> proper definition of reasoning must be provided

It will be related to the context. If a process tells me that "if the light is blue and the sound is sharp then the lever must be lowered", it outputs reasoning; if what the activity did was running on conditional paths like an Expert System, what it did was different from the content and only improper reasoning; if the activity produced semantic evaluations about the light, the blue, the sound, the sharpness, the lever etc., then it was reasoning even if it was a process that reasoned.

But all of this makes sense only when it is there for the proper implementation of reasoning. Otherwise, Dijkstra is valid.