Static typing helps IDEs show the programmer what each variable is. (Which is necessary, because unless you use Hungarian notation, it's not obvious.)
Without static typing, the user is just given a bunch of text files, and has to guess what each variable is by tracing backwards through the probable execution flow of the program to figure out where the variable is coming from.
In a Live programming environment, the programmer could be given a set of Probable Types for any variable. They could also drill into each variable and see the probable code paths that ended up passing the variable into the method.
The fundamental idea is to have the IDE automatically run the program many times with many different inputs, and allow the programmer to see what happened in different ways. Codepaths, types, runtime structures, etc. As the programmer modifies the code and test data, the statistical model will be differentially updated.
It seems likely that brain is doing this kind of online-learning and statistical-typing/model-reorganization all the time.