The lines are all the same shape

December 9, 2017 11:12 PM

A New Optical Illusion Was Just Discovered, And It's Breaking Our Brains - "Researcher Kohske Takahashi calls it the 'curvature blindness illusion' and it's very trippy."

also btw...
This is Akiyoshi Kitaoka messing with your brain - "He's a professor of psychology at Ritsumeikan University in Kyoto. He's spent a long time collecting and perfecting illusions." (previously, too)

posted by kliuless (40 comments total)

Piss off! I'm not having this! I blame the illusion upon an undigested bit of beef, a blot of mustard, a crumb of cheese and a fragment of underdone potato to boot!

Real eyes realise real lies 😭
posted by I'm always feeling, Blue at 11:20 PM on December 9, 2017 [25 favorites]

I mean, it's also an amazing discovery as it pertains to subtle misapprehensions arising from human perception.

But mostly just generalised grar about having my eyes tricked into doing a thing - see above.
posted by I'm always feeling, Blue at 11:25 PM on December 9, 2017 [3 favorites]

What species of satanic deceit is this??
posted by Chitownfats at 12:55 AM on December 10, 2017

He has the best job ever. Also, whoah white hearts yellow hearts.
posted by Literaryhero at 1:46 AM on December 10, 2017 [13 favorites]

Imagine if you printed this pattern onto a white/gold dress.
posted by greenhornet at 3:07 AM on December 10, 2017 [6 favorites]

I am curious, is this illusion effective on all humans?

I live in a (roughly, don’t get me started on buildings not actually having right angles) rectangular building, and have since I was born. Many objects around me have flat edges, too.

But that can’t be true for all humans everywhere, especially if you go back 10000 years. So, is this illusion a product of evolution, or of each person’s visual cortex optimizing to the world around them at an early age?

Suddenly I am really curious what illusions two year olds actually can see.

Also I am lying about objects in my world, as I am currently typing this on a rather fetchingly rounded rectangle.
posted by nat at 4:42 AM on December 10, 2017 [7 favorites]

I’m not sure. I remember there was at least one case of a child whose visual cortex didn’t develop the ability to recognise straight lines. Not getting into details because it was linked to a horribly abusive upbringing.
posted by um at 5:32 AM on December 10, 2017

What species of satanic deceit is this??

I dunno. I don't get the new Digg layout either.
posted by Ogre Lawless at 5:34 AM on December 10, 2017 [6 favorites]

Line detection in humans is la-la freakazoid. At one stage in the visual pipeline, as I understand it, all perceived ines are endless; it's only later that the cues which delineate (hah) the start and stop are added

My broken eyes/optical nerves don't always get that right these days. Words with descenders become underlined - in red yet, which makes spellcheck an amusing adventure - and clusters of letters with major vertical ascenders - t, l, i, j, etc - will merge or multiply, or grow/lose cross-strokes. It's a bit like late-onset dyslexia, and as one might imagine it makes life particularly entertaining for one who makes his living from the correct use of the written word.

A bonus, though, is a ringside seat in a living laboratory of visual perception. If you've ever doubted that all vision is an optical illusion, let me put those doubts to rest. The ones that are obviously weird to normal eyes, such as this one, are fascinating, but really - it's all like that.
posted by Devonian at 6:41 AM on December 10, 2017 [4 favorites]

My almost 6 year old sees the illusion. Haven't figured out how to ask my almost 2 year old yet.
posted by skynxnex at 6:41 AM on December 10, 2017 [1 favorite]

I don't feel like those sets of curves are the same shape over the white or black sections, either. I mean, they're not as clearly 'one set is pointy, the other set is curved' different, but they look like distinctly different curves to me. I'm not saying they are -- I trust that the illusion is what it says it is. I just find that it's also happening to some degree over the non-grey backgrounds.
posted by jacquilynne at 6:49 AM on December 10, 2017 [6 favorites]

This is why we frown upon pie charts, y’all.
posted by oceanjesse at 6:51 AM on December 10, 2017 [1 favorite]

is this illusion effective on all humans?

Well that's an empirical question that could be tested. But I'm gonna guess "yes". The corner features it's detecting aren't about square buildings; it's about shadows. And the sun casts straight line shadows everywhere.

