All Visual Perception is Illusory to Some Extent

May 26, 2011 4:41 AM

Top 10 finalists in the 2011 Illusion of the Year Contest.

Previously.

posted by bwg (29 comments total)

Kinda seems like they're running out of illusions (or that compelling nex examples are getting harder to come by); these are nowhere near as cool as past entries have been. (which is probably to be expected)
posted by ShutterBun at 4:53 AM on May 26, 2011

Those were neat, thanks!
posted by KaizenSoze at 4:55 AM on May 26, 2011

My favorite was the eleventh.
posted by fairmettle at 5:02 AM on May 26, 2011 [2 favorites]

The rotating red dot on the morphing face thing didn't work particularly well. I don't get what was going on with the rotating ring of short gray lines, either, since when I scrolled up and obscured the central dot, the "jumps" occurred anyway. Try it.

I can't get bistable illusions to flip for me, they look like a muddy blend of two different things and I start feeling like the guy in Mallrats unsuccessfully looking for a sailboat in a Magic Eye picture. Grouping by contrast was a nice trick. Contrast illusions are possibly the most startling illusions out of the bunch, and so consistent.

I sometimes wonder if, to get decent vision systems in AI, if the systems in question won't be susceptible to many of the optical illusions that fool humans.
posted by adipocere at 5:18 AM on May 26, 2011

Axl Rose could use these.
posted by Eideteker at 5:38 AM on May 26, 2011 [4 favorites]

I agree with adipocere: the morphing face just doesn't work.

The rotating rings don't work for me either. I can choose to view them as "sliding" or "sticking" at any point in the video. Since the video starts with the rings looking "transparent," I continue to view them this way even when they become one solid color.

I can't believe I couldn't get the Venetian mask. And now I can't go back to seeing it the way I did at first. Very interesting.

The triangle is too abstract for me. It's like an unconvincing version of M.C. Escher.

"Attention-induced motion displacement" is way too busy and frenetic to work for me. In fact, I wonder if it could induce epileptic seizures.
posted by John Cohen at 5:40 AM on May 26, 2011

I get the motion after-effect after playing Guitar Hero for a while. It is fucking weird as you look away from the screen and everything seems to be warping upwards.
posted by EndsOfInvention at 5:45 AM on May 26, 2011 [1 favorite]

As someone who used to work in a visual perception lab, the first questions I ask myself when I see these are "Do these surprise me? Could these teach us anything about how we perceive?" A couple of these are not terribly interesting, because we could tell beforehand how they would turn out. BUT there are some gems here.

I couldn't view "Attention-induced motion displacement" so I can't comment on it.

Silencing awareness of change by background motion:

This is very cool. Do the magnocellular (motion) and parvocellular (color) pathways "compete" for our attention? I would be interested to see a few variations on this: changing luminance or shape rather than color would be the first ones I'd look at.

After viewing it a second time, I'm slightly less impressed. You can still see the color changes if you pay attention to them, even when fixating on the central dot. The rotation just involuntarily captures your attention.

The Loch Ness Aftereffect

Very strong aftereffect, far stronger than I thought possible. It's so strong that I didn't "get" the illusion at first- I couldn't see what they were getting at because I was completely fooled. What's happening is that every once in a while the rotating ring of randomly-oriented lines is completely replaced with a new ring of randomly-oriented lines. It's not "jumping backwards" at all.

Why is this aftereffect so strong? What makes this illusion different from other aftereffect illusions? Would we still experience this if the characteristics (eg. length, width, color, distribution of orientations) of the randomly-oriented lines were appreciably different before and after the "jump" (that is, does the illusion depend on our brains thinking after the "jump" that "these must be the same bars, but how did they get there?"; if so, could we interfere with that by changing the characteristics of the bars during the "jump")?

This was easily the best of the bunch for me.
posted by a snickering nuthatch at 5:52 AM on May 26, 2011 [4 favorites]

I sometimes wonder if, to get decent vision systems in AI, if the systems in question won't be susceptible to many of the optical illusions that fool humans.

I am nearly certain of this. Vision starts with a flood of information at our retinas/cameras that must be transformed a great deal before it can be usefully acted on. For the perceiver to not be subject to illusions, their transformations would have to be 1:1.
posted by a snickering nuthatch at 5:58 AM on May 26, 2011

Illusions, Michael. Tricks are what a whore does for money.
posted by grubi at 6:03 AM on May 26, 2011 [12 favorites]

I like these, although the first one didn't "work" for me. Not in the sense that the animation doesn't play (it does), but in the sense that even when the ring of dots starts rotating, I still see the colours shifting like they did when it was stationary.

All the others worked for me including the Loch Ness one (and boy does the text under it ever move weirdly after watching that one!) so I wonder what percentage of the population doesn't experience the intended effect from each of these? And what does that tell us about differences in how our brains work?
posted by FishBike at 6:16 AM on May 26, 2011

I hate that I fall for almost of all of these.

My subconscious is incredibly gullible.
posted by MuffinMan at 6:18 AM on May 26, 2011

The most interesting illusion I've learned of recently is the McGurk Effect, which is not exactly an optical illusion but forced lip-reading (requires sound).
posted by justkevin at 6:27 AM on May 26, 2011 [1 favorite]

What's happening is that every once in a while the rotating ring of randomly-oriented lines is completely replaced with a new ring of randomly-oriented lines. It's not "jumping backwards" at all.

Aha! I didn't understand what was going on - thanks Jpfed!
posted by Sutekh at 6:30 AM on May 26, 2011

grubi: "Illusions, Michael. Tricks are what a whore does for money"

Or candy!

