The options are not the same.

If you have a small cube sitting top of a large cube the bottom of the small cube is imprinted on the top face of the large cube. The large cube now has a top with 2 faces, while the small cube has a bottom with 1 face. There are now two faces at the location of the bottom of the small cube. One belongs to the small cube and one belongs to the large cube. They are not shared and therefore could be meshed with different element sizes. Frictional contact could be defined between the two cubes and they could be separated by applied loads.

Shared deletes one of the two small faces and forces the small cube and the large cube to share one small face. Now there can only be one set of nodes on that face. The elements in the small cube are connected to the elements in the large cube. The cubes cannot be separated by applied loads.