As an expert in the science of visual perception, Damon Clark is always thinking about how animal brains process and repackage the information from their environment in order to make it as useful as possible. Recently, he’s been trying to unravel these processes using fruit flies, styrofoam balls, and an optical illusion—all part of an experiment to see if tricking flies into perceiving motion can offer valuable clues about how human vision works.
“There’s nothing inherently weird about illusions, I would say, because nothing we perceive is real,” says Clark, associate professor of molecular, cellular, and developmental biology at Yale University. He’s not espousing new-age theology or brain-in-a-vat philosophy: Clark knows well, from his years of studying the visual system, just how different our perceptions can be from what’s actually in front of us. Consider, for example, how frequently most people see faces in wood grain, or burnt toast, or a popcorn ceiling. There’s nothing special about those surfaces, but the human brain evolved to recognize faces far more quickly and readily than other objects. After all, a brain that incorrectly assumes that the grill of that car is a face is also a brain that can readily and rapidly detect people—who may be friends or foes—in any setting. So while the difference between what we see and what’s actually there might be most obvious when we look at optical illusions, the brain’s ability to jump to (sometimes incorrect) conclusions is also fundamental to our ability to successfully navigate our visual environment.
The strangeness of illusions, though, makes them a powerful tool for understanding how the sense of sight works. The optical illusions you may have paged through as a youngster—the Hermann grid, color aftereffects, and the Müller-Lyer illusion, to name a few—have helped scientists chip away at the mysteries of our visual system. These illusions powerfully demonstrate how sensitive our brains are to context and how readily they adapt to, and begin to ignore, unchanging stimuli.
Clark and his colleagues wanted to understand a class of illusions called peripheral drift illusions, in which stationary patterns seem to move when they are not looked at directly. Specifically, they focused on a relatively simple version of this illusion, in which sections of a ring transition smoothly from white to black and then suddenly back to white, like in the image below. If you look to the side of the illusion, so that it lies in your peripheral vision, you should see the rings rotate. Blinking quickly can enhance the effect.
Unlike most scientists who study illusions, Clark didn’t examine this illusion in humans and other primates. Instead, his team studied it in flies—specifically, the fruit fly Drosophila melanogaster, known among scientists for their usefulness in biological experiments and among the rest of us for their tendency to flock around compost bins.
Studying flies to understand a human visual illusion might seem like a strange approach, but to Clark, it makes a great deal of sense. “There are a lot of really deep similarities between fly and vertebrate visual systems, including human visual systems, in terms of how they detect motion,” he says. Humans and fruit flies are similar enough, apparently, that they perceive the same illusions: This week, Clark and his colleagues published a paper in Proceedings of the National Academy of Sciences demonstrating that these flies, too, are susceptible to the peripheral drift illusion.
To measure whether the flies saw the black-and-white ring as stationary or in motion, the researchers positioned them on top of styrofoam balls that sat on a cushion of air. As each fly moved, the ball would move as well, so the fly could run in any direction without actually moving in space. “It’s like a 2D treadmill,” says Ryosuke Tanaka, a graduate student in Clark’s laboratory and one of the paper’s lead authors. By then surrounding the fly-on-a-ball with a 270-degree screen, the researchers created a “fly VR” environment in which they could show the fly any visual stimulus and watch how it responded. If they showed the fly a scene rotating in a particular direction, the fly would move in that same direction to try to compensate. To test if the flies could see the peripheral drift illusion, then, they surrounded them with the illusion and observed whether they behaved as if they were trying to compensate for motion. And they did: In response to an illusion that a human would perceive as rotating clockwise, the flies also tried to execute a clockwise turn atop their styrofoam balls.
Initially, Margarida Agrochao, a postdoc in Clark’s lab and the paper’s other lead author, had been skeptical that they would find anything. “You run those experiments, and then you go to the computer and do some analysis, and maybe your hypothesis is that the flies won’t see anything. Maybe the behavior will be so subtle that we can’t really distinguish it from the noise in the experiments,” she says. But the experiment showed a clear result. “Seeing those data that actually indicate the fly consistently rotates the ball in that particular direction, and that if we reverse the stimulus, then the fly will now turn in the other direction,” she continues, “it was great. It was a start.”
“Just that in itself is a really stunning finding,” says Karin Nordström, a professor of neuroscience at Flinders University who was not involved in the study. “If you Google different visual illusions, it’s one of the ones that most people come upon. And I was really impressed that they were able to show that same illusion in flies.”
But Clark didn’t stop at simply demonstrating that flies experience this illusion: He and his colleagues broke down how it works not only in flies, but also, perhaps, in humans. Few scientists have been able to definitively demonstrate how particular visual illusions work. While theories exist for every common illusion, they are extremely difficult to prove. “There are very few cases in which we know a causal mechanism for any kind of illusion,” Clark says. That’s partly because the human brain is incredibly difficult to study, and impossible to study on the single-neuron level.
Tanaka knows this challenge well: He used to study humans before moving to flies. When neuroscientists try to figure out how a process works in the human brain, they use a technique called functional magnetic resonance imaging (fMRI), which tracks how much blood flows to different brain regions. Since blood carries the resources that the brain needs to operate, a region receiving more blood flow is typically more active. Tanaka compares this technique to how someone might naively go about trying to understand how a computer can play videos, by trying to determine which components are working particularly hard. “What MRI does is, basically, you just stick your hand into the computer, and touch lots of different parts,” he says. “And let’s say, ‘Oh, this particular box inside the computer is getting hot. So it might be relevant to showing video.’” But just like touch isn’t a particularly precise way to figure out how a computer works, fMRI is a limited technique for understanding the brain—it can only show regional information, not pinpoint which specific neurons are active.
