Maybe it’s just that I’m five weeks into a course on Functional Human Neuroanatomy, but Friday’s Wired article on Eyewire definitely caught my… well, eye. In short, Eyewire is a browser-based game where you trace neurons from a mouse’s retina through a series of cubes. Using pictures acquired by electron microscopy from the Max Planck Institute, the Eyewire AI selects a neuron within a set of slices, and your job is to help fill in the parts that the AI has left off. The slices are put together like a flipbook, and as you move up and down through each level, you can follow the parts of the neuron that are completed and fill in the the holes. Each cube is made of 256 slices, and each slice measures about 4.1 x 5.6 micrometers. For a sense of scale, a typical human hair ranges from 50 to 150 micrometers in diameter.

Diagram of one complete J cell.

There’s a tutorial to help train players to recognize structures within each slide, since most of us don’t work with electron microscopes on a daily basis. It does a good job of quickly orienting you to the task and slowly weaning the feedback you get as you practice, so you don’t need to be a neuroscientist to join the cause. Each cube you help complete will earn you points based on accuracy as compared on a consensus with fellow users and speed of completion. Since the game compares answers across a large number of people, one or two people’s errors will not hurt the end-goal of their research. There are daily, weekly and monthly leaderboards to keep track of progress, further reinforcing the game-like atmosphere.

But honestly, can tracing neurons really be all that addictive? Let me put it this way: I followed the link over to the Eyewire site to take a look before writing up the story, and then two or three hours later, I had run through the tutorial and a good number of cubes, but this article was still as blank as when I started. It seems perfectly geared for those who love puzzle games and mazes, and the few times I have tried it out so far, there has been plenty of friendly conversation going on in the chat sidebar, too.

Eyewire isn’t the first research project to be turned into a game, though it may be one of the best at capturing gamers’ attention. Projects like FoldIt  from the University of Washington or Neil Schmitzer-Torbert’s Quasar Quinn and Colonize  also tap into games for the purpose of research; however, these don’t seem to find that same relaxing abnegation that more traditional games offer. It also doesn’t hurt that the controls are fairly straightforward, so once you get rolling, playing can become almost zen-like.

A screenshot of the interface. On the left is a 3D view of the area you’re working in. On the right is the slice you are currently looking at.

Since its launch in December 2012, Eyewire has had over 70,000 users complete 1.5 million cubes and counting. Currently, it takes about 50 hours for Seung’s lab to reconstruct a single neuron, so with over 80 billion neurons to map in the human brain, the group will need a few more hands to get the job done sometime before the sun burns itself out. (Okay, okay. The project will only take half of a billion years, give or take a few millenia.) Right now, players are working on “Mystery Cell #8 – A mystery ganglion cell whose type should be found as tracing progresses.” So, not only are gamers helping trace out known neural pathways, their contributions will help us learn just what the cells do, as well. The crew puts one or two new cells up to explore each week, so there is always work to be done.

Eyewire is the brainchild of Sebastian Seung and colleagues at MIT, and while the current edition is working on mouse neurons, if all goes well, the same technology could be applied to human neurons in the near future. Their project is funded by the National Institutes of Health as part of the Brain Research through Advancing Innovative Neurotechnologies (BRAIN) Initiative. You can get more information about the game and the project at the Eyewire wiki.

Share this post: