Skip to content

Stevens's Power Law

Relate the perceived magnitude of a sensation to the physical intensity of the stimulus by a power function ψ = k·I^β, whose modality-characteristic exponent β compresses the sense (β < 1), expands it (β > 1), or leaves it near-linear — recovered as the slope of magnitude estimates plotted on log-log axes.

Core Idea

Stevens's power law (S. S. Stevens, 1957, 1961) states that the perceived magnitude of a sensation is related to the physical intensity of the stimulus by a power function — ψ = k · I^β — where ψ is the judged subjective magnitude, I is the measurable physical intensity, k is a scaling constant, and β is an exponent whose value is characteristic of the sensory modality and is the law's substantive empirical content. Modalities with β less than 1 show response compression: large physical changes map onto small perceptual changes (brightness: β ≈ 0.33, so a tenfold increase in luminance feels roughly twice as bright); modalities with β greater than 1 show response expansion: small physical changes map onto large perceptual changes (electric shock: β ≈ 3.5, so a modest increase in current intensity feels dramatically more intense). Intermediate modalities approximate linearity (taste of salt: β ≈ 1.3).

The law emerged from the method of magnitude estimation, introduced by Stevens precisely to access the full dynamic range of a sensory continuum: observers assign numbers in direct proportion to their felt intensity, producing a dataset that, when plotted on logarithmic axes, yields a straight line whose slope is β. This methodological move is what allowed Stevens to demonstrate that the power function fits across several orders of magnitude of stimulus intensity, where Weber–Fechner's logarithmic law had been understood to hold only over a middle range. The modality-specific exponents are reproducible across laboratories and moderately stable across individuals, which gives the law its empirical standing as a calibration tool in sensory engineering: audio volume controls, display gamma curves (γ ≈ 2.2 compensates for the brightness exponent), loudness standards such as LUFS, and haptic interfaces all rely on Stevens-derived compression or expansion curves to map physical steps onto perceptually uniform steps.

Structural Signature

Sig role-phrases:

  • the physical-stimulus continuum — the measurable input on a ratio scale (luminance, sound pressure, current, concentration), the I term
  • the perceived-magnitude response — the felt subjective intensity ψ, elicited by magnitude estimation or cross-modality matching
  • the magnitude-estimation method — the procedure that accesses ψ directly by having observers assign numbers proportional to felt intensity across orders of magnitude
  • the power-function form — the fitted relation ψ = k · I^β, a straight line on log-log axes whose slope is the exponent
  • the modality exponent β — the law's substantive content: a characteristic, reproducible value per sensory pathway
  • the compression/expansion reading — the sign of (β − 1): β < 1 compresses (large physical → small perceptual), β > 1 expands (small physical → large perceptual), β ≈ 1 near-linear
  • the threshold-to-saturation bounds — the dynamic range, above absolute threshold and below physiological saturation, within which the single-exponent fit holds
  • the calibration inversion — the engineering use: apply a correction inverting (β − 1) (gamma curves, volume tapers, haptic tuning) to make physical steps land as perceptually uniform ones, subsuming Weber-Fechner's log law as the middle-range special case it cannot extend to expansion