Gonna say it now; some junior engineer overtrained the neural network around layer 7 of the visual processing system. It's overcompressing and making incorrect assumptions.
posted by Nelson at 7:01 AM on December 10, 2017 [7 favorites]

"Who you going to believe, me, or your lying eyes?" - Kohske Takahashi
posted by Frayed Knot at 7:25 AM on December 10, 2017 [3 favorites]

Yes, it seems to be a lighting effect to me: For the curve that stays curvy, my brain can't figure out a way for it to be in light and shadow the way it does while having sharp corners, so curvy it is; for the one that "becomes" straight line segments, light's"coming" from one direction and it works perfectly. I think that's also why this is prominent over the gray, where the color value difference is the largest for both tones of the line.
posted by seyirci at 7:32 AM on December 10, 2017

One thing different from the classic optical illusions, usually when you "see" the effect the image changes, sometimes oscillating between the two different things. This seems to be in the sci-fi-horror area of "there are things in the world that humans are unable to ever see, like ghouls or germs.
posted by sammyo at 7:42 AM on December 10, 2017 [5 favorites]

My whole worldview changed when I was 10 or 11 years old in a way that I have only sice experienced with a combination of very deep emotional pain and very potent psilocybe mushrooms.

I was very bored in class staring out the window at a giant flowering royal poinciana tree, having fun with the kind of natural optical illusions that take place when looking at the fractal structure of the boughs and branches and flowers against a bright sky with swift clouds.

And then it got me: All perception happens in the mind. Not the skin or the ears or the retina. The mind. And the mind can be changed and altered, nevermind the sensory organs.

In that moment God died (I had been a rosary reciting god fearing good Catholic boy), my parents became humans, my constant fear and worry melted to little puddles. For a second there it felt like I had been dreaming all my life and had just opened my eyes for the first time.

I regressed to the mean after a few weeks, but my mind was never the same.

I was the kid with optical illusion books insisting that you take another look, getting frustrated that your mind was not sufficiently blown.

I know now that some dude under a different tree beat me to this by a few thousand years, but at the time it felt like I had discovered something new and important that I had to share with everyone.
posted by Index Librorum Prohibitorum at 7:53 AM on December 10, 2017 [25 favorites]

Not to dispute your epiphany but you are literally wrong about "all perception happens in the mind". The fun thing about these optical illusions is they happen in the retina itself, in the very first layers of neurons. They turn raw light sensors into features like edges and normalized brightness values. It is the very definition of and inspiration for neural network computing and it's really cool how you can trick it with various illusions that reveal how the neurons compute at this pre-cognitive level. Outside the brain.

On second thought though you can keep your epiphany, as long as your definition of "the mind' includes both the retina and the brain.
posted by Nelson at 7:59 AM on December 10, 2017 [16 favorites]

nat: I am curious, is this illusion effective on all humans? I live in a (roughly, don’t get me started on buildings not actually having right angles) rectangular building, and have since I was born. Many objects around me have flat edges, too.

I very vaguely remember a study in which a group of people which wasn't raised around endless right angles was tested for their ability to recognize perfect right angles; they scored much worse on that visual test than the average Westerner. I don't recall whether the researchers bothered to find out if they were better at other visual tasks.

In a talk by Geoffrey Hinton, he said that we're very bad at recognizing right angles if we're told that we're looking at a diamond shape... but take exactly the same shape and tell us that it's a turned square, and we're suddenly great at spotting whether it has right angles or not.

In short: Brain works in mysterious ways.
posted by clawsoon at 8:10 AM on December 10, 2017

Oh yeah, since the brain is the mind organ and the retina develops from the embryonic diencephalon and is made of brain tissue then it is part of the mind.
posted by Index Librorum Prohibitorum at 8:18 AM on December 10, 2017 [2 favorites]

Offering that I have just shown this to the members of my household. The six-year-old was correctly fooled by the illusion right away. The four-year-old was also fooled by the illusion, though needed help with words to identify the differences. The forty-year-old was confused about why I was demanding he look at an illusion early in the morning. I think he was fooled, too.
posted by Sauter Vaguely at 8:18 AM on December 10, 2017 [2 favorites]

Wow I mangled that sentence but you know what I mean.
posted by Index Librorum Prohibitorum at 8:21 AM on December 10, 2017

Here's a plausible (to me) mechanism that would give rise to this illusion, which I posted when the question came up on the maths StackExchange.

Suppose there is a part of the brain human visual system (thanks Nelson) that processes curves and tries to classify them: is this a straight line? an arc of a circle? some other weird shape? In the "zigzag" case, the light and dark segments are both close enough to straight lines that the HVS goes, yeah, they're probably straight, my vision is a little blurry anyway. In the other case, the light and dark segments are closer to circular arcs, so that's what we see them as. (Personally I see the whole curve as "rounder" than a usual sine wave, which is consistent with this hypothesis.)

