It is good policy to avoid “intuitive” thinking until one understands how observed phenomena arise or are described by the theory being applied. Quantum Electrodynamics is a very complete and mature theory. In this theory photons are neither particle nor wave but appear as the result of quantized fields and their interaction with matter. Thinking of photon at either wave or particle is very counter productive. Case in point I recently devised a particle model for a simple photon counting experiment I could run with a CCD camera and a single mode fiber. The model is completely wrong as any one who understands these things would know from the start. Basically my reasoning was this, photons emitted by an LED arrive at the sensor quite slowly. So slowly the probability of two being emitted from the fiber aperture at the same time is negligible. I figured I could model photon emission like a machine gun firing bullets at random intervals but at a constant rate. If this is true (which it isn’t) then if a photon is detected at pixel 1 it can’t be detected at pixel 2. Well, this leads to a very detectable anti-correlation between observed photon counts which we all know isn’t observed. The flaw in my argument is that though it’s very rare, an LED unlike a machine gun can emit multiple photons at the same time. This effects the state of the emitted light in a way that completely removes the correlation between pixels. I had to go through the math to understand this. I hope this helps.