What It Is Not

  • Not a causal mechanism. The law is a descriptive fitted relation between two ratio scales — physical intensity and judged magnitude — not an account of how sensory transduction produces the percept. The exponent summarizes the input-output curve; it does not explain the neural pathway that generates it, and reading ψ = k · I^β as a process misclassifies a measurement construct.
  • Not a single universal exponent. There is no one law-of-perception curve: β is modality-specific and is itself the substantive empirical content. Brightness runs ≈ 0.33, shock ≈ 3.5, salt ≈ 1.3 — characteristic per sensory pathway and reproducible across laboratories, so "the exponent" only has meaning relative to a named continuum.
  • Not always compressive, and not the logarithmic Weber-Fechner law. Weber-Fechner is monotone-compressive and holds only over a middle range; the power form additionally accommodates response expansion (β > 1) that shock and warmth genuinely exhibit and that a log law structurally cannot represent. Stevens subsumes the logarithmic relation as a middle-range special case, not the reverse.
  • Not valid across the entire stimulus range. The single-exponent fit holds in the dynamic range above absolute threshold and below physiological saturation. Near the floor and the ceiling the relation departs, so calibration curves derived from β should not be extrapolated into the saturation region.
  • Not the generic claim "perception is non-linear." Stripped of a physical continuum on a ratio scale, a magnitude-estimable percept, and a threshold-to-saturation range, the law reduces to that truism and loses its calibratable bite. So importing "Stevens's power law" into decision theory or economics — perceived inflation, utility curves — over-reads a psychophysical instrument: the general objective-to-subjective bending there is nonlinearity (or the prospect-theory value function), and a fitted "exponent" outside psychophysics has no comparable empirical standing.

Scope of Application

Because Stevens's power law is a fitted measurement relation between two ratio scales, not a sensory mechanism, it applies literally wherever its precondition holds — a physical-stimulus continuum measurable on a ratio scale, a perceived-magnitude response elicitable by magnitude estimation, and a dynamic range between threshold and saturation; the habitats below are genuine uses of the identical construct, not metaphor (its modality exponents do not travel, and loose objective-to-subjective bending elsewhere belongs to nonlinearity / the prospect-theory value function).

  • Psychophysics — the home turf: the magnitude-estimation paradigm and cross-modality matching spanning brightness, loudness, warmth, taste, and shock with their characteristic exponents.
  • Display engineering — gamma correction (γ ≈ 2.2) compensating the brightness exponent so pixel values map onto perceptually uniform brightness.
  • Audio loudness standards — loudness units (LUFS, phons) built on Stevens-shape compression of physical sound-pressure level.
  • Clinical pain assessment — visual-analogue and similar scales exploiting the high noxious exponent (β > 3) so small physical changes become clinically detectable.
  • Olfactory and gustatory / food science — taste and smell exponents used to calibrate flavor intensities.

Clarity

Naming Stevens's power law converts a qualitative truism everyone has felt — equal physical steps do not feel like equal perceptual steps — into a single calibratable parameter. The vague concession that "perception is non-linear" gives the engineer nothing to set; the power-function form gives them the exponent β, and with it a sharp question for every sensory continuum: what is this modality's exponent, and does it compress (β < 1) or expand (β > 1)? That one number predicts the qualitative shape of the whole mapping — brightness at β ≈ 0.33 buries a tenfold luminance change into roughly a doubling of felt brightness, while electric shock at β ≈ 3.5 magnifies a modest current change into a dramatic one — and it is what lets sensory engineering space physical steps so that perceptual steps come out uniform, the logic behind volume tapers, display gamma, and loudness units. The law's clarifying force is to make "subjective magnitude" a measurable quantity tied to physical magnitude by a known curve, rather than an unquantified given.

It also sharpens the relationship to the older Weber-Fechner law, which a coarse "perception is logarithmic" picture would leave as a rival. By exposing β as the discriminating parameter and the power function as the more general fit across orders of magnitude, the law reframes the logarithmic relation as an approximation valid in the middle of a modality's range — and, decisively, one that cannot represent the response expansion (β > 1) that real modalities like shock and warmth exhibit. The distinction it makes crisp is between a fixed law of perception and a modality-indexed one: there is no single exponent, only a characteristic value per sensory pathway, reproducible across laboratories, and the method of magnitude estimation is what gives the analyst access to it. That sharper question — not "is perception non-linear?" but "which compression or expansion exponent does this sense run, and over what dynamic range between threshold and saturation?" — is the one the law was built to answer.