I assume the reason it doesn't work on the white and black backgrounds is because the contrast of the segments relative to the background is much lower. Presumably there is another part of the HVS segmenting the curve into pieces, and in the white/black backgrounds it decides the whole thing is a single curve instead of lots of segments.
posted by a car full of lions at 8:28 AM on December 10, 2017 [2 favorites]

I've just noticed that this is quite different from the hypothesis proposed by Takahashi in his abstract. Well, he's the expert.
posted by a car full of lions at 8:32 AM on December 10, 2017

> I am curious, is this illusion effective on all humans?

Some optical illusions are effective on cats. SCIENCE!

One of the things I've noticed since getting cataract surgery is that some optical illusions don't work on me any more. The Curvature Blindness Illusion works very well, but others that work on the ways we defocus our eyes to discern patterns do not work. I assume it's because I have fixed lens implants and no longer have motor control over eye focus. So for example the White Heart and Yellow Heart (scroll down) illusion doesn't happen.
posted by Subaru drwxrwxrwx at 8:49 AM on December 10, 2017 [2 favorites]

That heart thing appears to be fake. I held my hand over one side and it still flipped. I copied it out of the webpage and it no longer flips. I think we've been GIFed.
posted by njohnson23 at 8:56 AM on December 10, 2017 [2 favorites]

The color heart illusions have a couple animated GIFs. I assume they're to demonstrate how the effect is perceived. Below each of the animations are static images you can try in isolation to test the illusion for real.
posted by Subaru drwxrwxrwx at 9:12 AM on December 10, 2017

Here's a nice animated version to further boggle your mind.
posted by malevolent at 10:11 AM on December 10, 2017

re: people not raised with right-angle buildings -- I remember that one of the illusions tested was whether diagonal stripes blear.

I felt like this illusion relates to the two pieces of fashion advice, 'wear all one color if you want to appear slimmer' and 'wear a contrasting belt to create a defined waist.' At least, when I think about it this way, I can kind of understand the illusion, though I still see straight lines.
posted by batter_my_heart at 10:30 AM on December 10, 2017

This illusion is good as hell.
posted by cortex at 10:32 AM on December 10, 2017

To me this looks like a slightly angled curve, and then a slightly curved angle no matter what color background, it looks like a shadow. A shadow on a curve would be a gradient, this has a blurred but not gradient dark line, so the contrast makes it more apparent.

They don't look that different unless I compare them to the curve with the dark line on top. In other words, if I look up and down the grey area it looks like angle-curve-angle-curve, but if I look side to side on the angle line it only changes slightly.
posted by bongo_x at 10:51 AM on December 10, 2017 [1 favorite]

In that moment God died

Phenomenology is a helluva drug. With me, it took a while to think through before I had that brain-blinding flash of conceptual restructuring - I'd spotted, around the age of seven or so, that under certain lighting conditions I saw subtly different shades of red in the same scene depending on which eye I used. My eyes are morphologically different anyway - I have what was called a lazy eye, which means in my case the left eye focusses slightly longer than my right. Early on in the development of the visual system in infancy, this condition means that the brain has to choose one eye over the other, and the visual cortex only properly develops for the dominant eye. These days, the condition can be spotted early enough that corrective optics can bring things together before that stage, but in the 60s/70s the best on offer was putting an opaque layer over the dominant eye's spectacle lens in an attempt to force the visual system to develop for the other eye - far too late to do any good, alas.

Anyway, I got into the habit of looking at things first with one eye, then the next, because the differences were intriguing. Mostly, it was that things were fuzzier with the left eye than the right, although I could see more stars at night with the left eye - it was somehow more sensitive in low light levels. One summer's day, though, I was looking at a red bush in the garden and noticed that the left eye saw it as subtly more orangy-yellow than the right. It wasn't much of a difference, but it was repeatable in the right conditions.

After a while musing on this, I came to the conclusion that if I couldn't say with certainty what 'red' was in my own head, how on earth could I know what other people were seeing? Hoe could they know? I assumed that because they didn't see different shades between eyes, it simply wasn't a factor for them. For a while. this didn't bother me - when you're growing up as someone who devours libraries, there are a zillion ideas to absorb and test. But I was in a (fortuitously liberal) very religious family and had been brought up as an active, involved Episcopalian, said the creed twice on Sunday and all that, and have never doubted what I'd been told. But if nobody could be sure what they were perceiving, then how could anyone be certain about the things the church said were certainly true? But people were certain about it, my parents included, so what was I missing?