The morphing face worked for me the first time before the dot stopped moving, but afterwards the effect was gone. I've seen the "Grouping by Contrast" illusion before and still can not make my brain see what is actually going on.
posted by Room 641-A at 6:31 AM on May 26, 2011 [1 favorite]

Heh, the guy that won is in my lab. And Jpfed, he's done the lumanince and shape change and they work just as well, there's a paper out on it (here) and he's working on his second and presented the findings at the conference this contest is part of (). He is working on the attention side of things too, I think.
posted by katers890 at 6:31 AM on May 26, 2011 [1 favorite]

I didn't understand the Loch Ness one, either -- the explanation implied that it was just the same ring going round and round and it would suddenly appear to jump. But if I traced specific little bits of the ring with my fingers, they'd disappear at the 'jump', and it seemed like the whole thing was kind of a fraud. Now I see that's the point.
posted by jacquilynne at 6:34 AM on May 26, 2011

The Loch Ness Aftereffect completely had me, but the description was so vague it was like a secondary illusion. I thought they were saying that nothing was happening when it appeared to reverse, and I was half convinced they were just playing a trick.


I sometimes wonder if, to get decent vision systems in AI, if the systems in question won't be susceptible to many of the optical illusions that fool humans.

I'm sure I've seen the face recognition google uses to blur people out of Street View getting false positives, just like we do. But I bet they'll be susceptible to plenty of their own optical illusions too.
posted by lucidium at 6:42 AM on May 26, 2011

He is working on the attention side of things too, I think.

The sudden jump in the speed of the rotation (between when it is not rotating and when it is rotating) may facilitate attentional capture. What happens if the ring smoothly accelerates to its final speed? I have a feeling the effect would be reduced, though of course the only way to know is to do it.
posted by a snickering nuthatch at 6:43 AM on May 26, 2011

I get the motion after-effect after playing Guitar Hero for a while. It is fucking weird as you look away from the screen and everything seems to be warping upwards.

This sounds like a variation of the Tetris Effect.
posted by BenS at 8:10 AM on May 26, 2011

I get the motion after-effect after playing Guitar Hero for a while. It is fucking weird as you look away from the screen and everything seems to be warping upwards.

I get this effect from looking at the ground pass by while riding in a train. Every stop, it seems like the train is slowly backing up.
posted by orme at 8:36 AM on May 26, 2011

The contrast one is the most amazing to me. I turn off the background and try my very best to get the rhythm in my head, I say "On, Off, On, Off" in time with the bottom row of circles and then turn the background back on and it still seems like they're different.
posted by straight at 9:44 AM on May 26, 2011

I get the motion after-effect after playing Guitar Hero for a while. It is fucking weird as you look away from the screen and everything seems to be warping upwards.

Several of the courses I took in college were done on a minicomputer with a bunch of green screen text terminals. These had a really nice "smooth scroll" function that could be enabled. We were warned that after hours of looking through code scrolling in one direction, the text would appear to scroll in the other direction whenever we stopped.

The interesting thing (to me) was that if smooth scroll was turned off, and the text therefore jumped instantly one line at a time, the after-effect never happened. Even though it was still obvious which way the text was "moving", the larger jumps totally defeated the effect.
posted by FishBike at 10:02 AM on May 26, 2011

Just last week I was so taken in by this optical illusion that I took a screenshot and pasted it into Photoshop so that I could eyedropper the two labeled squares and prove that they were NOT the same color. (Spoiler alert: they were.)

(Also: disappointed to learn that the McGurk effect has nothing to do with H. Jon Benjamin.)
posted by ErikaB at 10:03 AM on May 26, 2011

Yeah, some of these just don't seem to work for me, or at least it isn't clear what's illusory and what's real, as with the rotating sticks one.
posted by Decani at 10:11 AM on May 26, 2011

The illusions were interesting but the writing on the website was absolutely of English as a second language quality.
posted by Poet_Lariat at 10:53 AM on May 26, 2011

I was lucky enough to see these presented by their authors. This was actually quite a good batch of illusions, theoretically speaking. A brief commentary so I don't ramble on too much.

The Loch-Ness one is one of my favorites because it is so amazingly theoretically weird. In the usual visual aftereffect, motion-sensitive neurons in your brain adapt to, say, upward motion. Then when the upward motion disappears and you look at something static, the static object appears to move slowly downward. That's not new. This Loch Ness thing is so cool because the speed of the 'jump' moves faster than the speed of the thing you adapted to. This is a completely new, unique result that is not at all predicted by current models.

Attention-induced motion displacement is neat because it is such a huge displacement, and because it somewhat flies in the face of old-school models of motion perception. When we model vision we like to assume that an object's position is absolutely critical, and other stuff - the object's features like color or shape, or what else you're looking at in the world - are secondary. Perceiving position should be absolute. Here, the authors demonstrate that just paying attention to something else causes a huge misjudgment of position. Nifty.

As for the Impossible Illusory Triangle? It didn't make my top 3, but it's a fun fact that its author is Christopher Tyler, the same man who made the first austostereogram - better known as Magic Eye - in the 70s.

(@katers890 - he presented it very well!)
posted by nicodine at 1:46 PM on May 26, 2011

I noticed with the Loch Ness one that if you set the ring to be stationary, it still appears to jump in one direction or the other when it is randomized. If you watch it moving for a while, then stop it, the ring appears to drift slowly in the opposite direction. (That is the original Loch Ness illusion, right?) In that case, all the jumps appear to be in the same direction, opposite the direction the ring was moving.
posted by eruonna at 4:22 PM on May 26, 2011

The nice thing about this webpage is that normally when I look for optical illusions on youtube, I end up having a monster scream at me at the end of a video.
posted by Galaxor Nebulon at 5:05 PM on May 26, 2011