Manages Complexity

Psychophysics and the sensory engineering downstream of it would otherwise face every continuum as its own measurement problem: how brightness scales with luminance, loudness with sound pressure, warmth with skin temperature, salt with concentration, pain with current — each a separate empirical mapping with its own shape, its own calibration curve, its own threshold and saturation behavior, derived afresh for displays, volume controls, haptic motors, loudness standards, and pain scales. Treated that way, the field is a stack of unrelated intensity functions and the engineer re-measures the full input-output curve for each device. Stevens's power law compresses this by asserting that every such mapping has the same functional form, ψ = k · I^β, so that the entire qualitative behavior of a sensory continuum is carried by one number — the modality-characteristic exponent β. With the form fixed, the analyst no longer reconstructs each curve but reads it off a single parameter: β < 1 means response compression (large physical changes buried into small perceptual ones, brightness at ≈ 0.33 turning a tenfold luminance increase into roughly a doubling of felt brightness); β > 1 means response expansion (small physical changes magnified into large perceptual ones, shock at ≈ 3.5 turning a modest current change into a dramatic one); β near 1 means approximate linearity. A roomful of distinct intensity functions collapses to one form indexed by one scalar per sense.

What the engineer or researcher then tracks shrinks accordingly to a short list, none of it requiring the full mapping to be rebuilt. First, the modality's exponent β, obtained by magnitude estimation and read as a straight-line slope on log-log axes — a single reproducible number that fixes whether the sense compresses or expands and by how much. Second, the sign of (β − 1), which is the branch point: it determines the qualitative shape of the whole curve and therefore the direction of the calibration correction needed to make physical steps land as perceptually uniform ones — gamma curves for the brightness exponent, tapers for loudness, amplitude tuning for haptics, sensitive scales exploiting the high pain exponent. Third, the bounds of validity: the dynamic range between absolute threshold and physiological saturation, within which the single-exponent fit holds across orders of magnitude. From those the qualitative outcome follows — how a given modality will translate any physical step into a felt one, and how to space physical steps for uniform perception — without modelling the transduction pathway in detail. The compression also subsumes the older logarithmic account as the middle-range special case while extending past it to the expansion (β > 1) the log law structurally cannot represent, so the analyst's question sharpens from the unanswerable "is perception non-linear?" to the determinate "which exponent does this sense run, and over what range?" A high-dimensional family of separately measured sensory mappings becomes one parametric form read off a single, calibration-ready number.

Abstract Reasoning

Stevens's power law equips the psychophysicist and the sensory engineer with moves that all reduce a continuum's perceptual behavior to a single estimated exponent. The measurement-diagnostic move recovers β from data: collect magnitude estimates across orders of magnitude of stimulus intensity, plot on log-log axes, and read the exponent as the slope of the resulting straight line — so the analyst reasons FROM "the magnitude-estimate plot is linear in log-log with slope s" TO "this modality runs exponent β = s, characteristic and reproducible across laboratories." Once β is in hand, the predictive move reads the qualitative shape of the entire mapping off the sign of (β − 1): β < 1 predicts response compression — large physical changes buried into small perceptual ones, so the analyst forecasts that a tenfold luminance increase at β ≈ 0.33 feels like roughly a doubling of brightness; β > 1 predicts response expansion — small physical changes magnified, so a modest current increase at β ≈ 3.5 feels dramatically more intense; β ≈ 1 predicts approximate linearity. The analyst predicts a felt step from any physical step without re-measuring the curve, because the form is fixed and only the scalar varies.

The interventionist move is the law's engineering payoff and is governed by the same sign: to make physical steps land as perceptually uniform steps, apply a calibration correction whose direction is set by (β − 1) — gamma curves that pre-distort pixel values to compensate the brightness exponent, logarithmic-ish volume tapers for loudness, amplitude tuning for haptic motors, and clinical scales that exploit a high pain exponent so small physical changes become detectable. The analyst reasons that the correction must invert the modality's compression or expansion, so a compressing sense (β < 1) demands an expanding control law and vice versa. The boundary-drawing move fixes where the single-exponent fit is trusted: it holds in the dynamic range above absolute threshold and below physiological saturation, so the analyst predicts the power function describes the response across many orders of magnitude within that window and expects departures at the floor and ceiling — and reasons that calibration curves derived from the exponent should not be extrapolated past the saturation region. The classificatory move relocates the older Weber-Fechner logarithmic relation as the middle-range special case of the power form, and draws a sharp consequence: because the log law is monotone-compressive, it structurally cannot represent the response expansion (β > 1) that shock and warmth exhibit, so the analyst predicts that any modality with β > 1 will be mis-described by a logarithmic model and reasons that the discriminating test between the two accounts is whether expansion occurs. The compact set the analyst tracks is therefore just the exponent, the sign of (β − 1), and the threshold-to-saturation bounds — and from these the felt consequence of any physical step, and the control law that makes stepping perceptually uniform, both follow without modeling the transduction pathway.

