Posts Tagged ‘binocular vision’

Jorge Salazar of EarthSky.org recently interviewed me about my research, and you can find the podcast and text here. I got a chance to talk about the similarity between accents and color vision (how we all believe we have uncolorey skin and no accent), the function of color vision (it’s for giving you that empath sense you didn’t know you have), and why we don’t have eyes on the sides of our heads (it’s for seeing better in cluttered leafy habitats, just the thing for a primate).


Mark Changizi is Professor of Human Cognition at 2AI, and the author of The Vision Revolution (Benbella Books) and the upcoming book Harnessed: How Language and Music Mimicked Nature and Transformed Ape to Man (Benbella Books).

Read Full Post »

Ian Woolf of Diffusion Radio just reviewed The Vision Revolution, and you can hear the podcast here (15 minutes in).


Mark Changizi is a professor of cognitive science at Rensselaer Polytechnic Institute, and the author of The Vision Revolution (Benbella Books).

Read Full Post »

Recently I was interviewed Jovana Grbic of ScriptPhD about The Vision Revolution. She has a great knack for asking unusual questions, taking me out of my standard responses and making me think. (To find the podcast itself, scroll down within this link until you see it.) I also wrote a guest piece for them on idea-mongering and non-genius that you’ll find there.


Mark Changizi is a professor of cognitive science at Rensselaer Polytechnic Institute, and the author of The Vision Revolution (Benbella Books).

Read Full Post »

This first appeared on January 2, 2010, as a feature at the Telegraph.

You know what I love about going to see plays or musicals at the theater? Sure, the dialog can be hilarious or touching, the songs a hoot, the action and suspense thrilling. But I go for another reason: the 3D stereo experience. Long before movies were shot and viewed in 3D, people were putting on real live performances, which have the benefit of a 3D experience for all the two-eyeds watching. And theater performances don’t simply approximate the 3D experience – they are the genuine article.

“But,” you might respond, “One goes to the theater for the dance, the dialog, the humans – for the art. No one goes to live performances for the ‘3D feel’! What kind of low-brow rube are you? And, at any rate, most audiences sit too far away to get much of a stereo 3D effect.”

“Ah,” I respond, “but that’s why I sit right up front, or go to very small theater houses. I just love that 3D popping out feeling, I tell ya!”

At this point you walk out, muttering something about the gene pool. And you’d be right. That would be a rube-like thing for me to say. We see in 3D all the time. I just saw the waitress here at the coffee shop walk by. Wow, she was in 3D! Now I’m looking at my coffee, and my mug’s handle appears directed toward me.Woah, its in 3D! And the pen I’m writing with. 3D!

No. We don’t go to the live theater for the 3D experience. We get plenty of 3D thrown at us every waking moment. But this leaves us with a mystery. Why do people like 3D movies? If people are all 3D-ed out in their regular lives, why do we jump at the chance to wear funny glasses at the movie house to see Avatar? Part of the attraction surely is that movies can show us places we’ve never been, whether real or imaginary, and so with 3D we can more fully experience what it is like to have a Tyrannosaurus Rex make a snout-reaching grab for us.

But there is more to it. Even when the movie is showing everyday things, there is considerable extra excitement when it is in 3D. Watching a live performance in a tiny theater is still not the same as watching a “3D movie” version of that same performance. But what is the difference?

Have you ever been to one of those shows where actors come out into the audience? Specific audience members are sometimes targeted, or maybe even pulled up on stage. In such circumstances, if you’re not the person the actors target, you might find yourself thinking, “Oh, that person is having a blast!” If you’re the shy type, however, you might be thinking, “Thank God they didn’t target me because I’d have been terrified!” If you are the target, then whether you liked it or not, your experience of the evening’s performance will be very different from that of everyone else in the audience. The show reached out into your space and grabbed you. While everyone else merely watched the show, you were part of it.

The key to understanding the “3D movie” experience can be found in these targets. 3D movies differ from their real-life versions because everyone in the audience is a target, and all at the same time. This is simply because the 3D technology (sending up left and right eye images to the screen, with glasses designed to let each eye see only the image intended for it) gives everyone in the audience the same 3D effect. If the dragon’s flames appear to me to nearly singe my hair but spare everyone else’s, your experience at the other side of the theater is that the dragon’s flames nearly singe your hair and spare everyone else’s, including mine. If I experience a golf ball shooting over the audience to my left, then the audience to my left also experiences the golf ball going over their left. 3D movies put on a show that is inextricably tied to each listener, and invades each listener’s space. Everyone’s experience is identical in the sense that they’re all treated to the same visual and auditory vantage point. But everyone’s experience is unique because each experiences themselves as the target – each believes they have a special targeted vantage point.

