I would also add that you don't necessarily 'know' solutions, but rather make sense of the limited input you have and try to extrapolate in that. Obviously this fails hard at human interactions, which makes this the frustrating part. How to think of an alternative when the 'right' solution seems so beautifully simple?
