Illusions like this always remind me of failure modes of image classifiers - where an AI thinks a sunflower is a panda or whatever, because it's never seen a sunflower before.
That is, presumably the illusion works because it depicts something we've never needed to distinguish before, in such a way that it shares salient features with something more familiar (in this case, oddly-textured concentric rings and normally textured spirals, respectively). Or am I spouting nonsense?
Over millions and millions of years of evolution, many mammals (including us) adapted our vision centers to quickly process large chunks of environmental information by developing a series of beneficial hard-wired processing networks in the visual cortex. These neural pathways allow us to do things like subconsciously notice small movement or even judge distances fairly accurately without parallax. Most of the time it works pretty well; I mean, it kept us as a species alive up to today. However, illusions like this "exploit" the "visual hacks" our brain uses (that allow us to process such large amounts of visual data so quickly) by creating objects that don't (or can't) exist in nature. Our brains are fairly picky about things like depth-of-field and textural shading and oftentimes their limits can be demonstrated by simply playing with a few colors and changing a few angles.
Cats that grow up in an environment without any horizontal lines are unable to see horizontal lines [1]. This indicates that it might not be as much "evolution made me recognize these shapes" as it is "I learned to recognize these shapes since I see them often", which is more inline with OP's explanation. Of course evolution plays into what you are able to learn to recognize in the first place.
Minor correction: Predators like lions have no problem distinguishing individual members of a herd. The stripes pattern messes with the way blood-sucking, disease-spreading flies (such as horse flies or tsetse flies) see the world.
Yeah, that set is a great one. I've noticed the scrolling effect too, it definitely intensifies the movement. I suspect the cause of the movement is the saccades shifting where it is in the visual field slightly, and scrolling is a much bigger shift.
I noticed that the rotating snake illusion only seems to occur when my eyes are actually moving; if I focus on a specific spot for long enough, they seem to all stop spinning.
Are there examples of these patterns in use in real world logos or, e.g., website designs? Not designed to make people dizzy, but used to give an illusion of animation maybe?
> Yup. Those aren’t spirals. They’re all concentric circles.
Except they are spirals. The black patches form the implication of a spiral - this is where your visual system picks out the spiral from.
If you were to say that there are technically no spirals in that image because they're not contiguous, then you could equally say that there are technically no circles either, only an arrangement of square boxes that is strongly suggestive of a circle.
>If you were to say that there are technically no spirals in that image because they're not contiguous, then you could equally say that there are technically no circles either, only an arrangement of square boxes that is strongly suggestive of a circle.
There are circles though. The grey concentric rings.
If you removed them, you'd see the black and white pattern spirals towards the center.
This is something I've been tempted to blog about. As you and my sibling poster point out, the illusion is actually essentially composed of the things they claim aren't there. The same goes for the Kanizsa triangle, Fraser's illusion and many others. The simplest description of each image matches what our eyes see, but the creators say "Ah, but I created it this other way, so your eyes are wrong".
For me the illusion here is no trick. Looking at it I have considerable difficulty not seeing a single unbroken line spiraling out from the center to the perimeter. In reality there are huge disconnects, it is concentric rings and not a continuous line at all. Even when I force myself to see the circles they look like they are all trying to float away from each other in opposite directions. Almost poetic, really.
The spirals are actually there, partially concealed by plain circles, so seeing spirals is not really incorrect. Your visual system is good at seeing concealed things. That's why I disagree with it being classified as an illusion. Though not as much as with the Kanizsa triangle.
"The spirals are actually there" - the unbroken spiral is not there, and it is an illusion of one, not multiple. At least to me. I guess if you want to be particular, the circles are broken by the pixels on my monitor, but at 4k that isn't perceptible. I guess if one gets far enough away from the monitor that one can no longer resolve the space between the circles it could be said that the spiral is actually there.
Focus on the gray, and let your mind put it as the foreground, and circles, let them be a low frequency background. It may be your best chance to not see the illusion.
I also think that as you become more trained in graphics (and illusions in general?) you become more adept at stripping the component that makes up the illusion.
Also the challenge that comes with "Warning!! You cannot unsee this!!!" probably helps too.
Worked for me on the squared circle but took concentration on the spiral. It changed as I looked at it but not like before. Different ways of focusing on it made it go static vs moving.
