An illusion

September 26, 2013 • 8:23 am

Matthew Cobb sent me this image from Richard Wiseman’s Quirky Mind Stuff. It’s an example of “the Ponzo illusion.”

The images of all of these cars are the same size.  If you don’t believe it, measure them!


More proof that neither evolution (nor, according to Alvin Plantinga, God) gave us absolutely reliable belief-forming mechanisms.

Here’s another example of a Ponzo picture:


55 thoughts on “An illusion

  1. They are all the same size on the orthographic plane, but they shouldn’t be. Normally identical objects that recede in the distance get smaller. So something is amiss.

  2. That’s cheating! What if the question was: “Here is a picture of three cars, which is the largest car?” You can’t expect our perception to be different based on the question. If I cover up the background then the cars look like the same size.

  3. From the left back fender edge to
    the right front fender edge I do
    indeed measure 1.75 inches for each car.
    Before I measured I had faith that
    they were of equal size, but now I believe it.
    Is there a God of Quirky Illusions?

    1. What exactly is the illusion there? Is it that A is blackish while B is whitish? I ask because I could easily, conclusively prove to you that that is in fact the case and that there is no illusion whatsoever and you are perfectly right not to believe the “explanation”.

      1. The illusion is that although they look very different shades of grey, squares A and B are actually exactly the same grey level (in RGB terms, each component being about 47% brightness, according to Photoshop). The different perceived brightness is caused by the relative difference from the surrounding squares.

      2. I have just the following to say here. After that, if your curiosity is piqued, you can read my response to Ben’s comment (#11) and the links I give.

        > A checker board is by definition made of alternating black and white squares
        > This means that if you count an odd number of squares away from a black square you get a white and if you count an even number of squares you get another black.
        > A is black and B is 3 squares away, therefore B is white.

        Simple as that, really. All the complications that follow are due to ideological preconceptions about perception and philosophically half-baked twaddle.

        1. But relax, you’ll be in good company if you buy it because even Steven Pinker has nice section in his book, The Blank Slate, in which he discusses the Checker Shadow illusion and demonstrates just how unblank his own slate is in this regard.

        2. Gah! Not this again!

          Yes! That we all know what a checkerboard is supposed to look like is the source of the illusion!

          The point of the checkerboard illusion is not to make you think you’re mistaken about the construction of checkerboards. The point is that if you were to imagine drawing the picture yourself, it appears you would have two use two different colored pencils for the two highlighted squares when in reality you’d have to use the same colored pencil.

          Counting squares does not enter the equation.

        3. Thanks, Mr Beef, for your excellent reply! It was good to get a response I hadn’t heard from the herd before.

          Firstly, no, it does not appear to me, or any artist worth his salt, that one would use two different colours to shade A and B. This was obvious to me from the start and what got me into discussion about it a few years ago. It seemed dumb to me. Artists have known for hundreds of years how to see and paint shade and the same principle applies to perspective. This kind of “illusion” is for children and novices who haven’t learned the skill yet.

          And it makes no difference whether you are talking about an object or its representation.

          More importantly, I think you are are also wrong on the point of fact that the logic of the difference between A and B is irrelevant. You may be smart enough to avoid confusing the colour of a square and the colour you’d paint it, but you are a rare individual. Most, unable to make the distinction, are plunged into confusion on presentation of the “scientific” evidence from their colour pickers. They take themselves to be proved wrong about A and B being different, yet something in the back of their minds niggles at them (see the last line of comment #8 here). Many completely abandon the common sense that tells them A and B are different. It’s either-or for them. They need to be reminded of the logic, reassured that what common sense tells them is correct. They are not shown by the colour picker to be wrong about A and B being different – as Kathryn Schulz does in her Being Wrong book (she also uses the Checker-Shadow in her TED talk). What is needed is for the distinction to be explained but people tend to show no interest if the logic of A and B’s difference is not clear. They prefer to just abandon the difference and live in a state of suspended belief.

