Adaptive Dynamic Modularity in a Connectionist Model of Context-Dependent Idea Generation

Cognitive control -the ability to produce appropriate behavior in complex situations -is a fundamental aspect of intelligence. It is increasingly evident that this control arises from the interaction of dynamics in several brain regions, and depends significantly on processes of modulation and dynamical biasing. While most research has focused on explanations of behavioral responses seen in experiments and pathologies, it is reasonable to expect that internal functions such as planning and thinking would also use similar control mechanisms. In this paper, we present a connectionist model for an idea generation process that can rapidly retrieve old ideas in familiar contexts and search for novel ideas in unfamiliar ones. Based on a simple reinforcement signal, the system learns context-dependent biases that represent effective internal "response systems" for generating ideas from conceptual elements. A broad goal of the research is to show that preconfigured structural modularity, limited real-time selectivity, and adaptive modulation can interact to produce the flexible functionality necessary for cognition and intelligent behavior.

I. Introduction

Modularity has been recognized as a fundamental property of biological systems [1], [2], [3], and especially of biological networks [4]. In these systems, modularity facilitates the emergence of complexity in both structures and processes by providing combinatorial diversity, efficiency and robustness. Modularity is also an obvious feature of the brain, where the processes of perception, cognition and control arise from the interaction of distinct regions with specific functions. The neocortex, in particular, has a distinctly modular organization, with neurons arranged in horizontal layers and vertical columns that seem to be the primary computational elements of the cortex [5], [6], [7], [8], [9], [10], [11], [12]. However, rigid modularity also has drawbacks. It is not compatible with the flexibility required for cognitive function, where the system must continuously respond to a dynamic environmental context in producing its behavior. The main focus of this paper is to explore how flexible functionality can arise within neural architectures with fixed structural modularity in conjunction with real-time selectivity and modulation. A simple idea-generation scenario is used as the motivation for this study.