I do better on the spiral one when I imagine I am slowly floating down a tunnel, like a game of Luxuria Superbia. I can see an individual circle or two pop out at a time, except for the smallest ones :/.
It could be said that his explanations of the illusions are wrong: there is a spiral in the final image, it's just masked out by concentric circle outlines. Particularly toward the center of the image where the resolution is so low that it all blurs together anyway it is hard to argue that is not a genuine (if blurry) spiral.
When learning to parse the concentric circles with eighteen patches of mirror-spiraling dot sequences, my thought was: "Look at the gray areas, duh. Just like life. They're what's actually there, the rest is a distraction designed to deceive your natural perception." > woah.jpg
It's like generative adversarial samples for biological neural nets. I wonder is someone could weaponize this so if would cause lasting dizziness and disorientation after a prolonged exposure.
Similar device was used in the sci-fi novel Snow Crash. It appeared like a black and white image of static but in fact contained a subliminal message which tapped into some subconscious low-level cognitive functions of the brain. Just by looking at it, it reprogrammed the victim's brain and made them into a mindless zombie.
Well, it's Phil Plait, the Bad Astronomer, and it's kind of his style.
I like his style, but I knew him when we were in elementary school together back in the early 70's (he was into telescopes and stuff even then), so I cut him some slack.
Yeah. Despite the style, he's strictly no nonsense (after all, the original point of his blog and website was to refute bad astronomy in popular media.
If you cover one eye and look at the image of two concentric circles, it gives a slightly exaggerated illustration of how I see text with both eyes open due to misalignment. That's why I hate two-space indentation with a passion. Saving this for the next tabs vs. spaces debate.
I thought it looked like snakeskin. And then wondered about evolutionary adaptation, i.e. making it harder for predators to accurately see where the snake is.
Was reading about a dinosaur discovery that allowed them to project colouration, turns out the bottom was brighter than the top, an adaptation prey has so that their shadow is the same brightness as their top half, deer have the same adaptation, evolution is amazing!
I read that certain optical illusions are not experienced as strongly by native, non-city-dwelling populations. It was hypothesized that this was due to the relative absence of 90-degree angles and "carpentered" features in their environments [1]. Looking it up again now to refresh my memory, I see that this explanation is contentious and there may be other reasons for the variation.
The Wikipedia page also mentions papers about pigeons [2] and a parrot [3] reportedly being susceptible to the illusion. I'm pretty sure the birds used in those experiments live in carpentered environments that are heavy on the 90-degree angles. A Smithsonian magazine article on the same topic says that "a computer trained to mimic the perceptions of the human eye was also susceptible" [4].
Not necessarily bugs. These are limitations which are bugs or not, depending on requirements.
Camouflage does exploit limitations in our vision, but surely limitations, and therefore hiding strategies, will exist at some level, no matter what.
Overall, these seem like very minor limitations compared to others, like our propensity for near-sightedness. And in despite those much more significant limitations, it seems like human vision has been a massive success.
The reason is, of course, that there's nothing in nature that would trigger this false perception. And even if there were, it's difficult to come up with a scenario where that would cause enough selection pressure to fix it (given that it's probably caused by complex interplay of traits that are in other contexts extremely adaptive).
Perception can be tested in many other species with the kind of experiments that B.F. Skinner was famous for: condition them to respond a distint way in the presence of particular stimuli and then show them different things and see how they react. With enough carefully chosen training and observations it is possible to get a good sense of what they are perceiving. I don't have any links handy, but I think there was quite a bit of this type of thing done at one point.
I normally find optical illusions interesting and fun. Here it's like the fun guy at the party has had too much to drink and is now obnoxious and a bit nauseating.
raises hand I've been doing 3D modeling and photo manipulation for over a decade. I learned to dismantle these illusions, focusing more on the "how" than the "wow". I've dabbled in making illusions. http://imgur.com/J3kSB1c
This effect is kind of similar to what makes subpixel font rendering look horrible when the wrong LCD subpixel order is selected. Or why text looks jaggy on pentile displays, especially while scrolling.
That is, presumably the illusion works because it depicts something we've never needed to distinguish before, in such a way that it shares salient features with something more familiar (in this case, oddly-textured concentric rings and normally textured spirals, respectively). Or am I spouting nonsense?