Potency¶

Quantify how much drug a given effect costs by reading one coordinate off the dose-response curve: the dose that produces half the maximal response (the ED50), so a lower value means less drug suffices.

Core Idea¶

Potency is the pharmacological scalar that quantifies how much drug is needed to produce a given magnitude of effect — specifically, the dose required to produce half the maximal response a drug can elicit, termed the ED50 (effective dose fifty). A drug with a lower ED50 is more potent: less of it achieves the reference effect. The definition is relational: potency is always potency at a specific reference point on the dose-response curve, typically the midpoint of the sigmoid, not a property of the drug in isolation but of the drug-receptor-tissue system at that operating point.

The load-bearing structural commitments are three. First, potency presupposes a dose-response curve — the sigmoidal relationship between log dose and effect magnitude that characterises the drug's behaviour in a defined biological system. The curve's shape (slope, midpoint, plateau) is the empirical object; potency is a single derived coordinate of it. Second, potency is explicitly distinct from efficacy: efficacy is the maximum response the drug can produce at any dose (the plateau of the sigmoid, the ceiling), while potency is where on the dose axis the drug reaches the half-maximum. A drug may be highly potent — producing its effect at nanomolar concentrations — yet have low efficacy if its ceiling response is modest; another drug may require millimolar concentrations to be clinically useful (low potency) yet achieve a full response (high efficacy). Morphine and codeine illustrate the point: morphine is roughly ten times more potent as an analgesic than codeine (the ED50 for analgesia is reached at a much lower dose), but both are full agonists at opioid receptors and can in principle achieve equivalent maximum analgesia given enough dose. Selecting between them for a given clinical context depends on which dimension — potency, efficacy, or the combined profile — matters for that use. Third, the in vitro correlate of potency is receptor-binding affinity: the drug's affinity for its receptor (expressed as the dissociation constant Kd or its inverse) is a primary determinant of how little drug is required to produce a given receptor-occupancy and, from that, a given functional effect, making receptor pharmacology the mechanistic basis from which potency is predicted and engineered.

Structural Signature¶

Sig role-phrases:

the dose input — the quantity of drug or agent administered, the variable on the dose axis
the sigmoidal dose-response curve — the empirical object: log dose mapped to effect magnitude in a defined drug-receptor-tissue system, with its own slope, midpoint, and plateau
the half-maximal reference effect — the canonical operating point (50% of maximal response) at which the dose is read, making potency relational and not a free-floating molecular property
the potency scalar (ED50) — the derived coordinate: the dose producing the reference effect, so a smaller ED50 means higher potency (less drug suffices)
the receptor-affinity predictor — the in vitro correlate (dissociation constant Kd, or its inverse), the mechanistic basis from which the ED50 is predicted and engineered
the potency-versus-efficacy separation — the load-bearing guarantee: the dose-axis coordinate (ED50) held strictly distinct from the ceiling coordinate (maximal achievable response), two independent axes of one curve
the reference-point boundary — potency rankings are comparable only at the same reference effect in the same tissue system; a potency comparison says nothing about achievable ceiling
the pharmacology-bound intervention family — the characteristic limitation: select-higher-potency-compound, raise receptor affinity, adjust dosing regimen — levers that do not survive extraction past the dose-response precondition

What It Is Not¶

Not efficacy. Potency is where on the dose axis a drug reaches its half-maximal effect (the ED50); efficacy is how high the plateau sits — the maximum response achievable at any dose. They are independent coordinates of the same curve: a drug can be highly potent yet low in efficacy, or require large doses yet reach a full response. A potency comparison says nothing about achievable ceiling.
Not a property of the molecule alone. Potency is relational — always potency at a specified reference point on the dose-response curve, conventionally the half-maximal effect, in a defined drug-receptor-tissue system. A potency ranking made at a different reference effect or in a different tissue is not directly comparable, so treating it as a free-floating molecular constant is a category error.
Not the lay sense of "stronger." "Stronger" fuses two distinct things — working at a smaller dose (higher potency) and reaching a larger maximum effect (higher efficacy). Collapsing them into a single strength ranking is precisely the conflation the concept exists to forbid; the coherent question is which axis matters for the use case.
Not the maximum effect achievable. A potency advantage does not buy a higher ceiling. Increasing the dose of a more potent but low-efficacy drug climbs toward its plateau but cannot exceed it, so the more potent drug may be unable to reach a response that a less potent, higher-efficacy drug attains.
Not the speed or duration of action. Potency concerns how little drug is needed for a given fractional effect, not how fast the effect appears or how long it lasts. Onset and persistence are pharmacokinetic properties of a separate axis; a highly potent drug can be slow or short-acting.
Not receptor-binding affinity itself. Affinity (the dissociation constant Kd, or its inverse) is the in vitro correlate and mechanistic cause of potency — it predicts where the ED50 will fall — but potency is the functional dose-for-effect scalar read off the whole-system response, not the binding constant. Tissue and signal-transduction factors can separate the two.

