WHY IS THE STUDY of perception so appealing? One reason is that you can gain deep insights into the inner workings of your own brain by doing relatively simple experiments that any schoolchild could have done 100 years ago. More on those in a moment.
Your sensory experience of the world does not involve faithfully transmitting the retinal image to a screen in the brain so that it can be “seen” by some inner eye. One piece of evidence for this fact is that your perception of an object (in a, do you see two faces or a goblet?) can change radically even if the image on the retina is held constant, which implies that even the simplest act of observation involves judgment by the brain.
Less obvious, but equally important, is the converse. Your perception of the world—or an object in it—can also remain stable if the image is changing rapidly on the retina. One example is how you take in a scene when you move your eyes around. Every time you glance around a room, the image dances around the retina at warp speed, hundreds of feet per second. Yet all appears rock steady. Why?
Now, at first you might think the world does not appear to lurch because all motion is relative. The clouds glide in the twilight sky, but we assume they are stable and attribute the motion to the smaller object, the moon.
A simple experiment demolishes this idea. Close one eye—let us say the left. Then, keeping the right eye open, use the right index finger to displace the right eyeball, rocking it side to side slightly in its socket. (Gently!) You will see the world jump as if in an earthquake, even though there is no relative motion on the retina.
Why do we see a stable world when we swivel our eyes naturally but not when we jiggle an orb manually? The answer came from the great 19th-century physician, physicist and ophthalmologist Hermann von Helmholtz. He suggested that when the command to move the eyes is sent from the frontal lobes to the muscles of the eyeballs, a faithful copy of the command (like a “CC” for an e-mail) also goes to visual motion–detecting centers in the back of the brain. As a result, they are tipped off ahead of time: “You are going to get some motion signals, but they are not caused by real movement of the world, so ignore them.”
We can speak of two independent systems in the brain, either of which can signal a sensation of motion. Neuropsychologist Richard L. Gregory of the University of Bristol in England calls these the image/retina system (caused by image movement on the retina) and the eye/head system (generated by sensing the movement of the eyes). Ordinarily, the brain subtracts one signal from the other. When you move your eyes around, these two motion signals cancel each other out and the world remains stable.
We know that the image/retina system exists because of the experiment in which you jiggled your eye with your finger. But how do we know the eye/head system can independently evoke a motion sensation? Think about what happens when your eyes track a glowing cigarette tip moving across a completely dark room. You correctly see it moving several feet, even though the cigarette image does not move much at all on your retina. Instead your eyes are making a big excursion. So the brain “concludes” that the cigarette must have moved an amount equivalent to the eye movement. Again, we can speak of the final movement perceived as resulting from the subtraction of image/retina signals (close to zero because you are tracking it) from eye/head signals (large, because the eyes move a large distance to keep the cigarette’s image on the fovea, the area of the retina responsible for acute vision). The net result is that you see the glowing orange spot moving several feet.
You can produce a more striking version of this effect by having a friend take a photograph of you while you look directly at the flash. The result is a persistent afterimage of the bulb caused by continued activity of the receptors long after the light burst is gone. This flash image is “glued” to your retina; it cannot move even a tiny bit. Yet if you go to a dark room and move your eyes around, you see the afterimage moving vividly with the eyes. The eye/head system is signaling a large value, but the image/retina signal is zero—so as a result of the subtraction, you see the afterimage moving even though it is fixed and stationary on the retina.
You can create a similar fixed afterimage without a flash by staring for 30 seconds at the central X in the image in b; you will see the afterimage when you shift your gaze to a blank sheet of paper. (Blink your eyes to refresh the image if necessary.)
Forward and Back
Next question: What is the source of signals generated by the eye/head system? One possibility, called feedforward, is that a copy of the command from eye-movement centers is delivered to the sensory motion–detecting centers so that they will expect—and thus cancel—spurious image/retina signals. A second option, called the feedback theory, is that receptors in the eye muscles themselves sense the degree of eye movement and send the “cancellation” information to the sensory motion–detecting centers. Which is correct?
