Thermodynamically Evolving Systems

The spontaneous evolutionary ordering of the natural world suggests a general physical selection principle that accounts for the production of ordered states (macrostates) from disordered states (microstates) in natural systems across a wide range of scales. Cells, ecosystems and cognitive states of the brain (seat of intelligence) are a few examples of coherent macroscopic states of matter that progressively evolve some subset of accessible microstates from a larger set of initially accessible microstates. We believe the phenomena associated with intelligence and evolution can be understood as natural consequences of complex open thermodynamic systems.

The second law of thermodynamics suggests that for closed systems, entropy will increase until there is no free energy and therefore, no structure. Unlike closed systems, an open system exchanges energy with its environment. As energy flows into an open system, entropy increases (causing more disorder at microstates) and continues to increase until it reaches a critical threshold—consistent with reversible thermodynamics—where the system must either dissolve under the stress of entropic fluctuations or self-organize to form a new structure that dissipates entropy.

Open systems tend to self-organize macroscopic structure for the purposes of dissipating entropy into the environment. Thus, entropy and self-organization are intertwined, where entropy provokes self-organization, while self-organization enables the system to offload entropy.

We believe that naturally intelligent systems are openly evolving dynamic systems that solve problems by forming energy flow paths within their environments, resulting in a self-organizing emergent structure. We would like to establish a sound theoretical basis for engineering this “physical intelligence” inherent in natural systems.

Sections

Program and Sponsor

Project: Physical Intelligence (PI)
Sponsor: Defense Advanced Research Projects Agency (DARPA)
Partners: University of California Davis, University of Connecticut, University of Illinois Urbana-Champaign, Scripps Research Institute, Palo Alto Research Center, Wake Forest University