Scope of Application¶

Because potency is a scalar derived from a dose-response transfer function at a reference operating point, not a mechanism, it applies wherever its precondition holds — a defined sigmoidal dose-response curve in a drug-receptor-tissue system with a canonical half-maximal reference effect; the habitats below are real uses of the identical construct (the same ED50-type reading, potency-versus-efficacy split, and Kd predictor), variants of one pharmacological substrate. The bare leverage-at-an-operating-point shape that recurs cross-domain belongs to the parent primes leverage / sensitivity / elasticity (plus ceiling), not to the potency vocabulary, which is over-read past the dose-response precondition.

Clinical and receptor pharmacology — the home use: ED50 for therapeutic effect, comparative-potency rankings (morphine ~10× codeine), and receptor-binding affinity (Kd) as the mechanistic predictor of where the dose-response curve sits.
In vitro and screening pharmacology — IC50 for inhibition and EC50 in functional assays, the same half-maximal construct used to rank compounds in drug discovery before whole-organism testing.
Toxicology and risk assessment — LD50 for lethality, threshold doses, and potency ranking of carcinogens (the carcinogenic-potency database's dose-per-tumour scalar), the toxicological reading of the identical curve coordinate.
Endocrinology and hormone signalling — hormone potency in receptor binding and downstream cascade activation, the same potency-versus-efficacy separation applied to endogenous agonists.
Pharmacodynamic modelling — fitting the sigmoidal dose-response (slope, midpoint, plateau) so that ED50 and the ceiling are read as separate coordinates and the dose for a target fractional response is forecast.

Clarity¶

Naming potency makes the loose clinical word "stronger" precise and, in doing so, separates two ideas the lay term fuses. The single most operationally important thing the concept clarifies is that potency is not efficacy: "this drug is stronger" might mean it works at a smaller dose (lower ED50, higher potency) or that it can reach a larger maximum effect (higher ceiling, greater efficacy), and these are independent axes of the same dose-response curve. Holding them apart is what lets a pharmacologist say that morphine is more potent than codeine while recognising that both are full agonists capable in principle of the same maximal analgesia — a sentence that is incoherent until potency and efficacy are pried apart. Without the distinction, drug comparison collapses into a single ill-defined "strength" ranking; with it, the practitioner can ask the correct question, which is which axis matters for this clinical context — dose volume and side-effect-per-dose when potency dominates, achievable ceiling when efficacy dominates.

The concept also disciplines comparison by insisting potency is read at a reference point on the curve, conventionally the half-maximal effect, rather than treated as a free-floating property of the molecule. That makes "more potent" a checkable, quantified claim — a statement about the dose axis at a defined operating point — instead of an impression, and it makes potency rankings reproducible across compounds and studies. Finally, by tying potency to receptor-binding affinity as its in vitro correlate, the concept connects a clinical scalar to a mechanistic cause: the sharper question it licenses is not merely "how potent is this drug?" but "how little drug is required to occupy enough receptor for the functional effect, and can affinity be engineered to lower the ED50?" — turning potency from an observed number into something predictable and designable from receptor pharmacology.

