Issues in applying bio-inspiration, cognitive critical mass and developmental-inspired principles to advanced intelligent systems

Abstract

This Chapter summarizes ideas presented at the special PerMIS 2008 session on Biological Inspiration for Intelligent Systems. Bio-inspired principles of development and evolution are a special part of the bio-models and principles that can be used to improve intelligent systems and related artifacts. Such principles are not always explicit. They represent an alternative to incremental engineering expansion using new technology to replicate human intelligent capabilities. They are more evident in efforts to replicate and produce a "critical mass" of higher cognitive functions of the human mind or their emergence through cognitive developmental robotics (DR) and self-regulated learning (SRL). DR approaches takes inspiration from natural processes, so that intelligently engineered systems may create solutions to problems in ways similar to what we hypothesize is occurring with biologics in their natural environment. This Chapter discusses how an SRL-based approach to bootstrap a "critical mass" can be assessed by a set of cognitive tests. It also uses a three-level bio-inspired framework to illustrate methodological issues in DR research. The approach stresses the importance of using bio-realistic developmental principles to guide and constrain research. Of particular importance is keeping models and implementation separate to avoid the possible of falling into a Ptolemaic paradigm that may lead to endless tweaking of models. Several of Lungarella's design principles [36] for developmental robotics are discussed as constraints on intelligence as it emerges from an ecologically balanced, three-way interaction between an agents' control systems, physical embodiment, and the external environment. The direction proposed herein is to explore such principles to avoid slavish following of superficial bio-inspiration. Rather we should proceed with a mature and informed developmental approach using developmental principles based on our incremental understanding of how intelligence develops. © 2009 Springer-Verlag US.