Open-Ended Elaborations in Creative Metaphor - Computational Creativity Research: Towards Creative Machines - page 227

to find one. Similarly, recall the real example in the Introduction of a migraine being
described as a moving animal in the head. Much of the point is about the time-course
of the migraine developing and about the fact that the progress cannot be resisted.
Such content can be carried over from pretence to reality by VNMAs. For example,
difficulty of resistance amounts to difficulty of preventing continuation, and here
the difficulty, prevention and continuation are handled by three VNMAs.
But the ATT-Meta approach could be expanded to include the creation of
new correspondence rules using methods discussed in the analogy literature
(e.g., Falkenhainer et al. [ 18 ], Indurkhya [ 27 ]). If this were done then the approach
would actually allow two different ways in which an explicit metaphorical pair-
ing could be processed, such as in “Yesterday was Christmas Day for newspaper
editors” 20 or similarly “Yesterday was a feast day”—when yesterday was not a
special day. One of these ways, the one already available, and suitable in some cir-
cumstances, would be for such a sentence to be understood as a pretended equation
(the day is Christmas Day) or pretended categorization (the day is a feast day). Then,
within the pretence, an inference might be made that the day was very pleasurable,
and this could be carried over into reality, with the help of VNMAs, to create the
conclusion that the real day was very pleasurable. The other understanding approach,
suitable in other circumstances, would not be to put an equation or categorization into
the pretence, but rather to try to discover an analogy between the target item (the real
day) and the source item (Christmas Day, or a feast day, as an aspect of the pretence).
The analogy would be couched initially in newly-constructed correspondence rules
about the specific day, but might later be generalized to cover any day.
This potential double approach to explicit novel pairings resonates with and is
closely analogous to a prevalent concern in the psychological literature on metaphor.
Here there is dispute between the categorization theory (or class-inclusion or
property-attribution theory) and the so-called comparison theory (see, e.g., [ 11 , 23 ]).
Considerations in that dispute about when categorization or comparison are appro-
priate could illuminate the question of when ATT-Meta should seek a new analogy
across the pretence/surround boundary and when it should use a within-pretence
categorization or equation.
ATT-Meta's handling of creativity benefits from the fact that metaphorical views
are not reified in ATT-Meta: they are not themselves explicitly represented. This
allows great freedom in the nature of the correspondence rules. Metaphorical views
are merely a theoretical abstraction from the guards in the individual correspondence
rules possessed by an understander using the approach. Consider again correspon-
dence rule (8). Because the guard requires there to be a person (in the surrounding
space) that (within the pretence) has subpersons, the rules can be thought of as
belonging to the Mind as Having Parts that are Persons view. So can rule (9). But
there is no explicit connection between these rules, other than the fact that they have
the same guard, or any labelling of the rules as belonging to any view. Thus, the
use of correspondence rules does not require a prior, explicit decision that some
particular metaphorical view is in play. Rather, what needs to be inferred is certain
20 On BBC Radio 4, Today programme, 4 November 2005.