The difference, then, between a live show seen up close and a 3D movie of the same show is that the former pulls just one or several audience members into the thick of the story, whereas 3D movies have this effect on everyone. Part of the fun of 3D movies is not, then, that they are 3D at all. We can have the same fun when we happen to be the target in a real-live show. The fun is in being targeted. When the show doesn’t merely leap off the screen, but leaps near you, it fundamentally alters the emotional experience. It no longer feels like a story about others, but becomes a story that invades your space, perhaps threateningly, perhaps provocatively, perhaps joyously. No, we don’t suffer the indignity of 3D glasses for the “popping out feeling”. We enjoy 3D movies because when we watch them we are no longer mere audience members at all.

Mark Changizi is a professor of cognitive science at Rensselaer Polytechnic Institute, and the author of The Vision Revolution (Benbella Books).

Read Full Post »

Today I was on LateNightLive (of ABC Radio National) with host Phillip Adam.  In addition to divulging my enjoyment at running over pigeons, I got a chance to talk about sex, blood, the blind, writing, rabbit-heads and other topics from The Vision Revolution.

Late Night Live

Late Night Live

Check out the segment here. (Download the mp3.)

Mark Changizi is a professor of cognitive science at Rensselaer Polytechnic Institute, and the author of The Vision Revolution (Benbella Books).

Read Full Post »

I was on the Lionel Show / Air America this morning, which was a blast!  Got to talk about my recent book, and about evolution, autistic savants, intelligent design, color, forward-facing eyes, illusions, and more. I really must get off the elliptical machine next time I do a radio show. Here’s the segment with me (or mp3 on your computer).

Mark Changizi is a professor of cognitive science at Rensselaer Polytechnic Institute, and the author of The Vision Revolution (Benbella Books).

Read Full Post »

This first appeared on October 26, 2009, as a feature at ScientificBlogging.com

Later this evening I’ll be giving a talk to a group of astronomers on what its like to see like an alien. The beauty of this is that I can speculate until the cows come home without fear of any counterexamples being brought to my attention. And even if an alien were to be among the audience members and were to loudly object that he sees differently than I claim, I can always just say that the jury is out until we get more data, and then advise him not to let the door slam into his proboscis on the way out.

E.T. the extra-terrestrial

E.T.'s forward-facing eyes suggests its ancestors evolved in forests

Although it may seem wild-eyed to discuss the eyes of aliens, if we understand why our vision is as it is, then we may be able to intelligently guess whether aliens will have vision like ours.

And in addition to the fun of chatting about whether little green men would see green, there are human implications. In particular, it can help us address the question, How peculiar is our human vision? Are we likely to see eye to eye with the typical alien invader? Or does our view of the world differ so profoundly that any alien visual mind would remain forever inscrutable?

Let’s walk through four cases of vision that I discuss in my book The Vision Revolution and ask if aliens are like us.

Do aliens see in color like us?

Let’s begin with color. I have argued in my research that our primate variety of color vision evolved in order to sense the skin color signals on the faces, rumps, and other naked spots of us primates. Not only are the naked primates the ones with color vision, but our color vision is at the sweet spot in design space allowing it to act like an oximeter and thereby see changes in the spectrum of blood in the skin as it oxygenates and deoxygenates. (See the journal article.)

Aliens may be interested in eating our brains, but they have no interest whatsoever in sensing the subtle spectral modulations of our blood under our skin. Aliens will not see color as we do, and will have no idea what we’re referring to when we refer to “little green men.”

Little green men may not think they look green

This can take the wind out of many people, namely those who feel that their senses give them an objective view of the world around them. But evolution doesn’t care about objective views of the world per se. Evolution cares about useful views of the world, and although veridical perceptions do tend to be useful, little-white-lie perceptions can also be useful. We primates end up with colors painted all over the world we view, but our color vision (in particular the red-green dimension) is really only meaningful when on the bodies of others. Although we feel as if the objects in our world “really” have this or that color, no alien would carve the world at the color-joints we do.

Do aliens have forward-facing eyes?

How about our forward-facing eyes we’re so proud of? I have argued and presented evidence that forward-facing eyes evolved as an adaptation to see more of one’s surroundings when one is large and living in leafy habitats. Animals outside of leafy cluttered habitats are predicted to have sideways-facing eyes no matter their body size, but forest animals are predicted to have more forward-facing eyes as they get larger. That is, in fact, what I found. (See the article.)

So, would aliens have forward-facing eyes? It depends on how likely it is that they evolved in a forest-like habitat (with leaf-like occlusions) and were themselves large (with eye-separation as large or larger than the typical occlusion width). My first reaction would be to expect that such habitats would be rare. But, then again, if plant-like life can be expected anywhere, then perhaps there will always be some that grow upward, and want to catch the local starlight. If so, a tree-like structure would be as efficient a solution as it is for plants here on Earth. The short answer, then, is that it depends. But that means that forward-facing eyes are fundamentally less peculiar than our variety of color vision. Aliens could well have forward-facing eyes, but it would not appear to be a sure thing.

