The anatomy of an illusion -- and what it tells us about the visual system

Take a look at this amazing illusion created by Arthur Shapiro (you'll need the latest version of Flash Player to see it):



ResearchBlogging.orgYou're looking at two donut-shaped figures whose "holes" are gradually changing color from black to white and back again. It appears that the holes are changing in an opposite pattern -- when one is light, the other is dark, and so on. But if you click to remove the surrounding donuts, you'll see that the two holes are actually changing together.

If you're still not convinced, get a friend to help. One of you looks at the light donut and the other looks at the dark donut. Then each of you says "light" when your donut hole turns light. You'll soon be saying "light" simultaneously!

Shapiro calls this the Contrast Asynchrony illusion, and he argues that it can tell us a lot about how the visual system works. Below is an interactive version of the illusion. You can manipulate all sorts of variables to change the way the illusion appears. There's even more than one way to eliminate the appearance of the illusion entirely. Can you figure out what you need to do to make the illusion disappear?

Of course you can click the "antiphase" button to make the donut holes appear to change in sync, but this too is an illusion. If you remove the surrounds, you can see that the donut holes are out of sync with each other. One way to remove the appearance of the illusion entirely even when the centers are in-phase is to move either of the Color Angle Θ or Φ sliders. The amount you need to move the slider will depend on your computer monitor, but for most people about a 90-degree difference will make the centers again appear to be in-phase -- as they really are.

Shapiro showed this illusion to four volunteers and systematically varied the color angle of the donuts, then allowed these observers to adjust the donut-holes themselves to nullify the illusion. With few exceptions, the observers set the center at a 90-degree shift from the surround.

But what does a "90-degree shift" in color angle mean?

It's a reference to a way of describing the colors that the human eye is capable of seeing. In the 1980's, psychologists and neuroscientists began to realize that they could describe colors not in terms of the literal wavelength of light, but in terms of how our eye actually responds to those colors. Color vision is made possible by the fact that most people have three different types of cones in their eye. The cones respond differently to different wavelengths of light. While you may have heard the cones described as "red, green, and blue," they are really firing in response to long, medium, and short wavelengths of light. In addition, they fire more when the light they see is brighter, or has a higher luminance. If you combine the firing rates of the long and medium cones and then plot them against the short cones and luminance, you end up with a three-dimensional solid representing all the colors the human eye can see. Here's a two-dimensional representation of the solid -- think of it as "slices" through the solid block of colors:

i-00845edcb083679d9e110ca55edc412e-Shapiro1.gif

The donuts in this version of the illusion are always 180 degrees apart; when the color angle of the center of the donuts is about 90 degrees from those colors, then most people don't see the illusion.

This simple relationship between the colors used to create the illusion and the perception of the illusion has helped Shapiro come up with a model of how we perceive it. We see the illusion because our visual system relies not just on color information, but also on contrast information to make judgments about what we are seeing. When color values are plugged into Shapiro's mathematical model, when the difference between the color angle of the surround and center equals 90 degrees, then the plot of contrast for the left (red) and the right (blue) disk becomes identical, so the centers are seen changing together, while for all other values we see them changing out of phase.

i-4222c938b2ffd19943be20e742f975a9-shapiro2.gif

I mentioned that there's another way to make the illusion disappear -- that has to do with frequency. Set the color plane for both the centers and the surrounds to LM-S, set the angles equal, and change the amplitude to about 45 percent. If you then adjust the frequency to 3 Hz (oscillations per second), you'll see the illusion quite readily. But if you set the frequency to 1 Hz, most people will see the disks oscillating together (especially if you've been staring at the illusion for a long time). The illusion peaks in effectiveness at about 3 Hz. At extremely fast oscillations the illusion becomes difficult to see just because of the speed.

Notice, too, that the illusion does not rely on a sharp edge -- a blurred edge will cause the effect as well. A look at Figure 10 from Shapiro's paper demonstrates that the effect doesn't rely on a circle and disk configuration -- you can also see it with vertical bars.

Shapiro argues that this shows the visual system relies heavily on contrast to judge colors, in addition the wavelength of the light itself. Far from being a "minor side-effect" of color vision, contrast is fundamental to the process of vision itself.

If you enjoyed this illusion, you should also take the time to visit Arthur Shapiro's lab web site. There are several different variations of this illusion, as well as a bunch of other fascinating effects.

Shapiro, A.G. (2008). Separating color from color contrast. Journal of Vision, 8(1), 1-18.

More like this

Its interesting how your mind can misinterpret visual signals like that. It makes you wonder how much of our perceptual conciousness is "real". We take our own perception of the world for granted, but in reality it is just our own brain's interpretation of the actual world.

