The actual sensor device in a PIR module normally has two different sensors side-by-side inside it, so that they do not detect background IR, which is always present, the output is the difference between them. This means that if both sides see the same IR level, whatever it is, there is no output. If something with a different IR response, either emitting it or blocking the background comes into view, the radiation falling on one sensor will be different to the other so it produces an output. The fresnel lenses on commercial IR sensors are designed to optimize the effect, they direct IR radiation from different directions to one side or the other to make the range wider and to 'focus' the radiation so it is more sensitive. If you imagine you are inside a commercial sensor and looking outwards, (I'm using example angles) you might find 0 to 10 degrees falls on sensor A, 10 to 20 degrees falls on sensor B, 20 to 30 degrees falls on sensor A again and so on. This method increases the chance of a moving object being detected while allowing anything common to all zones being ignored.
If you move the sensor, the chances are you will get the same response as a stationary sensor detecting a moving object. As it rotates, any object in view will pass from zone A to zone B and back again which will make it detect movement.
Brian.