Do aliens suffer from illusions?

One of the more peculiar things our brain does to us is see illusions. I have provided evidence that these illusions are not some arcane mistake, but a solution to a problem any brain must contend with if it is in a body that moves forward. When light hits our eye, we would like our perception to occur immediately. But it can’t. Perception takes time to compute, namely about a tenth of a second. Although a tenth of a second may not sound like much, if you are walking at two meters per second, then you have moved 20 cm in that time, and anything perceived to be within 20 cm of passing you would have just passed you – or bumped into you – by the time you perceive it. To deal with this, our brains have evolved to generate a perception not of the world as it was when light hit the eye, but of how the world will be a tenth of a second later. That way, the constructed perception will be of the present. Although there is no room in this piece to describe the details, I have argued that a very large swathe of illusions occur because the visual system is carrying out such mechanisms. (See the paper.)

Are aliens buying books of illusions and “ooh”ing and “ah”ing at them like we are? If they are moving forward (and have non-instantaneous brains), then they probably are buying these books. This is because the optic flow characteristics that underlie the explanation of the illusions are highly robust, holding in any environment where one moves forward. Aliens are, then, likely to suffer from illusions. The illusions we humans suffer from, then, may not be due to some arcane quirk or mistake in our visual system software, but, instead, a consequence of running the efficient software for dealing with neural delays.

Is alien writing shaped like ours?

I have provided evidence that our human, Earthly writing systems “look like nature,” in particular so that words have object-like structure. And I have shown that for writing like ours where letters stand for speech sounds, letters look like sub-objects, namely object junctions. Certain contour-combinations happen commonly in natural scenes, and certain combinations happen rarely. I have shown that the common ones in such environments are the common letters shapes found in human writing systems. Culture has selected writing to have the visual shapes our illiterate brains can see, which is why we’re such capable readers. (See the paper, a popular piece, and an excerpt from The Vision Revolution on this.)

Would alien writing look like this?

In this light, would alien writing look like nature as well? It depends on how specific one is when one says “like nature.” If, say, our human writing looks specifically like a savanna – i.e., if our writing mimicked signature visual features of the savanna – then it would appear very unlikely that aliens would have our kind of writing. But what if human writing looks like a very general notion of nature, so general it is likely to apply to most conceivable aliens? In my research I have provided evidence that the “nature” that appears relevant for understanding the shape of human writing is, indeed, highly general: namely, “3D environments with opaque objects strewn about.” Although highly general, aliens could float in a soup of cloudy transparent blobs, which is a kind of “nature” radically different than the one that human writing looks like. But it does seem plausible that most aliens will be roaming around opaque objects in 3D, and if that’s the case, then (so long as their culture has selected their writing to harness their visual object recognition system) their writing may look similar to human writing. Alien writing, if thrown into a pile of samples across our human writing, might just fit right in!


So let’s take stock.

Would aliens have our color vision? No. Definitely not. Ours is due to our peculiar hemoglobin.

Would aliens have forward-facing eyes? Maybe. If they evolved in leafy habitats and were large.

Would aliens see our illusions? Probably.   If they move forward.

Would aliens have writing that looks like ours? Probably. If they live in a 3D world with opaque objects.

Mark Changizi is a professor of cognitive science at Rensselaer Polytechnic Institute, and the author of The Vision Revolution (Benbella Books).

Read Full Post »

Originally a piece in ScientificBlogging, September 17, 2009…

Reading pervades every aspect of our daily lives, so much so that one would be hardpressed to find a room in a modern house without words written somewhere inside. Many of us now read more sentences in a day than we listen to. Not only are we highly competent readers, but our brains even appear to have regions devoted to recognizing words. A Martian just beginning to study us humans might be excused for concluding that we had evolved to read.

But, of course, we haven’t. Reading and writing is a recent human invention, going back only several thousand years, and much more recently for many parts of the world. We are reading using the eyes and brains of our illiterate ancestors. Why are we so good at such an unnatural act?

Here I describe recent evidence that, although we have not evolved to be good at reading, writing appears to have culturally evolved to be good for the eye. More specifically, recent research supports the exciting hypothesis that human visual signs look like nature, because that is what we have evolved over millions of years to be good at seeing. This ecological hypothesis for letter shape not only helps explain why we are such good readers, but answers the question, Why are letters and other visual signs shaped the way they are?

