Mind-Controlled Motion Perception

What you think is what you see.

Posted Feb 18, 2019

The brain has an astonishing capacity to detect patterns in the world. So much so, it will often create patterns where none exist.

Pareidolia is the tendency to see familiar shapes in random objects. Seeing a bunny in the clouds, an angry face on a bell pepper, or Jesus on a piece of toast are all classic examples of pareidolia.

It turns out that we also have pareidolia for motion, or motion pareidolia. For this, I will ask the reader to observe the following display for a few seconds:

Can you see any coherent motion from one frame to the next?

If you can't, try saying out loud, "Right, Left, Right, Left, Right, Left," as each frame changes. Did that change your perception? Now, try saying, "Up, Down, Up, Down, Up, Down," while observing frames. Did you see up and down motion this time?

In reality, there is no motion in the display—no coherent motion, that is. Sure, there are pixels moving around in random directions, but any perception of coherent motion is purely constructed by your brain.

How does the brain do this? In a recent paper (Davidenko, Heller, Cheong, and Smith, 2017) we proposed that the brain creates coherence from these noisy pixel arrays using three separate processes: the first is flexible correspondence. When you look at the display above, you might focus on a particular cluster of pixels in one frame, and check for a corresponding cluster of pixels in the next frame. As long as there is some cluster that roughly matches up with the cluster on the previous frame, your visual system will consider that to be a match. Of course, with enough flexibility, any cluster can be made to match up with any other cluster, so the possibilities for perceiving motion are nearly endless.

The second process is local-to-global propagation. The idea is that by focusing our attention on one part of the display, we can't attend very well to any other part of the display. That means that the rest of the display will be assumed to move the same way as the part we are attending to. In 1985, V. S. Ramachandran and Stuart Anstis ingeniously demonstrated this phenomenon using ambiguous motion quartets. If you look at the dots below on the left, you can perceive them as moving either up and down, or right and left. In fact, you can control with your mind which way they appear to move. (Note: if you are stuck on one interpretation, try covering up part of the display with your hand). Amazingly, when you look over to the whole array of quartets on the right, they all seem to behave the same way. Either they all move up and down, or they all move right and left, in perfect synchrony.

The third process is a form of top-down control, or confirmation bias. By repeating the words "up, down, up, down", you are guiding your perceptual system to find clusters of pixels that conform to the expected motion pattern. If this verbal guide is strong enough, it will override other motion signals that may be present in the display, and you will end up seeing the motion you are repeating to yourself.

Put the three processes together, and you get motion pareidolia.                                                      

But why does the brain do this? Why does the brain fool us into perceiving motion that is not there? Isn’t this a major flaw of the visual system?

Actually, seeing illusory motion patterns in noise may be an evolutionary advantage. A vulnerable animal is better off to assume that an ambiguous movement behind a tree is a predator and be wrong, than to incorrectly assume the movement is not a predator and be eaten. Overreacting to a false alarm is far less costly than ignoring a real threat. In order to arrive at the wrong conclusion, however, the brain has to actively construct possible scenarios from incomplete sensory input. And this process includes the formation of illusory percepts.

In our day-to-day life, such illusory percepts are usually fleeting. Generally, there is a stable reality out there that we can rely on to verify whether our initial percepts are correct. Motion pareidolia works because the information disappears from one frame to the next. We cannot verify whether we really saw an upward motion of the pixels, or whether that motion was illusory.

Rather than indicating a flaw of the visual system, motion pareidolia simply reveals what our “unverified percepts” look like, and how we can influence them with our thoughts.

References

Davidenko, N., Heller, N. H., Cheong, Y., & Smith, J. (2017). Persistent illusory apparent motion in sequences of uncorrelated random dots. Journal of vision, 17(3), 19, 1-17.

Ramachandran, V. S., & Anstis, S. M. (1985). Perceptual organization in multistable apparent motion. Perception, 14(2), 135-143.