By Cathleen O'Grady

Spiders have multiple eyes scattered around their heads, whereas vertebrate predators have just two eyes, facing forward. Mantis shrimp can see more colors than humans, who in turn see more colors than our dogs. Why did so many different types of eye evolve? A novel, gamelike computer simulation in which eyes evolve over generations suggests different eye types may have emerged to solve specific tasks, and that better eyes needed to evolve along with better brains.

The new approach to studying eye evolution, described today in Science Advances, is “supercool,” says Lauren Sumner-Rooney, an evolutionary biologist at the Leibniz Institute for Evolution and Biodiversity Science who was not involved with the work. Older models of the evolution of vision simulated the sense in isolation from other abilities, but the new simulation looks more broadly at how creatures learn about their world and the role vision plays in that process, which “much more closely approximates how biology really works,” she says.

Researchers studying evolution generally need to start at the end: They can look around at different species and try to understand what caused them to develop their specific traits. But Massachusetts Institute of Technology Ph.D. student Kushagra Tiwary and his colleagues wanted to simulate the starting line, creating a visual environment and observing what kinds of vision emerged as eyes adapted over multiple generations.

The team built that environment using software that mimics the physics of the real world, much like a video game, Tiwary says. Then they populated the world with hundreds of digital “agents”—simple characters with one eye that could see one pixel, and a computational system known as a neural network to represent a basic brain that could learn from its environment. The researchers then adjusted these features slightly so that each agent started with different biological makeup, to mimic genetic diversity in the real world.