The Variety of Visual Signs

Given the tremendous variety of visual signs over human history, it may at first glance seem that there could be no simple answer to the question, Why are visual signs shaped as they are? After all, we have been making visual signs for at least 40,000 years, starting with tool decorations and cave paintings. The evolution of ornamentation, art, painting, and other non-linguistic visual signs (i.e., signs not part of language) has gone on unabated, diversifying into millions of non-linguistic symbols used over the ages, and occupying nearly all aspects of our lives, including pottery, body art, religion, politics, folklore, medicine, music, architecture, trademarks and traffic.


Writing (i.e., visual signs distinguished by use as a means of visually recording the content of spoken language) has also undergone an evolutionary explosion in variety. The earliest writing appeared several thousand years ago, and occurred independently in Sumer, Egypt and China (and much more recently in the Americas). These earliest linguistic visual signs were pictograms, evolving later to logograms (where a character denotes an object, idea or action), and a single logographic writing system (such as Chinese or Linear B) can have many thousands of distinct visual signs. It wasn’t until about 2000 years ago in Egypt that phonemic writing was invented and used, where each character stands for a constituent of speech rather than having a meaning as in logographic writing. Many hundreds of writing systems have evolved and diversified from this ancestor (e.g., Latin, Arabic, Avestan, Mongolian, Phags-pa), varying widely in geometrical shape and style, and in the aspects of speech the characters represent (e.g., alphabets represent consonants and vowels, abugidas represent just consonants, and syllabaries represent syllables).

Amongst both non-linguistic and linguistic signs, some visual signs are representations of the world—e.g., cave paintings and pictograms, respectively—and it is, of course, not surprising that these visual signs look like nature. It would be surprising, however, to find that non-pictorial visual signs look, despite first appearances, like nature. Although writing began with pictograms, there have been so many mutations to writing over the millenia that if writing still looks like nature, it must be because this property has been selectively maintained. For non-linguistic visual signs, there is not necessarily any pictorial origin as there is for writing, because amongst the earliest non-linguistic visual signs were non-pictorial decorative signs. The question we then ask is, Why are non-pictorial visual signs shaped the way they are?

Previous efforts at answering this question have primarily concentrated on the differences. In particular, some of the shape differences among different (non-pictorial) visual signs are due to the kind of writing implement used, whether impressions in clay tablets with a blunt reed, rounded writing on leaves, or the physical details of a modified feather-tip point. Little attention has been devoted to uncovering the similarities, however, and as we will see here, there are deeper visual regularities that hold across human visual signs, independent of the writing mechanism (regularities that are also found in nature). It is as if someone had noticed that throat size causes male and female voices to sound differently, without noticing that male and female speech possesses a critical deeper regularity, namely that they utter the same set of phonemes, morphemes, words and sentences as one another (within a single language speaking community). We will find that, despite superficial differences in their shapes, visual signs appear to possess similar underlying “visual phonemes.”

The Shapes of Visual Signs

Uncovering deeper visual regularities that might govern visual signs is crucial in any attempt to explain why visual signs are shaped as they are, and, in turn, to explain why we are so good at reading. After all, we cannot explain why visual signs are shaped as they are if we do not first determine how visual signs over history are, in fact, shaped! A difficulty in trying to find such regularities is that it is not straightforward to describe how even a single letter is shaped, for a single letter can undergo considerable distortion (e.g., from person to person, or from font to font) without losing its identity. How can one hope to scientifically address the kinds of shapes found among visual signs, when it is awkward to even rigorously say what the shape of a single letter is?

To solve this problem I decided to use a topological notion of shape, where the details of the geometry do not matter, and what matters is only the manner in which strokes intersect, or join, with other strokes. A straight line, a C and an S have the same topological shape because each is topologically just a single stroke. L, T and X are the three distinct kinds of topological shape having two strokes. For example, a V has the same topological shape as an L because each consists of two strokes meeting at their endpoints. This notion of shape will be helpful later in measuring the shapes of nature, because while geometrical shape can change quickly as a function of a person’s viewpoint, topological shape is more viewpoint invariant, providing a more robust characterization of the shapes in nature. This topological notion of shape is not merely useful, but possesses psychological justification as well: experts in psychology (e.g., Irving Biederman’s work on intermediate-level representations) and computer vision believe that our visual systems may represent shape in a topological manner.


In earlier research I had shown (along with my collaborator Dr. Shinsuke Shimojo) that letters have on average three strokes, and this average does not vary as a function of the number of letters in the writing system. Because of this, I considered topological shapes with three or fewer segments, in particular the 36 topological shapes that can be drawn with three straight lines (even though each topological shape covers curvy geometrical shapes as well). In addition to a single-stroke, and L, T and X, there are five three-stroke configurations having a single junction (i.e., a single point of intersection of the strokes) exemplified by Y, K and Y. There are 11 configuration types having three segments and two junctions, exemplified by characters such as ] (or, equivalently, Z), 1, F, I, p, and ≠. Finally, there are 16 topological shapes with three segments and three junctions, such as D and A.