This is a striking illusion. The key is a form of nonlinearity. The visual system is not reporting lightness, but contrast. And not straight contrast, but rectified contrast. This is common in many models of early visual processing. The significant perceptual event reported downstream is when the center and surround are maximally different, and because one surround is dark and the other light, this occurs at opposite times in the cycle.

Very Cool. Easiest way to get convinced is by taking a screenshot.

Here's an interesting thing: when I shifted my eyes to the left two inches or so, and stared at that point (just below dave & greta's picture), after a few seconds my peripheral vision started seeing the "lights" going off/on together. Then, when when I looked directly at the circles again, the illusion resumed. Odder still, when I tried looking to the right, top, or bottom by the same distance, the illusion persisted. Only looking to the left worked. I'm right-handed, if that is relevant.

This points out the importance of contrast in our perception (I don't think it needs to be limited to visual in a more general discussion). Try to override your perceptual tendencies by make a conscious effort to say "white, black, white, black" to yourself while they change and you can "force" the truth. It is surprisingly difficult to break this illusion through conscious regulation! In fact, when I first saw the animation but before I read the caption, I thought, "What's the illusion here?" Normally, they're immediately apparent, but this one is hard to see through on an automatic level. There must be something very adaptive about attending to contrasts (sudden noises contrast a normalized level of sound with new information, visual contract-- or changes in contrast-- indicate movement or new shadows). I wonder how much of it is due to our post-hoc processes "seeing" the phases of the holes being opposite after seeing the opposite contrasts as more important on a pre-cognitive level. So instead of white-black-white it's important-not-important-not in reference to the entire body/hole system.

I wish we could edit :) I hadn't finished reading the entire post before I started blathering about contrast... then, of course, there is a section on contrast. Yeesh.

tbiggs, that's great!

Ok, firstly, I don't think this illusion is about color at all. My brain is telling me "there's a dot appearing and disappearing here". This is some sort of geometry perception (greatly affected by contrast), a higher level of the visual perception system.

As tbiggs pointed out, when I use the far side of my peripheral vision the illusion changed (all angles worked for me, perhaps he needs to look further off-screen). There is no longer a geometry perception working, and I saw only a change in brightness. Brighter corresponding to appearance.

Frustratingly, even at the far reaches of my peripheral vision, once I had looked for a minute or so my brain "figured it out" and the illusion partially returned.

The comment about contrast brings up several things:

a ) isn't it fairly accepted that contrast affects perception of color? In other words, the surround affects what color we see.

b ) continuing with the idea of "contrast", years ago, when I was taking linguistic classes, we were acquainted with the notion of the "unmarked case." That is, in phonetics for example, there was an argument that the USUAL case did not have to be "marked", only the exceptional or less often case. This relies on the notion of contrast, of "which out of a set", rather than of the absolute value. There might be similar examples in sound, pitch perception, etc.

Nice illusion.

It's not easy, but I've found it is possible to "defeat" this illusion. By concentrating on the central dots and "tuning out" on the surrounds, I can see that the dots are flashing in unison. But it's an effort to keep looking at the picture in this way - as soon as I relax, vision shifts back to a default mode of attention, and the dots appear to be in antiphase again.

Question: What happens when it's the surrounds that flash? The same illusion, but easier to break?

Your flash animation is off for the cycling dots.
It's easing in on the bright flash, making the animation curve like a "bouncing sine wave" instead of a sine wave which is a true cycling animation.

This webpage is a lie, it's not an illusion. You can test this on your own, without asking a friend for help. Just cover one of the sides with your finger and say to yourself "dark" whenever the dark appears. You'll get into a rhythm that you can continue without looking at the screen. Then keep saying "dark" in that rhythm, cover up the circle and uncover the other one. You'll see that you are saying "dark" when the other circle is light, exactly the way it looks when you look at it. The button "Add/Remove Surround" resets the blinks so that they blink in unison. It's cheating. It's not an illusion at all.

I just wanted to point out something I noticed that interests me a lot. When I looked away from the two circles, ie, down at the bottom of my monitor, it was easier to see the two lights in rhythm. Something about viewing them with my peripheral vision changed the way I perceived the circles. Try it and see, I think I might be onto something.

It's such a strong illusion that you've convinced yourself it's a lie! Cut two holes in a piece of paper and hold it up to the screen to convince yourself.

Use the second example and slowing the illusion down makes it a little easier too, but it is still very compelling.

You are responding to the "object" rather than the luminance. What is bouncing back and forth is the "dot" - in the white background the dot is present when the center is dark, in the black background the dot is present when the center is light, so the "object" seems to move back and forth.

