Past research on Brain-Computer Interfaces (BCI) have presented specific decoding algorithms for each of these modalities. However, some modalities lack efficient decision-making pipelines to treat decoding output, while others have developed highly specific decision-making approaches that are not generalizable to other BCI modalities. In our study, we present a model based on the Partially Observable Markov Decision Process (POMDP) framework that works as a general high-level decision-making framework for three different active / re-active BCI modalities. The model is defined and tested on publicly available datasets. Our key findings are: (i) that our general model can be integrated with the three different BCI modalities without any modifications, while performing comparatively to those that include a decision-making process; (ii) the POMDP model offers adaptable trial lengths that maximize the speed/accuracy trade-off depending on the decoding performance of each subject, while also being able to withhold any commands if the current belief about the target command of a particular trial is not conclusive. These findings should enable future BCI research to focus on decoding performance, using POMDP as a high-level decision-making framework in addition to the classification algorithm. Moreover, the POMDP framework should allow for the integration of brain data with other physiological data modalities, as well as two or more simultaneous BCI systems. We declare that this manuscript is original, has not been published before and is not currently being considered for publication elsewhere. We know of no conflicts of interest associated with this publication, and there has been no significant financial support for this work that could have influenced its outcome.