304 THE LANGUAGE AND CONCEPTS OF CONTROL 



though the best informed opinion today is that it will be possible, but diffi- 

 cult, to construct a machine to do such work. On the interesting subject of 

 self-control or self-determination, which implies judgment of what is good 

 and bad, and free choice to do either, nothing can be said about what a 

 machine' of the future will be able to do. Today's machines are completely 

 deterministic — as are many of man's acts. 



The question of whether creativity and the emotional, psychic, and reli- 

 gious experience of man can be contained within the physical structure of the 

 human brain is unanswerable from the framework of science, because the 

 extrapolation from experimental test is simply too far to be reliable. This 

 will be especially evident to those doing experimental work even in heavily 

 experimented subject matter: the results are, even there, always full of sur- 

 prises! To assume an answer to this question, then, would be unscientific, 

 since experimental verification is not yet possible. 



A more useful question for control biophysics is: "How far can physical 

 equipment be made to go toward reproducing the functions and behavior of 

 man's brain and mind? How does the brain actually do the job of con- 

 trolling so finely the human body? The answer seems to he in models or 

 representations. 



This is the interest of biophysics in Samuel's checker-playing machine; 

 Shannon's chess-playing proposal; the U. S. Naval Research Laboratory's 

 self-replicating machine; psychologist Ashby's homeostat, which adapts it- 

 self into compatibility with a new environment; Walter's Machina Speculative 

 and MIT's mechanical hand— robots which have component parts which 

 give them many of the response characteristics of animals; and other ma- 

 chines, some much more complex. 



Within the past few years there has been considerable effort expended in 

 making models of the nervous system. The work falls roughly into two 

 forms In one, man attempts to represent or reproduce the biological phe- 

 nomena as closely as possible. In the other he explores the behavior of simu- 

 lators-electronic elements, for example, whose electrical behavior is similar 

 to that of the nervous system. For example, M. L. Babcock, F. Rosenblatt, 

 B. G. Farley and L. D. Harmon have all done intriguing pioneer work. 

 Farley et al. have simulated the firing pattern of a two-dimensional array of 

 neurons (Figure 11-4) by programming their TX-2 computer with cor- 

 relative information on 256 circuits, each of which can do several of the 

 tricks that a single nerve cell can do. An input (stimulus) at some point 

 causes a firing pattern to occur throughout the network; and, if properly 

 displayed on a television screen, this firing pattern can be watched as it 

 progresses. With such an apparatus a study can be made of the characteris- 

 tics which lead to different firing patterns. There and elsewhere the follow- 

 ing have been simulated: the all-or-none firing pattern of the axon, the slow 