The actual aha moment came when I saw Dark Star one quiet weekend afternoon on the local TV station. I'd never come across the term 'phenomenology' before - despite my unrestrained bibliophagy - but it turned out that not only did my realisation have a name, it did indeed have serious consequences. And whether there was or was not a god, you simply couldn't decide on the matter using internal, non-empirical methods. Raising belief to knowledge was really not an option. No way around that - and the people who were telling me they actually knew the truth were making statements they couldn't possibly support.

I had to decide for myself, and entire shells of unexamined acceptance fell away like the dust around a new star that's just turned on.

(In next week's episode, Devonian's introspections about the nature of awareness, perceptions and knowledge take a sharp left turn following his discovery of certain applications of the indole group... but that's another story)
posted by Devonian at 11:14 AM on December 10, 2017 [6 favorites]

Do you mean phenomenalism, or phenomenology?
posted by thelonius at 11:31 AM on December 10, 2017 [1 favorite]

is this illusion effective on all humans?

You're right to ask, and we can't find the answer by armchair guesswork. Some optical illusions don't work on people from hunter gatherer societies. The only way to know if this novel illusion works on them would be to ask them.
posted by justsomebodythatyouusedtoknow at 11:40 AM on December 10, 2017 [2 favorites]

On second thought though you can keep your epiphany, as long as your definition of "the mind' includes both the retina and the brain.

Def, the mind is an epiphenomen of biological activity that cannot be located in time or space.
posted by os tuberoes at 1:34 PM on December 10, 2017 [1 favorite]

which is why I always get lost and show up places late THANK YOU YOU'VE BEEN GREAT TIP YOUR WAITER
posted by cortex at 2:54 PM on December 10, 2017

Well sir, I don't like it. My brain functions poorly enough as is without deliberately sabotaging it.
posted by Justinian at 2:57 PM on December 10, 2017

I've just noticed that this is quite different from the hypothesis proposed by Takahashi in his abstract. Well, he's the expert.

my initial reaction was the same as yours - it feels to me like the primary effect isn't the curves being corners, but rather the intervening segments being straight instead of s-shaped. the "corners" effect just seems like a required result of the segments being straight. but what do I know about the human visual system? my perception of my perception could all be backwards.

stupid brains.
posted by russm at 6:58 PM on December 10, 2017

Right angles in nature are not that common, but in the more generalized category of "sharp/pointy corners" you can find a lot of things.

For an interesting experiment, I opened the image in Photoshop, and copy/pasted a layer of the image over itself. I set the opacity of the new layer at 50%, and then dragged it up and down. The apparent pointy lines seem to stay pointy until the curvy lines are exactly on top of them. When they are just a hair apart they still seem like pointy vs. curvy.
posted by polecat at 12:19 AM on December 11, 2017

As someone who studies the perception of form through painting I think this has to do with both value grouping and perceived lighting. Over the light background we read a dark line the silhouette is of primary importance so the internal value changes are visible but aren't what we use to interpret form. Thus we see the shape of the line accurately.

over the dark background that simple figure/ground relationship breaks down. The line is both lighter than and darker than its environment so more complex visual interpretation must be engaged to decide what is what. the line that still appears curvy is essentially broken up by value changes in to simple curved shapes with no lighting implied so we can easily read the accurate shape of a light curved piece, then a dark curved piece and so on to construct the line.

The line that appears zigzag is broken up by values in a way that imply lighting from the side. We don't read a light piece then a dark piece, we perceive lit and unlit surfaces of a continuous object. We know that lit surfaces that curve get lighting gradients, only flat surfaces can be lit with a continuous value. Only sharp corners create a hard edge change in light value. So the lighting tells us there are corners and flat surfaces. The perception of corners is enhanced by the lines being doubled. You don't just see a zigzag lines, the pair of lines creates the perception of a zigzaging strip of surface. The value transitions become corner points of a lit quadrilateral. 4 corners and a flat lit surface means a straight surface even if the edges of the silhouette are a little curvy. Silhouettes being quite variable they are subservient to lighting in our perception of form.

Similar effects are commonly used in pixel art. When working on very low res art every pixel is lined up on a grid so one has to vary colors in ways that imply lighting or grouping with nearby pixels to imply positioning more subtle than the grid will allow. The primacy of value to imply form over actual position is essential. It can be the difference between a character with a bold gaze and one that is just cross eyed.
posted by subtle_squid at 1:59 PM on December 12, 2017 [3 favorites]