Artificial Chemistry

Suzuki, Hideaki, and Peter Dittrich. “Artificial chemistry.” (2009): 1-3.

During the long history of science, theory and experimentation have often worked together like the two wheels of a bicycle. In the last century, the most representative examples can be found in physics. Today, this is also true of modern biology, as in the case of brain science and artificial neural networks or of biochemistry and systems biology. Theoreticians in neuroscience construct artificial neural networks with a computer, and through simulations, they try to understand the mechanisms of the brain, whose elementary units— neurons— and their switching properties are already known by modern neurophysiology but whose entire functions, such as thought processes and consciousness, have not yet been clarified. Models of artificial neural networks are useful not only for understanding the design principles of the brain, but also for constructing powerful engineering/computational systems that have desirable features in common with living things. The same has happened to the recent relationship between systems biology and biochemistry— two modern biosciences that study cells at a molecular level. Systems biology’s researchers construct fine models of molecular activities and try to simulate the biological cell system as a whole. Unlike artificial neural network research, however, systems biology research is currently contributing only to biology. A biological cell is such a huge system that the computational cost to simulate the entire cell on a molecular level is enormous. When we try to examine the origin and evolution of biological cell systems or when we want to engineer a bio-inspired computational system imitating molecular activities, we have to design more abstract models than those currently used by systems biology. The theme of this special issue, artificial chemistry (AChem), is a research field that complements systems biology from this point of view. In a typical AChem study, we construct a model of a biological system at a molecular level, but we minimize the computational cost by dramatically simplifying the model in order to make self-organizing phenomena or functional emergence happen on a computer. The biomolecular system is one of the few systems whose elementary processes (molecular reactions) are well known and that exhibit such phenomena as self-organization or selfassembly. AChem facilitates one of the most promising approaches toward the design of computational systems with emergent characteristics. Although the term ‘‘artificial chemistry’’ itself was coined by Rasmussen [4], Fontana [1], and others around 1990, the methodology of AChem (i.e., abstracting biomolecular processes computationally) has some other roots. Approaches like Turing patterns [6], typogenetics [2], and Laing’s molecular machines [3] should capture fundamental properties of (bio)chemistry without trying to model specific chemical processes in detail. In Tierra [5]—as another, more recent example—two self-replicating programs can interact with one other by complementary nop matching, which we can now regard as a typical example of an invention for AChem.

Cited by 10
Related articles