Knowledge Transfer

Stevens's power law is a functional/measurement construct — a fitted parametric relation between two ratio scales, not a causal mechanism — so the "mechanism within, metaphor beyond" framing does not apply to it the way it does to a process. What it has instead is an instrument's reach: the construct ψ = k · I^β transfers literally wherever its precondition holds, namely a physical-stimulus continuum measurable on a ratio scale, a perceived-magnitude response elicitable by magnitude estimation or cross-modality matching, and a dynamic range between threshold and saturation. Within psychophysics and the sensory engineering downstream of it, that precondition holds across every modality, so the form, the exponent-as-slope recovery, the compression/expansion reading off the sign of (β − 1), and the calibration correction all carry intact and literally. In psychophysics it spans brightness, loudness, warmth, taste, and shock with their characteristic exponents. In display engineering it is gamma correction compensating the brightness exponent. In audio it underlies loudness units (LUFS, phons). In clinical pain assessment it exploits the high noxious exponent to make small physical changes detectable. In food science it calibrates flavor intensities. These are not analogies — the same measured relation applies wherever the two scales and the estimation paradigm exist, which is the proper reach of an instrument.

The boundary to mark, then, is not mechanism-versus-metaphor but construct-reach versus over-reading, and it has two parts. First, the modality-specific exponents are facts about particular sensory transduction pathways and do not transfer: 0.33 for brightness and 3.5 for shock are not portable numbers, and there is no reason a fitted "exponent" in some non-sensory setting carries comparable theoretical status. Second, where genuine cross-domain nonlinearity-of-the-subjective does recur — prospect-theory value functions, probability weighting, perceived inflation, social judgments of magnitude — the portable content is the general commitment that the mapping from objective to subjective is nonlinear, which is shared abstract mechanism carried by parents like nonlinearity, the prospect_theory value function (whose curve has a Stevens-like shape), and anchoring-style scale dependence. Stevens's specific power-function form is one parameterization among many of that umbrella, not the thing that travels; importing "Stevens's power law" into decision theory or economics is over-reading a psychophysical instrument as if its precondition (two well-defined ratio scales plus magnitude estimation) were met where it is not. Stripped of that setup the law reduces to "perception is non-linear" — true but no longer the calibratable construct. The discipline is therefore to use Stevens's power law literally wherever a physical continuum, a magnitude-estimable percept, and a threshold-to-saturation range all exist (its instrument-reach), to carry the general nonlinearity / value-function parent — not the power-law form or its exponents — wherever only a loose objective-to-subjective bending recurs, and to refuse the temptation to treat a fitted exponent outside psychophysics as having the law's empirical standing (see Structural Core vs. Domain Accent).

Examples

Canonical

The magnitude-estimation derivation is the defining construction. An observer is shown a reference light and told it has brightness "10," then judges a series of test luminances by assigning each a number proportional to its felt brightness. Plotting the mean judgments (ψ) against physical luminance (I) on log-log axes yields a straight line, and its slope is the exponent β — for brightness, about 0.33. The fitted law ψ = k·I^β then makes quantitative predictions. A tenfold increase in luminance produces a perceived-brightness ratio of 10^0.33 ≈ 2.1 — the light feels only about twice as bright though it is ten times as intense (compression, β < 1). Contrast electric shock, β ≈ 3.5: merely doubling the current yields a felt-intensity ratio of 2^3.5 ≈ 11, a dramatic jump (expansion, β > 1).

