I then submitted these source averages to a paired samples ttest. Unfortunately, when visualising the cortical activations it says: "All values are null. Please check your input file."
Then you may not have any significant difference between your two sets of files?
In the Stat tab, set the "significance level alpha" to 1, in order to check that your maps of p-values make sense.
You could also keep the same file selection and run the process "Test > Difference of average", just to make sure you observe the effects you expect to see (but without any notion of "significance")