I have a deep desire now to know if it's possible to train one's self to "override" this phenomenon. I gave it my best for a minute or so with no progress - only a little disorientation.

If I can digress - the post that first attracted me to repeat visits to Cognitive Daily was an article about using vision tracking equipment to analyze how trained & untrained artists' eyes moved when looking at art work. My hardening cynicism as art faculty began to lift as if a deep mystery about my students' inabilities (and my job in helping them through them) had just been revealed.

I've also seen the color constancy "illusions" such as the ones at lottolabs.org and know that a person can train to overcome that phenomenon through practice in observational painting.

So, I wonder: can a person can develop a functional awareness and/or control of contrast asynchrony as one can with color constancy?

The flash animation is correct. You can see that by crossing your eyes or taking a screenshots. But the reasoning is bogus.

When the two inside circles are bright you see:
( ) (o)
and when dark you see:
(o) ( )

So this has precious little to do with contrast and everything to do with movement... the small circle alternates between the two sides. You aren't seeing the color at all you are seeing the presence / absence of the inner circle.

When the colors of the outside circles are changed to blend in to a greater or lesser degree with the inside circle then the inner circle only partially disappears within its outer circle so what you see ranges from partial movement to no movement when the colors are different. It's easy to detect the center colors being the same since when you are no longer detecting movement so you are seeing *two* circles instead of *one moving circle*.

Gah. The only thing this 'tells' us about vision is that movement trumps color -- duh.

By A Bad Coder (not verified) on 20 Feb 2008 #permalink

Eugene: You're incorrect, the flash animation is not out of sync, nor does it change when you click to remove the surrounds.

I've tested this personally.

What I did was measured the distance between the center dots, then cut out little holes (smaller than the dots themselves) in an opaque envelope. Then I held it up to the screen, therefore manually removing the surrounds (instead of clicking the buttons to do so).

Guess what I found? They're perfectly synchronized.

The only reason you think they're not synched is because of the illusion. Even when you tried to train yourself to say "dark" in rhythm, your mind messed up your tempo when you switched to the next dot.

Try it yourself. Cut holes in opaque paper and see.

Excellent explanation, as usual. I love this illusion and Adelson's checkerboard - with these two illusions, even knowing what I'm actually seeing, I can't talk myself out of it. But by analogy with the Adelson (which I can see accurately when squinting), I blurred my vision and stared at a point between the two dots - and was able to see them blinking properly.

I found that if I concentrated on the rhythm of the change, counting a beat each time one of the inner circles went black, I could after a few cycles, begin to see that the two were changing together without removing the background circles.

I made couple of circles out of paper, cutting holes in the middle, as my wife thought that clicking the button made them flash together. Place them over the surrounds and sure enough, flashing in unison. A cool way to experience this one I thought.

Well this sucks (for me). I love optical illusions and generally get them every time, but I'm not seeing what the deal is here. The pulsing is the same with or without the surrounds. I don't get it. Perhaps this says something about how I'm processing the visual input, but I wouldn't have thought so as I "fall" for pretty much every other illusion.. but haven't come across this one before.

OK, is something wrong with my brain? It was easy for me to see from the start that both inside circles are the same tone all the time. In fact, I had to keep rereading the page to make sure I wasn't missing something!

I'm an artist and I know I don't perceive visual input like most people do. This is the best example of that I've seen so far. I assuem it's because I'm very focused on detail?

By Ken Jacobsen (not verified) on 20 Feb 2008 #permalink

Nah. I'm an artist too (Graphic designer) and I am focused on detail as much as anyone and this one baffled me. Probably has something to do with how you perceive contrast. How do you do on those color blindness tests with the numbers "hidden" amongst the dots?

This one was not hard for me to see. Once I knew the truth, I looked for it, and saw it.

Also, for the guy who posted the links above- The first one, the chessboard, is a very good one. I can fix it in my head only when I use my hands to cup out the surrounding white. Then I see that they are the same colour.

The second one, however, is very odd to me... it says it should appear spinning clockwise, and then eventually will switch? And that he then had a hard time having it switch back, and wasn't able?

I never saw it clockwise. She was always counter-clickwise to me. I can't imagine it any other way, even though in real life I have come across such illusions.

Odd.

.K

Um, I made it go away on its own, without the adjustment of color angle and what not. It started out different....I think, but then I realized that they indeed were changing at the same time. It's sorta freaking me out.

as long as the holes blend alternately with the surrounding colors, you will see the illusion. otherwise, you would see them flashing in sync.

