Proof of stake systems attempt to create red light - green light types of things with non-gameable attributes (eg collaborative random numbers). This can't be done with mining because mining is completely random - its not possible to know which miner will mine a block. If it were, it wouldn't be proof of work, but something else. What you describe sounds like proof of identity, which isn't possible in a decentralized adversarial environment. In fact, one of the primary achievements of the Proof of Work consensus mechanism is to work around the Sybil issue, where (like ZmnSCPxj mentioned) a single user can have many identities.
There can be hybrid systems that use both proof of work and proof of stake, but my conclusion after having done a lot of research and thinking about it (
[1],
[2]) is that the security mostly boils down to the weakest piece of the hybrid system, and so its not very effective to have hybrid systems like you mentioned.