Memory is one of the most extraordinary phenomena in biology. The mammalian brain stores billions of bits of information but the most remarkable property of memory is perhaps not its capacity but the speed at which the correct information can be retrieved from a pool of thousands or millions of competing alternatives. Despite more than hundred years of systematic study of the phenomenon, scientists are still largely ignorant about the mechanisms that enable mammalian brains to outperform even the best search engines. One of the greatest challenges has been the dynamic nature of memory. Whereas memories can be retrieved over time periods as short as milliseconds, underlying coding principles are normally inferred from activity time-averaged across many minutes. In the present proposal, I shall introduce a new ¿teleportation procedure¿ developed in my lab to monitor the representation of past and present environments in large ensembles of rat hippocampal neurons at ethologically valid time scales. By monitoring the evolution of hippocampal ensemble representations at millisecond resolution during retrieval of a non-local experience, I shall ask(i) what is the minimum temporal unit of a hippocampal representation,(ii) how is one representational unit replaced by the next in a sequence,(iii) what external signals control switches between alternative representations,(iv) how are representations synchronized across anatomical space, and(v) when do adult-like retrieval mechanisms appear during ontogenesis of the nervous system and to what extent can their early absence be linked to infantile amnesia.The proposed research programme is expected to identify some of the key principles for dynamic representation and retrieval of episodic memory in the mammalian hippocampus.