Mapped back: Luminance and current are the physical-stimulus continuum (I); felt brightness and felt shock are the perceived-magnitude response (ψ), accessed by the magnitude-estimation method. The log-log straight line is the power-function form, its slope the modality exponent β. That 0.33 buries a tenfold change into a doubling while 3.5 magnifies a doubling into elevenfold is the compression/expansion reading off the sign of (β − 1).

Applied / In Practice

Display gamma encoding is Stevens's law doing daily engineering work. Human lightness perception compresses luminance with an exponent near one-third (CIE L* uses a cube-root of luminance), so equal steps in physical luminance are perceptually crowded in the shadows and wasted in the highlights. To store images efficiently in 8 bits without visible banding in dark tones, display standards (sRGB, and broadcast video) encode pixel values through a power function with gamma near 2.2 — roughly the inverse of the perceptual compression — so that equal code-value steps map onto approximately perceptually uniform lightness steps. The scarce code values are thereby allocated where the eye is most sensitive.

Mapped back: Luminance is the physical-stimulus continuum and perceived lightness the perceived-magnitude response, related by the power-function form with a compressive modality exponent β near ⅓. Gamma ≈ 2.2 is the calibration inversion: because the sense compresses (β < 1), the encoding applies the opposite, expanding transform so physical (code-value) steps land as perceptually uniform ones — the engineering payoff read straight off the sign of (β − 1).

Structural Tensions

T1: Universal form versus non-portable content (one law whose substance is a per-pathway number). The law's compression is that every sensory continuum shares the same functional form, ψ = k·I^β — a roomful of distinct intensity functions collapsed to one shape. But the shape is the empty part; the substantive empirical content is β, and β is modality-specific, a fact about a particular transduction pathway. 0.33 for brightness and 3.5 for shock are not portable numbers, and knowing the form tells you nothing about any sense until its exponent is measured. The tension is that what is universal (the power-function form) carries no information, and what carries the information (the exponent) is irreducibly particular — so the law is simultaneously a sweeping generalization and a demand for a fresh measurement per modality. Diagnostic: Is the claim resting on the shared power-function form (universal but contentless) or on a specific exponent (contentful but valid only for the one continuum it was measured on)?

T2: Predictive fit versus explanatory silence (a curve that forecasts without accounting). Because the fitted relation predicts the felt consequence of any physical step precisely — a tenfold luminance change feels like a doubling — it delivers everything sensory engineering needs to space steps for perceptual uniformity. But it is a descriptive fit between two ratio scales, not an account of how transduction produces the percept; the exponent summarizes the input-output curve without explaining the neural pathway that generates it. The tension is that the law's calibration power comes bundled with explanatory silence: it tells you exactly what will be felt and nothing about why, and there is no mechanism inside it to intervene on — only a curve to invert. Reading ψ = k·I^β as a process misclassifies a measurement construct as a theory of the sense. Diagnostic: Is the exponent being used to predict and calibrate the mapping (its proper work), or being read as an explanation of the sensory mechanism (which it is not)?

T3: Subsuming Weber–Fechner versus the difficulty of the discriminating regime (superiority proved only at the edges). Stevens's law claims to relocate the older logarithmic relation as a middle-range special case and to extend past it, because the log law is monotone-compressive and structurally cannot represent the response expansion (β > 1) that shock and warmth exhibit. That is the decisive argument for the power form's generality. But over the middle range both models fit acceptably, so the discriminating evidence lives precisely where measurement is hardest — in the expansion regime and across the many orders of magnitude near threshold and saturation. The tension is that the power law's superiority over the log law rests on the regions (expanding modalities, extreme intensities) where the data are noisiest and the fits least stable, so the claim is strongest exactly where its support is weakest. Diagnostic: Does the modality in question actually exhibit response expansion (β > 1) or wide-range behavior that separates the power fit from the log fit, or is the middle range where the two are empirically indistinguishable?

