A presentation by Dr. Liane Gabora, an interdisciplinary cognitive scienust trom the University of B.U.!
In both psychology and artificial intelligence, it is widely thought that the principal task of a cognitive system is the minimization of prediction error. While prediction error minimization can explain how we reduce mismatch between expectation and observation (world-tracking), it cannot explain how we develop and maintain coherent, selt-organizing world-models that constrain and enable future learning and action (world-shaping) by telling us not just what is but what's possible! I propose that this capacity constitutes a central aspect of cognition in its own right, and argue that autocatalytic networks are uniquely suited to model it.
I present a theory of creativity, honing theory, according to which creativity is the process by which a world-model reorganizes itself to regain and maintain autocatalytic constraint closure. This account explains how internally coherent structure can emerge, persist, and transtorm over time, thereby enabling conceptual integration rather than mere incremental adjustment. It helps explain phenomena that are otherwise difficult to account for, including cross-domain transfer, conceptual change, and the often-therapeutic nature of creative activity.
I present both computational and empirical evidence consistent with this view from studies of musicians, artists, creative writers, dancers, and comedians, and show how it offers a new way of understanding convergent and divergent thought.
Finally, I outline an early application of the autocatalytic network framework to artificial intelligence, illustrating how learning systems that preserve internal coherence can support forms ot interence and adaptation unavailable to prediction error minimization alone.