          But there is a deeper level to this. Some current philosophical trends have carried many people off to a point of denying, on principle, the colour of the square, denying, on principle, the distinction and falling into a radical subjectivism / relativism regarding colour and perception. The distinction between the colour of a square and the colour you’d paint it is denied on principle. It is asserted there is no such thing as the colour of an object and that colour is subjective, and things are said like “There is actually no colour. That is part of the illusion. If you think there is colour, you will need to tell me where it is.” The faulty philosophy here seems to underpin what some scientists, including neuroscientists are saying. Despite what you might think from my swipe at Steven Pinker I’m with him where, in chapter 12 of his Blank Slate he attacks the relativist view. I just think that in being a bit of a reductionist, he doesn’t go far enough. And I’m not going to go into detail at this point because I’ve had enough to say for now. I just want to point out for now that there are plenty of good reasons why the logic of A and B’s difference is interesting and important.

          1. I think the way to cut through the whole mess is to note that color scientists use two different (relevant) instruments to measure scenes such as the one in the Adelson checkerboard. The instruments work quite differently, though it’s at least mathematically possible to transform the results from the one to the other, possibly with the help of other instruments.

            The first instrument is the contact spectrophotometer. It has its own built-in light source that it shines off an object. The reflected light is broken up into a spectrum by a diffraction grating (a prism would work, too) and photodetectors measure the absolute illumination of discrete parts of the spectrum. Because the spectral power distribution of the light source is known, it’s straightforward to compute the difference of that and the reflected spectrum to calculate what spectral distribution the object would reflect in any other light source as well.

            The second instrument is generally much more expensive: the spectroradiometer. It has no light source and instead simply reads the absolute spectrum and intensity of whatever you point it at, with a narrow field of view. If you illuminate an object with the same light source as used by the contact spectrophotometer, the spectroradiometer will record the exact same absolute results. But if you increase or decrease the amount of light put out by your light source, the spectroradiometer will report greater or lesser intensity.

            So, if you were to set up the Adelson checkerboard in real life, the spectrophotometer would read two very different values for the two squares, but the spectroradiometer would read the same value. If you were to illuminate the two squares evenly, the spectrophotometer would still read two very different values, and the spectrodadiometer would have readings that at least proportionally matched that of the spectrophotometer.

            Note that, through some very careful lighting, it’s possible to create a three-dimensional real-world example of this illusion by “painting on” the shadow. See this video for a particularly well-done example:


            In this case, the lighting across the scene actually is uniform, provided by that softbox on the right in the final pull-back shot. The spotlight to the left, behind the cylinder, is aimed up and the cylinder’s shadow is being cast on the far wall, not on the checkerboard.



          2. That video is *extremely* well done! I’d never have thought of making a loose square that could be lifted from being a ‘white’ square in the (apparent) shadow and placed on a ‘dark’ square in the lighted area to physically prove beyond doubt that they’re the same shade.


            By the way, I don’t think the illusion has anything to do with it being a checkerboard per se. It has to do with the fact that (a) our eye tends to average out the brightness of an area of the view (so the ‘light’ squares appear lighter by contrast with the ‘dark’ squares around them, and vice versa; and (b) the eye is adjusting automatically for the ‘shadow’ by lightening all the area in the shadow.

            You can see a similar effect to (a) – but with colours – here:

            With respect to (b), every photographer who has taken photos of scenery in bright sunlight knows to his cost that the eye compensates for shadows – without us ever realising it – far better than the camera does.

          3. With respect to (b), every photographer who has taken photos of scenery in bright sunlight knows to his cost that the eye compensates for shadows without us ever realising it far better than the camera does.

            Actually, the problem with dynamic range isn’t the camera nearly so much as it is the print, and the fact that all our image-related software assumes the print is the final output. In an original scene, it’s trivial to have something “brighter than white” — say, light sources or clouds or specular highlights. But paper can never be brighter than white, so you either have to clip those highlights or do a bit of compression and make things a bit darker than they are so there’s still a bit of detail left when you scale down that which is brighter than white. And every image format you’ve ever worked with sets the maximum brightness at paper white, even when you’re working on a display that can create images significantly brighter than ambient lighting.

            Modern cameras — at least, high-end DSLRs such as the Canon 5DIII — are actually capable of recording more dynamic range than the human eye. But good luck processing the resulting file once you’ve got it on your computer.