T4: Holds across orders of magnitude versus the threshold-and-saturation bounds (validity that fails at the interesting extremes). The single-exponent fit is celebrated for describing the response across several orders of magnitude of intensity — its advertised advantage over the middle-range log law. Yet it holds only within a window: above absolute threshold and below physiological saturation, with departures at the floor and the ceiling. The tension is that many of the engineering situations that most need calibration — the darkest shadows a display must render, the loudest signals, near-saturation stimuli — sit exactly at the bounds where the exponent-derived curve is not to be trusted and must not be extrapolated. The property that makes the law attractive (wide-range validity) is qualified precisely at the extremes where wide range would matter most. Diagnostic: Is the stimulus range of interest safely between threshold and saturation, or does it push into the floor/ceiling regions where the single-exponent fit breaks down?

T5: Autonomy versus reduction (a psychophysical instrument or an instance of nonlinearity). As a fitted measurement construct, Stevens's power law transfers literally wherever its precondition holds — a physical continuum on a ratio scale, a magnitude-estimable percept, a threshold-to-saturation range — so gamma correction, loudness units, and pain scales are genuine instances of the identical construct, not analogies. But that instrument-reach is bounded by the precondition, and beyond it lies over-reading: prospect-theory value functions, probability weighting, and perceived inflation share only the general commitment that the objective-to-subjective mapping is nonlinear, which belongs to nonlinearity and the prospect-theory value function, not to the power form. A fitted "exponent" outside psychophysics has no comparable empirical standing. The tension is between a construct that applies exactly and literally where two ratio scales and magnitude estimation exist, and a loose objective-to-subjective bending that its parents carry everywhere else. Diagnostic: Resolve toward nonlinearity / the prospect-theory value function when only a general objective-to-subjective bending recurs without ratio scales and magnitude estimation; toward Stevens's power law when a measurable continuum, a magnitude-estimable percept, and a threshold-to-saturation range are all literally present.

Structural–Framed Character

Stevens's power law sits in the mixed region of the spectrum, with the profile of a measurement instrument rather than a causal mechanism. Its evaluative_weight is nil: it is a descriptive fitted relation between two ratio scales, and the exponent it reports praises and condemns nothing. It is human_practice_bound only through its object: the law concerns perceived magnitude, so it requires a perceiving subject and the magnitude-estimation procedure to elicit ψ — but the modality exponent it recovers is a reproducible fact about a sensory transduction pathway, stable across laboratories, more a regularity of the perceptual system than a convention. Its institutional_origin is none: the exponents are measured and discovered (Stevens), not stipulated. On vocab_travels and import_vs_recognize it has an unusual profile the entry stresses: as a functional/measurement construct it transfers literally — not by analogy — wherever its precondition holds (a ratio-scale physical continuum, a magnitude-estimable percept, a threshold-to-saturation range), so gamma correction, loudness units, and pain scales are genuine instances of the identical construct; but the modality-specific exponents do not travel, and importing "Stevens's power law" into economics or decision theory is over-reading, since the general objective-to-subjective bending there belongs to a parent, not to the power form.

The portable skeleton is a nonlinear mapping from objective magnitude to subjective magnitude — the parent nonlinearity, with the prospect_theory value function (whose curve has a Stevens-like shape), probability weighting, and anchoring-style scale dependence as the sibling instances. Stevens's specific power-function form ψ = k·I^β is one parameterization among many of that umbrella, and it is the umbrella — not the power form or its exponents — that carries wherever only a loose objective-to-subjective bending recurs. Its distinctive cargo — the modality exponents, the compression/expansion reading, the magnitude-estimation method, the calibration inversion — stays within psychophysics and its sensory engineering. Its character: an evaluatively-neutral, discovered psychophysical measurement instrument whose structural core is the general nonlinearity-of-the-subjective belonging to its parent, mixed — structural in that borrowed nonlinearity and domain-specific in the per-pathway exponents and magnitude-estimation apparatus that give it its calibratable bite.

