A lot has been written about the relative merits of decoherence versus the many-worlds theory in answering the question: where does the extra information go when a superposed state "collapses" to a definite state. However, what about the reverse: when the definite state returns to a new superposed state, where does the new information come from? For the decoherence advocates, I can imagine an answer that it comes from the environment, but for the many-worlds theory, I cannot imagine an answer. I suspect I am on the wrong track altogether, and would like to be put straight. Thanks.