Skip to main content
CoolColor Space

How vision evolved: see the world through early eyes

Step from a microbe that only knows light from dark to a lens eye and color. Each stage turns a sample photo, your own photo or video, or your live camera into roughly what that kind of eye can resolve. Everything runs on your device: nothing is uploaded. The stages are a ladder of eye designs drawn from different branches of life and ordered by eye type; a box jellyfish is not our ancestor. How each step happened, with sources: how vision evolved, the guide.

Assumption: the picture spans 60° from side to side, our own choice. Angles from the research are turned into pixels on that basis, so a wider or narrower lens would change the scale.

Archaea with sensory rhodopsin I

Age
Living microbe, used as a model of the simplest light sensing; when it arose is not dated in our sources
Senses
Light versus dark only — no direction, no image (Nilsson’s class I, detection angle 360°).
Drawn with
One brightness value for the whole picture; the 360° field is drawn in the inset (wider than the picture, so it cannot change it)
Source
Spudich & Bogomolni (1984), Nature 312:509.; Olson, Zhang & Spudich (1995), sensory rhodopsin I signalling.; Nilsson (2013), Visual Neuroscience 30:5–20.

Its first “color” is a behavior, not a picture: sensory rhodopsin lets the cell tell colors apart — orange light attracts it, and two-photon signalling repels it. That cannot be drawn on a screen, so it is only described here.

Approximate. The picture is assumed to span 60° from side to side. The patch, spacing and blur angles (25°, 10°, 20°, 11.25° and 1°) are converted to pixels on that basis; the 360° and 180° fields and the 21° wedge are drawn as symbols, not to scale. Brightness, noise and what the animal makes of the image are not modelled.

How the pictures are made

The stages follow the four classes of eye function in Nilsson (2013), from no direction at all to a sharp image. Several stages are living animals that stand in for an early design, so the order is by eye type, not strictly by date.

An angle is treated as the width of a blur spot at half its height. With a 60° picture, a 25° spot covers 42% of its width. Brightness is averaged in linear light: every pixel is decoded from sRGB before it is averaged, in the patches, the ommatidia and the blur alike. Gray stages show brightness only. The early mammal is drawn from the two cone pigments reconstructed for the mammalian ancestor, with the same two-cone method as our animal pages. The lamprey is shown in human color, because its cone peaks are not in our sources.

Nilsson’s four classes and their detection angles (Table 1 of Nilsson (2013), Visual Neuroscience 30:5–20.)
ClassWhat it can doAngle
I, non-directionalLight versus dark only360°
II, directionalWhich way the light comes fromAbout 180° (receptors “typically collect light over 100–180°”)
III, low-resolution visionA crude image25°, 40° in the relaxed limit
IV, high-resolution visionA sharp image1°, 5° in the relaxed limit

Stages we do not draw

  • Algal eyespots (Chlamydomonas, Euglena). A single cell finds the direction of light by shading its photoreceptor as it rotates. It would look like the cyanobacterium’s arrow, so it is not drawn separately. (Kreimer (2009), Current Genetics 55:19.)
  • Warnowiid “ocelloid”. A single-celled plankton organism with analogues of a cornea, lens and retina inside one cell. What it does with them is unknown, so any picture would be invention. (Gavelis et al. (2015), Nature.)
  • First animal opsins. The first animal opsin arose between 755 and 711 million years ago and might not have been light-sensitive; the main opsin families split about 711–700 million years ago. A date, not a view. (Feuda et al. (2012), animal opsin evolution.)
  • Anomalocaris, about 515 million years ago. At least 16,000 lenses per eye are known from fossils, but its acuity and its color vision are not, so we do not draw what it saw. (Paterson et al. (2011), Nature.)

What this does not show

  • The colors of extinct animals: opsins do not fossilize, so no Cambrian palette is shown.
  • How bright or noisy each eye’s image was. The Nautilus view is darkened for illustration only.
  • What the animal experienced, or what its nervous system did with the signal.
  • Ultraviolet or four-cone color, which an RGB screen can only fake with false color.

The story behind the stages

When each kind of eye appeared, what the sources do and do not say, and why the order is by eye type and not a family tree: how vision evolved, the guide.

Sources

  1. Spudich & Bogomolni (1984), Nature 312:509.
  2. Olson, Zhang & Spudich (1995), sensory rhodopsin I signalling.
  3. Nilsson (2013), Visual Neuroscience 30:5–20.
  4. Schuergers et al. (2016), eLife.
  5. Boden et al. (2021), Nature Communications 12:4742 (molecular clock).
  6. Nilsson (2009): the 180° field of the Platynereis larval eye.
  7. PNAS (2017), the compound eye of the trilobite Schmidtiellus.
  8. Bielecki et al. (2013), citing Nilsson et al. (2005) and Garm et al. (2011).
  9. Garm et al. (2007), box jellyfish opsin.
  10. Muntz & Raj (1984), J Exp Biol 109:253.
  11. Lamb, Collin & Pugh (2007), Nature Reviews Neuroscience 8:960: evolution of the vertebrate eye.
  12. Yokoyama et al. (2005), Genetics 170:335: cone pigments of the mammalian ancestor.
  13. Hunt et al. (2009), review: mammalian cone genes reduced from four to two.
  14. Genome study of mammalian cone opsins.
  15. Dartnall, Bowmaker & Mollon (1983), human cone peaks.
  16. Frontiers in Ecology and Evolution (2017): primate trichromacy from duplication of the LWS gene.
  17. Kreimer (2009), Current Genetics 55:19.
  18. Gavelis et al. (2015), Nature.
  19. Feuda et al. (2012), animal opsin evolution.
  20. Paterson et al. (2011), Nature.