mlloyd1 said:
I susepct the diodes are in the circuit to protect the caps from excessive reverse voltage. Under normal operating circumstances, the diodes do not conduct at all - the caps wouldn't allow enough forward voltage to bias the diodes.
I initially thought as you did; then I though some more.
I also tried an experiment using much lower voltages than power line levels. The diodes did not turn on.
Michael

Let me think twice :
First, let's us assume DC only :
- as soon as the DC voltage on the upper mains terminal is greater than two diode drops (say 2 V with big rectifiers), D1 and D4 will conduct and bypass the capacitors...
- If the DC is reverse, i.e. negative on the upper terminal, D2 and D4 will be direct polarized and then bypass again the capacitors...
Then AC only : If I consider the bottom terminal as a reference, the voltage at the top terminal will change from +325V to -325V peak, assuming a 230V mains (european practice...). The diodes will be polarized as soon as the voltage across them is greater than two diode drops, but now they are shunted by the impedance of the capacitor. If the capacitor is big enough, this low impedance will keep the diodes from conducting...
Now let's us consider DC + AC :
-The AC component will pass through the capacitors, if they are large enough, but the DC component will be blocked by them, as they have a very high impedance and will take the less impedent path at zero frequency, through the diodes, except the first 2 V...
Maybe I missed a point and I would be happy that someone shows me where... A spice simulation showed the same blindness as me...
I fully agree with you that the diodes can be non-conducting under normal circumstances, i.e. no DC, but DC seems not to be blocked !