You have just so much information stored in those smaller images.
When you try to enlarge an image, you're basically pulling the pixels apart, and the program has to "guess" what is to occupy the additional space between the pixels. Most programs aren't very good at "guessing". I imagine, there are some high-end (expensive) image editing programs that are better at "guessing", than other programs, but I can't say for sure. Even then, I'd imagine the results would be mediocre, at best.