colleagues discovered that such direct stimulation must be continued 

 for at least 350 milliseconds for anything to be experienced at all. 



The main discovery of the investigation was that conscious 

 awareness of a tap does not develop until about 500 milliseconds — 

 half a second! — has elapsed. This surprising result was arrived at 

 by comparing the order in which the patient perceived two taps, a 

 real one to the back of one hand and one elicited by direct stimu- 

 lation of the cortical area corresponding to the other hand. (Details 

 of the experimental protocol are rather complicated, and we shall 

 not consider them here.) This half-second delay should be compared 

 with the maximum time taken for signals to travel — using the nerve 

 cells as "stepping stones" — across the entire length of the brain, via 

 the shortest route, namely about 100 milliseconds. 



Why should consciousness take such an extraordinarily long time 

 to develop? What are the nerve signals doing during that half-sec- 

 ond? The most obvious conclusion is that they are running around 

 the brain's neural circuit (see fig. 2), and from what was stated 

 earlier, it is apparent that a prime candidate for their route is the 

 closed loop observed by Shigeo Kinomura and his colleagues (see 

 above) to be activated by attention. Those signals would not be 

 undertaking their circuitous journey just to consume time, however. 

 They are modified as they travel, and one can imagine them pro- 

 voking recall of past memories as they go. This seems to be a safe 

 conclusion because the muscle-directing region is a major compo- 

 nent on the closed-loop route, and as we noted earlier it is only 

 through motor sequences — actually executed or merely imagined — 

 that we can ever learn anything, that is to say acquire memories. 

 Physicists use the term self-organization for such a process, and the 

 system uses it in order to lock on to the sequence of muscular move- 

 ments corresponding to a specific cognitive element. 



A case in point was that third sentence we considered earlier, with 

 its two different pronunciations of the six-letter string p-e-r-m-i-t. It 

 can easily be shown that it takes a minimum of about 50 millisec- 

 onds to articulate a single phoneme, so even if we read the sentence 

 as quickly as possible, we would not reach the first occurrence of 

 that string of letters until after half a second had elapsed. During 

 that period, our system would be articulating other unambiguous 



14 



