Technically you are correct when measuring the output of any single LED chip especially at a very close distance of less than 12 inches. Let's just say that there is only 1 LED that is emanating light. That light wants to spread out 360 degrees from it's point source, and the photons from the side not visible are being absorbed as heat in the heat sink of your lamp. You can take measurements of scattered photons from the periphery and they will be far less than those emanating from the exact center of the LED light. Your example of saying that you are measuring the output of only the central LED is not totally correct due to the scattering phenomenon of light mentioned earlier. It is impossible to measure the output of one LED unless you cover all the rest so that light does not emanate from the shield and then your reading will be much different. When you consider an entire light field composed of hundreds of LEDs as a "single point source" or whether you are measuring a small area or single LED the Inverse Square Law
always works and the values you get are going to be determined by
the location of the group of LEDs that you are taking your measurments under. All adjacent LEDs are contributing scattered photons into the meter when you are even as close as 1 inch to the central LED due to the inherent scattering of photons that is characteristic of all light visible and invisible, which means that this reading will even fluctuate a tiny bit due to the amount of scattering that is hitting the receptor of the measuring device at any given second in time. So all PAR readings are set by the mfg based on an average of readings and may vary with any two lamps of the same model and manufacturer. Here is an article that deals with PAR that you might find interesting.
http://www.compareledgrowlights.com/par_and_lumens/