Dear Laurent Bouvier  

Let's NOT say "homeostatis and, behind it, cybernetics"; that makes no sense to me.

Homeostasis is control in intricate biological contexts, including that OF BEHAVIOR PATTERNS ** (behavior is biological functioning). If we understand this (homeostasis), as in works in behavior systems (including IN conceptual systems -- including IN abstract thought systems), then you will have a better form of control to model.  [ There is much more gain to be had from this -- and more readily -- than from language modelling.  (Language is very helpful but is, in a way, an add-on to basic yet very sophisticated abilities. Language is involved in the fruition of some cognitive abilities, but not with the basic understandings) ]  Homeostasis IS about control and optimization (adaptation is a form of optimization -- just ask Darwin).

Understanding behavior in a way you can see homeostasis involves a close understanding of behavior.  This includes finding the perceptual/attentional SHIFTS which occur at the inception of new cognitive abilities (new behaviors and qualitatively new types of learning).  A psychological theory which allows for this is ETHOGRAM THEORY (see "A Human Ethogram ... " in the "Human Ethology and Development" Project, here on researchgate).  [ This theory is in a sense NEW, since only recently can its hypotheses be tested, with eye-tracking technology, etc.  This theory is thus not well-known (to say the least); one might say this theory has not yet been understood enough to even be recognized.  So, be happy with this perspective provided because "new" is good when it would provide useful findings when/if the hypotheses are investigated and verified; the hypotheses are testable and can be shown true or not -- how difficult thorough testing would be is an open question. ]

** FOOTNOTE: Display of real empirical observable behavior PATTERNS FOUND through research is TOO rare nowadays.  But, such can be seen in the work of classical ethology (summarized in Eibl-Eibesfeldt's Ethology: the Biology of Behavior (second edition in English, 1975)).  Also note, though: unfortunately Eibl-Eibesfeldt's latter book on Human Ethology book is off-track because he did not define or recognize the second form of Piaget's equilibration, the "balance" between making a 'stage shift' and not-yet "making it".  This is key to the real and larger qualitative changes in human behavior (conceptual behavior).

• You should see which type of cybernetics is your topic.
• Cybernetics is inspired by Swinth through a plan from thermostat. Actually the most completed cybernetics is "Double Looped Cybernetics" that means there's two loops for controlling respective process as Quality Control (QC for internal control) and Quality Assurance (QA for external control). 
• The main and important question is if cybernetics could advance by new loops for more control or not?