          4. @ben

            It’s not just prints. I’ve taken photos of waterfalls with the aim of showing them on a webpage – the thing about waterfalls is they tend to inhabit cliffs and narrow valleys, which means that on a sunny day half the (white) water is in sunshine, while the vegetation at the sides of the gorge is in shadow. Possibly the worst subject for contrast you can get, other than snowfields. These (waterfalls) are frequently several miles from the nearest road which tends to limit ones available time on site (unless I want to spend a night in the bush – no thanks!) so I’m stuck with doing the best in the conditions at hand – which usually means bracketing exposures then combining them and tweaking the daylights out of the result with Gimp… – but the point is, my eyes have no trouble in viewing the whole scene. Of course my eyes are actually scanning the scene (while automatically adjusting exposure) while my brain builds up an overall picture – thus (I guess) increasing the contrast range of my eyes. So I think I can see everything quite clearly all at once.

          5. You remind me of another typical problem with modern image processing software.

            By default, 99 44/100% of image processing applications “helpfully” apply an S-curve to the tone map, which increases contrast in the midtones at the expense of highlight and shadow detail. This “enhancement” is often applied very early in the processing pipeline, long before the user ever sees the data, and generally can’t be turned off and certainly can’t be undone after the fact. Those highlights get blown and the blacks get crunched, all for the sake of “punchier” images. A similar thing happens to saturation and thus color fidelity.

            If you don’t mind going down a maze of very twisty little passages, all alike, you can actually get the camera to function not unlike a high-resolution spectroradiometer. And, when you do so, you wind up with all the highlight and shadow detail preserved. The midtones obviously aren’t as “punchy” as one might be used to, but they actually look natural.

            The catch? You need to build a high-quality ICC profile for your camera (a much more challenging proposition than it might naïvely seem) and you need to use RAW development software that isn’t necessarily the most user-friendly in the world.

            Of course, many people prefer the “painterly” approach of tone-mapping with local inversions and discontinuities in the tone curve. And that certainly has a very appealing aesthetic to it. It’s a not-bad model of the perception one gets from focusing in sequence on different parts of the scene, but at the cost of the overall gestalt.



        4. Here is a nice quote from the chapter of Pinker’s book I mentioned above. The second part makes perfect sense and is at odds with what Jerry has to say about the accuracy of our”belief-forming mechanisms”. But the first sentence is confused; Pinker’s unsuspecting observer does not arrive at an incorrect conclusion but at an incomplete one. And it does not make any difference whether we are talking about the real world or a representation of it:-
          “Like a policeman framing a suspect, Shepard and Adelson have planted evidence that would lead a rational but unsuspecting observer to an incorrect conclusion. If we were in a world of ordinary 3-D objects that had projected those images onto our retinas, our perceptual experience would be accurate. Adelson explains: “As with many so-called illusions, this effect really demonstrates the success rather than the failure of the visual system. The visual system is not very good at being a physical light meter, but that is not its purpose. The important task is to break the image information down into meaningful components, and thereby perceive the nature of the objects in view.””

  4. I’m just reading about this in “Thinking Fast and Slow,” by Daniel Kahneman. As an artist, I’m accustomed to looking at a picture as a 2-D surface with shapes on it, but most people use a perspective heuristic to try to make sense of the image (Type 1 thinking).

  5. I went to his website and had a major LOL. Check out his post for Sept. 12. It’s titled “Message from God?” and shows clouds in the shape of a hand…with the middle finger extended!

    And, no, even after measuring the cars and people, I still don’t believe it. 🙂

  6. The problem here is that we’re conflating definitions of the term, “size.”

    Do the abstract two-dimensional shapes all have the same measurements? Yes, sure.

    But, in the context of the two-dimensional projection of the three-dimensional scene being represented, they all subtend the same angle while being at different positions on the Z axis. If you were to reconstruct the physical scene that the two-dimensional projection is representing, you would wind up with the rear car about twice the size as the front car.

    What’s remarkable isn’t that you “fail” at recognizing that they’re all the same “size.” What’s remarkable is that your brain can perform that dimensional reconstruction on the fly in such a way as to correctly determine the relative sizes of the three-dimensional objects.

    What’s even more remarkable is that, even though this is a very small scale image and the actual angles subtended by the time the image reaches your eyes are much narrower than what the original photographer experienced, you can still accurately and intuitively estimate the actual real-world sizes of the objects. The nearest is a bit bigger than a Shriner’s car; the middle is about a 3/4 scale model; and the far one is about the same scale as an oversized SUV.



    1. Spot on, Ben! Well said. I see no illusion. All I see is a row of cars each dramatically bigger than the one before. And since my studio overlooks that scene, I can pop out and measure them for you and provide proof that they are not in fact the same size. As to measuring the picture, what would be the point of that, other than to work out the sizes of the cars? (You’d have to know the lens and sensor size.)

