Let's begin with individual cells, these are the entities which need to control metabolic substrate availability.
You eat some food. Plasma glucose and chylomicrons/FFAs rise, delivering energy to peripheral tissues. In the early stages of food absorption both glucose and FFAs enter cells under the facilitation of insulin. They do this easily, the cells are "hungry".
As an individual cell becomes replete it has to signal that it doesn't want any more metabolic substrate. This is achieved via the CoQ couple acting as the master sensor for metabolic energy status. It is being reduced using NADH from the cytoplasm (the glycerol-3-phosphate shuttle), by FADH2 input via complex II (acetyl-CoA in the TCA) and via FADH2 input from ETFdh (from beta oxidation of saturated fats). And of course from mitochondrial NADH via complex I. Given a high delta psi (ie minimal consumption of the proton motive force because ATP is already plentiful) this CoQ reduction eventually facilitates RET through complex I to give superoxide generation in order to stop insulin signalling. Which then limits cellular caloric ingress. This can be thought of as the "cellular satiety" signal. It is ROS generated. Let's say that again:
Satiety in peripheral cells is an ROS signal. It is generated in the mitochondria. This is pure Protons.
Now let's scale that up.
As more and more peripheral cells decide that they no longer need to respond to insulin then there is less and less of a "sump" available for absorbed calories to drop in to. The availability of calories which no longer have anywhere to go is the whole-body driver of the need to signal satiety. This surfeit of calories will be sensed in the VMH and the cessation of eating will be ROS mediated.
These people have the correct sort of idea:
Fuel utilization by hypothalamic neurons: roles for ROS
and so do these
Role for mitochondrial reactive oxygen species in brain lipid sensing: redox regulation of food intake
I'll make that clear: Satiety occurs when the brain senses that calories are no longer being accepted by the peripheral tissues using an ROS signal. Superoxide will be that signal.
That's what I consider to be the normal physiology.
We can now apply this to the various studies using intranasal insulin.