For a mechanistic understanding of brain function, it is important to understand the relation between patterns of activity and connectivity in neural networks. My lab is studying this relation in the retina by classifying its neurons into cell types, and mapping the connections between types. I will describe preliminary results concerning the connections of the J type of ganglion cell, and what they suggest about the mechanism of its direction selectivity. To enable our neuroscience research, we have used machine learning and social computing to build systems that analyze light and electron microscopic images through a combination of artificial and human intelligence. The most exciting recent example is EyeWire, an online community that mobilizes the public to map the retinal connectome by playing a coloring game. I will conclude by describing our beginning efforts to search for the cell assembly, a pattern of connectivity hypothesized by Hebb in 1949 as a structural basis of long-term memory.
