On the Role of Computers in Creativity-Support Systems - Computational Creativity Research: Towards Creative Machines - page 108

On analyzing many of Rodari's techniques to stimulate creativity and imagination,
we find that they are essentially elaborate ways to get some unrelated combination
of words or concepts. When people try to connect these unrelated concepts in their
imagination, their creativity is stimulated. For example, in one method, two children
are asked to go to different rooms, close their eyes, open a dictionary at a random
page, and put their finger at some point on the open page. This produces two random
words, and then the children make a story connecting these two words. In another
activity, each child is asked to bring a picture from some newspaper or magazine. The
children sit in a circle, and all the pictures are put face down in the middle. Children
take turns at random (by drawing lots), and the child whose turn it is turns over one
of the pictures, and starts a story based on that picture. Each child (in random order)
repeats this process, except that all the subsequent children have to continue the story
generated so far by incorporating the picture they just turned over. This process is
akin to generating novel and creative metaphors by combining unrelated words or
concepts together [ 29 ].
In terms of computational modeling, at least this aspect of creativity is easily
modeled algorithmically [ 31 ]. This is because it is very difficult to model common-
sense conceptual associations on the computer, and it has been a challenging research
areas since the advent of Artificial Intelligence. This implies that it not difficult for
a computer program to break or ignore these conceptual associations and generate
a combination of two unrelated words or concepts. For us humans, on the contrary,
conceptual associations are an unalienable part of us, and so elaborate methods have
to be devised to look past those existing associations.
6.2.2 Modeling Creativity in Legal Reasoning Computationally
We now move to a completely different topic and present another piece of previous
research where we studied creativity in legal reasoning, and sought to model it
computationally [ 23 , 25 ]. The main idea behind this approach was that creative
insights often come from applying the high-level structure (or gestalt) of one situation
to the low-level details of another situation. The distinction between the levels (high
and low levels) is important, for if both the situations are considered from a high level,
then only traditional analogy can result, which, as far as creativity is concerned, is
counterproductive (see [ 24 , 29 ] for detailed arguments with examples). We explain
this approach with an example below.
We focused on situations where new categories are brought in to analogize or
distinguish between prior cases and a new case in order to argue for a particular
resolution of the new case. Our domain was a particular tax law in the US, which
allowed taxpayers to deduct their home-office expenses from their taxable income
under certain conditions. These exception conditions were based on factors such
as whether the home office was the principal place of business ( PPB ) (e.g. when a
doctor saw patients regularly at a home clinic), whether the employer provided office
space for the employee, and so on.