Tanaka eventually grew frustrated with the technological obstacles involved in human research. “It felt a little bit like going in circles, studying humans,” he says. For him, researching flies is a way to study interesting visual phenomena at a much more detailed level. “Flies are a nice compromise between behavioral complexity or, I would say, interestingness and tractability,” he says. “We can really get down to the resolution that really matters for understanding computations.”
So once the team had established how flies behave when faced with the rotating ring illusion, the researchers could move from studying the “what” to the “how.” And in flies, unlike in humans, they had a whole arsenal of tools at their disposal. “In flies, we can do this magical thing, where we can turn off neurons,” says Michael Reiser, a senior group leader at the Howard Hughes Medical Institute’s Janelia Research Campus, who was not involved in the study. By using genetic manipulations that have been designed to work in fly brains, it is possible to selectively deactivate a very specific subset of neurons. If turning off a set of neurons changes a certain behavior, it’s a pretty safe bet that those neurons play a role in it.
Specifically, Clark and his team focused on two populations of neurons in the fly brain called T4 and T5 neurons. Both are sensitive to moving edges, but T4 neurons are sensitive to the lighter side of the edge, and T5 neurons are sensitive to the darker side. Importantly, even when an edge is static, the T4 and T5 populations are both active— but they mostly cancel each other out.
When the researchers presented the illusion to flies whose T5 neurons had been genetically turned off, the flies appeared to perceive motion even more strongly in the same direction as before—they moved even more enthusiastically in that direction atop the styrofoam ball. But when the team experimented on flies with inactivated T4 neurons, the flies moved in the opposite direction. This led the researchers to the conclusion that, somehow, the T4 neurons were dominating the T5 neurons.
Why would this be the case, if T4 and T5 neurons generally have a net-zero effect in response to static images? Clark and his team have a theory: In most natural scenes, edges are equally likely to face in either direction. Light areas lie to the right of dark areas just as often as dark areas lie to the right of light areas. So even if the T4 neurons dominate just a bit for each of these edges, some of those T4 neurons are suggesting that there’s motion to the left, and just as many are claiming that the motion is really to the right. The overall result is that the image appears static. But in the peripheral drift illusion, all the sharp edges face in the same direction around the ring. The color will transition gradually from dark to light, and then suddenly back to dark, over and over again. And those sharp light-to-dark edges mean that the T4 neurons are all indicating motion in the same direction—so they all work together to inform the fly that it’s moving.
But what’s going on in people? While the human brain doesn’t contain T4 or T5 neurons, it does have neurons that play essentially the same role. Associating those neurons with the peripheral drift illusion is difficult—it’s impossible to turn off sets of neurons in people, since any genetic modification of that sort would have to happen before birth. With a bit of creativity, however, it’s possible to get a similar silencing effect reversibly and non-invasively: If people see a particular stimulus for long enough, their brains will stop responding to it. That process is called adaptation. It's “a little bit like a poor man’s silencing experiment,” Clark says.
So Tanaka, with his background in human neuroscience, coded up an experiment in which people would see the light or dark sides of edges advancing (“silencing” T4- and T5-like neurons, respectively) and then report the motion that they saw in a subsequent viewing of the peripheral drift illusion.
He didn’t have to wait long to see whether the experiment would work. “What’s beautiful about human psychophysics experiments is that you can just test on yourself and you get the result in an hour,” he says. “I just quickly wrote that over a weekend and tried it in an isolated room no one was using in the lab.” Tanaka “silenced” his own T4-like neurons by watching light shapes advance around a screen, then looking at the peripheral drift illusion on the same screen. He perceived that the ring seemed to be moving in the opposite direction as it normally appeared to move. “I basically saw my brain doing the same thing as flies. So that was really, like, the most exciting moment in this project,” he says. When Tanaka tested the experiment on 11 other people, he got the same results.
Tanaka and his coauthors stop short of asserting that this experiment proves that the illusion works exactly the same way in humans and in flies. But they still think their results are very suggestive. “We can say that a mechanism similar to what we see in fruit flies … based on our experiments, could be happening in humans as well,” Agrochao says. “Can we prove exactly that it’s happening? No. But the experiments indicate that a mechanism like that could also work in humans.”
Nordström was impressed, though not necessarily surprised, by the connection that the researchers were able to make between fly and human vision. “It’s very, very similar, how flies visually encode the world and how humans do that, which is really striking because our eyes are so different,” she says. “We have camera-type eyes. They have these crazy big compound eyes with thousands of lenses, and the photoreceptors are different. But as soon as you start getting into the brain, it’s very, very similar between flies and humans.” Clark and his team, she says, were able to take advantage of these similarities in order to put forward a promising hypothesis—and some evocative evidence—about how the peripheral drift illusion works in humans.
The story could end there. But there’s a deeper mystery at work in this study—why do human and fly vision work similarly at all? “We don’t believe there was a common ancestor that had something like an eye—like a proper eye—that could do motion detection,” Reiser says. For some reason, then, flies and humans separately evolved somewhat similar systems for perceiving motion. Why? Maybe it’s because flies and humans evolved to see the same natural surroundings.
“The fact that we evolved to deal with the same sort of visual environment made us develop similar-ish computations in our brains,” Tanaka says. For some yet-to-be-determined reason, slight differences between the way that light-sensitive and dark-sensitive motion-detecting neurons work helped evolving animals survive. And while these asymmetries are normally impossible to detect, strange, unnatural stimuli like optical illusions can draw them out.
There is more than one way to solve every biological problem—flies and humans could very well perceive peripheral drift illusions quite differently, or perceive them similarly for very different reasons. And yet, Clark and his team may have taken a major step toward showing that, at least in this case, flies and humans are very much the same.