" It appears that the holes are changing in an opposite pattern"

ummm, no it doesnt.
seriously, you have to be pretty dense not to be able to work that out for yourself

By tom_mandory (not verified) on 21 Feb 2008 #permalink

Tom_Mandory:

What are you saying? That you don't see the illusion? There are few illusions that work for everyone, but that doesn't mean that those who experience the illusion are "dense."

LindseyK

Keep checking the rotating girl illusion. If you dont see it ask friends which way she is rotating. You will get mixed answers. With practice you can get her to rotate any way round at will, but once you get her spinning in one direction its almost impossible to imagine how she can rotate the other way round. Its a fantastic 3d illusion.

S

By SevereCircle (not verified) on 21 Feb 2008 #permalink

The answer is actually a lot simpler than people are saying (sorry if someone has already realized this). The key is in observing the actual blinking. If you look closely you can see that the blinker on the left actually fades between black and white whereas the one on the right is just changing from black to white. So it doesn't matter that they are both showing the same colors at the same time, it's the fading one which messes with our head

Hoober,

I think your mistaken. When you remove the surrounds each inner circle is cycling through from balck to white and versa the same.

When I look at it I find that it looks more like the black ball is jumping from the image on the right to the image on the left and then back. Once I tell my mind this is not happening I see them fading in and out in sync.

By SevereCircle (not verified) on 21 Feb 2008 #permalink

Am I weird for seeing through the illusion? The colors aren't shiofting to me. I see two grey cricles turning black at the same time and then back again.

DON'T SEE ANY ILLUSION,JUST LOOKS LIKE A BLINKING LIGHT WTF

There is no illusion.

"It appears that the holes are changing in an opposite pattern -- when one is light, the other is dark, and so on."

The holes _are_ changing in an opposite pattern, but not simultaneously, as implied by the above quote. They alternate. One center appears to flash, then a fraction of a second later, the other appears to flash, and so on, back and forth. In fact, both centers have to be the same colour at the same time in order to get the sensation of alternating flashes.

Why do they appear to flash? Let's start with the white surround. When the center of the white surround is white, the center disappears. This is what you would expect. When the center turns to black again, it re-appears or "flashes". Again, just what you would expect. The same thing is happening with the black surround, but opposite.

So when you see the "flash" on the left side, both centers are black. You are seeing the one on the left, but the black center on the right has disappeared into its black surround. When you see the "flash" on the right, both centers are white, only you don't see the white center on the left because it has faded into its white surround. The sensation of alternating flashes happens precisely because the centers are the same colour at the same time.

That's why when you remove the surrounds, the rhythm of the flashing feels different. The flashes are in synch, they don't alternate anymore.

There's no illusion here at all.

Ack! Try this: take a sheet of paper and put it perpendicular to your screen between the two circles in the first arrangement. That is, make a baffle so that when you but your face up against the other edge of the paper, one eye can see one circle, and the other eye can see the other. When I tried this I first saw the two circles overlapping, venn diagram style, as usual when the stereo is broken, but the illusion was intact. However, over the next few seconds, the two circles merged into one and the illusion vanished. The process was actually a bit uncomfortable, like crossing your eyes for too long, and I guess that's exactly what happened, but I would swear it was automatic.

This is dumb. The holes don't appear to have out-of-phase colors, they appear out of phase. Since appearing means "coming to contrast the surrounding color", and since they have opposite surrounding colors, the fact that they appear out of phase means that their color are in phase.

So contrary to the claim made here, those holes don't appear to be the opposite color. They appear to be the opposite visibility (i.e. one matches its background color and therefore "disappears" while the other, with the same color but a different surrounding color, simultaneously appears).

By Clay Shentrup (not verified) on 25 Feb 2008 #permalink

To the couple of people who posted about not seeing the illusion in your periphery, and that this appears to not have anything to do with colocr, that would be expected given the explanation.
Your peripheral vision has very few of those cones that respond to color, and instead largely contains only receptors the respond to luminance; rods. So by using your peripheral vision, you're essentially removing one of the factors that causes this to happen.

If you look at it and blink in the same rhythm as the flashing, you see the same inner donut colours appear at the same time, so then illusion is broken.

i may be being a bit obvious here but is it not just black not showing on black and white not showing on white?

By Lisa Harman (not verified) on 29 Mar 2008 #permalink

The illusion is presented wrongly: You write "It appears that the holes are changing in an opposite pattern -- when one is light, the other is dark, and so on."
But no, what you see is simply: when one hole is visible, the other is not. So yes, the holes are changing in an opposite pattern, but that pattern is perfectly consistent with the holes being light at the same time and being dark at the same time.

So it doesn't matter that they are both showing the same colors at the same time, it's the fading one which messes with our head