The eye progress


While I know we only have done a few completed cells, but I was wondering with the current speed of progress, how long will it take to complete the full retina?


Oh, interesting question. Altough I would rather like to know when you think you will have enough cells colored.
Otherwise my bet would be: 3500 years

I think the point of this work is to get an idea of reoccuring patterns within those different types of cells. Something like - hey, there is a junction in those amacrine cell connections every 0.02µm… and 98% of those are connected with some other cells of layer X of the mices retina… (Sounds great, right? Did you notice that I have no idea what I am talking about? :D)


I do somewhat, the um( sorry don’t have the correct character for the “u”) is the measurement for neurons and Layer X I’m guessing is one of the slices of the retina. Besides that, will the discovery of these re-occurrences help speed up the eventuality of finding the human Connectome?


A large part of the goal of this project is to improve an algorithm for segmenting the data automatically. When we trace in Eyewire, we are utilizing a part of that algorithm to break the data into chunks, and identifying which chunks belong to the neuron being worked on by coloring them. This provides a lot of data that can be used to improve both that part of the algorithm and other parts that are being worked on.


nkem is right that it would take a long long time to do the whole retina.  The goal for now is to do enough to write a paper about the JAM-B retinal ganglion cells and how their connections to Starburst Amacrine Cells leads to upward direction selectivity in mice.  We’re hoping to have enough cells done to write it in the next few months.


I see that the last response to this question was 2 years ago.  Does anyone have an update?


We reached our previous goal and are now working on cataloging ganglion cells - you can read about our next goal, Countdown to Neuropia. The countdown will trace every cell within a section of our dataset - and it’s going to take us over a year to complete! So yes, under the current system it would take us quite some time to do the whole retina. We are also preparing a new dataset in the olfactory system for tracing.

We’re still working on training the AI and improving the segmentation algorithm which will help speed up everything in general. These both are projects that go on in the background at the Princeton Seung lab.


Here is my request:

  • Change the name.
  • Change dataset.
  • Port over all the player’s data (like cubes, points, etc). 


Hi whitefieldcat,

The items on your list are already considerations for the new dataset, but it’s still being prepped in the lab.  As usual with science, things are taking longer than predicted.  You’ll know for sure when we release it!



Almost another year has gone by and yet another years worth of data has been gathered. While I know that what we have discovered is no small feat, especially considering the little time is took relatively, we are still using Mr. McMouse’s generous donation. While I know that genetically (a mammal) and his basic neural “mapping” are the same as a humans, has there been any plans to use a human’s retina, a Mr.McMan if i may? If so, then when might we see this “closer-to-home” contribution?


Hi deadxdying,

Certainly, in the future we hope to be able to use data gathered from humans. However, as I’m sure you know, this is a bit complicated both ethically and scientifically. In order to image data from the brain, we would need a recently deceased person to generously donate their brain to science. However, dye injections needed for the imaging process would need to be performed while brain functions are still happening before removing the brain from the patient, so gathering imaging data is a very time sensitive issue. I have heard that perhaps one way of obtaining a human brain sample is from seizure patients who need to have a certain part of their brain removed. Hopefully, scientific advancement may help us better gather this data in the future. For the moment, even our tiny dataset on Eyewire is keeping us rather busy.


Only as an example, let’s say someone who was blind in one eye (because of a malfunction or permanent damage to the brain or eye, leaving the retina 100% healthy), was of sound body/mind, and wished to donate the unused retina. If this case were to arise, would you be willing/able/allowed to accept this donation?


Hi deadxdying,

I’m afraid I’m not quite qualified to answer that question, but I’ll try my best to point you in a direction. (Eyewire’s dataset e2198 was obtained from the Max Planck Institute for Medical Research in Heidelberg, Germany). In such a case as you described above, I would suggest contacting your local research facility (hospital, university, etc.) to inquire about such donations. Due to the nature of such donations, it is common for a donor to specify to whom donations should be sent.

On a personal note, I have a family member who has an agreement that the closest research facility will receive their eyes for research purposes upon their death. I believe they arranged that thru their doctor/ophthalmologist.

Hope that helps answer your question.


Hi guys! It’s will be great if there are some picture of overall neurons that was traced up to date in this region. And updating with each new neurons. this will help everyone trace progress.


Hi Andrey,

There are a few ways you can check out our progress so far. Previously, we searched thru our dataset, e2198 and traced only specific types of neurons. Most of those cells were Starburst Amacrine Cells and Bipolar Cell types.

Eyewire’s current goal is the Dig, where we reconstruct all the neurons found in a specific zone of the dataset. You can read more about it here:

We’re currently working on finishing up Sector 3 in the Dig. You can see screenshots of all the cells we’ve traced in that zone in an album on our facebook page here:

Keep an eye on our facebook feed as we post a screenshot every time we finish a cell in Eyewire. We also make announcements on social media and our blog with updates on our current goals. You’ll probably see one soon for Sector 3 of the Dig.

If you’re looking for a more scientific tool, the Princeton lab and Eyewire HQ recently completed this site:

This is a tool for researchers to help classify the cells we have traced in e2198 (this includes Eyewire cells and cells traced in our lab). Some of those cells are already classified in the museum. You can view cell types using the museum viewer. Additional info/stats such as location in the location, layer, and size are available for each cell.

Note: The museum is periodically updated, so it might not have the most recent cells we’ve completed in Eyewire.

Hope that answers your questions.


PS. Due to how the science world works, some of the research we’ve completed isn’t released until we publish a scientific paper.


Almost 3 year gone by. Many changes, many players joining, and many cubes completed. I once again want to know how much of our total dataset have been completed. Also how much more has the mapping AI learned and improved?


like @sorek_m did say in the post over we are now tracing in the dig zone of the retina, we are most likely going to finish sector 13 of 16 sometime in march. and i think we can finish all of the sectors this year. From what i can see of the pictures the 16 sectors we are going to trace is about 40 % of the total sample we have. So if we will trace cells outside the sectors after we finish our current project or if we will move on to something else is still not determined so far as i know.

You also ask about how much the ai have learned? so far as i know the ai used in eyewire are not really learning from our playing and will be on the same lvl as when this game started. with some exceptions where the devs have made changes in the confindense lvl the ai are using to make seeds.

On the other hand the technology behind AI is in continuous develoment and the lab now have AI that can mostly trace cells on its own from other datasets and only needs human help to find and correct mistakes. The new game the lab is making NEO will use this AI. This will probably need better trained players to spot mistakes but, the progress will be considerable faster.


Maybe it’s time for a neural network to learn Eyewire @amy lol