The Weber-Fechner law is often taught as one rule: noticeable change depends on proportion, and sensation grows with the logarithm of stimulus intensity. The combined name, however, joins two different claims.

Weber's law describes an approximate pattern in discrimination. Under some conditions, the difference needed to distinguish a stimulus from a reference grows in proportion to that reference. Fechner's law is a stronger proposal about sensation magnitude: given additional assumptions, sensation can be represented as a logarithmic function of physical intensity.

Neither relation is exact across every sense, person, procedure, or intensity range.

A fictional planetarium worksheet

Everything in this example is invented. The planetarium, trainer, control scale, constants, readings, and exercise are not study data, physical light measurements, equipment settings, or safety guidance.

A trainer creates an abstract light-desk worksheet to keep the two laws separate. For the Weber exercise, the sheet assumes a Weber fraction of k = 0.08.

At a reference value of 100 abstract units:

ΔI = kI = 0.08 × 100 = 8 units

At a reference value of 400 units:

ΔI = kI = 0.08 × 400 = 32 units

The larger reference requires a larger absolute increment in this model, although the assumed ratio remains 8%. The value 0.08 is a teaching choice. It is not an estimate of anyone's visual sensitivity.

The trainer then opens a separate Fechner exercise. It assumes K = 2 and a positive reference level I0 = 25, giving:

ψ = 2 ln(I / 25)

The modeled values are:

  • ψ(100) = 2 ln(4) ≈ 2.7726
  • ψ(200) = 2 ln(8) ≈ 4.1589
  • ψ(400) = 2 ln(16) ≈ 5.5452

Each doubling adds about 1.3863 modeled sensation units because 2 ln(2) ≈ 1.3863. That is a property of the chosen logarithmic equation, not evidence that actual viewers experience the fictional controller this way.

Weber's law concerns difference thresholds

Let I be a positive reference stimulus and ΔI the increment required to reach a measured difference threshold. Weber's law is usually written:

ΔI / I ≈ k

or equivalently:

ΔI ≈ kI

The quantity k is the Weber fraction. It has meaning only when the sensory dimension, stimulus range, observer group, adaptation state, task, and threshold procedure are specified. The approximation says that a proportional change can preserve discrimination performance over some range. It does not say that the same fraction works for every kind of stimulus.

Weber's relation dates to nineteenth-century work on sensory discrimination. A modern mathematical treatment writes the just-noticeable increment in the same proportional form and shows how a logarithmic transformation follows under suitable assumptions. (Drösler, 2000)

Most importantly, Weber's law does not itself measure how intense a sensation feels. It concerns the difference needed to tell two physical levels apart.

A JND is an estimate, not a hard switch

The phrase just-noticeable difference, or JND, can sound as if one exact increment is invisible and the next is always seen. Psychophysical thresholds do not work that cleanly.

Researchers estimate a threshold from repeated responses under a defined procedure. They choose a performance criterion, such as a particular point on a fitted response curve. Internal noise, attention, adaptation, the order of presentation, and response tendencies can all affect the estimate.

Signal detection theory makes this uncertainty explicit. It separates discrimination sensitivity from a person's tendency to choose one response and requires the task and criterion to be part of the threshold definition. A methods review in vestibular psychophysics shows why different detection and recognition procedures can produce different threshold considerations. (Merfeld, 2011)

So a JND is not a universal sensory atom. It is a measurement tied to a reference, method, and criterion.

Fechner's law adds a sensation scale

Fechner's proposal asks a different question: can physical stimulus values be mapped onto a quantitative scale of sensation?

The familiar logarithmic form is:

ψ = K ln(I / I0)

Here:

  • ψ is the modeled sensation magnitude.
  • K sets the scale. Changing the base of the logarithm changes this constant.
  • I is a positive stimulus magnitude.
  • I0 is a positive baseline. By convention, ψ is zero there; the baseline is often linked to a threshold.

Both I and I0 must be greater than zero. If I0 is treated as an absolute sensory threshold, values below it should not be turned into literal "negative sensation" merely because the algebra produces a negative number.

One familiar route to the formula begins with an infinitesimal version of Weber's relation:

dψ = c(dI / I)

Integrating gives a logarithm. Interpreting that mathematical scale as sensation also requires a scaling principle, often summarized as treating equal discrimination steps as equal increments of sensation.

Fechner's own account set sensation relative to a finite threshold and acknowledged limits to Weber's law in external stimulation. (Fechner, 1860/1912) Modern historical analysis adds another nuance: Fechner gave ratio-based derivations that do not reduce simply to counting JNDs, while the link between Weber behavior and JND counts is approximate. (Dzhafarov and Colonius, 2011)

The logarithmic model has an elegant ratio property. For a fixed multiplier r:

ψ(rI) - ψ(I) = K ln(r)

Equal physical ratios therefore produce equal modeled sensation increments wherever the model's assumptions hold.

Where Weber's approximation breaks

A constant Weber fraction is not expected from zero to the largest possible stimulus. Near an absolute detection threshold, the reference is already competing with sensory noise, so a simple proportional rule commonly fails. At high intensities, receptor behavior, adaptation, saturation, apparatus limits, or physical constraints can also alter discrimination.

The failure is not merely a modern technical objection. In a 1924 analysis of visual intensity discrimination, Selig Hecht reported that the ratio of threshold increment to reference intensity was not constant across the tested range. It first decreased and then increased as intensity rose. (Hecht, 1924)

That result does not show that Weber's law is useless. It shows why it should be described as a bounded empirical approximation, often most informative across a middle range under stable conditions.

The fraction can also change with the sensory dimension and task. A value estimated for one kind of intensity cannot be copied to frequency, duration, size, motion, or number. Differences among observers, training, sensory status, adaptation, and attention matter too.

Stevens' power law and other countermodels

Fechner's logarithm is not the only proposed relation between physical stimulus magnitude and reported sensation magnitude.

S. S. Stevens used direct magnitude-estimation methods and argued for a power function. In its simplest form:

ψ = aI^n

The exponent n changes with the sensory dimension and method. Values below one produce compression; values above one produce expansion. Stevens' 1957 article established the power law as a major competing account of suprathreshold scaling. (Stevens, 1957)

This does not mean that one power function universally replaces the logarithm. Over a limited range, different curves can resemble one another. The result can depend on how people report magnitude, which levels are sampled, and whether the research target is discrimination, subjective report, or neural coding.

A review by Kenneth Johnson, Steven Hsiao, and Takashi Yoshioka makes the counterevidence especially concrete. The authors compare logarithmic, power, and linear proposals and report tactile-roughness studies in which subjective reports related linearly to a particular neural coding measure. (Johnson, Hsiao, and Yoshioka, 2002) That bounded tactile result is not a law for every sense. It is enough to reject the shortcut that neural firing must always be a logarithmic copy of physical intensity.

What the law does not explain by itself

The Weber-Fechner law is a psychophysical model family, not a cognitive-bias diagnosis. Failing to detect a small increment does not show irrationality.

It also does not establish:

  • one logarithmic receptor or neural-firing mechanism;
  • the true sensation curve for every modality;
  • a universal value of k, K, or the power-law exponent n;
  • a perceptibility guarantee for a display, alarm, control, or product change;
  • accessibility or safety for a particular user population;
  • how a person will value money, judge a price, or make a decision.

A logarithmic engineering scale can be useful without proving that the mind follows Fechner's equation. Likewise, a constant ratio can be a sensible design hypothesis without being an empirical threshold.

Sources