To find out, Helmholtz performed a heroic experiment. He paralyzed his eye muscles using a local anesthetic instilled around the eyeballs. Every time he then tried to move his eyes (unsuccessfully, of course), the world appeared to move in the opposite direction—even though neither the image nor the eyes were moving. He concluded that the feedforward model was correct. His brain could not have relied on feedback, because his eye muscles were paralyzed. It is as if a copy of the intention to move the eyes is sent (feedforward) to the motion-sensing areas to be subtracted from the expected image/retina movement. But because there is nothing to subtract, the net result is motion perceived in the opposite direction.
Another bit of evidence. Create an afterimage on one retina using a flash (keep the other eye closed). What happens if in a dark room you now jiggle the eyeball with your finger? The answer is... absolutely nothing. You do not see the afterimage jiggling. The reason is that in the dark when you jiggle the eyeball the afterimage remains perfectly still on the retina. So there are neither image/retina signals nor any command signals from the eye-movement motor centers. Subtract zero from zero, and you get zero. The experiment is also indirect evidence for the feedforward theory and against the feedback theory (because when you push your eyeball around, stretch receptors in the eye muscles are activated—albeit not in a coordinated manner).
Now consider an extreme example. Create an afterimage of a flash in one eye. Now imagine (do not actually try it!) that you pluck the eye from its socket, keeping the optic nerve undamaged. Holding the eye in your hand, turn it so it is looking behind your shoulder. Where do you think you would see the afterimage? You would still see it in front even though the eye is pointing backward because there is no way the visual centers could know that the eye is pointing backward.
The Joint Is Jumping
Let us imagine another scenario. You walk into a discotheque lit by a strobe light. Given the right strobe rate, if you just move your eyes around, the entire world—including people and furniture—will appear to be jumping. When you move your eyes, the commands from the eye/head system go to the motion-sensing areas. Usually these messages would be canceled by image/retina motion signals. But your eyes in effect take static snapshots with each strobe, sampling the image. These samples behave effectively like afterimages. The ensuing failure to subtract retinal signals from commands results in a net perceived movement of the world.
Better still, have a friend hold a tiny luminous spot—like a lit cigarette or tiny-wattage penlight—motionless. Move your eyes, and it will, of course, look stationary. If you now strobe the room, every time you move your eyes, your friend will appear to jump around, but the glowing point will remain exactly where it is. This is because the light, being self-luminous and continuously visible, generates image/retina motion signals that are canceled by eye/head commands. Yet the rest of the room and your friend, being “sampled” with the strobe, do not generate retinal motion and therefore appear to jump with the eye. The astonishing paradoxical perception you see is the penlight flying away from the person.
Our former mentor Fergus W. Campbell, who was a physiologist at the University of Cambridge, found an ingenious practical application for this effect in a London nightclub. He had the cabaret women wear skimpy luminous bikinis as they danced in a strobe-lit room. When patrons moved their eyes around, they would see the luminous bikinis flying off tantalizingly, yet they revealed nothing. The illusion was a hit and was perfectly legal because there was no real nudity. We sometimes wonder whether science itself is the same way; each time you think you are unveiling the truth, all you get is a teasing glimpse of what turns out to be yet another veil.
The intelligent reader who has followed our reasoning so far will inevitably ask the following question: When I move my eyes intentionally, the “volition” signals get sent to the sensory motion areas to cancel out the spuriously produced image-on-retina motion. But why can’t the same type of cancellation or subtraction occur when you voluntarily use your finger to jiggle the eyeball? Why can’t you send “finger movement” signals to the visual-image motion centers? After all, you know you are moving your eyeball.
The answer tells us something very important about perception. Even though it appears “intelligent” at times and can benefit hugely from high-level stored knowledge, it is by and large on autopilot, because it has evolved to do things quickly and efficiently. Even though you know you are pressing on your eyeball, no cancellation occurs because—unlike the eye-movement command centers—the finger-movement centers in the brain simply do not send the CC message to the motion-sensing areas. Our forebears apparently developed connections between eye-movement command centers and sensory-visual areas because we often do move our eyes. But our ancestors did not, we can be sure, walk around tapping their eyeballs with their fingers. Hence, there was never any evolutionary selection pressure to evolve such connections.