The transition from photopic to scotopic vision is accompanied by a shift in the peak of spectral sensitivity

What it is

Switching from a bright environment to dim light involves more than an overall rise in sensitivity (dark adaptation) — there's an easily overlooked detail: which wavelength band the eye is most sensitive to also changes. In bright conditions the eye is most sensitive to yellowish-green light; once dark vision takes over, the most sensitive band shifts overall toward blue-green — a phenomenon called the Purkinje effect (or Purkinje shift).

The everyday version of this is observable: red and blue flowers that look equally vivid and equally eye-catching in daylight will, as dusk falls and light dims, show the red flower fading noticeably faster into the background than the blue one, even though the two were evenly matched in salience under bright light. This isn't the red physically getting dimmer — it's a systematic reversal in the visual system's relative sensitivity to the two wavelength bands.

Why it happens

The direct cause of the Purkinje effect is that the cone system's and the rod system's spectral sensitivity curves simply don't peak at the same place to begin with: the cone system (dominant in photopic vision) is most sensitive near yellowish-green wavelengths; the photopigment rhodopsin carried by rods has an absorption spectrum that peaks at a more blue-green wavelength. As light gradually dims, dominance in the visual system gradually shifts from the cone system to the rod system (this shift is itself the same turning point on the dark-adaptation curve where rod sensitivity overtakes cone sensitivity), and the whole "currently operative sensitivity curve" correspondingly transitions smoothly from the cone-dominated curve to the rod-dominated one, with the peak sliding from yellowish-green toward blue-green along the way.

This explains why red objects lose their salience especially fast in a dimming environment: red light's wavelength sits not far from the cone sensitivity peak, but far from the rod sensitivity peak (rhodopsin's absorption peak); once dominance shifts to the rod system, which barely responds to long-wavelength light, the response a red object generates drops sharply. Blue-green objects are the opposite case — their wavelength sits closer to the rod sensitivity peak, so they come out relatively less disadvantaged after the shift, and can even appear brighter than a red object of equal physical luminance under pure scotopic conditions.

Studying it

The standard approach is spectral sensitivity measurement across adaptation states: under fully light-adapted (cone-dominated) and fully dark-adapted (rod-dominated) conditions respectively, detection thresholds or subjective brightness matches are measured for a series of different wavelengths, and two sensitivity curves are plotted and compared to directly quantify the difference in peak wavelength position and the magnitude of the shift. A brightness-matching paradigm can also be used: participants judge which of two colors (a red and a blue, say) looks brighter under both photopic and scotopic conditions, checking whether the judgment reverses with adaptation state. The independent variables are test wavelength and the participant's current adaptation state; the dependent variable is detection threshold or subjective brightness judgment.

This phenomenon rests on solid, well-understood evidence and belongs to fairly classic basic vision science; applied measurement specifically targeting how much the Purkinje effect actually matters in interface or product scenarios is comparatively rare — existing design recommendations are mostly reasonable expectations derived from this basic mechanism rather than the results of dedicated interface experiments.

Where it stops holding

• The effect is easiest to notice in the transition zone between photopic and scotopic vision (mesopic vision). Under pure photopic vision, rods aren't dominant and the effect doesn't appear; under pure scotopic vision, cones are essentially not participating, so a color judgment like "which is more salient, red or blue" already falls outside what rod monochromacy can support, and the comparison loses its meaning.
• This entry is about a change in relative brightness/salience, not about color vision itself increasing or decreasing. The Purkinje effect occurs during the mesopic stage where the visual system can still draw on cones to some degree — a different issue from the more complete limitation of having no color vision at all under full scotopic vision, and the two shouldn't be conflated.
• The exact magnitude of the shift and the illumination range where it occurs vary between individuals, tied to factors like an individual's dark-adaptation speed and age-related changes in photopigment density; there is no single, precise turning-point illumination shared by everyone.

Applying it

• For key indicators that need to stay salient in a dimming environment (safety markers, emergency status indicators, elements that must keep drawing attention through dusk or a gradually darkening scene), don't rely solely on a red-family color scheme tested only under bright conditions — red-family indicators are exactly the category whose salience drops fastest under the Purkinje effect, so the risk shows up first as light dims.
• Blue-green elements tend to hold onto their salience better in a dimming environment; if a scenario itself goes through a continuous transition from light to dark (an in-vehicle interface used around sunset, outdoor equipment operated at dusk), consider shifting key status colors somewhat toward blue-green, or providing a color scheme that switches as ambient light changes.
• Don't conclude that a color which "tested very visible under office lighting" will be equally visible in a real dimming-light use scenario, especially for red-family critical indicators.
• Verification: test candidate critical indicator colors under lighting that simulates a continuous transition from light to dark, and watch for a noticeable, premature drop in salience as the light dims; if that occurs, prioritize switching to a blue-green color or adding a non-color salience cue (shape, flashing, position).

Related

• Same group: A1.18.1 Rods handle dark vision without color discrimination; cones handle bright-light vision and color
• Nearby: A1.07 Light/dark adaptation and adaptation time constants · A1.05 Color vision and opponent-channel mechanisms
• Search terms: Purkinje effect · Purkinje shift · mesopic vision · spectral sensitivity