In the days before a newborn mouse opens its peepers, nerve impulses that have been sweeping randomly across the retina since birth start flowing consistently in one direction, according to a paper published in Science today (July 22). This specific pattern has a critical purpose, the authors say, helping to establish the brain circuitry to be used later in motion detection.

“I love this paper. It blew my mind,” says David Berson, who studies the visual system at Brown University and was not involved in the research. “What it implies is that evolution has built a visual system that can simulate the patterns of activity that it will see later when it’s fully mature and the eyes are open, and that [the simulated pattern] in turn shapes the development of the nervous system in a way that makes it better adapted to seeing those patterns. . . . That’s staggering.”

The thread of this concept may be looped, but to unravel it, Berson says, it helps to think of the mammalian visual system, or really any neuronal circuitry, as being formed by a combination of evolution and life experiences—in short, nature and nurture. We might expect that life’s visual experiences, the nurture part, would begin when the eyes open. But, much like a human baby in the womb practices breathing and sucking without ever having experienced air or breastfeeding, the eyes of newborn mice appear to practice seeing before they can actually see. Motion detection is important enough to mouse survival that evolution has selected for gene variants that set up this prevision training, says Berson.

Researchers had been aware of retinal waves in the unopened eyes of newborn mice for decades and had suggested that the waves, which form and flow every minute or two, play a part in developing the visual system’s circuitry. In the new study, neuroscientist Michael Crair of the Yale School of Medicine and colleagues found that when mice are 8 to 11 days old—shortly before their eyes open at around day 13—the waves adopt a  “remarkable directionality,” Crair says. Naturally, “we were curious as to whether this directionality played any role in the functional development of the brain,” he says. “Is the directionality happenstance, or does it actually do something, mean something?”

The waves of neural activity, which the team observed via fluorescent calcium imaging of the newborn pups’ retinas, traveled from the animals’ temples toward their noses. This “really piqued our interest,” says Crair, because it “closely mimics the optical flow—the directionality you would see, or an animal would see, if it was moving forward through space.” Because the eye inverts images landing on the retina, forward motion creates the reverse flow.

“It’s like the animal is dreaming about running forward” before it has ever done so, says Crair.

Next, the team teased apart the circuitry underlying the waves’ direction. By inhibiting a variety of neurotransmitters in the eyes of the newborns and seeing if any disrupted the wave direction, they zeroed in on gamma aminobutyric acid (GABA), which in turn suggested the cells responsible—a type of interneuron called a starburst amacrine cell that releases GABA. Sure enough, Crair explains, ablating the function of starburst cells also led to the loss of retinal wave directionality in the newborn mice.

The team went on to show that such loss of retinal wave directionality in the newborns—whether from GABA signal inhibition or amacrine cell dysfunction—caused impaired motion detection once the animals had opened their eyes.

Fluorescence microscopy video of typical directional retinal waves in an unopened newborn mouse eye, shown at 10x real speed

Two-week-old mice were shown images of black and white bands that moved in different directions while the researchers recorded neuronal activity in a part of the animals’ brains called the superior colliculus, a region with specialized cells that respond to certain directions of visual movement. The brain activity of mice with disrupted retinal waves—whether caused by temporary GABA inhibition or permanent amacrine cell dysfunction—looked different from that of control mice, says Crair. Fewer cells responded to directional stimuli, he says, and those that did had poorer direction selectivity. “Normally you have cells that respond to one direction and no other,” he explains, “but in these animals the cells would respond to many different directions.” 

The paper contains “a massive amount of work” that’s “definitely interesting,” says David Feldheim, who studies sensory information processing at the University of California, Santa Cruz, and did not participate in the study. “It both confirms and extends what we were thinking about how retinal waves work,” he says, “and then there’s some unanswered questions.”

It would be great to know, for example, the mechanism by which the wave direction shapes direction selectivity in the superior colliculus, he says, and to find out whether mice whose retinal waves were temporarily disrupted before eye opening (via GABA signal inhibition) can later recover from the impaired motion detection or are stuck with it.

Whatever the findings on questions like this, they will likely have implications for early neural training beyond the visual system, says Berson. “I’m sure this is happening on the motor side; I’m sure it’s happening in other sensory systems as well,” he says. “This is just the tip of the iceberg.”