Unfortunately, there is no way to have SuperLab consider one subset of areas for some events and then a different subset of areas for other events. One workaround would be to indicate that both “Yes” and “No” are the correct responses for certain events, and then specify that the event can only end upon a correct response. This means that SuperLab’s datafile would always show “C” (for correct) on these events, but you can apply your own scoring formulas in Excel (or similar program).
It might be helpful to shrink the “Yes” and “No” areas until they no longer overlap. Also, the label that is provided in the datafile output is the label of whichever mouse area matches first. If it is more important to have emotion labels (Happy, Disgusted, etc), then you should move “Yes” and “No” to the very bottom of the list of mouse responses. This way, the emotion labels will always match first. That will ensure that no “FERCHOICE” event will ever contain “Yes” or “No” in the Participant Response column of the output.
Regarding using the serial port, I am not aware of any better options out there then what you are currently using.
There is not a specific setting for initial position. However, as a workaround, it would be possible to create a trial that requires clicking on a particular stimulus. As soon as that trial is successfully completed, you would know for certain that the mouse pointer had just been positioned at the desired spot.