Zenith Grant Awardee

Susanne Still

University of Hawaii at Manoa

Co-Investigators

Gavin Crooks, Lawrence Berkeley National Laboratory

Project Title

Foundations of information processing in living systems

Project Summary

Despite the huge amount of new data on biological systems, the question of \"What is Life\" remains outstanding. In particular, are there organizing principles whose action clearly separates living from nonliving systems? One candidate is energy efficiency. A more subtle possible principle is informational in nature, that of predictive inference. If a life form can predict some aspect of its environment, then it can respond more appropriately. We have recently pointed out that these two concepts are fundamentally related physically, by exposing the direct tie between energetic inefficiency and inefficient use of system memory: retaining memory that is not useful for prediction. We will develop these ideas further, paying particular attention to processing speed. Intuitively, there should be a trade-off between energy efficiency, and how quickly an organism has to act. Predictive inference is further affected by an organism\'s ability to change its own environment. Finally, we will explore if, on a microscopic level, quantum effects could allow for more efficient information processing. Altogether, the proposed research might lead to a sharper understanding of what it means to be alive, by providing an operational definition, based on information and the processing thereof.

Technical Abstract

Nature is shaped by processing and transforming information on all levels of scale and complexity. Living systems show particularly striking information processing abilities: they adapt to their environment and learn from experiences, making inferences about the future. Are there general principles that characterize living systems? We will explore the idea that predictive inference may be a defining property of living systems, and ask for physical implications thereof. We recently tied predictive inference to thermodynamic efficiency, by exposing the relationship between dissipation and instantaneous nonpredictive information. In our approach, system dynamics contain an implicit model of the fluctuating environment. The efficiency of this model is measured in terms of predictive inference: a good model has large predictive power at fixed memory, i.e. small nonpredictive information. Efficiency arguments often have the flaw that efficiency is maximized when a system changes arbitrarily slowly. Living systems do not have this option. What are the fundamental energetic and information processing limits on systems evolving at finite rate and temperature? Are quantum effects relevant, and do they provide unique advantages? How does a system\'s ability to change its environment affect energetic and informational constraints?