There could be two types of diffusion. One is retrograde, from
postsynaptic to presynaptic. Another type is from one postsynaptic
site to another postsynaptic site nearby (inside the same dendrite).
Im suggesting this latter form might lead to some interesting learning
The secondary messengers of metabotropic receptors should
serve some purpose, otherwise they could be replaced entirely
by electrical synapses. Thats why Im suggesting that there
could be some downstream events after their activation, and
2nd messengers might diffuse to a nearby synapse.
Lets say there are 1-3 synapses per um (micrometer) of dendrite.
That means synapses are about 1-0.3 um apart. "Fast" axonal
transport is 2 to 5 um per second. Diffusion might even be faster