Computers guzzle energy. A big chunk of the money spent on computers pays for power, and an estimated four percent of the total energy consumed in the United States is used to keep our computers computing. Physicists have long been interested in understanding the physical laws that describe that tradeoff: What’s the thermodynamic cost of processing information?
This isn’t just a question for physicists, though. It resonates with living systems, too, which have evolved to perform specific jobs with access to only limited resources. “Our brain requires quite a bit of food whether we think anything useful or not,” says computer scientist and SFI External Professor Peter Stadler. Similarly, cells carry out biochemical reactions that might be regarded as information processing, akin to computing.
Artificial, human-built digital computers are always out of equilibrium: They need a steady source of power to keep running. So it goes with cells. Without energy, they stop functioning and die – so equilibrium, for a living thing, is death. In the last decades, breakthroughs in the rigorous study of systems far from equilibrium — a field called nonequilibrium statistical physics – have led to the development of new tools. Now, researchers from a range of disciplines can better analyze such systems, whether they’re used on our smartphones or keeping organisms alive.
Notably, recent studies suggest that cells carry out some biochemical “calculations” at a level of efficiency orders better than modern, artificial computers. That comparison raises a number of provocative questions. Do efficient biological systems look like any existing ideas in computer science theory? How did they evolve such efficient ways of computing, and what can we learn from them?
“Thinking of biological systems as computing or information-processing entities immediately begs the question, What do they actually compute? It’s an open question,” says Stadler.
In an effort to start probing those questions, Stadler and SFI Professor David Wolpert have organized a working group, “Thermodynamic and Computational Efficiency in Cellular Chemical Reaction Networks,” happening at SFI April 23-24. They’ve invited researchers from a range of disciplines — chemistry, physics, molecular biology, mathematics, computer science — who are interested in investigating the connection between energy and information processing, or computing.
The meeting may help researchers gain insights into whether or not living cells should be seen as inspirational for future low-power devices. Are cells more like vending machines with simple inputs and outputs and limited functionality? Or do their functions speak to a broad range of calculations?
“One potential outcome,” says Stadler, “might be that biological systems are a very bad model for building computers.”