ahrkron,
"Radial propagation" is linear trajectories from or to a common point.
The term "nonradial" I used only as actual re Descartes' vortices.
We assume there are photons, or for that part, fermions which are not physically provable. The only photons, or information about other matter, that we can be sure of are those which impinge upon us radially, i. e., at all. Our picture of the universe is much like Plato's cave allegory in this regard, where only two dimensions of space experienced at each moment.
Picture a Minkowski cone. At its vertex, the observer receives his total light along radial lines in space, as is true locally even in general relativity. The radial spatial geometry of the cone relies fundamentally on the speed of light being constant. Likewise, physical law in a local inertial frame remaining unchanged relies upon the isotropy of the light cone.
Individual photons removed more than one order of interaction from the observer lose all information about their position, momentum, time, energy, angular momentum and other quantum observables from our perspective.
Wavefunction probabilities, based upon orthogonal axes, have no effect on the radial nature of light, I surmise.
"Radially isotropic (first order) observation" can refer back to the Minkowski cone. Juxtapose two such cones at random orientations, and you can see that each additional cone adds a new (classical) "order" of photon interactions. A separation between two events of two or more paths is reducable to total isotropy relative to the initial observer. Nonclassical uncertainty disrupts a composite path when considering the single photon. Although the individual plural-order interaction photon may be completely uncertain, you are right that the total interactions themselves yield definite probabilities upon quantum considerations.[zz)]