With these 36 kinds of topological shape primitive in hand, we (Dr. Shinsuke Shimojo, and two Caltech undergraduate students, Qiong Zhang and Hao Ye) set about rigorously measuring how common these shapes occur among visual signs. We began by measuring from three distinct classes of non-pictorial visual sign: phonemic writing systems (non-logographic), Chinese characters (logographic), and non-linguistic symbols. The set of phonemic letters were taken from about one hundred phonemic writing systems over history, and the topological shape of the entire letter was measured (if it was one of the 36 types in our repertoire). For Chinese characters and non-linguistic symbols, the signs typically have more than three strokes, and we measured all the topological shapes that occur as part of the whole sign.

What we discovered is that the shapes across these three very different kinds of visual sign are similar. For example, Ls and Ts are in each case common, but Xs rare. And across the 32 different kinds of topological shape with three segments, these three classes of visual sign highly correlate with one another. For example, Ys tend to be common relative to Õs, Zs and Fs more common than1s, and Hs more common than ps. That is, despite the seemingly unrestrained variability in shape among these visual signs, they in fact possess a similar topological shape “signature.” Now we are in a position to more meaningfully ask why visual signs are shaped as they are. Namely, why do visual signs have this signature?


They don’t get this signature by chance. For example, if one were to randomly place strokes onto the writing surface, the most common two-segment topological shape would be X. Ts would be rarer because they require a coincidental alignment of one endpoint along the edge of another. Ls, in turn, are even rarer because they require the double-coincidence of two endpoints touching. Among the topological shapes with two or three junctions, the shapes with more Xs will be more common in a randomly generated sign, and the shapes with more Ls the rarest; e.g., ≠ is the most common topological shape with two junctions, and Z the least common. This is not at all the case for the visual sign signature, where Ls and Ts are more common than X, and where, for example, ≠ is actually much rarer than Z. Another mechanism for the random generation of visual signs would be the act of scribbling, which is similar to the random-stroke case just mentioned, except that for scribbles Ls are now common, not rare. That is, for scribbles Ls and Xs are much more common than Ts, leading to a distribution of topological shapes unlike that of human visual signs.

[Figure 4 to be put near here.]

Designed for Reading or Writing?

Thus far, we have seen that human non-pictorial visual signs appear to possess a characteristic signature, and we have seen that this signature is not a result of chance. Before attempting to explain this signature, a natural first question is, Does this signature appear to be good for the eye, or good for the hand (or any other writing mechanism)?

There are at least two reasons for expecting that visual sign shapes are designed (by cultural selection) for ease of reading, not ease of writing. First, visual signs are written once, but can be read many times. Second, writing speed is typically limited not by the motor system, but by the time taken for the writer to compose the sentence; that is, writing is not like talking, where we can talk effortlessly without feeling as if we are composing our thoughts.

Shorthand is an example kind of visual sign that violates both of these reasons—it is typically not read more than once, and it is written without the writer having to compose the sentences (instead, the boss is orally dictating). Accordingly, shorthand is designed for the hand at the expense of the eye. We measured the topological shapes across six different shorthand writing systems, and found that their topological shapes are radically different from that found in visual signs more generally.

In contrast, consider trademark logos, which are designed to be seen at the expense of being written (they are, in fact, typically not written at all). We discovered that trademark logos possess the same shape signature found in visual signs.

That is, when we look at signs we know are designed for the eye at the expense of the hand, the signature matches the general signature we saw earlier, but when we look at signs we know are designed for the hand at the expense of the eye, the signature is altogether different.

As further support for this, we found that the visual sign shape signature correlates well with the number of angles in the shape (a measure of visual complexity), but does not correlate at all with the number of hand motions required to write the topological shape (a measure of motor complexity).

Together this makes a strong argument that the topological shapes of visual signs have been selected for reading, not writing.

Natural to the Eye

The topological shapes of non-pictorial visual signs are, then, for the eye, not the hand. But we are still left with the question, Why does the eye like these shapes? Here is where the evolutionary, or ecological, hypothesis enters into the story. Because over millions of years of evolution our visual systems have been selected to be good at processing the conglomerations of contours occurring in nature, I reasoned that if visual signs have culturally evolved to be easy to see, then we should expect visual signs to have natural topological shapes.