      But I disagree with you about this being a conflation of two definitions of size. No, not definition, but yes, certainly a matter of conflating or equivocating between two different references made by what is a clearly defined word. One has to keep a handle on whether one is referring to the size of the cars themselves or the size of the representations of the cars. It’s the same definition of “size”, different reference. But I think that’s what you’re trying to say, so this is just a disagreement on semantics.

      Essentially, the presentation of an “illusion” like this is like a magician’s act – it trades on shifting your attention without your knowing it. In this case one’s attention is shifted to a different reference and you don’t realise it because the word used for the different reference remains the same. It’s called equivocation, and I’m surprised to find someone of Jerry’s calibre falling for it.

      Typically, discussions about this kind of thing come down to the person defending the existence of an illusion by saying that the image is only a representation (good example here: (comment 29) and also the discussion page of this: ). I tend to bow out at that point because if someone can’t or won’t think as far as asking themselves the obvious, begged question (a representation of what?) then they’re unlikely to get it no matter how hard you try.

      1. I disagree with Ben and Andrew as to whether there are illusions present. I can agree that there would be poor ways to ask the question that brings out the illusion (“Are the cars the same size?”, “Are the squares the same color?”), but there are also correct ways (“Are the images of the cars the same size?”, “Do the pixels in square A and square B of this image have the same RGB value?”) (Note that Jerry did talk about the size of the images, not the size of the cars.)

        Many people will have great difficulty in quickly answering those questions correctly when confronted with these images, even if you explain in great detail that you’re talking about the size of the image, and RGB values in the image, rather than hypothetical sizes and colors in the “real” scene represented by the image.

        While this failure may not be “remarkable”, I believe it’s interesting and fun (and, as Jerry pointed out, an example of the unreliability of our belief-forming mechanisms).

        Maybe if you somehow marked questions about images as illegitimate (“Things Man was Not Meant to Know”?) you could define away these images as illusions, but I don’t see any reason for forbidding such questions.

        1. Seems to me that if you want accurate answers about the pixel colors or dimensions of various 2D shapes, the sensible thing to do is to get out your instruments and start measuring. These are not the sort of tasks for which our unaided visual system is optimized, simply because there’s no demand for such skills in the wild. So it’s no surprise that we perform poorly on them.

          But I don’t see how a demonstration of this fact counts as an “illusion” or impugns the reliability of our belief-forming mechanisms with respect to tasks for which we are optimized.

          1. I entirely agree with your first paragraph.

            When I look at the checkerboard image, I have the strong impression that the pixels in square A of the image are darker than the pixels in square B of the image, even though measurement shows they have the exact same RGB values. This seems to exactly match the first dictionary definition of “illusion” I found, “something that looks or seems different from what it is”.

            I agree that these illusions don’t really impugn “the reliability of our belief-forming mechanisms with respect to tasks for which we are optimized”, which suggests that we might restrict ourselves to “tasks for which we are optimized” to take advantage of presumed better built-in belief-forming mechanisms. Unfortunately there are a few problems with that.

            First, we may often want or need to perform tasks that we’re not optimized for.

            Second, we don’t even know what tasks we are optimized for. (I spend a large fraction of my time reading. Am I optimized for reading? I’m not sure what this question really means, let alone how to answer it.)

            Third, it’s not necessarily the case even in tasks we’re optimized for that this optimization includes the most accurate possible belief-forming mechanisms; for instance, consider that we seem to be hard-wired to recognize faces so strongly that we see faces where there really aren’t any (pareidolia).

            All in all, it seems best to just accept that our built-in belief forming mechanisms range from working poorly to working quite well, and to augment them with learned methods to improve the accuracy of our beliefs. One of the best and most well-known such methods is of course the Scientific Method.

          2. This seems to exactly match the first dictionary definition of “illusion” I found, “something that looks or seems different from what it is”.

            But what does “it” refer to? In the case of a real checkerboard with real shadows, “it” presumably refers to the underlying pigmentation of the squares, not the intensity of reflected light reaching our eyes. In that case things are exactly as they appear, so there’s no illusion.

            Now suppose we take a digital photo of that scene and display it on a monitor. Do the rules now change so that “it” refers to pixel color rather than pigmentation? Why?

            I suggest that if something is not an illusion when we view it directly, then it’s not an illusion when we view a picture of it, pixel colors notwithstanding.

