Bolt shear is a limit state for the bolt, whereas bearing stress is more a limit state for the connected plates. Therefore, it makes sense to evaluate bearing stress on a plate-by-plate basis. When two or more plates are sharing the load, you can consider the total thickness in each loaded direction versus the total applied load.
I suppose that in principle, when two or more plates are sharing the load you could split the load between the plates based on relative bearing area and evaluate adequacy on a plate-by-plate basis. However, since the loaded 'width' is the same for all plates (i.e. the bolt diameter) this approach would give the same result.
As a point of interest, when you start to consider real-world behavior with more than two shear planes, or when some of the plates involved are filler plates, the true behavior can become quite complex.
A little qualitative thought experiment:
There are two shear planes and three bearing 'areas'
The stress on the two shear planes is equal
The $Plate_1$ load + $Plate_2$ load = $Plate_3$ load
Load in $Plate_1$ > load in $Plate_2$ (because $t_1$ > $t_2$)
Bearing stress at $Plate_1$ = bearing stress at $Plate_2$ = $\frac{P}{bolt diameter(t1 + t2)}$
If ($t_1$ + $t_2$) = $t_3$ then the bearing stress is uniform along the bolt length, otherwise it will vary