Manages Complexity¶

Comparing drugs to choose among them is, in full detail, a comparison of entire dose-response curves — each a sigmoid with its own slope, midpoint, and plateau, measured in a particular drug-receptor-tissue system — and the lay vocabulary of "stronger" offers no way to rank them, fusing several independent features of those curves into one undifferentiated impression. Potency compresses one whole axis of that comparison to a single derived coordinate: the dose at which the drug reaches half its maximal effect, the ED50, read at a fixed reference point on the curve rather than treated as a free-floating property of the molecule. An entire curve's position along the dose axis collapses to one number, and "more potent" becomes a checkable, reproducible claim — a smaller ED50 — instead of an impression, so a shelf of compounds can be ranked by how little drug each requires to reach the reference effect.

The decisive part of the compression is that potency is held apart from efficacy, so that what the pharmacologist tracks is not one "strength" but two independent scalars read off the same sigmoid: the ED50 (where on the dose axis the half-maximum falls) and the ceiling (the maximal response the drug can produce at any dose). Those two coordinates summarise the comparison, and drug selection reads off a clear branch from them. When dose volume or side-effect-per-dose is what matters, the potency axis dominates and the lower-ED50 compound is preferred; when the achievable maximum effect is what matters, the efficacy axis dominates and the higher-ceiling compound is preferred — and the two can diverge, a drug being highly potent yet low in efficacy, or requiring large doses yet reaching a full response, so the branch genuinely has to be taken rather than assumed. The morphine-versus-codeine comparison is exactly this: a roughly tenfold potency difference read off the dose axis, against a shared full-agonist ceiling read off the plateau, the choice falling to whichever axis the clinical context selects. A third reduction connects the clinical scalar to a mechanistic cause: receptor-binding affinity (the dissociation constant Kd, or its inverse) is the in vitro correlate of potency, so the question of how little drug suffices reduces to how tightly the drug occupies its receptor — making potency not merely an observed number but one predictable and engineerable from receptor pharmacology. The high-dimensional task of comparing curves collapses to tracking two coordinates plus an affinity predictor, with selection read off which axis the use case makes binding.

Abstract Reasoning¶

Potency licenses a set of inferential moves within pharmacology and toxicology, all reading a drug-comparison or dosing question off the dose-response curve and its two independent coordinates — where the half-maximum falls (ED50) and how high the plateau sits (efficacy).

Diagnostic — infer the operating mechanism and the right axis from where a drug sits on the curve. The signature move is to decompose the lay claim "this drug is stronger" into the two distinct things it could mean and read which one the curve actually supports: a smaller ED50 (the half-maximal effect reached at a lower dose, higher potency) versus a higher ceiling (a larger maximal response, greater efficacy). From a measured curve, the analyst infers a compound's profile — nanomolar ED50 with a modest plateau is read as high-potency, low-efficacy; a high ED50 with a full plateau as low-potency, full-efficacy — and these readings are independent, so one cannot be substituted for the other. Running the inference toward mechanism, a low ED50 is read back to tight receptor binding: because receptor-binding affinity (the dissociation constant Kd, or its inverse) is the in vitro correlate of potency, a drug requiring little to occupy enough receptor for the functional effect is inferred to have high affinity, and a measured affinity predicts where the ED50 will fall before the functional assay is run. The morphine-versus-codeine comparison is exactly this diagnostic: a roughly tenfold lower analgesic ED50 for morphine is read off the dose axis, while a shared full-agonist plateau is read off the ceiling, so "morphine is more potent but both can reach equivalent maximal analgesia" is a coherent, curve-grounded inference rather than a contradiction.

Interventionist — choose the lever for the axis the clinical context selects, and engineer potency from affinity. Because potency and efficacy are separable, the concept routes a drug-selection decision to whichever axis the use case makes binding, with a predicted consequence. When dose volume or side-effect-per-dose is what matters, the potency axis dominates and the lower-ED50 compound is chosen, predicted to deliver the reference effect at less drug and so with less mass-dependent burden; when the achievable maximum effect is what matters, the efficacy axis dominates and the higher-ceiling compound is chosen, predicted to reach a response the more potent but lower-efficacy drug cannot attain at any dose. The crucial paired non-prediction is that increasing the dose of a low-efficacy drug will not lift its response above its plateau — pushing the dose climbs toward the ceiling but cannot exceed it — so a potency advantage does not buy a higher maximum. At the design stage, the lever is receptor affinity: engineering a tighter-binding analogue is predicted to lower the ED50 (shift the curve leftward along the dose axis) without necessarily changing the plateau, making potency a designable target derived from receptor pharmacology rather than an observed accident.

