The thesis is divided in three chapters. In the first chapter, we reveal that magnetic tunnel junctions fit in the recently emerged field of memristors and memristive systems. We describe how this 'hot topic' emerged from the combination of two research fields and take a step further by including magnetic systems. The somewhat illustrative data presented in this chapter has been chosen to demonstrate the most striking features of memristive magnetic tunnel junctions. Not only the smooth switching between two well defined resistive states is presented, but also - and this is the novelty introduced by magnetic electrodes - additional states due to spin-dependent tunneling. At the end of the first chapter, the limits of the applicability of the memristor theory are discussed.

In the second chapter, the resistive switching is identified as the physical mechanism responsible for the memristivity of magnetic tunnel junctions. First, the main models and theories proposed in the literature are summarized. The data presented thereafter has been chosen to give a complete phenomenological picture of the memristivity of magnetic tunnel junctions. Based on this experimental evidence, a model is developed particularly suited to describe the physical mechanisms responsible for the memristivity of magnetic tunnel junctions.

The third chapter deals with neural networks and synapses in particular. In the beginning, a brief introduction to biological neural networks is given. The experimental results needed for a comparison of memristive magnetic tunnel junctions with biological synapses are provided in the middle part of this chapter. Finally, the astonishing equivalency of both systems is revealed by a discussion of the measurements on memristive magnetic tunnel junctions and by comparison with recent results from the neuroscientific literature.