          3. Well, if we’re to look at an image or scene and see “something that looks or seems different from what it is”, we’re probably going to need the extra information of “what it is” to know that it actually is an illusion — otherwise we’ll only see what it “looks or seems” like and won’t notice the illusion at all.

            For example, if you just look at the first image here: , there’s nothing terribly surprising about it… it’s only when you’re also told that the head of the moth is facing down that the image is revealed to be an illusion (according to the above dictionary definition).

            Similarly, if somebody set up a physical scene with a real checkerboard and real shadows, the extra information that would make the scene surprising — and hence an illusion — would be something like “If you took a digital picture of this scene, then the pixels of square A would be the same gray value as the pixels of square B” or “If you painted this scene as realistically as possible, you’d need the same color of paint for square A and square B” — so these sentences are the “it” from the dictionary definition. On the other hand, if you said “Square A on this checkerboard is darker than square B”, then that’s just what the viewer would expect, so there would be no surprise and no illusion would be revealed.

            In the image, the “it” would be the difference between the RGB values of the pixels in squares A and B of the image, which is actually 0 (no difference) but “looks or seems” nonzero.

            At this point, we seem just to be arguing about the definition of the word “illusion”, which is off-topic for this website and may not be interesting for other readers; so let me make a few final points:

            Many people would and do call these images illusions, as evidenced by the phrase “Ponzo illusion”, the phrase “checker shadow illusion”, and the inclusion of Ponzo-illusion images in many books named “Optical Illusions”. (I suspect it would be “most people”, but I can’t prove it.)

            I think I’ve made a reasonably good case that these images do fit the dictionary definition I’ve quoted, but if you disagree it won’t hurt my feelings. 🙂

            I understand that you have a definition for “illusion” under which these images are not illusions. I have no particular opinion on whether your definition is “better” than mine, but if (as I suspect) most people would classify these images as illusions, you may be better off finding a new word or phrase to represent your definition than trying to hijack the word illusion.

            Having said all that, I will endeavor not to make any more posts here, at least on the topics of the definition of “illusion” or whether or not these images are illusions. 🙂

      2. “Illusion” is exactly the term used to refer to the misdirection and switcheroo technique of magicians. So why not here?

        I think what’s happening here is we’re arguing about the definition of “illusion”. I’m wondering what would count as a bona fide illusion in your estimation. I guess I’d rather take it to the “all 2D representations of 3D scenes are illusions” extreme, which Gregory below implies would be silly, than the “there are no illusions; what we see is what is there; perception = reality” extreme.

        1. I can’t speak for Andrew, but in my book there are such things as illusions. Among them I would count 2D representations that lead us to build incorrect models of the 3D scenes they represent. Examples of this sort of illusion include Escher’s circular waterfalls, rooms with deliberately distorted perspective that cause us to misjudge the size of objects they contain, these Buzzfeed photos, and the like.

          But the Ponzo and checker-shadow “illusions” are not in that category. They’re 2D representations that lead us to build correct 3D models. And the coincidences of pixel color or 2D measurement they embody are just meaningless coincidences (even if deliberately engineered) and not faulty perceptions.

          I think that’s a distinction worth making. YMMV.

    2. Agreed. The only “illusion” here is that 3D scenes can be represented by 2D arrays of pixels. But that’s exactly what happens on our retinas, so if we buy that definition, then everything we see is an illusion.

      1. Well put. To you and I it is a good reductio ad absurdum but, believe it or not, there are many who bite that bullet.

    3. Well, that saved me a lot of typing (well, some).

      It absolutely is the case that if this photograph were reproduced in the real world, our brains would be completely correct.

      I’ll add that the same is true of the checkerboard with shadow “illusion”. If you were to reproduce that scene in real life, then remove the source of the shadow, the apparently lighter square would, in fact, be lighter.

      It’s a general trend with such illusions to confuse our visual system, which is designed to model the world, with a simple photometric device. The fact that our brain can still create a three-dimensional model from a static two-dimensional model of a three-dimensional world really is rather remarkable.

      1. Remarkable or not, the important point is that it’s evidence in favor of (not against) the proposition that evolution endows us with “reliable belief-forming mechanisms”, to borrow Jerry’s phrase.

        1. You could argue that, or not. These ‘illusions’ serve to illustrate to people that we are not sentient cameras taking accurate snapshots of the world, but are constructing internal maps of our surrounds and there are assumptions in the processing. This certainly has a bearing on how we perceive real objects or events. Things can happen too fast for us to process accurately, leaving us to fill in the blanks. Our brains can reconstruct the part of a scene we are focusing on with some elements missing ,whereas shifting focus away will bring those elements into view. This is not to mention the effect that prejudice, emotion and expectation have on perceptions.

          Our ultimate belief forming mechanism, the scientific method, is designed to work around our faulty mental reconstructions.

  7. Great illusion but watch out — that front car’s going to get badly scraped by the next truck that drives by in that direction.

    Since Alvin Plantinga was mentioned: Do I misunderstand his argument? I take it to be this: if you conclude that you are the result of evolution, then you cannot trust your own evolved brain to draw conclusions such as that, because it is fallible. It would seem that there are two easy answers:

    1. By arguing with each other we can come to less error-prone conclusions than any one brain can achieve unaided. And we do this all the time.

    2. If we are the result of evolution, then so is Alvin Plantinga, and thus his conclusion that we didn’t evolve is also called into question by his argument.

    (Though I think Plantinga is coy about whether evolution has happened).

    1. Good one, Joe!

      If evolution is true, then we are geared into the rest of life with very tight precision. Those who aren’t, including those those who perceive their God- or evolution-given perceptions to be fatally flawed, are more likely to have their genes eliminated from the pool.

    2. Plantinga’s fatal flaw lies in assuming that if you declare yourself infallible then you actually are infallible. Specifically, he’s declared that, through the power of YHWH as revealed in the Bible, he knows with absolute certainty that the Universe began in an enchanted garden with talking animals and an angry wizard. Because he is absolutely certain of this, his certainty trumps the wishy-washy belief of the scientismists who can’t even pin down the age of the Universe to within a million years.

      How anybody can consider him and his ilk “sophisticated” is beyond me, unless “sophisticated” is nothing more than a synonym for “polysyllabic.”



  8. If I want to give people the impression that I have great height, perhaps I’ll go to a railway station and stand like the taller appearing suited fellow in the picture and force their perspective. Of course from this view, they’ll be hit by the train but then they’ll never learn the truth!

  9. If you go for this stuff, you have got to visit the Purves lab site. I was taken there by an American Scientist article some years ago. Such “optical illusions” reveal how we have evolved to process light and shadow. If our ancestors didn’t see things that way, they’d have been dead, and we’d have never been.

  10. I’ve had that illusion myself, although up to now I didn’t know it had a name.

    I was standing looking with binoculars along a beach (parallel to the shoreline) at a flock of sandpipers. And although they were all the same species of sandpiper, the ones which were farther away looked larger than the nearby ones.

    That taught me to be very cautious about using size as a fieldmark to identify bird species.

  11. Also known as the ‘railway lines’ illusion.

    What surprises me is how powerful the illusion can be – that is, the extraordinary difference in (apparent) size of the cars, and how convincing the illusion is – even when you know exactly what the illusion is, and you have all the experience in the world that cars are all pretty much the same size, the illusion still persists. Try as I might, I cannot see the three car images as the same size (unless I cover up the background).

    By the way, for an excellent and interesting account of optical illusions of all sorts, I can recommend Eye and Brain by R L Gregory (I first came across it decades ago and I was fascinated by it).

  12. That reminds me of that misshapen room – I can’t remember what the room is called now. The Museum of Science and Industry in Chicago used to have one when I was there 30 years ago – I wonder if it’s still there.

      1. There are several good illustrations of the Ames room (which is really just the Ponzo illusion in three dimensions) on Youtube. Just google ‘youtube ames room’.

  13. There’s another brilliant illusion here:

    When I first look at the silhouette, although it’s ambiguous, the overwhelming perception I get is that she’s spinning clockwise. As soon as I look at her shadow on the ground (which is unambiguously anticlockwise) she starts spinning anticlockwise to match it and – this is the weird thing – even when I stop looking at the shadow, she *won’t stop spinning anticlockwise* as long as the shadow’s in my peripheral vision. I have to physically hide her shadow to get her back to clockwise spinning.


  14. All these fancy elegant ideas on visual illusions.

    Let Pantinga, spend his remaining days with children born blind.

    Maybe this will disabuse him of his inane illusions

Leave a Comment

Your email address will not be published. Required fields are marked *