Where are these topological shapes in nature? What were conglomerations of strokes for visual signs are now conglomerations of contours for natural scenes. Contours are the edges of objects (as seen by the eye), not, of course, strokes in the world. For example, an L occurs in the world when exactly two edges of an object meet at their endpoints, like an elbow. A T occurs in the world when the edge of an object goes behind another object in the foreground. A Y occurs, for example, at the inside corner of a rectangular room. We measured how common these and the other topological shapes occur in natural scenes, and were stunned to find that nature possesses the shape signature we saw earlier for visual signs. That is, visual signs are shaped like nature, confirming our ecological hypothesis for the shapes of visual signs.

If visual signs look like nature, one might first suppose that the shape signature of nature depends significantly on which natural environment one considers. However, to our surprise, we found that the shape signature is highly robust, differing hardly at all whether we measured images of ancestral environments (e.g., tribal villages, savannas) or urban environments (buildings, walkways). Because of the robust topological notion of shape we used in our analysis, any environment with opaque objects strewn about will tend to have the same shape signature. This underlies why the diverse kinds of visual sign have a similar signature despite the diverse environments from which they spring, and one may speculate that aliens might for this reason possess visual signs that look reminiscent of our own.

changizi_sciblog_topographyOfLanguage_figs_Page_07We have been considering visual signs generally, but let us now specifically consider letters in phonemic writing system, for there is an additional question one might have about letters. We saw a moment ago that letters look like natural object junctions. Our ecological hypothesis expects letters to look natural, but why natural junctions? Why not have letters shaped like natural single contours? Or, alternatively, why not have letters shaped like whole objects? Instead of one stroke or a dozen strokes, letters in fact tend to have about three strokes (independent of the size of the writing system), and thus are at an intermediate level between edges and objects.

The answer may lie in the following pair of facts: (i) we wish to read words, not letters; and (ii) we have evolved to see objects, not object-junctions. In this light, we expect culture to select words to look like objects, so that words may be processed by the same area in visual cortex responsible for recognizing objects. Logographic characters (e.g., Chinese) and non-linguistic symbols do tend to be more object-like, possessing many more than three strokes. For phonemic writing, however, there are severe limits to how closely words can match natural objects, for the manner in which letters combine is determined by speech. However, by having letters shaped like natural object-junctions—rather than natural contours or natural whole objects—written words become combinations of natural junctions, and thus more similar to objects and more easily processed by our visual system.


Evolution by natural selection is too slow to design our brains for reading, and so cultural selection has come to the rescue, designing (without any designer) visual signs for our brains. Because our visual systems have evolved to be good at perceiving natural objects, cultural evolution has created non-linguistic symbols, logographic symbols, and written words in phonemic writing that tend to be built out of object-junction-like constituents, and are thus object-like. In particular, this explains why letters tend to have around three strokes and have the topological shapes they do. We expect that these insights will be useful in designing optimal alphabets or visual displays.

Because culture is capable of designing for the eye, the visual signs of our culture are a fingerprint of what our visual systems like. Akin to the linguistic study of the auditory productions humans make, the “visual linguistic” study of the visual productions people make is a currently under-utilized tool for vision research. There is every reason to believe that the study of visual linguistics will aid traditional lab experiments on vision and brain design as much as linguistics has supplemented lab experiments on cognition.

changizi_sciblog_topographyOfLanguage_figs_Page_09Mark Changizi is a professor of cognitive science at Rensselaer Polytechnic Institute. This research – and other work of his on the evolution of color, illusions and stereo vision – are the topic of his new book, The Vision Revolution (Benbella Books).

[A relevant ScienceDaily piece.]

Short Bibliography:

Biederman I & Cooper EE (1991) Priming contour-deleted images: evidence for intermediate representations in visual object recognition. Cognitive Psychology 23: 393–419.

Changizi MA & Shimojo S (2005) Character complexity and redundancy in writing systems over human history. Proceedings of the Royal Society of London B 272: 267-275.

Changizi MA (2006) The optimal human ventral stream from estimates of the complexity of visual objects. Biological Cybernetics 94: 415-426.

Changizi MA, Zhang Q, Ye H & Shimojo S (2006) The structures of letters and symbols throughout human history are selected to match those found in objects in natural scenes. The American Naturalist 167: E117-E139.

Changizi MA (2009) The Vision Revolution (Benbella Books, Dallas).

Daniels PT & Bright B (1996) The World’s Writing Systems. New York: Oxford University Press.

Read Full Post »

This is the Wall Street Journal review of The Vision Revolution that appeared earlier this year.

Why the Eyes Have It
We can read words, gauge distance and see color. How did that happen?

By Christopher F. Chabris

Why are we ­humans so good at seeing in color? Why do we have eyes on the front of our heads rather than on the sides, like horses? And how is it that we find it so easy to read when written language didn’t even exist until a few thousand years ago—a virtual millisecond in evolutionary time?


