Conjecture: reorganization within a control hierarchy is a result of collective control by the neurons.
Neurons are living control systems in a shared environment with some genetically determined constraints. Environmental stabilities result from and are maintained by collective control. This has been demonstrated and does not require additional special explanation. Rather, some special explanation would be required if this did not occur among neurons (or any population of living control systems) in a shared environment.
Neurons seem to be predisposed (genetically or epigenetically?) to active reorganization of synaptic connections with one another. This is seen in vitro and in early development of a (mammalian) brain, as described in the paper “Imaging Synapse Formation and Remodeling In Vitro and In Vivo” (cited in the topic " The pervasiveness of reorganization").
This algorithm is based on observed behavior of slime molds. In principle, it could be done with interacting control systems, in the manner of the CROWD demo.
A model of collective control constraining innate reorganization would involve more variables.
The same neurochemical variables are involved in functions of the reticular activating system. That is another discussion, but insofar as attention is coincidental to and perhaps instrumental to reorganization it is relevant.