The stock oil pressure gage is a bimetallic servo type. Picture a circuit where 12 volts comes from the ignition switch then goes through a little heater coil that's wrapped around a bimetal strip in the gage. As you may know bimetal is a sandwich of two metals with differing coefficients of expansion. This difference causes bimetal, when heated, to physically deform.
The strip is mechanically connected to the gage's pointer. If the other end of the gage coil is grounded current will flow through it, heating the bimetal strip which in turn warps and causes the gage pointer to move. Course, if the current is left flowing not only will the pointer go to max but the coil may burn out. What is needed is a ground only long enough to move the pointer to a position representative of oil pressure.
The sending unit does that through a contact on another heater-wrapped bimetal strip in the sender that is mechanically biased by a metal diaphragm sensing oil pressure. At a given pressure the diaphragm pushes up on the strip causing the contact to close and connecting the gage's coil to ground. This causes the gage's pointer to begin deflecting upward.
This same contact causes current to flow in the sender's own heater coil which warps the strip in the sender. The warping continues until the the bias of the diaphragm is exceeded. At that point the strip will open the contact, breaking the circuit and causing not only the gage to begin falling back towards zero but also causing the sender strip to cool and quickly close the contact again. The process repeats itself at a fix rate as long as the diaphragm bias on the sender strip (how much it's pushing up on the strip) remains constant.
The result is an output signal from the sender to the gage that's an on/off pulse whose duty cycle varies with oil pressure. In simple terms the warping of the strip in the gage (and hence pointer deflection) closely follows the warping of the strip in the sender. That's why it's called a servo.
You can see this behavior on a cold start by closely watching a stock gage. It should rise to some value and then fall slightly before rising again. This slight up and down pointer deflection will continue until the heating in each strip reaches equilibrium. The relatively slow response time (about one second) of the design is what prevents the gage from having noticeable oscillations once this happens but the bottom line is the stock gage is always turning off and on and, as the TRSM points out, so will a bulb connected in place of it.
It was done this way because the oil pressure input on the ECU (used to trim idle speed) requires this type of "digital" signal. Toyota chose to do it this way rather than use an analog equivalent. Why I don't know.