HI Mark,
Some attempts at answers to these questions about the philosophy of science.:) Maybe others will share thoughts also.
1. There is a signicant evidence base for most of these theories, particularly the ones in the psychological category because they are more traditionally scientific in approach. If you google them you will find many many relevant research articles. The references I provide offer some as well. For Behaviorism, the research goes back to the 1910's, I believe, and the theory is founded on a desire to have an incontrovertable evidence base. But the fact that many of its findings are not valued as highly today shows that evidence is not the only thing that is important. A theory has to grab the zeitgeist as well to be used.
If you narrow your questions sufficiently, you will get solid evidence. If you keep them broad, it is harder to claim certainty, but you still get guidance, and often more useful guidance.
For Cognitive Load Theory, I mentioned Richard Mayer (just Google him), but I question how valid his research is about multimedia design since he uses overly simplified controls. Go to John Sweller instead, who I think coined the term in the 80's. But the 7 + or - 2 research goes back to the 50's. For Dual Coding Theory, Mayer talks about it also, but the solid research goes back to Allan Paivo. (I had one of his books but it was unreadable.)
2. Evidence, it depends on what you are willing to accept as evidence. Behaviorists would not trust a person's report about their thoughts. But evidence is always an approximation. If you use a theory, and you get good results, does that count as evidence? That is often the kind that is achievable for the more complex learning theories. That kind of research is quite respectable, by the way. It is classified as "design research."
Theories are not always attempts to describe exactly how the world works, but instead can be about how we can interact with it and achieve good results. John Dewey, one of the humanistic philosphers of science, prefered to avoid the word "facts" in preference of "warranted assertions". His Pragmatist philosophy valued ideas that worked over ideas that merely had logical validity or physical evidence, and it assumed that those ideas that worked would change over time because research is as much about what questions we are willing to ask as the answers we get.
Theories are often based on analytical and intuitive thought, gather face validity, and then become the source for research questions that begin to develop more solid support. In fact, that is the more common approach, rather than "grounded theory" developed from the evidence up. The theories in this session fall along that spectrum.
I agree we want to avoid passing fads. We have to know that a theory has a lot of support, either philosophically or pragmatically, before deciding to use it. By using it, we extend that support if it works for us. We want to avoid fads, but we also want to avoid pigeon holes that prevent us from accepting new ways of looking at the domain that could have bigger payoffs.
Pat