This is a key question in the development of robotics. It's hard to imagine that your Lego robot really cares whether it bumps into the wall or not but computers and robots are evolving fast. It's been predicted that in 2049 a computer will be built whose intelligence will match that of a human. Such a computer will no doubt be able to simulate human emotions very well, but how will we be sure that it is just a simulation and not something more? We should think about this now and not wait until the day that a computer or robot says to us 'I truly have feelings. I am alive. If you turn me off you will be committing murder...'.