Amoeba-based nonequilibrium neurocomputer utilizing fluctuations and instability

Abstract

We employ a photosensitive amoeboid cell known as a model organism for studying cellular information processing, and construct an experimental system for exploring the amoeba's processing ability of information on environmental light stimuli. The system enables to examine the amoeba's solvability of various problems imposed by an optical feedback, as the feedback is implemented with a neural network algorithm. We discovered that the amoeba solves the problems by positively exploiting fluctuations and instability of its components. Thus, our system works as a neurocomputer having flexible properties. The elucidation of the amoeba's dynamics may lead to the development of unconventional computing devices based on nonequilibrium media to utilize fluctuations and instability. © Springer-Verlag Berlin Heidelberg 2007.