Page 87 - Deep Learning

70 Creativity

The question of limitations is also unanswered. if the mind can generate
novel combinations indefinitely, there is no reason why a problem that requires
a creative response should be hard. all it takes is to keep generating new com-
binations until one is found that passes the relevant test. The only limitation
on creativity according to generate-and-test theories is the time it takes to find
the desired new idea. along the way, the mind is frantically busy generating
and testing endless combinations, or so this type of theory implies. There is no
explanation for the experience of being at an impasse; that is, of being stuck, of
having exhausted all possibilities and not knowing what to do next.
in short, the principle of combination of cognitive elements cannot
answer the creativity questions as long as it is embedded in a generate-and-test
mechanism with a random generator, so the three combination theories are
unsatisfactory. However, the combination principle itself is powerful, and the
notion of nested envelopes of evaluation resonates with the concept of mul-
tiple system levels. Both play central roles in Chapter 4.

Novelty Through accumulation
Novelty can arise through the accumulation of work. When a familiar step is
executed as part of a sequence of steps, it might interact with other steps in
such a way that the final outcome is novel. Chess provides an example. The
individual moves in a chess game are guaranteed to be familiar, because the
legal moves are specified by the rules of the game. Nevertheless, every chess
match unfolds differently from any match played before it, and the particular
sequence of moves played in a particular match might be called creative by chess
experts. Similarly, a mathematical theorem can be novel even though every step
in its proof consists of a familiar algebraic transformation. each step along the
solution path contributes some small amount to the gradually widening gap
between the product under construction and prior products of the same type.
Cognitive psychologist robert W. Weisberg puts this position as follows:
On the whole, it seems reasonable to conclude that people create solu-
tions to new problems by starting with what they know and later modify
it to meet the specific problem at hand. at every step of the way, the
process involves a small movement away from what is known, and even
these small tentative movements into the unknown are firmly anchored
in past experience. 29
The accumulation-of-work principle has the advantage of concurring with two
of the most salient facts about the production of novelty: Creative individuals