Wave-particle duality.

QM needs wave-particle duality to explain why the same quantum objects sometimes behave as waves and sometimes as particles, depending on the measurement setup.

All the “schizophrenic” behaviour appears because we apply our spacetime concepts to instantaneous pointlike events. For carriers, the whole process happens in an instant, at detection time. Being just a point, they can't decide about travelling a path or a thousand. Every possible path they could traverse at c from start to finish is “real” (much in the same way as the Feynman path integral formulation). We use wave-particle duality because we don't know how to make sense of dimensionless interactions from our spacetime perspective.

Interactions shouldn't be pictured as particles or waves propagating through space, but as spacetime connections “unfolding” or “decompressing” from detection time into the past, conforming each observer's spacetime lattice. Only the complete emission-flight-detection process is fully equivalent to the carrier's instantaneous experience. This “spacetime projection” would be equivalent to what the carrier would feel if it could decelerate. It's also remarkably similar to the Huygens principle, a rule applied to build waves and explain propagation, diffraction and interference.

So carriers get wave behaviour when we “map” their instantaneous existence into our spacetime lattice. Trying to get an answer in the middle of this process is worthless, as we're missing a lot of information that isn't really available until detection time. That's why we are forced to explain the quantum world with waves, propagation, probabilities, superposition and collapses. We can't predict quantum behaviour better than we do, because detection point is always ahead of us in time.