Page 135 - Deep Learning

118 Creativity

EVALUATION

The evaluation of a complicated scientific theory is a multifaceted affair that
includes assessments of the conceptual clarity and internal coherence of the
theory, its explanatory power, completeness, simplicity, support from empiri-
cal studies, relations to other theories, practical usefulness and other factors as
well. The evaluation process sometimes takes decades, and widespread accep-
tance of a theory is rarely due to the work of a single person or laboratory. The
evaluation of the redistribution theory has hardly begun. The previous sections
implicitly argue for its clarity, internal consistency and explanatory power.
The present section discusses completeness, simplicity, empirical support and
relations to other theories.

Completeness and Simplicity

The redistribution theory reduces the insight sequence to cognitive processes
that are so simple that they can be completely explained in mechanistic terms.
All of the basic processes and structures postulated by the theory (layered per-
ceptual processing, memory retrieval via spread of activation, heuristic search,
limited working memory capacity, mental look-ahead, feedback propagation,
48
etc.) have been modeled by running computer simulations. Breakdown into
processes of yet smaller scope would take us into neurophysiology. Raising or
lowering the activation of a memory link cannot be analyzed further at the
cognitive level, but has to be explicated in terms of neural matter. The task of
analyzing insight into processes that are so simple that they do not cry out for
further analysis has therefore been completed. There is no homunculus wait-
ing to be discharged in the redistribution theory, no black insight box in the
flow diagram and no leftover component to be explicated another day.
Furthermore, the theory is innocent of exotic claims. None of the processes
presupposed by the theory were postulated specifically to explain insight phe-
nomena. On the contrary, cognitive psychologists routinely draw upon them
to explain a variety of phenomena, many of which have little to do with the
production of novelty. That perception and comprehension are constructive,
49
interpretive processes guided by prior experience is the standard view, sup-
ported by countless studies and a variety of phenomena from visual illusions
and priming effects to perspective shifts during reading.
Also, there is nothing controversial about the claim that information in
memory might fail to come to mind when needed. The everyday experience
of trying to recall a person’s name, failing to do so but then spontaneously