Several photos have surfaced where identical copies of pictures of a road placed side-by-side look different. What causes our brain to misinterpret something like that?
Over the past couple of weeks, a particular image has been doing the rounds on my Twitter timeline. A screenshot of a reddit post, the image contains what appears to be two pictures of roads laid side by side, with a text on top that says that the two pictures are in fact identical.
My brain is hurting pic.twitter.com/IA5b8Anmat
— tag (@mxdric) February 6, 2018
It’s true, the two images are actually identical. There are several methods to verify this, of course. The easiest would be measuring the angles of the roads to the base of the image. If you have experience editing images, the two can be layered on top of each other for comparison.
If you’re a command line geek, the UNIX command ‘cmp’ compares two images pixel-for-pixel. Or you can do the lazy test I did without getting off my couch: take multiple screenshots of the roads, crop them, look at the thumbnails in your gallery.
Ooh nice. I went a little further and took screenshots of them and tried different combos. Then blocked the left most one with my thumb each time and they all look identical 🙂 pic.twitter.com/d2UPTZiiSt
— Sandhya Ramesh (@sandygrains) February 6, 2018
If you think this illusion has been around forever, you would be surprisingly wrong. It was discovered just 11 years ago by the trio — Fredrick AA Kingdom, Ali Yoonessi, and Elena Gheorghiu — from the ophthalmology department of McGill University, Canada. This illusion is called the ‘leaning tower illusion’, and comes from this picture that sparked the discovery in November 2007:
This illusion is a classic visual illusion where the point of confusion stems in the mind and not from the lighting or shadows in the image itself. Thus, it is not an optical illusion, like ‘The Dress’.
The leaning tower illusion is a prime example of how our brain processes visual information in the absence of context. We see the world in 3D and our brains are hardwired to correctly identify perspective and build a world around the object we see. That’s why an object farther away looks small, and therefore when you see it on paper, you expect it to look small too.
Our brains cannot distinguish the two lines from the rest of the image. Similarly, our brains can also not distinguish the two leaning towers as two different images. We see them in 2D but our brain reconstructs them in 3D as a single image in real time. If we follow the outlines of each tower upwards, we know that they would eventually join into two points on the 2D plane. From our perspective, they would recede from view by converging into a point. But our brain knows that the towers are not conical and don’t converge in real life.
Kingdom, Yoonessi, and Gheorghiu explain this in their paper with the example of a photo of the Petronas Towers.
The two towers are receding from our point of view in the image. Their individual outlines, if extended beyond their top, will converge to the same point. But our brains know that in reality the towers do not converge; so it diverges the towers to make up for this perspective flaw, enabling us to understand that the towers rise at the same angle from the ground and will never converge together.
This is exactly what happens in the leaning tower illusion as well. The top halves of the image look identical if the bottom is blocked off, but the instant we start seeing the lower parts of the two leaning towers — the part that is visually close to us in 2D perspective, our brains start diverging the towers.
Perhaps what’s even more interesting is that the illusion is not accumulative. In the screenshot of my own phone’s gallery, I could see that the roads in each image don’t diverge increasingly. Only the first one appeared different. And when I blocked the first image, the next one took its place. This can be seen here as well:
The simplest reasoning for this, according to the authors of the paper, is that if each subsequent tower tilts more than the previous one, the rightmost will ultimately topple. But if you remove the rightmost one, it’s the one to its left that should topple. This process can be repeated over and over again, with no way to arrive at exactly which tower should topple. This goes against the brain’s ability to reconstruct 3D from 2D to be accurate to real life. Therefore our brain doesn’t do all that unnecessary computation.
The illusion is easy to recreate. All that’s required is an image with a receding outline or diagonal lines. Here are some more examples created by Akioshi Kitaoka, an experimental psychologist who specialises in visual illusions.
Have you any images in your gallery with receding lines that you can duplicate and place side by side? Tweet your results to us at @ThePrintIndia.