Hearing, Vision & Perception Curves

In the 1930s, Harvey Fletcher and Wilden A. Munson at Bell Labs measured equal-loudness contours: pure tones at many frequencies, adjusted until listeners judged them equally loud. The result is not flat. Human hearing is most sensitive in the midrange (roughly 2–5 kHz) and less sensitive at low and high frequencies, and the shape changes with level. Modern charts use ISO 226 (this page plots the ISO 226:2003 formula with the standard 29-frequency coefficient table)—the same idea as Fletcher–Munson, refined by later multi-lab data. That is why A-weighting (dB(A)) and loudness models exist: an unweighted SPL meter lies about how loud something sounds.

Vision science did the parallel job for light. The CIE 1924 photopic luminosity function V(λ) (and the scotopic V′(λ) for dark-adapted vision) weights spectral power by how bright it looks to a standard observer—peaking near 555 nm in daylight vision, not at the blue or deep red ends. Separately, the CIE 1931 color-matching functions (x̄, ȳ, z̄)—built on Guild and Wright’s trichromatic matches—reduce any spectrum to three numbers (XYZ). Luminance Y is essentially the integral of SPD × V(λ). So “brightness” in imaging is already a Fletcher–Munson-class curve for light: physics filtered by a standardized eye.

They are not the same curve and not the same variable (Hz vs nm; loudness vs brightness/chromaticity). The correlation is the measurement idea: both are standardized ways to quantify human perception of energy across a spectrum. Equal energy at different frequencies or wavelengths is not equal experience. So both fields invent (1) a standardized observer (average listener / CIE standard observer), (2) weighted meters (dB(A), luminance cd/m², colorimetry), and (3) delivery systems that spend bits and power where the curve is steep (audio codecs and equal-loudness contours; gamma/PQ and camera log for vision). When an audio engineer says “flat isn’t neutral,” a colorist already knows the sentence.

Harvey Fletcher & Wilden Munson (Bell Labs, equal-loudness, 1933). Later contours: Robinson–Dadson; modern reference ISO 226. On the vision side: CIE photometry (V(λ)), CIE 1931 standard observer (Wright & Guild data), Judd–Vos and CIE 2006 refinements for some applications. Neither program “invented hearing or vision”; both quantified the weighting so engineering could stop arguing adjectives and start meeting numbers.

Broadcast: program loudness (LKFS/LUFS), legalized levels, A-weighting on set and plant noise, speaker and control-room monitoring that is not “flat to a mic.” Cinema: theatrical mix reference, show LUFS practice, and picture pipelines where luminance is V(λ)-weighted, grades live in CIE-based spaces, and HDR PQ follows contrast sensitivity—another perceptual curve. Virtual production: LED volumes, XR stages, and hybrid rooms with PA and picture in the same bubble fail when one side measures physics and the other measures perception. This module is the handshake diagram: same human, two spectra, weighted meters on both.

Not a claim that vision and hearing share the same shape. Not medical audiology or clinical color vision. Not a substitute for ISO 226 tables or CIE publications in regulated work—the plots are teaching reconstructions of classic contour/sensitivity shapes. For metrology, use the current standard documents and calibrated instruments.