Boundary-drawing — potency is read at a reference point, and only the potency axis, not the strength axis. The concept draws a strict boundary that potency is always potency at a specified reference point on the curve — conventionally the half-maximal effect — and is a coordinate of the drug-receptor-tissue system at that operating point, not a free-floating property of the molecule. This bounds comparison: "more potent" is a checkable, reproducible claim about the dose axis at a defined operating point, and a potency ranking made at a different reference effect or in a different tissue system is out of scope and not directly comparable. A second boundary separates the potency axis from the efficacy axis cleanly, so the analyst knows that a potency comparison says nothing about achievable ceiling and an efficacy comparison says nothing about dose required — collapsing them back into a single "strength" ranking is precisely the error the boundary forbids. The concept also bounds which questions potency answers: how little drug is needed for a given fractional effect, not how large the effect can become and not how fast or how long it acts.

Predictive and order-of-events. Reading a position on the sigmoid supports forward prediction along the dose axis: from the ED50 and slope, the analyst predicts the dose needed for a target fractional response and anticipates the diminishing gain as the dose climbs toward the plateau. The affinity-to-potency link supports prediction in the other direction — a structural modification that raises receptor affinity is predicted to manifest, downstream, as a lower ED50 and a leftward-shifted clinical dose-response — so the order from receptor binding to occupancy to functional effect lets the pharmacologist forecast a compound's potency from its receptor pharmacology rather than waiting on the whole-organism result.

Knowledge Transfer¶

Potency is a scalar derived from a transfer function at a reference operating point (the ED50), so the boundary to mark is construct-reach — where its dose-response apparatus is meaningful — versus over-reading. Wherever its precondition holds — a defined sigmoidal dose-response curve in a drug-receptor-tissue system with a canonical half-maximal reference effect — the construct transfers literally, carrying its full apparatus. That precondition is met across pharmacology and toxicology, where potency ports without translation: ED50 for therapeutic effect, IC50 for inhibition, LD50 for lethality, the carcinogenic-potency database's dose-per-tumour ranking, and hormone potency in receptor binding and downstream cascade activation. In all of these the same two-coordinate reading holds literally — where the half-maximum falls (ED50) versus how high the plateau sits (efficacy) — together with the potency-versus-efficacy separation and the receptor-binding-affinity (Kd) predictor of where the curve sits. The transfer is literal, not analogical, because dose-response curve, half-maximal reference effect, and receptor affinity are well-defined in each; these are variants of one pharmacological substrate (drug/toxin/hormone acting on receptors), which is exactly why the entry is domain-specific rather than a prime.

Beyond that precondition the honest verdict is shared abstract mechanism carried by a higher prime, with the named scalar over-read when imported directly. The structural content that genuinely recurs cross-domain is leverage — output-per-unit-input around an operating point — and its near relatives sensitivity (the derivative of output with respect to input at an operating point) and elasticity (the unit-free percent-change form); the dose-response curve itself transfers only as the generic response curve / transfer function. So "input efficiency for a given effect" — how much marketing spend to lift sales 1%, how much subsidy to shift adoption by X%, how many keystrokes to cause a given UI displacement — is real, but it is leverage / sensitivity / elasticity doing the work, already substrate-independent primes, and calling it "channel potency" adds no structure. The useful potency-versus-efficacy distinction does port, but as the unbundled leverage-versus-ceiling pair (per-unit-input leverage separated from maximum-achievable-output), which lives more cleanly in a combined leverage + ceiling framing and does not require the potency vocabulary to carry it. What does not survive extraction is precisely the pharmacological apparatus that gives potency its precision: the canonical half-maximal (ED50) reference effect, the agreed sigmoidal dose-response curve, and the potency-versus-efficacy split — because most non-pharmacological systems have no canonical half-maximal reference and no agreed response-curve shape, so the analyst importing "potency" is left with bare leverage and the distinction collapses into the already-available leverage-versus-ceiling one. (One adjacent pharmacological idea travels as its own distinct pattern, not as potency: therapeutic-window / operating-margin thinking — the gap between effective and toxic dose — ports to operating-band reasoning in engineering and policy, but that is a margin pattern, structurally distinct from this leverage-scalar.) The disciplined move is therefore to carry the cross-domain lesson with leverage / sensitivity / elasticity (plus ceiling for the efficacy axis), recognizing potency as the pharmacological instance — a leverage scalar specialized to dose-response substrates with a canonical reference effect. This is the boundary drawn in Structural Core vs. Domain Accent: the leverage-at-an-operating-point skeleton lifts to leverage / sensitivity / elasticity; the pharmacological accent — ED50, the sigmoid, potency-versus-efficacy, receptor affinity — stays home and is over-read when applied past its dose-response precondition, while remaining a literal measure wherever that precondition holds.

