I think that doubling up a style part would be too "heavy". As it is, I already like to thin out the styles by switching parts off.
However, I wonder if there's a way to make something like this happen. If you connect MIDI out to MIDI in, you can effectively use the Song parts to play along with any MIDI data being transmitted (this is how the V Console product works). So, if you did this, then set a Song part to receive the same MIDI channel as one of the style parts is sending, you could achieve this result by assigning whatever 'layer' voice you wanted to the song part.
This should work on any PSR, Tyros or Genos. Give it a shot and see!