What do they understand ?

• They understand a decomposition of their visual input space as a hierarchical-modular network of convolution filters. source
• They understand a probabilitistic mapping between certain combinations of these filters and a set of arbitrary labels. source