Gelernter's basic argument is that types of human thinking can be laid
out in a continuous spectrum, from what he calls 'high focus' at one end to
'low focus' at the other. He says that at the high focus end thought is
logical and penetrating, that the mind might move from one thought to
another according to logical connections, and at the lower end the mind
move from one thought to another according to emotional connections.
While I doubt that this is really how the human mind works ( it seems
odd for a fuzzy neural net to function along a linear scale ), it does
describe a relatively simple way of emulating emotions in an AI.
BTW, I wouldn't reccomend buying the book; Gelernter goes off on a
religious tangent many times which spoils it as a whole work. But
the first 3 chapters contain the gist of the idea - a mere 64 pages
which can be read while browsing in a bookstore or library.