Examples¶

Canonical¶

Morphine and codeine are the textbook potency-versus-efficacy pair. Both are full agonists at the μ-opioid receptor, so given enough dose each can in principle reach the same maximal analgesia — equal efficacy, the same plateau on the dose-response sigmoid. But morphine reaches its half-maximal analgesic effect at a much lower dose: it is roughly ten times more potent, meaning its ED50 sits about a factor of ten lower on the dose axis. Illustratively, if morphine's analgesic ED50 were X mg, codeine's would be near 10X mg for the same fractional effect. The clinical shorthand "morphine is stronger" then resolves cleanly — stronger in potency (lower ED50), not in efficacy (equal ceiling).

Mapped back: The analgesic dose is the dose input; the sigmoid relating dose to analgesia is the sigmoidal dose-response curve; half-maximal analgesia is the half-maximal reference effect; morphine's roughly tenfold-lower ED50 is the potency scalar; and the shared full-agonist plateau set against the differing ED50 is the potency-versus-efficacy separation.

Applied / In Practice¶

In regulatory toxicology the same half-maximal construct is read for lethality as the median lethal dose, LD50 — the dose killing 50% of a test population — introduced by J. W. Trevan in 1927. Chemical-hazard frameworks such as the UN Globally Harmonized System classify acute toxicity by banding the oral LD50 (its most hazardous category, for instance, covers substances with an oral LD50 at or below 5 mg/kg body weight). A lower LD50 means less substance is lethal — higher toxic potency — exactly paralleling a lower ED50 meaning less drug suffices for effect, and the ranking is only comparable across substances measured by the same route and endpoint.

Mapped back: The administered amount is the dose input; the lethality curve is the sigmoidal dose-response curve; 50% lethality is the half-maximal reference effect; LD50 is the potency scalar in its toxicological guise; and the requirement that comparisons hold route and endpoint fixed is the reference-point boundary.

Structural Tensions¶

T1: Potency versus efficacy (the separable axes can be optimized against each other). Prying potency (where the half-maximum falls, ED50) apart from efficacy (how high the plateau sits) is the concept's load-bearing move, and each axis is genuinely useful — a lower ED50 delivers the reference effect at less drug and so with less mass-dependent burden. But the two are independent, and optimizing the one can mislead about the other: a highly potent, low-efficacy drug reaches its effect at nanomolar doses yet cannot exceed its ceiling at any dose, so choosing on potency alone can select a compound that never attains a response a less potent, higher-efficacy drug reaches easily. The tension is that "more potent" and "more effective" feel like the same virtue but can point at different molecules, and pushing the dose to buy more effect from a potent drug climbs toward the plateau without crossing it. Diagnostic: Does the clinical context make dose-volume/side-effect-per-dose binding (potency axis) or achievable-ceiling binding (efficacy axis) — and could optimizing one pick a drug that fails on the other?

T2: A single coordinate versus the whole curve it compresses (equal ED50 is not equal behavior). Collapsing a drug's entire dose-axis position to one number makes "more potent" checkable and reproducible — a smaller ED50 — and lets a shelf of compounds be ranked at a glance. But the ED50 is one derived coordinate of a sigmoid that also has a slope and a plateau, and two drugs with identical ED50 but different slopes behave differently across the dose range: one may be forgiving, the other switch from sub-therapeutic to maximal over a narrow window. The scalar that buys comparability discards exactly the curve-shape information that governs dosing safety away from the reference point. The tension is that the compression to a single half-maximal coordinate is the source of both the concept's rigor and its capacity to hide clinically decisive differences at other points on the curve. Diagnostic: Are the compounds being compared on ED50 alone, or does the slope (and plateau) of the full dose-response curve differ in ways the single coordinate conceals?

