Failing to See the Illusion

Every year, there is a competition for who can come up with the best ‘illusion’:

This year’s winner is a dynamic variant of the Ebbinghaus Illusion (by Christopher D. Blair, Gideon P. Caplovitz, and Ryan E.B. Mruczek at University of Nevada Reno):

There are several runners-up including one called ‘A Turn in the Road’ (by Kimberley D. Orsten and James R. Pomerantz, Rice University, Houston):


I can see the dynamic Ebbinghaus effect (although not sure of the difference in seeing it from the side).

However, try as I might I have not been able to see the illusion in the ‘Turn in the Road’.

I don’t see how rearranging the different images alters which one seems to be the odd one out. When the video finishes with an overlap it is unsurprising.

But I think I know what I am supposed to experience. And, perhaps, what is being assumed to be behind this alleged experience.

1. Call the actually odd one out ‘odd’ and the actually identical ones  ‘even’.

Then I think

2. When I look at the odd one in between the evens, it is supposed to seem to me that the odd and one of the evens look alike while the other even looks like the exception. But it doesn’t look like that to me.

3. Here is a possible explanation of why it would look the way it is supposed to in ‘2’. It is based on a hypothesis about (at least, visual) perception. This is notably promoted by Richard Gregory in his writings — in particular in his classic Eye and Brain. It also has roots in earlier work by Gestalt psychology.

(i) Visual experience of the world before us is (at least partially) a guess at how things are arranged before us in three dimensional space.

(ii) There are rules of interpretation in such experience (why? Evolution — they make running away from deadly things and running toward lively things easier).

(iii) These rules either

— Determine the visual experience, e.g., they determine what it is like to see something.

— Determine the judgements about visual experience, e.g., they determine how we interpret, or what we infer from, what it is like to see something (but not experience itself)

— Or both: depending on how you think about judgement and experience, e.g., experience is itself a judgement constrained by such rules. Or the rules constrain both experience and thinking about experience (but experience is not a judgement).

(iv) One such rule is that one interpret things as lying before one in three-dimensional space. There are principles that are followed which are very different from seeing something lying in a two-dimensional space, e.g., a pair of converging lines which, in two-dimensional space, make a triangle might be parallel lines receding away in three-dimensional space. Look at any picture of a road heading towards the horizon — or just look at a road heading towards the horizon. This is what gives perspective in art.

(v) So, with the ‘Turn in the Road’, our perception interprets what we see in three-dimensional terms. And what we see are three pairs of converging lines. In two dimensions, they meet at an angle, e.g., such as the vertex of a triangle. In three dimensions, though, they need not meet at an angle, but may just be receding parallel lines.

(vi) Part of the recession is about lines meeting at infinity, receding toward the same point at the far horizon. Multiple lines can recede to the same point. Also, one can have multiple points to which different lines recede:

E.g., look at any picture of a skyscraper presented as being from a storey in the middle of its height. The lines running down the building’s frame will recede toward two points — a point below you (toward the depicted ground) and a point above you (toward the depicted sky).

(vii) This hypothesis (which looks like a good one, explaining a wide range of phenomena), is why you might get this alteration in apparent exceptions/twinning in the ‘Turn in the Road’:

— When the odd lies between the evens, its lines recede to the same point as one of the evens. This twins them — they are similar because they are receding to the same place in three-dimensional space. The other even does not recede to that point, and so is the exception.

— Rearrange the odd and evens and this disappears. At best, all the images seem to recede to the same point (somewhere to the right); in that case, in the absence of this discriminating point of infinity, the different shape of the odd distinguishes it from the others.

But I don’t see the illusion. So, why not?

A. The Limits of the Presenting Medium

Perhaps if I saw this on a cinema screen, I might be affected by the illusion. But I saw it on a YouTube video in a 10cm frame. There is no sense for me of recession in seeing them this way. The images look like triangles, two of which are bent away at an angle. It just looks like tiny wire frames moving around.

I think this can be expected given the three-dimensional hypothesis. Something seen as that small does not recede like this; it would only recede in my vision if I was just as small. I’m certainly not tempted to see them as roads (which the dotted lines seem to suggest they are supposed to be). So probably the illusion of the year competition was judged using much larger screens than mine.

I could try to claim a more sophisticated explanation.

B. I Do Not Experience Stereopsis

Perhaps it is because I might not experience proper stereopsis, i.e., 3D visual experience due to some kind of fusion of the stimulus to both eyes.

