From what I can understand, it describes the phenomenon of when neurons detect the same features. Why does this happen?
Since the weights are not initialized properly and groups of neurons end up in the same local minima, according to their (similar) initialization.
To overcome this, you could use dropout / drop connect to break symmetry.
Hinton, G. E., Srivastava, N., Krizhevsky, A., Sutskever, I., & Salakhutdinov, R. R. (2012). Improving neural networks by preventing co-adaptation of feature detectors.