T3: Relational reading versus molecular-constant temptation (the number invites context-stripping). Potency is defined strictly at a specified reference point in a defined drug-receptor-tissue system — a coordinate of the drug-at-an-operating-point, not a free-floating property of the molecule — so a ranking made at a different reference effect or in a different tissue is not directly comparable. Yet the very fact that potency reduces to a single quotable number (morphine ~10× codeine) invites treating it as an intrinsic molecular constant, portable across contexts it was never measured in. The tension is that the scalar's convenience and reproducibility, which are real, are exactly what tempt the category error of detaching it from the system and reference effect that define it. A potency comparison is only as valid as the fixed route, endpoint, and tissue behind it. Diagnostic: Are the potencies being compared read at the same reference effect in the same tissue/route, or is a single ED50 being treated as a context-free molecular property?

T4: Affinity as predictor versus the binding-to-function gap (the mechanistic handle is only a correlate). Tying potency to receptor-binding affinity (Kd) is what makes it designable: affinity is the in vitro correlate and mechanistic cause, so a tighter-binding analogue is predicted to lower the ED50 before any functional assay is run, and potency becomes an engineering target rather than an observed accident. But affinity is a correlate, not potency itself — tissue and signal-transduction factors (receptor reserve, downstream amplification) can separate how tightly a drug binds from how little is needed for the functional effect, so a system with spare receptors reaches its half-maximal response well below full occupancy. The tension is that the same affinity link that makes potency predictable and engineerable will mispredict exactly where receptor pharmacology and whole-system response come apart. Diagnostic: Is functional potency being read off Kd directly, or does this system have receptor reserve or amplification that decouples binding affinity from the dose-for-effect?

T5: Autonomy versus reduction (a dose-response scalar, or leverage-at-an-operating-point specialized to receptors). Wherever its precondition holds — a sigmoidal dose-response curve with a canonical half-maximal reference effect in a drug-receptor-tissue system — potency transfers literally, carrying its full apparatus across ED50, IC50, LD50, and hormone potency; these are variants of one pharmacological substrate, which is why the entry is domain-specific rather than a prime. But beyond that precondition the structural content that recurs is bare leverage — output-per-unit-input around an operating point — with sensitivity and elasticity as its near relatives and ceiling for the efficacy axis. Most non-pharmacological systems have no canonical half-maximal reference and no agreed sigmoid, so "channel potency" adds no structure over leverage, and the useful potency-versus-efficacy split just becomes the already-available leverage-versus-ceiling pair. The tension is between a precise pharmacological scalar and the recognition that its portable core is leverage-at-an-operating-point, over-read the moment it leaves dose-response substrates. Diagnostic: Resolve toward leverage/sensitivity/elasticity (plus ceiling) when there is no canonical half-maximal reference effect; toward named potency where a defined dose-response curve with a half-maximal reference genuinely holds.

Structural–Framed Character¶

Potency sits close to the structural pole — best read as mixed-structural, analogous to planarity and pointer: a genuine, evaluatively-neutral quantitative measure of a natural relationship, held short of the pole by domain-pinned vocabulary. Four criteria run structural. Its evaluative weight is nil: a lower ED50 means less drug suffices — a quantity, not a verdict; potency praises and blames nothing. It is essentially not human-practice-bound: the dose-response relationship is a fact of the drug-receptor-tissue system, and a drug's ED50 exists in that system whether or not anyone measures it — the effect runs in the body observer-free (the one practice-flavoured element is the convention of reading at the half-maximal point, addressed below). Its institutional origin is minimal: potency is a coordinate of a natural transfer function, not an artifact of an agency or tradition — though the canonical 50%-of-maximum reference is a measurement convention pharmacologists adopted rather than a physical landmark, which lends the scalar a thin conventional seam. And within pharmacology and toxicology cross-domain reuse is literal recognition: ED50, IC50, LD50, carcinogenic potency, and hormone potency are the identical construct with the same two-coordinate reading and Kd predictor, variants of one drug/toxin/hormone-on-receptor substrate.

