| |
Abstract:
Here, the brain is considered a distributed intelligent
system; intelligence is assumed to arise mainly from the way
cerebral agents enroll to solve different kinds of problems and (c)
this enrollment is proposed to obey defined entropic rules. The EEG
associated to game playing was used to test these assumptions
(Rocha et al, 5th Ann. Meeting Cognitive Neurosci. Soc., 1998). The
possibility pi,j that the activity at each two i,j EEG site
recordings were correlated was assumed to be equal to the
regression coefficient for the corresponding ERAs calculated to
defined game events. The entropy of correlation hi,j of the
activity recorded at the i,j sites was calculated as: hi,j = - (
pi,j log2 pi,j + (1- pi,j ) log2 (1- pi,j ) ). and the mean
possibility pi as hi = - pi log2 pi + (1- pi ) log2 (1- pi ) The
complexity Hi of the cerebral processing at the site i was
calculated as a function for hi,j for the other 19 sites: 19 Hi = S
hi - hi,j j=1 Multiple regression analysis showed that game score
and IQ were linear functions of Hi. The present results strongly
support the proposal that intelligence arises when pi tends to .5
and increases the actual value of Hi.
|