The Vision Revolution

By Mark Changizi

Ben Bella, 215 pages, $24.95

Read an excerpt of “The Vision Revolution”

Most of us, understandably, have never given much thought to questions like these. What is surprising is that most cognitive scientists ­haven’t either. People who study the brain generally ask how it works the way it does, not why. But Mark Changizi, a professor at Rensselaer ­Polytechnic Institute and the author of “The Vision ­Revolution,” is indeed a man who asks why, and lucky for us: His ideas about the brain and mind are fascinating, and his explanations for our habits of seeing are, for the most part, persuasive.

Mr. Changizi takes care not to call himself a practitioner of evolutionary psychology. This is the one discipline of the mind sciences that focuses on why questions, but it often answers them by telling just-so stories that cannot be ­disproved. (Why do men have better spatial ability than women? Because a long time ago, in Africa, men needed spatial skills to track prey and to kill at a distance—a plausible theory but one that is difficult to test with experiments.) Instead Mr. Changizi calls ­himself a “theoretical neuroscientist,” seeking explanations for the design of the mind that are based on mathematical and physical analysis. He has his own stories, it is true, but they are grounded solidly in neuroscience, and they are backed up by data about a surprising range of human activities, from ­the colors found in historical ­costumes to the ­correspondence between the shapes found in written letters and the shapes found in ­nature.

Let’s start with the question of color. It is such a natural part of our visual experience that we don’t stop to wonder why we can see it at all. ­Without color television there would have been no “Miami Vice,” of course, but were we really missing out on so much when we had only black and white? The consensus explanation for our superior ability to perceive color is that primates evolved it to see fruit—you can’t order dinner if you can’t read the menu.

Mr. Changizi thinks otherwise. He proposes that color vision is useful for distinguishing the changes in other ­people’s skin color—changes that are caused by shifts in the volume and oxygenation levels of the blood. Such shifts, like blushing, often signal emotional states. The ability to see them is adaptive because it helps an observer to “read” states of mind and states of health in others, information that is in turn useful for predicting their behavior.

Our brains evolved in a time when people lived their entire lives without ever seeing someone with a skin color different from their own. Thus the skin color we grow up seeing, Mr. Changizi says, is “neutral” to us: It serves as a kind of baseline from which we notice even minor deviations in tint or hue. Almost every language has distinct words for some 11 basic colors, but none of them aptly describe the look of skin, which seems colorless (except in our recent multicultural societies, where skin color is newly prominent). As one might expect, primates without color vision tend to have furry faces and hands and thus less need to perceive skin color; ­primates with color vision are more “naked” in this respect, humans most of all.

James Steinberg

Conventional wisdom may be similarly misleading when it comes to binocular vision. It is said that we have two forward-facing eyes, which send our brains two separate images of almost everything in our field of view so that the brain can compare those images to estimate the distance of objects—a generally useful thing to know. But people who are blind in one eye, Mr. Changizi notes, can perform tasks like driving a car by using other cues that help them to judge distance. He offers a different explanation: that two eyes give us a sort of X-ray vision, allowing us to see “through” nearby objects to what is beyond.

You can experience this ability yourself by closing one eye and holding your forefinger near your face: It will appear in your field of vision, of course, and it will block what lies beyond or behind it. If you open both eyes, though, you will suddenly perceive your finger as transparent—that is, you will see it and see, ­unblocked, the full scene in front of you. Mr. Changizi observes that an animal in a leafy environment, with such an ability, gains an advantage: It can lurk in tall grass and still see what is “outside” its hiding place. He correlates the distance between the eyes and the density of vegetation in the habitats of animals and finds that animals with closer-set eyes do tend to live in ­forests rather than on plains.

As for reading, Mr. Changizi stops to observe how remarkable this ability is and how useful, giving us access to the minds of dead people (i.e., deceased writers) and permitting us to take in words much faster than we can by merely listening to them. He claims that we learn to read so easily because the symbols in our written alphabets have evolved, over many generations, to resemble the building blocks of natural scenes—­exactly what previous millennia of evolution adapted the brain to perceive quickly. A “T,” for example, appears in nature when one object overlaps ­another, like a stone lying on top of a stick. With statistical analysis, Mr. Changizi finds that the contour patterns most common in nature are also most common in letter shapes.

Mr. Changizi has more to say about our visual experience—about optical illusions, for instance, which he sees as artifacts of a trick the brain uses to cope with the one-tenth of a second it takes to process the light that hits our eyes and to determine what is actually in front of us. He calls for a new academic discipline of “visual linguistics,” and he tells us why there are no ­species with just one eye.

What does all this add up to? Provocative hypotheses but not settled truth—at least not yet. As a theoretician, Mr. Changizi leaves it to others to design experiments that might render a decisive ­verdict. Someone else will have to study how accurately people can perceive mental states from shifting skin tones, and someone else will have to ­determine whether, in most cases, looking at another ­person’s skin adds any useful information to what is easily known from facial expression, tone of voice and body ­language.

