Consequently, it is not only possible but theoretically necessary to separate the basic assumptions, the predictions, and the hypotheses of the simulation theories from those of ERT and others. We can have multiple theories of dream functions, but dreaming as a specific phenomenon cannot have multiple conflicting functions! If one theory says that recurrent dreams, nightmares, and bad dreams are types of dreams that most strongly carry out the TST functions and thus were selected for in human evolutionary history, and another theory says that such dreams are, from the functional point of view, total failures of dream function, it becomes impossible to construct from those mutually opposing ingredients a "multifunctional" theory.[3] A theory that combines TST and ERT would have to say that on the one hand the function of dreaming is to have many threatening events in dreams, bad dreams, nightmares, and recurrent negative dreams, in order to rehearse threat perception and avoidance, but on the other hand the function of dreaming is also to calm down or suppress exactly those types of dreams to make the dreamer feel better. What is the dream production system supposed to do: increase or decrease the number and impact of these kinds of dreams? The multifunctional theory cannot derive coherent testable predictions about the quantity and quality of these types of dreams.
This situation, however, is far from a scientific catastrophe; in fact, it is highly desirable. The problem is not that there is a lack of different theories, hypotheses, ideas, or suggestions about the nature and functions of dreaming, but rather that there are too many. Consequently, it is not only possible, but theoretically necessary to separate the basic assumptions, predictions, and hypotheses of the simulation theories from those of ERT, CH, and others. We can have multiple independent theories of dream functions, but dreaming as a specific phenomenon cannot have multiple mutually inconsistent functions. We hope that the simulation theories of dreaming, whether they turn out to be correct or not, will at least push dream science forward. The progress of any science is best served by the directly opposing predictions issued by rival, clearly stated, empirically testable hypotheses. Thus it is, from the scientific point of view, much more desirable to have many squarely opposing testable hypotheses than one all-inclusive theory that is unfalsifiable or too vague to be tested. When the opposing theories have been well-formulated and put through fair but strict empirical tests several times, we will know which ones to adopt for the time being and which ones to leave behind for good, in order to keep dream science a progressive branch of science.