Quick History of Geography
Ancient times to late 1800s: Exploration and cataloguing. Western geography was about finding new places and documenting what was there -- the coast is shaped like this, the people look like this, such-and-such plants grow there, etc.
Late 1800s to 1920s: Environmental determinism. European explorers had explored and catalogued most of the world, so geographers figured it might be time to try explaining what was going on. The theory they picked -- which dates back at least to the ancient Greeks -- was that human societies are determined by their biophysical environment. In particular, they decided that climate determines personality (e.g. people from hot climates are lazy and have high libidos). This made geography a SCIENCE because it was based on evolution, which was the paradigm of science at the time.
1930s to 1950s: Regional geography. So it turns out environmental determinism is entirely based on racism, and has no empirical support whatsoever. Oops. Geographers decided to give up on that whole embarassing "theory" thing, and go back to cataloguing. This time, they focused on cataloguing everything that's happening at one very narrowly defined place. And they used that information to try to determine where the "real," objective boundaries of "regions" are.
1950s to 1960s: Quantitative revolution. Regional geography is kind of dull, and nobody cares about it besides other geographers. Geography needs to be a SCIENCE. And since the paradigm science is now physics, that means lots and lots of math. Geographers decided they could be objective and scientific by just analyzing the spatial patterns of things.
1960s to present: Social theory. The quantitative revolution produced some useful analytical tools (including the forerunners to GIS). But if all you do is analyze patterns, you should go join the math department, because you're doing geometry, not geography. Geographers realized they needed some theories to explain how and why things were interacting in space. Luckily, there were a variety of theories available in other disciplines that could be geographized -- systems theory, neoclassical economics, Marxism, phenomenology, social psychology, postmodernism, etc.
The near future, hopefully: Geographical theory. I'd like to hope that geographers can move from borrowing other disciplines' theories to creating theories of their own (that other disciplines can also use).