A dynamical connectionist model of idea generation

Abstract

Abstract:In this paper, we present a model for the generation of ideas within a creative thinking/brainstorming context. In the model, ideas emerge as conceptual combinations from the interaction of complex dynamics at several semantic levels: Features, concepts, categories, and previously generated ideas. This dynamics is shaped by external information on task context, constraints and goals, and is modulated by evaluative feedback from an internal critic working through reinforcement. While the model is abstract, it attempts to capture the interplay between semantic representations in the temporal, frontal and parietal cortices, working memory in the prefrontal cortex, attentional selection by the basal ganglia, and modulation from the dopaminergic reward system. We show that a context-specific itinerant search for novel but meaningful conceptual combinations (ideas) emerges naturally from the dynamics of this system. We also briefly describe a computational model for ideation in groups using a multiagent formalism. The initial focus of this model is on studying the potential benefits of cognitive diversity in agent groups, e.g., the presence of convergent and divergent thinkers, or agents with different semantic organizations.

Effects of relevant and irrelevant primes on idea generation: A computational model

“…The features allow all concepts to be embedded in a single consistent semantic space, while the categorical clustering provides a coarsegrained repertoire of basis functions from which useful context-specific semantic spaces can be constructed [29].…”

Neurocognitive spotlights: Configuring domains for ideation

“…Models of ideation based on this have been proposed by Brown and Paulus [8] and Nijstad and Stroebe [5]. Recently, we have developed a larger model in which associative search plays a crucial role [13], [14], [17], [16], [18].…”

“…This, of course, projects a variable pattern of disinhibition on the CN, ranging from a very narrow search space at high inhibition to a much broader one when the inhibition is low. Because of this flexible functionality, the attractors implemented in the DSN are termed graded attractors [17]. A key requirement for graded attractors is that the order in which the disinhibition targets new categories in the CN should correspond to the pattern of participation for categories within the given context.…”

“…These experiments have produced a wealth of observations and several theories about the basis of creativity [6], [7], [8], [3], [5]. However, as the neural mechanisms underlying the more response-oriented functions have become better understood, it has become possible to infer similar mechanisms for internal processes such as ideation, and a number of computational models have been proposed [9], [10], [11], [12], including one by our research group [13], [14], [15], [16], [17], [18]. These models have been able to explain, albeit qualitatively, the phenomena observed during behavioral experiments in individual and group ideation (brainstorming), and have begun to suggest possible neural mechanisms for the ideation process.…”

Learning Complex Population-Coded Sequences

“…In particular, our system can learn highly similar sequences by sequence-specific selection of neural groups within the network. This is an example of switched modularity, which we have proposed as a general mechanism for information processing in complex biological networks [39][40][41][42] 3 Model Description A complex drawing, e.g., a letter or shape, is modeled as a sequence of elementary actions starting from a canonical initial point. Each elementary action results in the drawing of a short segment in a specific direction on a 2-dimensional workspace.…”