Clare explains: What's up with this optical illusion?
This is a bit different from my usual blogs in that I’ve used my own knowledge to explore why an optical illusion works rather than integrating information from different sources to tell you a story. I hope you like it!
If you’re on Twitter, you’ve probably seen this picture by now.
A friend alerted me to it hoping that I could explain what was going on, because as someone with a PhD in the psychology of human perception I am pretty well placed to do so. I can’t honestly say that I’ve got a definitive explanation, but I do have some theories.
1. The layout of the eye
You might remember from secondary school biology lessons that there are two types of cells in your eye that can detect light, called rods and cones. Rods are really good in low light, but they can’t detect colour and they don’t have very good acuity (that is, they don’t give a very sharp picture of the world). Cones, on the other hand, are good in daylight, can detect colour and do have good acuity – but we have very few cones and lots of rods.
Cones and rods also aren’t evenly distributed across the retina, the sheet of cells at the back of the eye. One part of the retina, the fovea, has almost all the cones and none of the rods. Further away from the fovea, this situation is reversed.
When we want to look at something in detail, we move our eyes so that whatever we want to focus on is available to the fovea to get a pin-sharp picture. Outside the fovea, things are a lot blurrier because there are so few cones.
You might notice that the illusion of the whole picture being coloured is stronger if you’re not looking directly at it. This is because we haven’t got many cones available to tell us what colour something is or exactly where it is, so our brain has to play fill-in-the-gaps in the same sort of way that it does in the blind spot. However, when I did a small Twitter poll, it was clear that not everyone sees the illusion more strongly when they’re looking at it from the corner of their eye. Something else must be going on, too…
In his original tweet with the photo, @page_eco says What you “see” is what your [brain] predicts the reality to be, given the imperfect information it gets. I think he’s referring to what psychologists typically call top-down processing.
In top-down processing, what we perceive is heavily influenced by our earlier experiences, so to some extent what we perceive is what we expect to perceive. For most people, the world is in colour and these days so are most photos, so when we’re shown a photo of the world, we expect it to be in colour.
Here’s a beautiful illustration of someone experiencing top-down processing in two different ways – first based on the expectation of seeing a colour photo, then based on the expectation that the tweet itself creates!
We can contrast top-down processing with bottom-up processing, in which the information that comes into your senses is built up into a perception of what’s there in the world. In the case of the photo, that means you’ll probably see the photo as black and white with a coloured grid overlay – and if you look through the replies to @page_eco you’ll see a lot of people saying that that’s all they can see!
Why such different responses? In reality, our brain probably does a mixture of both top-down and bottom-up processing. It’s useful to have some flexibility about which we use because top-down processing saves time, but the world isn’t always as we expect so it’s important to have the capacity to perceive surprising things like large predators heading towards us.
3. Different colours
One last thing I noticed was that the illusion works better in some areas of the picture than others, which also seems to be the case for other people. Specifically...
Luckily, I spent a large part of my life as a researcher dealing with how people perceive colours, so I was actually able to do some maths that will help us understand why this is.
To start with, you need to know that we can map all the colours that we can perceive in a ‘colour space’, which is kind of like a 3D graph with a rainbow in it:
There are quite a few different colour spaces, but the one I like the best is called CIELUV. The reason I favour it is that it’s a ‘perceptually uniform’ colour space, which means it’s tailored to how humans tend to perceive colours rather than, say, how much light a screen needs to emit in order to create a colour.
Looking at the CIELUV graph, it should be clear that colours that are further apart are more different from each other than ones that are close together - for example, yellow is more different/further from blue than blue is from purple - and we can use maths to figure out the exact distance between two colours. (If you want to see the specific equation I used, please check out this website.)
Since several people said that the ‘red’ t-shirts in the image didn’t work as well for them as the other parts, I compared a pixel from the red line with a nearby pixel from the greyscale part of the picture, then did the same with pixels from t-shirts with different colours. The red pixel and its nearby grey pixel were much further away in colour space than the yellow, blue or green pixels were from their nearby grey pixel. This means the colour and grey parts are more different from each other in the red areas than in other areas, and since we’re likely to notice this large difference, the illusion doesn’t work as well in these areas. Of course, this is only based on a few pixels, so we can’t be sure that it’s the case for all the pixels, but it gives us a good indication.