Monday, June 30, 2008

Four Squares

What to notice: The patterns in the four squares are stationary, yet they appear to move.

Brief Comment: This week’s illusion is a motionless image that creates the perception of motion. In this respect, the illusion is similar to the Rainbow Boom/Rainbow Bust illusion that I posted a few weeks ago. People differ in how they see peripheral drift illusions: for some people, the effect is very strong, and others do not see it at all. The effect seems to become less strong as you age.

This week’s illusion is based on a design principle made famous in Akiyoshi Kitaoka’s “rotating snakes.” To make this type of illusion, you start with an individual element made of segments that follow the pattern “dark, bright, less bright, darkest” (or, to put it into numbers, the luminance levels of the segments follow a pattern something like 0.3, 1.0, 0.7, 0.0—there are nuances concerning the luminance ratio and the size of segments). You then repeat the element enough times to form a ring. The effect appears to become stronger if you include multiple rings near the same location.

These patterns are fun to manipulate in programs like Adobe Illustrator and Adobe Photoshop. Here, I have created a pattern and placed the resulting rings into a four-square configuration with bright colors (Kitaoka-like elements in an Andy Warhol-like display with Keith Haring-like colors).

Why does this effect occur? Most explanations of peripheral drift illusions suggest that the illusory patterns fool the eye into sending to the brain information that mimics a real motion signal.

The eye sends information to the brain through different types of neurons. Some of these neurons transmit information faster than others, and some of these neurons respond more quickly to high-contrast parts of an image than to low-contrast parts of the image. Because of these differences, the response of the eye to one part of the illusory pattern reaches the brain at a slightly different time than the response of the eye to another part of the illusory pattern. The difference in the arrival time is exactly the same type of event that would occur with “real” motion, and so motion detectors in the brain signal that motion has occurred.

The details of these events for the peripheral drift illusions have not yet been settled. One explanation is based on the idea that the early visual system responds faster to high-contrast information than it does to low-contrast information. Variations on this idea were put forward by Ben Backus and Ipek Oruc (here is a link to their paper) and by Bevil Conway, Akiyoshi Kitaoka, Arash Yazdanbakhshm, Christopher Pack, and Margaret Livingstone (here is a link to their paper). Another explanation suggests that the motion occurs because the neural systems that respond to positive contrast changes (ON channels) respond faster than the neural systems that respond to negative changes in contrast (OFF channels). This idea was put forward by Maria Del Viva, Monica Gori and David Burr to explain a slightly different illusion (here is a link to their paper).

I will be on the road in California most of July. I will try my best to keep up with the blog while traveling.

Sunday, June 22, 2008

Swimmers: a gradient-gradient illusion

You are looking at a collection of “eggs” that are placed in front of a striped pattern. As the pattern moves from left to right across the screen, the eggs appear to swim up and down, and to change from light to dark.

Press the “Click to toggle background” button to replace the striped pattern with a stationary background. As you can see, the eggs are completely stationary.

The lever on the bottom left of the display allows you to rotate the eggs. Try rotating the eggs by about 45 degrees, and notice how the orientation of the eggs changes the pattern of the motion.

Brief comments: The motion of the eggs is surprising, but it really shouldn’t be. In the real world, we frequently encounter conditions where relative information makes stationary objects appear to move. For instance, when clouds drift by the moon, the moon may look as if it is shifting in the opposite direction (this is called “induced motion”). Also, there are motion aftereffects: if you view motion in one direction and then view a still object, the object will appear to drift in the opposite direction (here is another motion aftereffect).

Motion information from stationary objects can also be created whenever you move your eyes or your head, or whenever you move closer or further away from the stationary object. (Do stationary objects appear stationary when you move your camcorder?)

There are so many sources of motion information in our environment that I often find it surprising that anything ever appears stationary.

The motion in the swimmers illusion seems to arise from the contrast between the eggs and the background. The eggs are shaded light on top and dark on the bottom. When a light portion of the background crosses the egg, the egg appears to move upward; when a dark portion of the background crosses the egg, the egg appears to move downward. Does this type of motion sound familiar? The shift toward the point of low contrast is the same in some illusions in previous posts: the window shade illusion; Lucy in the sky; and the grouping by contrast illusion.

So, why should the contrast motion in the swimmers illusion surprise us? Motion from contrast seems to contradict an object-centered representation of the world. We have many words for objects, but not so many ways to describe the relationship between the objects and their surrounds—I am stuck with “high contrast” and “low contrast.” Perhaps the motion seems surprising because we have an impoverished ability to talk about (and consciously represent) relative stimulus information, compared to our ability to talk about objects.

The swimmers illusion was created when I was investigating effects of combining different sources of gradients in the environment (shadows on shadows). In 2007, this illusion was among the top ten in the third annual Best Illusion of the Year contest and was presented by Emily Knight, then an undergraduate in my lab.