Structural Core vs. Domain Accent

This section decides why Stevens's power law is a domain-specific abstraction and not a prime, and it carries the case for its domain-specificity in one place.

What is skeletal (could lift toward a cross-domain prime). Strip away the psychophysics and a thin relational structure survives: the mapping from an objective magnitude to the subjective magnitude it evokes is nonlinear — equal objective steps do not evoke equal subjective steps, and the mapping can either compress or expand. The portable pieces are abstract — an objective input scale, a subjective output that tracks it monotonically but unevenly, and a curvature whose sign says whether the subjective response lags or outruns the physical one. That skeleton is genuinely substrate-portable, which is exactly why the entry names the general parents it instantiates: the nonlinear objective-to-subjective mapping is nonlinearity, and it recurs as the sibling curves of the prospect_theory value function, probability weighting, and anchoring-style scale dependence. But this is the core Stevens's power law shares with those siblings — not what makes Stevens's power law the particular thing it is.

What is domain-bound. Almost all the worked content is psychophysics furniture that does not survive extraction. The law requires a physical-stimulus continuum measurable on a ratio scale (luminance, sound pressure, current, concentration), a perceived-magnitude response elicited by the magnitude-estimation method (or cross-modality matching), and a dynamic range between absolute threshold and physiological saturation. Its distinctive cargo is tighter still: the specific power-function form ψ = k · I^β recovered as a log-log slope, and above all the modality exponent β — the reproducible per-pathway number (brightness ≈ 0.33, shock ≈ 3.5, salt ≈ 1.3) that is the law's substantive empirical content — together with the calibration inversion (gamma curves, loudness tapers, haptic tuning) that puts β to work. The decisive test: remove the ratio-scale continuum and the magnitude-estimation paradigm and there is no slope to recover, no exponent to name, no calibration to invert — only the truism that "perception is non-linear," which sets no parameter and calibrates nothing. Crucially the exponents themselves are the least portable part: 0.33 and 3.5 are facts about specific sensory transduction pathways, not numbers that mean anything in another substrate, so a fitted "exponent" imported into economics carries none of the law's empirical standing.

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. Stevens's power law has an instrument's reach, and its transfer is bimodal in a construct-specific way. Within its precondition — anywhere a ratio-scale physical continuum, a magnitude-estimable percept, and a threshold-to-saturation range co-exist — the whole construct transfers literally: display gamma, loudness units, clinical pain scales, and flavor calibration are genuine instances of the identical relation, not analogies, because the same two scales and the same estimation paradigm are present. Beyond that precondition it does not travel as mechanism: importing "Stevens's power law" into decision theory or perceived-inflation talk over-reads the instrument, borrowing the power-law shape where the two ratio scales and magnitude estimation are absent. And when the bare structural lesson is needed there — "the objective-to-subjective mapping bends nonlinearly" — it is already carried, in more general form, by the parents it instantiates: nonlinearity, the prospect_theory value function, and anchoring-style scale dependence, of which the power form is merely one parameterization. The cross-domain reach belongs to those parents; the modality exponents, the magnitude-estimation apparatus, and the calibration inversion are baggage that should stay home in psychophysics.

Relationships to Other Abstractions

Local relationship map for Stevens's Power LawParents appear above the current abstraction, mutual partners to the right, and children below. Node labels state whether each abstraction is prime or domain-specific; colors identify relation types.Stevens's Power LawDOMAINPrime abstraction: Allometry and Scaling Law — is a decomposition ofAllometry andScaling LawPRIMEDomain-specific abstraction: Psychophysical Scaling — is a kind ofPsychophysicalScalingDOMAIN

Current abstraction Stevens's Power Law Domain-specific

Parents (2) — more general patterns this builds on

  • Stevens's Power Law is a kind of Psychophysical Scaling Domain-specific

    Stevens's Power Law is the magnitude-estimation species of Psychophysical Scaling, fitting perceived intensity to physical stimulus intensity over a bounded sensory range.

