In a capacitor start motor there is a winding, connected in series with a capacitor, used to produce the rotating field required to start the motor. This circuit is then opened by a centrifugal switch when the motor speed exceeds a set value. Obviously the timing is critical here, as is the time-constant and resistance of the bulb in series with the motor.
Too small a bulb "wattage" will result in the bulb burning at near normal current and the motor not even starting; a too large bulb will allow the motor to start normally and run at near normal speed. It would be interesting to instrument such a set-up.
With a normal "fan motor", which uses a shaded pole configuration, the bulb acts as a resistance, its value depending on the final current, and acts to slow the motor. One would not expect any conditions which could cause bulb failure.
Deriving a "Spice" model would be tricky, but perhaps doable.
Unfortunately we don't have any more details of the original posters experiment. He lost interest after failing several bulbs. It's interesting that a fairly common device can result in a seemly impossible result.