Title:Plants as a Window on Decentralized Computation and Mechanical Intelligence
Abstract: Plants continuously solve complex navigational problems, negotiating heterogeneous and changing environments while lacking a brain or nervous system. How can a completely decentralized organism sense its surroundings, process information, and generate coherent behavior? In this talk, I present plants as a model system for decentralized computation, where information processing emerges from the physics of growth rather than centralized control. I first show how plants integrate sensory information. We find that plants represent directional light as vectors and respond to a nonlinear vectorial sum of multiple light sources, giving rise to systematic perceptual distortions analogous to optical illusions. I then turn to the role of mechanics in plant behavior. I present recent work demonstrating that climbing plants actively assess the mechanical stability of newly encountered supports. Circumnutations generate controlled loading on candidate supports, allowing plants to evaluate their stiffness before committing to twining, in a manner reminiscent of whisking in rodents. Together, these examples illustrate how sensing, computation, and behavior can emerge from the coupled dynamics of growth, transport, and mechanics in decentralized living systems.