There are many different versions of Moog 'Poly' modes across the Phatty and Taurus line of products but finding the right combination to address your requirement/need is tricky.
The only one you can really depend on is the very newest algorithm which leverages a round robin scheme (using v3+ code) and the standard low-note/high-note priority scheme.
Because there is no notion of a 'split' within the firmware you really need to hold/drone one of the keys in order for the 'priority' scheme to work and that is not really suitable for the type of playing that would traditionally involve a single bass note and a single lead note.
The best advice I can offer (especially since you have a Tribute, assuming it's not upgraded with the 2+ upgrade) is to use an out-of-the-box Poly module or a piece of software. I use Midi Patchbay for the Mac; it's free (
http://notahat.com/midi_patchbay) and can do a simple split with ease.
Essentially you turn local control off, send midi into your mac and have the midi data captured by Midi Patchbay with one patch to take notes lower than a certain split point and re-map back to the Tribute's midi address (for example) and above a certain point to another channel/device. If you have a Poly module you can send these notes above the split point and then play this way.
The Poly mode which is built into the newest Phatty lets you configure x # of devices and give each mono synth an address. Each subsequent note is sent to the next Phatty in the list. I do not have multiple Phattys but this is my understanding of the way that the algorithm works (perhaps somebody else can chime in or correct me).
It is very much similar to the earliest hardware based algorithms which were developed for Oberheim 4 voice SEM systems. More modern synths leverage voice 'stealing' modes either by taking the OLDEST (Oberheim OBX's worked this way) held note or by possibly taking the note which will have the least harmonic effect (Kurzweil 250 worked this way).
When each 'voice' has it's own Filter and ADSR, it gets complicated. For some applications, you want all voices to be processed through a commong ADSR and filter arrangement for others you want dedicated. Still others, such as the Matrix 6 'stop' at the Nth note played and refuse to play the next note.
Then there are solutions which are DAW based, quasi-DAW based such as the Expert Sleepers solutions, and completely hardware based (such as MIDIpal which looks excellent and can be found here:
http://mutable-instruments.net/midipal). If you are into modulars, there are products by Synthtec (MOTM), Vermona qMI (Euro), and COTK's Pentaphon (MU) but the Modular solutions tend to favor CV synths.
Personally, I'm considering picking up one of the Mutable Instruments MidiPal units; the bloke behind it will shortly be offering assembled versions (originally it was only in kit form, particularly problematic since affixing the surface mount ARM processor/controller to the board required a fair amount of soldering skills that I don't have [nor do I have the eyesight for such small components; I'm old!])
Hope this lengthy explanation helps somewhat.