It is well known that Wheeler proposed several delayed choice experiments in order to show the impossibility to speak of the way a quantum system behaves before being detected. In a double-slit experiment, when do photons decide to travel by one way or by two ways? Delayed choice experiments seem to indicate that, strangely, it is possible to change the decision of the photons until the very last moment before they are detected. This led Wheeler to his famous sentence: No elementary quantum phenomenon is a phenomenon until it is a registered phenomenon, brought to a close by an irreversible act of amplification. Nevertheless some authors wrote that backward in time effects were needed to explain these results. I will show that in delayed choice experiments involving only one particle, a simple explanation is possible without invoking any backward in time effect. Delayed choice experiments involving entangled particles such as the so called quantum eraser can also be explained without invoking any backward in time effect but I will argue that these experiments cannot be accounted for so simply because they rise the whole problem of knowing what a measurement and a collapse are. A previously presented interpretation, Convivial Solipsism, is a natural framework for giving a simple explanation of these delayed choice experiments with entangled particles. In this paper, I show how Convivial Solipsism helps clarifying the puzzling questions raised by the collapse of the wave function of entangled systems.