  • Stevens's Power Law is a decomposition of Allometry and Scaling Law Prime

    Removing the sensory modality and magnitude-estimation procedure leaves a power-law relation whose characteristic exponent governs how one quantity scales with another.

Hierarchy paths (2) — routes to 2 parentless roots

Not to Be Confused With

  • Weber–Fechner law. The older psychophysical law holding that perceived magnitude grows with the logarithm of stimulus intensity, ψ = k·log I. It is the direct predecessor and rival, and Stevens's law subsumes it as the middle-range special case it cannot extend past: because the log form is monotone-compressive, it structurally cannot represent the response expansion (β > 1) that shock and warmth exhibit, whereas the power form accommodates both compression and expansion across orders of magnitude. Tell: does the modality show expansion (small physical change felt as large, β > 1)? If so it is Stevens territory and the log law fails; if the response only ever compresses over a middle range, the logarithmic account may suffice.

  • Weber's law. The proportionality that the just-noticeable difference scales with intensity — ΔI/I is roughly constant — a statement about discrimination thresholds, not about the full perceived-magnitude curve. Fechner integrated Weber's law to derive the logarithmic law; Stevens instead used direct magnitude estimation to build the whole scaling function. The confusion is that all three share the "Weber/Stevens/psychophysics" neighborhood. Tell: is the claim about the smallest detectable change at a given intensity (Weber's law) or about the magnitude of the sensation itself across the range (Stevens's power law)?

  • Power-law distributions (Zipf, Pareto, scaling laws). Statistical distributions in which frequency or size follows p(x) ∝ x^(−α) across a population of events or objects. Pure word-sharing: both say "power law," but a power-law distribution describes how often things of a given magnitude occur, while Stevens's power function relates a single stimulus's physical intensity to the perceived magnitude it evokes. Tell: is "power law" describing the frequency/size distribution over many items (Zipf/Pareto), or a functional stimulus-to-percept mapping for one continuum (Stevens)?

  • Gamma correction. The display-encoding transform (γ ≈ 2.2) that pre-distorts pixel values so physical luminance steps land as perceptually uniform ones. This is not a rival law but the calibration inversion — an engineering application of Stevens's law that inverts the brightness exponent (β ≈ ⅓). Part-vs-whole relation: the law states the perceptual mapping; gamma is one device that corrects for it. Tell: are you naming the perceptual relation between luminance and felt brightness (the law) or the inverse-power encoding applied to compensate it (gamma correction)?

  • Prospect-theory value function / probability weighting. Nonlinear mappings from objective quantities (monetary gain/loss, probability) to their subjective counterparts, whose curves have a Stevens-like bend. These are sibling instances of the same nonlinearity umbrella, not the psychophysical law: their inputs are decision quantities, not ratio-scale sensory stimuli, and they lack a physical continuum, magnitude estimation, and a threshold-to-saturation range — so a fitted "exponent" there has no comparable empirical standing. Tell: is the subjective quantity a felt sensory intensity elicited from a measurable stimulus (Stevens) or a subjective value/weight attached to money or probability (prospect theory)?

  • Nonlinearity (the parent it instances). The substrate-general pattern — an objective-to-subjective mapping that bends rather than scaling linearly — that Stevens's power law instantiates with one specific parameterization (ψ = k·I^β) inside psychophysics. Not a confusable peer but the umbrella carrying the cross-domain reach; the law adds the ratio-scale continuum, the modality exponents, and the calibration apparatus, which stay home. Tell: if the lesson needed is the bare "equal objective steps don't feel like equal subjective steps," that is the parent; Stevens's power law is the psychophysical measurement instrument that pins it to a per-pathway exponent. (Treated fully in earlier sections.)

Neighborhood in Abstraction Space

Stevens's Power Law sits in a sparse region of the domain-specific corpus (81st percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Psychophysical Laws of Perception (10 abstractions)

Nearest neighbors

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