Histories without collapse

This paper is a comparison of two theories of the probability of a history in quantum mechanics. One is derived from Copenhagen quantum mechanics using the projection postulate and is the basis of the "consistent histories" interpretation; the other is based on a proposal by Bell, originally for the "pilot state" theory but here applied to pure unitary quantum mechanics. The first can be used for a wider class of histories but depends on the projection postulate, or "collapse", which is widely held to be an unsatisfactory feature of the theory; the second can be used in a theory of the universal state vector without collapse. We examine a simple model based on Wigner's friend, in which Bell's model and the projection postulate give different probabilities for the histories of a sentient system. We also examine the Frauchiger-Renner extension of this model, in which comparison of the two calculations of histories throws light on the contradiction found by Frauchiger and Renner. By extending the model to equip the observer with a memory, we reduce the probability of histories to the use of the Born rule at a single time, and show that the Born rule, with the memory, gives the same result as applying projection in the course of the history, because of entanglement with the memory: entanglement implements collapse. We discuss the implications of this for the use of histories in quantum cosmology.