All effects have precedence. The swimmers illusion is, in many respects, a two-dimensional version of Stuart Anstis’s “footstep illusion” (2001). Click here for a .pdf of the original footsteps article, and click here to see a demonstration of the footstep illusion.

Stuart Anstis gives extraordinarily engaging lectures that are both humorous and informative. Here is a link to his lecture entitled “Colours, Faces and Mrs. Thatcher's Bikini,” which was given December 2007 and posted by Cambridge Research Systems. You really should see the lecture. It is an excellent opportunity to watch a maestro at work and to see some absolutely fantastic new illusions.

Sunday, June 15, 2008

Curtains: spotlights and shadows

You are looking at two curtains. At the center of the curtain on the left is a circular region that looks like a shadow; at the center of the curtain on the right is a circular region that looks like a spotlight. As would be expected from our understanding of the real world, the region in the shadow looks darker than the region in the spotlight.

But this is so only when the disks are vertically oriented. Actually, the two disks are physically the same.

Move the slider to rotate the disks. When the disks are turned, they no longer “line up” with the background, and as a result, they no longer appear to be a shadow and spotlight. In this view, the disks appear to be filled with the same gradient pattern (and actually are).

The disks can be moved. You can get a similar effect by clicking on and dragging a disk to place it next to the other, on the same background.

Brief Comments: One of my favorite places on the web is The Situationist, a blog that explores how the “situation” (or context) affects interpretation. The site has numerous examples of how objects, people, and events in one context are interpreted differently from the same objects, people, and events placed in another context.

The visual display above presents an example of the effects of the visual “situation.” In one situation (vertical orientation for the disks), the viewer interprets the disks with reference to the background context (i.e., the two curtains). One disk looks like a shadow on the curtain, and the other looks like a spotlight. The disks are therefore interpreted as a dark spot and a lighter spot on the curtains. In another situation (horizontal orientation), the viewer is able to separate the disks from the context of the curtains and therefore will identify the disks as having the same shading.

From my perspective, the most surprising aspect of this display is that the spots look the same when oriented horizontally. Why should it be surprising that the disks look the same? Consider the relation of this visual display to an old and well known situational effect: “simultaneous contrast.” In a simultaneous contrast display, a gray patch appears brighter when placed against a dark background than when placed against a bright background. In the visual display above, the left curtain is brighter than the right curtain, and so it makes sense that the left disk appears darker than the right disk when the disks are oriented vertically. But when the disks are oriented horizontally, it is as if one situational effect – simultaneous contrast – disappears in the presence of a different situational effect (orientation).

Why does the effect of orientation apparently supersede the effect of simultaneous contrast when we interpret the appearance of the disks? The illusion above is a response to research conducted by Bart Anderson and Jonathan Winawer (here is a link to their 2005 Nature article, and to Bart Anderson’s demonstration webpage). Please read their article to learn the details of their interpretations.

My interpretation differs from theirs. I will write more about interpretations of simultaneous contrast and orientation in future posts.

Visual scientists have argued for more than a century about the causes of simultaneous contrast. If you are curious about this topic, here is a link to a recent book by Alan Gilchrist that gives a history of the research into lightness illusions.

Saturday, June 7, 2008

Rainbow Boom/Rainbow Bust

You are looking at a pattern of rectangles sitting in front of a multicolored background.

For most people, the pattern appears to expand or, if you press the “click to reverse direction” button, to contract. The perception of expansion or contraction (of “rainbow boom” or “rainbow bust”--an economic metaphor) occurs even though the image is completely still.


The key to the motion is the contrast between the colored edges on the rectangles and the gradient background. If the yellow edges are on the outside of each rectangle, the pattern appears to expand; if the yellow edges are on the inside, the pattern appears to contract.

The effect is reminiscent of the Pinna-Brelstaff illusion, which consists of diamonds with two white edges and two black edges, in ring formation:

Stare at the blue star, and move your head closer to and further from the screen. As you do so, the rings should appear to rotate.

Unlike motion in the Pinna-Brelstaff illusion, the perceived motion in Rainbow Boom/Rainbow Bust does not depend on the observer moving back and forth. I achieved this effect by placing the rectangular pattern on a gradient background. In this way, even though the rectangles are the same throughout the display, the contrast between the rectangles and the background changes from the center outward. It is the contrast that seems to be driving the perception of expansion and contraction.

What causes the motion? One possibility is that small eye movements lead to changes in the contrast response; the visual system creates motion from the differences in contrast across edges. Another possibility, put forward by Ben Backus and Ipek Oruc, is that the brain’s motion detectors fail to compensate for the dynamics of local adaptation. The Backus-Oruc theory was created to account for the perception of motion in a different static illusion (here is a link to their paper). If it is correct, then it is quite likely that their theory applies to Rainbow Boom/Rainbow Bust as well.

One last note: Akiyoshi Kitaoka has created many wonderful and novel variations on the Pinna and Brelstaff theme. Take a look!

/*********google analytics*/ /*********google analytics end */