty is not in a definite moral state but in a weighted combination of moral states.

Let |U⟩ represent the evaluation under utilitarianism and |D⟩ the evaluation under deontology. An agent with 60% credence in utilitarianism is in the state:

|ψ⟩=√(0.6) |U⟩+√(0.4) |D⟩

This is a vector in a theory space, not a scalar. The agent’s moral situation cannot be captured by a single number (overall credence) but requires specification of the full vector — its components, their relative phase, and their transformation behavior under changes of basis.

When the agent acts, the vector must be contracted to a scalar choice — but the contraction is not unique. Expected value maximization is one contraction; other approaches perform different contractions. The choice of contraction is itself a moral commitment, distinct from the choice of theory.

The analogy with quantum superposition is structural, not merely metaphorical. In quantum mechanics, a state vector |ψ⟩ encodes the full information about a system; measurement collapses it to a definite outcome. In moral uncertainty, the credence vector encodes the full structure of the agent’s evaluative commitments; action collapses it to a definite choice. The mathematics of superposition, projection, and measurement carries over to the moral domain — a connection developed formally in Chapter 13 (Quantum Normative Dynamics).

Moral Hedging as Covariance Sensitivity

The sophisticated treatment of moral uncertainty involves hedging: choosing actions that are reasonably good under multiple theories, even if not optimal under any single one. This is analogous to portfolio diversification in finance.

In geometric terms, hedging is sensitivity to the covariance structure of moral uncertainty. Let Σ be the covariance matrix of the agent’s moral beliefs. Let Δ_a=(a_U-a_D) be the vector of differences between what each theory recommends for action a. The “risk” of action a is:

σ_a²=Δ_a^TΣ Δ_a

Actions with low σ² are robust to moral uncertainty; actions with high σ² are risky bets on particular theories. This is the moral analogue of portfolio variance — and it requires the full tensor structure of uncertainty, not just scalar credences.

Synthesis: What the Precursors Share

Across two and a half millennia, these thinkers share a recognition that moral reality resists scalar reduction:

None of these thinkers used the language of differential geometry. But they all developed conceptual tools to handle phenomena that geometry formalizes:

Multi-dimensionality: Moral evaluation involves multiple independent considerations that cannot be reduced to one.

Transformation behavior: What happens to moral claims when we shift perspective, permute agents, or change framing?

Metric structure: How do we compare values, measure moral distance, identify incommensurability?

Context-dependence: The same abstract principle yields different concrete prescriptions in different situations.

Structured combination: When moral considerations combine, they do so geometrically (with alignment, orthogonality, opposition), not arithmetically.

Discrete transitions: Some moral changes are discontinuous — phase transitions, not gradual slides.

What the Geometric Framework Adds

If these insights are already present in the tradition, what does the geometric framework add?

Precision. The tradition offers metaphors and intuitions; the framework offers definitions and theorems. “Duties interact” becomes “obligations combine as vectors under the moral metric g_μν.” “The mean is relative to us” becomes “virtue is determined by an agent-parameterized metric on disposition-space.” Precision enables analysis, criticism, and extension.

Unification. The tradition offers disparate insights from apparently incompatible systems. The geometric framework reveals common structure beneath surface disagreement. Aristotle’s context-sensitivity and Kant’s universalizability are not opposed but complementary: both constrain the transformation behavior of moral claims, Aristotle at the level of the fiber (agent-dependence), Kant at the level of the base (situation-independence). Ross’s plurality and Rawls’s symmetry are both structural features of the moral tensor — Ross specifying its components, Rawls specifying its symmetry group.

New questions. The framework suggests questions the tradition did not ask:

What is the signature of the moral metric — is it positive-definite (Euclidean), indefinite (Lorentzian), or degenerate?

What are the symmetries of moral space — which transformations leave the structure invariant, and what conservation laws do they imply?

What curvature does moral space have — how does parallel transport of an obligation around a closed loop change the obligation vector?

What is the holonomy group of moral connections — which transformations arise from cycling through moral perspectives?

These are not idle questions. Chapter 10 develops the curvature of moral space and shows that non-trivial holonomy corresponds to path-dependent moral evaluation — the geometric content of the familiar intuition that “where you start affects where you end up” in moral reasoning. Chapter 12 shows that the symmetry of re-description invariance implies a conservation law for harm — the geometric analogue of Noether’s theorem.

Computability. The tradition offers wisdom; the framework offers implementation. Tensors can be represented in computers; geometric operations can be implemented in algorithms; geometric equations can be solved numerically. A philosophical framework, however profound, is inert if it cannot be computed. The geometric framework bridges the gap between moral insight and moral engineering — not by replacing judgment with calculation, but by making the structure on which judgment operates explicit and manipulable.

Conclusion: The Geometry Was Always There

The history of moral philosophy is, in significant part, a struggle against the reductionism of scalar ethics. Each thinker we have examined recognized that moral reality has structure that scalars cannot capture, and developed conceptual tools to articulate that structure.

Geometric ethics is not a break from this tradition but its continuation — and, in a mathematical sense, its formalization. The conceptual tools developed by Aristotle, Kant, Ross, Rawls, Sen, Nussbaum, Hohfeld, and theorists of moral uncertainty find their natural mathematical expression in the language of manifolds, tensors, metrics, connections, and stratified spaces.

This is not to say that the tradition was secretly doing mathematics. It is to say that mathematics, properly understood, is the science of structure — and the structures that matter for ethics are precisely those that differential geometry was developed to describe: transformation behavior, multi-linear combination, coordinate invariance, metric geometry, and the interplay of continuous and discrete structure.

The tradition provides the insights. The framework provides the language. Together, they enable a moral philosophy that is both faithful to the complexity of ethical life and precise enough to be stated, tested, and refined.

In the chapters that follow, we develop the framework in full. But we do so in the company of these predecessors, whose insights we are formalizing, not replacing.

❖

Aristotle sought the mean.

Kant sought the universal.

Ross sought the balance.

Rawls sought the fair.

Sen sought the capable.

Hohfeld sought the discrete.

They were all, in their different ways, seeking the geometry.