Sebastian Seung, a professor of computational neuroscience at MIT, developed EyeWire, an addicting computer game with an ambitious scientific agenda. The objective of EyeWire is to build a connectome—a generalized visual map of connections between neurons that govern vision, memory, and disease in the brain. The completion of such a connectome will establish a normative model of these connections. From this normative model, theoretically a neuroscientist will be able to compare a connectome of a normally functioning individual and an individual with a mental disorder, such as Alzheimer’s disease, thus offering insight into the role neural structure plays in mental abnormalities.
The connectome’s possibilities don’t end there. Professor Seung has proposed the idea that every individual’s personal identity “is encoded in the pattern of connection between your neurons. If this hypothesis is true, then any kind of personal change is ultimately about changing your connectome”.
Of course, a project of personal brain mapping is presumably in the very distant future. The technology is still developing, and the only completed connectome of any creature so far is that of the C. elegans, a tiny worm that has a nervous sysem comprised of only 302 nerve cells . With the additional data from EyeWire, scientists are constructing connectomes of more cerebrally complex creatures, such as lab mice. The particular connectome that EyeWire is working on constructing is composed of neuronal connections within the retina of a lab mouse.
Due to the fact that there are about 100 billion neurons and at least 100 trillion connections in the brain, it would be impossible for a single neuroscientist to map all of the neurons alone: Seung has calculated that it would take one person working on this project around the clock 100,000 years to complete the connectome. Even with the assistance of advanced computer programs, this endeavor would take about 1,000 years [3}.
Due to its massive time requirement, Seung is attempting to distribute the work among as many people as possible: a tactic called crowdsourcing. By transforming his data into an engaging computer game, Seung hopes to reach a large number of people to make the realizability of the connectome more temporally feasible. EyeWire’s goals of creating a connectome rely on the generation and analysis of massive amounts of data, so the more people playing the game and generating new data, the closer Seung and his team at EyeWire can come to analyzing a connectome.
Creating a connectome requires visually tracing the connections of neurons in certain areas of the brain at a microscopic level. Artificial intelligence that is a part of the EyeWire program recognizes an incomplete neuron. By “filling in the blanks” for the artificial intelligence, you can construct three-dimensional, anatomically correct neurons. It’s exciting to see how your actions of identifying missing parts of the neurons visually manifest themselves, and the process becomes surprisingly addicting.
The website provides all new members with a comprehensive tutorial, teaching you how to interact with the slides of retinal tissue and construct neurons. No prior knowledge of or experience with neuroscience is necessary to have fun playing EyeWire, or even do well at it. You earn points by filling in all of the parts of the neuron that the artificial intelligence overlooked. The more accurately you complete the neuronal structure, the more points you get.
Along with participation from the general internet-using public, the artificial intelligence (AI, for short) that is involved in EyeWire helps acquire a lot of the images needed to create the connectome. The team of researchers behind EyeWire teaches computers to learn by example. “Instead of specifying the details of how the computer does something, you give it an example of what you want [it to do]” . Remarkably, EyeWire is the first project in which AI has been effectively taught to segment any kind of image.
EyeWire isn’t the first crowd-sourced research project; in fact, it has successful predecessors in scientific-research-turned-video-games, such as protein-folding research game “FoldIt.” This computer game assisted in figuring out the three-dimensional structure of an enzyme involved in replication of the HIV virus .