The image below depicts a strawberry tartlet. Colour is a bit on the greenish-blueish side, but the strawberries still are red, no? Check what happens if you move the mouse over the image (or tap it).

In fact, there is not a single pixel that's red here, or even reddish. If the image is masked, leaving only what for me were some of the reddest parts, the seeming red (mostly) disappears.

That there is no red here can also be seen from the distribution of all colours (left): they are all left of the central colour neutral. To make it more obvious, I have added 3 rectangles with mor saturated colours below (left), and that shows up clearly in the colour distribution (right).

I had developed this program, plotting the gamut of all colours from an image, for “The Dress”. It can be downloaded there.


On the left (bottom), the added rectangles are obvious. On the right, they leave their sign in colour space. The reason that there is not just one colour locus for the rectangles is JPEG compression which I did before running my colour analysis program.

How can we explain this?

Many people suggest that we see strawberries automatically as red. Indeed there is evidence that expectency affects colour perception to some degree (Hansen et al. 2006), my simple modification below demonstrates that there is more at play here.


On the left I replaced the strawberries with similarly coloured ovals. No red in the picture, as seen above. The ovals still appear reddish, although these are no strawberries no more…

Comments

This is a nice & strong demonstration of colour constancy, created by Akiyoshi Kitaoka in 2015. I see 3 mechanisms at work here:

  1. The predominance of green-blue shifts the perceived white point, giving rise to a reddish perception of areas which are objectively white/grey.
  2. Simultaneous colour contrast also plays a role. I deduce this from the finding that I had to make the cutouts for the mask above very specific –  as soon as a shadow is included (e.g. in the dimples of the strawberries) some red appears.
  3. Object-specific memory is also involved: From our extensive experiments with deuteranopes (n=1) we find that some ovals (above) only appear brownish, while the strawberries appear clearly red.
    However, this might be in another way object-specific, namely due to the dimples providing local shadow-colour information. More research seems necessary…

Sources

Kitaoka A (2015) Apparent reddish strawberries. ECVP contribution

Hansen T, Okkonen M, Gegenfurtner K (2006) Memory modulates color appearance

 

Created: 2017-03-05


|  Impressum·Legal notice  | 
Last update 2017-03-06 by Michael Bach (G+)