Since infancy, our world has been like a vibrant painting, mixed from a palette of optic primary colours: red, blue and green. But for birds, the planet is a much more lurid place. Our feathery friends cannot only see our spectrum but beyond into the UV. Colour photoreceptors in the retina called cones are stimulated by light. In Man, the relative activity of three cone types measures the amount of red, green and blue wavelengths, and determines which `colours' we see. The combined output of these cones measures total brightness. Some birds like quail have a fourth blue cone with sensitivity into the UV, whereas others like starlings posses a fourth cone for UV alone. It has long been known that birds can detect UV, but no one was sure whether birds see it as colours or as increased brightness. Emma Smith, Verity Greenwood and Andrew Bennett set out to investigate this dilemma by giving starlings and quail a series of UV `colour-blindness' tests, and found that birds can actually see UV colours (p. 3299).

The birds were shown eight tiled squares, covering food wells, and were taught to associate a certain colour with a food reward. Smith was disappointed early on when the birds refused to learn but soon realised that if she weighted the squares, the birds became more precise in their behaviour. Moving a weight introduces a `cost to being wrong,' she explains. But progress was still painstaking, as the starlings dashed about like excitable children and the quails got fed up and fell asleep!

Firstly, the birds had to learn to choose between a tile with a UV tint and a tile without. Tiles like this would look identical to you and me but are discernable with UV vision. Some birds were trained to select the UV tile and some the non-UV tile. The fact that the birds could learn to pick the correct tile proves that they can tell the difference between them. In other words,the birds can detect the presence of UV. Nevertheless, this test cannot determine whether the brids are seeing UV colours or just a brighter looking tile.

The second experiment also used UV versus non-UV tiles, but this time the absolute brightness was randomly variable, so that there were no rules to learn. Only UV colour was controlled and could be used as a cue to tell which tile was which. These tiles still look identical to Man and can only be told apart by an animal that can see UV as a colour.

Smith and colleagues watched as both bird species successfully learnt all experimental tasks, confirming the birds' UV colourvision. Countless extra tests guaranteed that the birds were really seeing colours, and were not cheating the system by using smell or texture cues instead.

Although the group were originally teased for choosing quail, often seen as dim-witted animals, both bird species were very successful at learning the visual tasks. Smith feels this work provides the first watertight study showing that birds can see UV as a colour and hopes it will lead on to more colourful research.