What keeps it off the structural pole is vocab_travels, which it fails, and the import_vs_recognize flip beyond the substrate: the operative vocabulary — ED50, the sigmoidal dose-response curve, potency-versus-efficacy, receptor-binding affinity — is pharmacology furniture that does not float free, and past the dose-response precondition "potency" is over-read, adding no structure over the general primes. The portable structural skeleton is leverage-at-an-operating-point: leverage (output-per-unit-input around an operating point), with sensitivity and elasticity as its near relatives and ceiling carrying the efficacy axis. That skeleton is exactly what potency instantiates — a leverage scalar specialized to dose-response substrates with a canonical reference effect — not what makes "potency" travel: the cross-domain reach belongs to leverage/sensitivity/elasticity (and the potency-versus-efficacy split becomes the general leverage-versus-ceiling pair), while the ED50, the sigmoid, the half-maximal convention, and receptor affinity are the domain accent that stays home. Its character: a real, evaluatively-neutral pharmacological scalar reading a natural dose-response transfer function at a canonical operating point, structural in the leverage-at-an-operating-point skeleton it instantiates but pinned by its ED50/sigmoid/receptor-affinity vocabulary to dose-response substrates — mixed-structural, close to but short of the pole.

Structural Core vs. Domain Accent¶

This section decides why potency is a domain-specific abstraction and not a prime — marking where the portable leverage-at-an-operating-point skeleton ends and the dose-response machinery begins.

What is skeletal (could lift toward a cross-domain prime). Strip the pharmacology and a thin relational structure survives: read a single scalar off a response curve at a chosen operating point — the input required for a given fractional output — so that less input for the same effect is higher leverage, and hold that input-efficiency coordinate strictly apart from the ceiling coordinate (the maximum output the system can reach at any input). The portable pieces are abstract — a response/transfer function, an operating point, an output-per-unit-input reading, and a separate maximum-output reading. This is leverage (with sensitivity, the derivative of output with respect to input at an operating point, and elasticity, its unit-free percent form, as near relatives), and ceiling carrying the efficacy axis. The skeleton is genuinely substrate-portable — input-efficiency-at-an-operating-point recurs wherever there is a response curve (marketing spend per 1% sales lift, subsidy per X% adoption shift) — which is why the entry names leverage/sensitivity/elasticity as what actually travels. That portable core is what potency shares, not what makes it potency.

What is domain-bound. Almost everything that makes the construct this scalar is pharmacology furniture and none of it survives extraction: the canonical half-maximal (ED50) reference effect (50% of maximal response is the agreed operating point); the agreed sigmoidal dose-response curve (log dose to effect, with its slope, midpoint, plateau) as a shared empirical object; the potency-versus-efficacy vocabulary; and the receptor-binding-affinity (Kd) predictor that makes potency mechanistically designable. These are the worked vocabulary, the instruments, and the empirical cases (morphine/codeine, LD50 banding) the discipline uses. The decisive test: most non-pharmacological systems have no canonical half-maximal reference and no agreed response-curve shape, so an analyst importing "potency" is left with bare leverage and the potency-versus-efficacy split collapses into the already-available leverage-versus-ceiling pair — "channel potency" adds no structure over leverage. Remove the canonical reference effect and the sigmoid and what remains is leverage-at-an-operating-point, the parent, not potency.

Why this does not clear the prime bar. A prime is a relational structure whose vocabulary travels and whose cross-domain transfer is recognition of the same mechanism, not analogy. Potency's transfer is bimodal. Within pharmacology and toxicology it transfers literally, carrying its full apparatus — ED50 for therapeutic effect, IC50 for inhibition, LD50 for lethality, carcinogenic potency, hormone potency all use the same two-coordinate reading, the potency-versus-efficacy separation, and the Kd predictor, because dose-response curve, half-maximal reference effect, and receptor affinity are well-defined in each: variants of one drug/toxin/hormone-on-receptor substrate, recognition not analogy. Beyond that dose-response precondition it is over-read: applying "potency" to a marketing channel or a UI displacement renames bare leverage and adds no structure, and the useful distinction it carries is already the general leverage-versus-ceiling pair. And when the bare structural lesson is needed cross-domain — input efficiency for a given effect around an operating point, separated from the achievable maximum — it is already carried, in more general form, by leverage/sensitivity/elasticity (plus ceiling). The cross-domain reach belongs to those parents; "potency," as named, is the pharmacological instance, carrying ED50, the sigmoid, and receptor-affinity machinery that stays home wherever the dose-response precondition fails.