Still, the novel ideas that Mr. Changizi outlines in “The Vision Revolution”—together with the evidence he does present—may have a big effect on our understanding of the human brain. Their implication is that the environments we evolved in shaped the design of our visual system according to a set of deep principles. Our challenge now is to see them clearly.

Mr. Chabris is a psychology professor at Union College in Schenectady, N.Y.
Mark Changizi is Professor of Cognitive Science at RPI, and the author of The Vision Revolution (Benbella, 2009) and The Brain from 25000 Feet (Kluwer, 2003).

Read Full Post »

This is the article that appeared in 2008 in the New York Times about my perceiving-the-present theory of illusions (the topic of Chapter 3 of The Vision Revolution)…

Anticipating the future to ‘see’ the present

By Benedict Carey

Staring at a pattern meant to evoke an optical illusion is usually an act of idle curiosity, akin to palm reading or astrology. The dot disappears, or it doesn’t. The silhouette of the dancer spins clockwise or counterclockwise. The three-dimensional face materializes or not, and the explanation always seems to have something to do with the eye or creativity or even personality.

That’s the usual cue to nod and feign renewed absorption in the pattern.

In fact, scientists have investigated such illusions for hundreds of years, looking for clues to how the brain constructs a seamless whole from the bouncing kaleidoscope of light coming through the eyes. Brain researchers today call the illusions perceptual, not optical, because the entire visual system is involved, and their theories about what is occurring can sound as exotic as anyone’s.

In the current issue of the journal Cognitive Science, researchers at the California Institute of Technology and the University of Sussex argue that the brain’s adaptive ability to see into the near future creates many common illusions.

“It takes time for the brain to process visual information, so it has to anticipate the future to perceive the present,” said Mark Changizi, the lead author of the paper, who is now at Rensselaer Polytechnic Institute. “One common functional mechanism can explain many of these seemingly unrelated illusions.” His co-authors were Andrew Hsieh, Romi Nijhawan, Ryota Kanai and Shinsuke Shimojo.

One fundamental debate in visual research is whether the brain uses a bag of ad hoc tricks to build a streaming model of the world, or a general principle, like filling in disjointed images based on inference from new evidence and past experience. The answer may be both. But perceptual illusions provide a keyhole to glimpse the system.


Orbison illusion, or Ponzo illusion variant.

When shown two images in quick succession, one of a dot on the left of a screen and one with the dot on the right, the brain sees motion from left to right, even though there was none. The visual system has apparently constructed the scenario after it has been perceived, reconciling the jagged images by imputing motion.

In an experiment originated by Nijhawan, people watch an object pass a flashbulb. The timing is exact: the bulb flashes precisely as the object passes. But people perceive that the object has moved past the bulb before it flashes. Scientists argue that the brain has evolved to see a split second into the future when it perceives motion. Because it takes the brain at least a tenth of a second to model visual information, it is working with old information. By modeling the future during movement, it is “seeing” the present.

Changizi and his colleagues hold that it is a general principle the brain applies to a wide variety of illusions that trick the brain into sensing motion.

“It’s likely that there are many different neural mechanisms involved in perceptual illusions,” said Jacob Feldman, a Rutgers psychologist. “But the idea that there may be some overarching explanation that accounts for these separate mechanisms is compelling and satisfying to some scientists.”

Timothy Hubbard, a psychologist at Texas Christian University, said the principle of perceiving the present was sound, adding, “If a person’s response to an object, to catch, hit, block, whatever, is to be optimal, that response should be calibrated to where the object would be”— not a split second earlier, when the perception occurred.

This is why identical squares arranged around the center of a spoked-wheel image appear misshapen, said Changizi, who writes about it in a book due in 2009, “The Vision Revolution.” The sides of squares closer to the center appear to bulge. The sides farther out appear shorter. The radiating lines in the pattern trick the brain into perceiving motion forward, so it projects objects forward, making those nearer the center appear closer to the eye.

The same effect can be seen by leaning forward toward a precise checkerboard. The image seems to bulge forward, this time because the eyes are moving.

Changizi says such illusions can also occur in real life. When a golf ball or baseball rolls through the grass and suddenly drops into a hole, the brain sometimes perceives a trace of the ball on the other side of the hole.

“But these are things that we don’t experience very often,” he said, “because the brain is so good at covering up its mistakes.”


The paper itself is here. A ScienceDaily piece, and an interview in Scientific American.

Mark Changizi is Professor of Cognitive Science at RPI, and the author of The Vision Revolution (Benbella, 2009) and The Brain from 25000 Feet (Kluwer, 2003).

Read Full Post »

Older Posts »