Two recent papers appeared in FOOP disagree regarding the role played by decoherence in quantum physics. On the one hand, Elise Crull [1] considers that decoherence, by itself, solves many conceptual problems in quantum physics, with no need of interpretative considerations. On the other hand, Antonio Vassallo and Michael Esfeld [2] reply by correctly claiming that, although decoherence is a powerful tool to deal with conceptual problems, it does not dispense us from interpreting the formalism. In this brief note we want to contribute to the debate with further considerations from another viewpoint.Vassallo and Esfeld stress the fact that, since in a standard Schrödinger's cat measurement the state of the whole system remains in superposition, the assumption that decoherence solves the measurement problem requires the assumption, among others, that the cat is a quantum system completely described by its own individual state. In fact, this assumption is not irrelevant at all, but brings to the fore the issue of how open systems are conceived. In particular, it requires to reconsider the status of the reduced states that describe their behavior.In classical statistical mechanics, the standard answer to the irreversibility problem in the Gibbsian framework relies on coarse-graining: whereas the statistical state of the system, represented by a density function, evolves obeying the Liouville theorem, the evolution of coarsegrained states is not constrained by the theorem and, under definite conditions of instability, may approach a definite limit for t . Of course, there are deep disagreements about the interpretation of the so-obtained irreversibility. But, independently of such disagreements, nobody ignores the difference between the statistical state, which evolves according to the dynamical postulate of the theory, and the coarse-grained state, which may tend to a final stable state. The situation in quantum mechanics is quite different: the distinction between the different kinds of states appearing in the quantum discourse is usually not sufficiently emphasized. For instance, sometimes it is said that the dynamical postulate of quantum mechanics only applies to closed systems, whereas reduced operators actually represent quantum states of open systems that evolve according to a different, non-unitary dynamical law. Although it is admitted that reduced states may cancel quantum correlations and, as a consequence, cannot be used for computations in certain cases, the states of closed and open systems are usually treated on equal footing.