At a first glance, we have a lot of negative evidence about the existence of macroscopic superpositions: after all, everybody has experienced several living cats and perhaps also a few dead ones, but nobody has ever encountered a half-alive/half-dead cat. This remark is not false but, as it stands, is ill formulated. For quantum states are not properties attached to an object. From a technical point of view, they are just the formal tool which allows one to draw predictions on results obtained under well specified experimental conditions. That zombies are not observed is a well-established experimental fact. But it does not imply that the superposition principle doesn't hold at the macroscopic scale. On the contrary, some recent experiments have shown that small 'classical' systems (like a few individual atoms or a classical electromagnetic field equivalent to a spark) can be forced to behave in a non classical way. This requires a very good isolation, i.e. no entanglement between the observed systems and the environment. Under such circumstances, the whole formal machinery of quantum theory, including ‘exotic' superpositions of states, must be employed to account for the observed results.

To give an example, let's consider the case of an isolated ion (an ion is an atom missing one electron). Exploiting their positive net electric charge, ions can be individually trapped by means of electromagnetic fields, and their position inside the trap can be visualised (see picture). Therefore, in this context, the ion behaves like a ‘classical' (if not literally macroscopic) body occupying a well-defined position in space. From an operational point of view, the sentence 'the ion is located in region R_i' means that the probability to find the ion within R_i is 1 whereas the probability to find it elsewhere is 0. Let's call the state vector corresponding to such statistical distribution. We are now in a position to understand the claim that a superposition of two localised quantum states, each corresponding to a distinct portion of space, has been prepared. Such preparation amounts to manipulate the ion in such a way that the statistical results of subsequent measurements are correctly predicted by a state vector having the form (c₁ and c₂ being two numbers). The observation of interference effects in measurement involving observables incompatible with position testifies that such a manipulation has been successfully carried out.

Creating ‘superpositions' of macroscopically distinct configurations and testing the empirical predictions of the corresponding state vectors (in particular interference effects) is a challenging task. Experiments analogous to those carried out with ions have been performed using classical fields trapped in a superconducting cavity. These fields are manipulated by means of individual atoms. In this case the superposed state vectors are the ones corresponding to two field configurations that are classically distinct, for example a bright state (field's intensity of a few photons) and a dark one (field in the vacuum state = darkness).