300 



THE LANGUAGE AND CONCEPTS OF CONTROL 



some narrow field may each have rejected as noise some information border- 

 ing on the subject which would be more pertinent to their discussions than 

 either realizes. This is one of the reasons for disputes, sometimes very 

 heated ones, between logicians who are specialized in different fields. Then 

 of course there are man's errors in logic — and they are a fact too. Over even 

 an hour's test, the adding machine will demonstrate man's errors in logic 

 very vividly. 



We have seen that there is variation in nature. There is also order. There 

 is variation in the physical structure of man's sensory organs. Therefore 

 the nonverbal impressions which two men have of the same object may be 

 quite different. The verbal impression each would give — thanks to training, 

 experience, and definition — would, however, be about the same. It is gen- 

 erally accepted that the essentials can be abstracted by one and com- 

 municated to another by words. The variations can be described also, if 

 they can be observed. Further, McCulloch and Pitts showed in a famous 

 deduction that if anything can be described fully in words, the description 

 can be programmed accurately into a man-made computer, provided the 

 computer is comprehensive enough. Therefore our own "built-in" com- 

 puter, as well as the man-made one, should have the physical capability to 

 receive (as well as give) a complete description. Yet language has a drift in 

 meaning over a course of time. Does the concept also drift? 



Feedback 



Control of a system by its computer is accomplished by feeding back into 

 the controller the result of the measurement of difference or error (Figure 

 11-2). The computer can then dispatch the corrective order, the order which 

 when carried out will reduce or eliminate the error. This is accomplished 

 in mechanical and electrical machines through what is called a control am- 

 plifier, a device which takes the determined error, amplifies it, and inverts 

 it as the corrective "order" to the process. In the living thing this is ac- 

 complished either by the conditioned reflex of the autonomic nervous sys- 

 tem, or the voluntary control by the central nervous system. 



(a) 



(b) 



Figure 11-2. The System Diagram, I. (a) General feedback only; (b) General plus 



particular feedbacks. 



