Dear Georgina,
maybe i should explain it a little more in detail.
You assume the photon to be a particle with a certain electromagnetic flux around it.
As you know from the double-slit experiment with single photons, over time, there will be an interference pattern at the observation plane. This pattern consists (usually) of dark and light bars. The lightest bar of them is found approximately in the middle of the two slits (projected to the observation plane).
If we measure with two detectors in the area whe are used to find the lightest bar, we can 'deduce' 'which way' each photon went (slit one or slit two). The relative frequencies of particle impacts at detector 1 and 2 are different from what we could expect if we do not measure 'which way' information.
In other words: By measuring 'which way' information, the inferference pattern i spoke of above is altered over time at the areas where the the two detectors are situated: Statistically, much more photons do arrive there than in the case of a measurement method that does not extract which-way information at these two points in the observation plane.
Note that until the observation plane (with or without the two detectors standing there) is reached by every single photon, the conditions for the physical mechanisms described by you (two slits open, particle, flux) from the slits to the observation plane remain unchanged. This must also include the mechanism that is responsible for every single particle's path to contribute properly to the whole interference pattern.
But as i remarked, by observing several regions of the observation plane with a device capable of extracting 'which way' information, the expected interference pattern gets altered. Means, now there are more photons that find their ways to these measurement locations than without the two detectors.
The question therefore is, how the two detectors in sufficient distance to the action at the slits can alter this action by remote.
For an interesting discussion of the so called "separation fallacy" you may wish to take a look at this interesting paper:
The 'Past' and the 'Delayed-Choice' Double-Slit Experiment
Best wishes,
Stefan