Videogamers And Researchers Join Forces And Make Brand-New Discovery About The Retina


Robin Andrews

Science & Policy Writer


Videogamers helped to find 6 new types of retinal neurons - and that's just the beginning. Eyewire

There are 100 billion neurons in our body, and each neuron is part of a huge web forming the body’s nervous system. Despite having a cartographical outline of this bioelectrical infrastructure, much of the map remains undetailed or fuzzy, and researchers across the world are racing to fill in the gaps.

One such project – spearheaded by Princeton’s Seung Lab – has taken a rather novel approach to this grand endeavor. Using an online video game named Eyewire, they’ve used the game data of around 265,000 players to create beautiful, detailed maps of thousands of neurons. Of these, six of them are brand-new types, never before seen or at least documented by scientists.


Eyewire, as the name suggests, (currently) focuses on neurons in the eye, those that represent the tip of the spear when it comes to visual perception. It’s not all that clear what the intricacies of the structures of these retinal ganglion cells (GCs) are – those that link the retina and the brain – so the team, quite rightfully, decided to take an unconventional approach to mapping them out.

Using data from the retina of a mouse, the online game (launched in 2012) allows players to solve 3D puzzles that effectively lets them trace the outline of these GCs. The more puzzles you solve, the more accurately you solve them, and the faster you solve them, the higher ranking you have in the game – a competitive element that encourages people to play more frequently.

That’s somewhat surficial compared to what the game is trying to accomplish, however. By playing it, you’re doing something machine learning – which also plays a role – cannot do as well as we can: visually identify neuron branching patterns.

The combined effort of these so-called Eyewirers, then, has led to the mapping of over 3,000 neural cells, and 1,000 are displayed in a virtual museum for all to see.

A mapped synapse (structural connection allowing for an electrical or chemical signal to be transmitted) between a ganglion neuron (blue) and an amacrine cell (yellow). Alex Norton/Eyewire

Citizen science, the recruitment of members of the public to help crunch some data and make those vital connections, has worked wonders before – arguably most famously when it comes to hunting exoplanets – but this particular project has clearly demonstrated that videogamers searching for new neurons types can be just as effective as those looking to the stars.

The team's new paper describes almost 400 GCs mapped from a single microscopically-viewed patch of mouse retina, using the data of 29,276 Eyewirers. Carefully cross-checking these new 3D interactive views of each of them with pre-existing research, the team stumbled on far more revelations than they bargained for.

Not only are there far more types of GCs than anyone could have guessed – 35-50, up from 15-20 suspected – but “six of these types appear novel in the sense that we have been unable to find any matching reports in the literature,” they note in the journal Cell.

“At this point, we know those 6 new types have different anatomical properties from other types… [and] we know how the cells in those types responds to the visual stimuli we have experimented,” this paper’s lead author, Alex Bae, a graduate student in electrical engineering at Princeton, told IFLScience.


“Unfortunately, we are not able to figure out what kind of role they serve in the retina with only this research.”

Clearly, this neuronal atlas project is proving its worth, but that doesn’t just apply to its ability to create detailed renderings. It’s a huge open-access boon too: those outside the lab can not only view the morphology of these cells, but find out what their functions are. It’s a bit like a jigsaw puzzle, where spatial relationships between neurons are revealed as branches are progressively filled in.

The gallery is full of carefully mapped neuron types, including six new to science. Eyewire

Remember, this is just the beginning. Apart from discovering more about the neuronal nuances of the eye, Eyewire’s success suggests it’s suitable for mapping other parts of the nervous system too.

Bae told us that the AI originally designed to handle the task is vastly improved, but that doesn’t mean that crowdsourced input is no longer required.


“Machine reconstructed cells still contains errors so we could use a great help from the community of Eyewirers on this,” he explained.

Fancy becoming an Eyewirer too? Then click here to join the community.

  • tag
  • retina,

  • open access,

  • nervous system,

  • map,

  • discovery,

  • new,

  • eye,

  • ATLAS,

  • princeton,

  • ganglion cells,

  • Eyewire,

  • 6 types,

  • crowdsourcing science