Here is a very brief and self-piteous story:

Several years back, exhausted for various reasons,  I was experiencing flashing lights in the corner of my vision. The university doctor sent me to a university ophthalmologist — but before he did, he gave me a brief visual test and concluded that I was amblyopic. I had thought I was long-sighted but, he told me, my eyes were fine (as the ophthalmologist later agreed). It seemed that instead the processing of stimuli from one eye was cancelling out or overriding processing from the other. What I saw with each didn’t fuse into one image; each eye competed with the other, one dominated, and so I was effectively seeing with one eye. I had a kind of internal eyepatch (an innargh pirate. Sorry.)

So I couldn’t see 3D depth properly. (“I knew it!” I thought. “That’s why I can’t see those 3D pictures and used to walk into poles at night.”)[1]

Let’s say, then, that I don’t have stereopsis. Thus, I don’t see things receding. That is not part of my visual experience. What I do instead is see things two-dimensionally and make judgements about the resulting two-dimensional experiences. That might explain my failure to see this illusion: for those with strong stereopsis (such as Oliver Sacks; go here for a video discussion about it between Sacks and a once-stereo-less patient), the stereopsis gives a powerful sense of recession. They experience the illusion. I do not.

But do I really need that here? I don’t think so. I don’t see the need to be so sophisticated and complicated in this case. (And I think sophistication/complexity in the absence of necessity is perverse).

C. Projection of Experts’ Expectations on their own Experience

A third possibility is perhaps ungenerous to the illusion-creators and the judges. Yet, given the hypothesis on which it’s based, it seems to be a possibility. It’s this: the judges, creators etc. are all experts and aware of the what might be called “(visual-experience = hypothesis)+(the-hypothesis-is-about-depth). Their tacit acceptance of this theory affects their judgements about what they see. So,

(a) Their visual experience is altered by their expectations (a variant of the prediction model of perception — e.g. a talk last February by Lange). They see the difference because their beliefs cognitively penetrate (Susanna Siegel paper) their perception. By my own hypothesis, my beliefs also penetrate my perception — but I don’t know what those beliefs are; they aren’t now that I will see 3D depth.


(b) Their interpretation of visual experience is altered by their expectations

Both look to be just examples of the Gestalt/Gregory idea of perception as hypothesis. It also explains perhaps why I don’t see what I do. I don’t have those expectations about what I see.

There is perhaps one serious problem with this. If I understand the hypothesis right, it is not supposed to be that perception is susceptible to recent or higher-order beliefs. The beliefs that penetrate perception are evolved over millions of years or developed in childhood (for example, with respect to the Muller-Lyer illusion, see McCauley and Heinrich 2006). That I have a quite sophisticated view about perception should not alter what I seem to see when I am looking at something.

Then, again, maybe what I am saying here is a counter-example to that claim. The claim here is that this is an example of a case where experts about illusion generate illusions non-experts can’t experience. This might be just the kind of example you’d expect if you thought knowledge or thought alters experience.

And again, this might have a precedent. In his book ‘Aspects of Motion Perception‘, the perceptual psychologist P.A. Kölers (not the Gestalt psychologist Wolfgang Kohler) studied in depth the phenomenon called apparent motion — commonly, the appearance of one thing moving where there is actually only a sequence of unconnected stimuli — and so no motion as such (the phi phenomenon is an example).

One claim he made there was that the apparent motion effect grows stronger through practice and repetition — people see illusory motion after such practice/repetition in cases where the unpractised do not (p.158). It’s a different kind of idea of expertise — more habituation, but it involves adults’ perceptions changing ‘live’, as it were, in their responses to potential illusions.

As a final anecdotal example — which will convince no-one needing convincing — I advance this: until I started considering it in University, I don’t remember ever seeing the difference between the lines in the Muller-Lyer illusion. I think I had to be told what was going on before I started seeing it. Then again, perhaps I am having a false memory of that (thus further proving the hypothesis that hypotheses alter experience).

Anyway, for now, I prefer ‘A’.


1. At the same time, though, I never checked with the ophthalmologist and have never followed up on it. So it might all be just a lot of self-indulgent guff.

Also, I could still have a little 3D vision anyhow; I remember reading (though I cannot find a source now) that each eye alone has some degree of 3D vision. A dominating eye should still provide its own stereopsis (albeit, the name would be misleading).


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.