So my layman's understanding is that a photon with more than around ~1 MeV of energy has a possibility of "acting like" an electron-positron pair in its interactions with other particles. I say "acting like" because I know that there is no actually observable event of the photon spontaneously converting into the pair. Additionally, AIUI the state also does not "cycle" into the pair and back (even imperceptibly), it would remain constant over time if left to itself. I'm imagining this as a sort-of superposition between the two, although I realize that that seems wrong but its hard to specify why. I know that it nonetheless acts like a pair in that there are Feynmann diagrams where the other particle experiences interactions from the charged pair (as opposed to the photon) that must be taken into account to properly calculate the outcomes.
I've also heard - but don't really understand the implications of - that the pair, being virtual, is allowed to have "wrong" particle statistics, like masses not matching the actual electron mass. I particularly don't understand how this manifests if the photon actually does have enough energy to cover the "real" masses the pair ought to have vs when it doesn't. Presumably the former case allows for interactions that separate the virtual pair into real (anti) electrons and the latter doesn't.
Now the question - imagining the particle as this not-quite-superposition of high-energy photon and $e\bar{e}$ pair, why does the "time spent" (in a metaphorical sense) as the massive pair not affect the particle's net/average velocity, or at least change the shape of the wavefunction? I know that neutrino flavours' varying masses do mean that a superposition of neutrino flavours has a net velocity that's dependent on how much each flavour participates in the superposition, so is this case not analogous?