Relationships to Other Abstractions¶

Current abstraction Potency Domain-specific

Parents (1) — more general patterns this builds on

Potency is a decomposition of Intrinsic Ceiling vs Input Prime

Potency is the pharmacological input-to-approach coordinate, defined by holding that coordinate independent from the response ceiling.

Hierarchy paths (2) — routes to 2 parentless roots

Potency → Intrinsic Ceiling vs Input → Dose-Response Relationship → Function (Mapping)

Show alternative path (1)

Not to Be Confused With¶

Efficacy. The other coordinate of the same dose-response curve — the maximum response achievable at any dose (the plateau/ceiling), not the dose at which half-maximum is reached. Potency is where on the dose axis the half-maximum falls; efficacy is how high the ceiling sits. They are independent: a drug can be highly potent yet low in efficacy. Prying them apart is the concept's central work. Tell: is the claim about how little drug achieves the reference effect (potency), or how large an effect is achievable at any dose (efficacy)?
Receptor-binding affinity (Kd). The in vitro correlate and mechanistic cause of potency — how tightly the drug binds its receptor. Affinity predicts where the ED50 will fall, but potency is the functional dose-for-effect scalar read off whole-system response; receptor reserve and signal amplification can decouple the two (a system with spare receptors reaches half-maximal effect well below full occupancy). Cause-versus-functional-readout. Tell: is the quantity a binding constant measured in vitro (affinity/Kd), or the dose producing a functional half-maximal effect in the tissue system (potency)?
Therapeutic index / therapeutic window. The safety margin between the effective dose and the toxic (or lethal) dose — a gap between two points, a ratio like LD50/ED50. Potency is a single leverage coordinate (one ED50); the therapeutic index is a margin pattern between two such coordinates. A very potent drug can have a narrow or wide window independently. Structurally distinct (margin vs leverage-scalar). Tell: is the referent one dose-for-effect coordinate (potency), or the distance between effective and harmful doses (therapeutic index)?
Bioavailability / pharmacokinetics. How much administered drug reaches systemic circulation, and how fast it is absorbed, distributed, and eliminated — the pharmacokinetic axis (what the body does to the drug). Potency is pharmacodynamic (what the drug does at the receptor per unit dose) and concerns neither absorption fraction nor onset/duration. Different axis entirely. Tell: is the concern how much drug reaches its site and how fast (pharmacokinetics/bioavailability), or how little is needed at the receptor for a given effect (potency)?
EC50 / IC50 / LD50 (the family guises). These are the same half-maximal construct read for different endpoints — functional effect (EC50), inhibition (IC50), lethality (LD50) — not rivals to potency. LD50 in particular is often loosely called "toxicity," but it is toxic potency: the identical dose-for-half-maximal-endpoint coordinate. They are variants of the one construct, comparable only at the same endpoint/route/tissue. Tell: is the endpoint therapeutic effect, inhibition, or lethality — each a guise of the same potency reading, not a separate quantity?
Leverage / sensitivity / elasticity (parent primes). The substrate-neutral core potency instantiates — input-per-unit-output (or output-per-unit-input) around an operating point, with ceiling for the efficacy axis. This is what travels beyond dose-response substrates; "potency" over-reads the moment there is no canonical half-maximal reference. Treated more fully in the Knowledge Transfer and Structural Core vs. Domain Accent sections. Tell: strip the ED50/sigmoid/receptor machinery and what remains — input efficiency for a given effect around an operating point — is the parent, not potency.

Neighborhood in Abstraction Space¶

Potency sits in a sparse region of the domain-specific corpus (95^th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Pharmacokinetics & Drug Response (19 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-07-12