Biomimetic Design in Augmented Cognition

Abstract

Biomimetic (“life mimicking”) systems automate functionality by replicating biological forms and processes (e.g., early airplane designs were trying to model winged flight in birds). In this way, organic systems not only supply proof of concept (“heavier-than-air flight is possible”), but also inform attempts to implement functionality by reproducing form in automated systems. Important sub-disciplines within the science of machine learning have developed in this same way. The most widely used cognitive architectures closely resemble their biological starting points. Obvious examples are neural networks, expert systems, and genetic algorithms. The application of sophisticated optimization techniques to these systems can supersede unscalable aspects of the underlying biological metaphor (“airplanes do not flap their wings”). However, biomimetic principles inform possible approaches to solving many problems. This paper describes the application of biomimetic design to augmented cognition and analyzes machine learning developments from a biomimetic lens. An experiment follows comparing Neural Networks and Bias-Based Learning and assesses their performance against human subjects. The last part synthesizes the results of the experiment with the principles underlying biomimetic design to foster effective problem-solving.