9/15 AN INTRODUCTION TO CYBERNETICS 



introduced when one considers how fast in time the chain is being 

 produced by some real physical process. So far this aspect has 

 been ignored, the sole graduation being in terms of the chain's own 

 steps. The new scale requires only a simple rule of proportion for 

 its introduction. Thus if (as in S.9/12) the insects' "unit time" for 

 one step is twenty seconds, then as each 20 seconds produces 0-84 

 bits, 60 seconds will produce (60/20)0-84 bits; so each insect is 

 producing variety of location at the rate of 2-53 h\ts per minute. 



Such a rate is the most natural way of measuring the capacity of a 

 channel, which is simply anything that can be driven by its input to 

 take, at each moment, one of a variety of states, and which can 

 transmit that state to some receiver. The rate at which it can trans- 

 mit depends both on how fast the steps can succeed one another and 

 on the variety available at each step. 



It should be noticed that a "channel" is defined in cybernetics 

 purely in terms of certain behavioural relations between two points; 

 if two points are so related then a "channel" exists between them, 

 quite independently of whether any material connexion can be seen 

 between them. (Consider, for instance, Exs. 4/15/2, 6/7/1.) 

 Because of this fact the channels that the cyberneticist sees may be 

 very different from those seen by one trained in another science. 

 In elementary cases this is obvious enough. No one denies the 

 reality of some functional connexion from magnet to magnet, 

 though no experiment has yet demonstrated any intermediate 

 structure. 



Sometimes the channel may follow an unusual path. Thus the 

 brain requires information about what happens after it has emitted 

 "commands" to an organ, and usually there is a sensory nerve from 

 organ to brain that carries the "monitoring" information. Monitor- 

 ing the vocal cords, therefore, may be done by a sensory nerve 

 from cords to brain. An effective monitoring, however, can also 

 be achieved without any nerve in the neck by use of the sound 

 waves, which travel through the air, linking vocal cords and brain, 

 via the ear. To the anatomist this is not a channel, to the com- 

 munication engineer it is. Here we need simply appreciate that 

 each is right within his own branch of science. 



More complex applications of this principle exist. Suppose we 

 ask someone whether 287 times 419 is 118213; he is likely to reply 

 "I can't do it in my head — give me pencil and paper". Holding 

 the numbers 287 and 419, together with the operation "multiply", 

 as parameters he will then generate a process (a transient in the 

 terminology of S.4/5) which will set up a series of impulses passing 

 down the nerves of his arm, generating a series of pencil marks on 



180 



