Anticipatory Systems

Robert Rosen (1934 to 1998) was an American theoretical biologist and professor of biophysics at Dalhousie University in Halifax, Nova Scotia.⁷ Anticipatory Systems: Philosophical, Mathematical and Methodological Foundations was published by Pergamon in Oxford in 1985, as the first volume of the IFSR International Series on Systems Science and Engineering.⁴ Rosen writes that its first ideas came in 1972, when he was a visiting fellow at the Center for the Study of Democratic Institutions in Santa Barbara, California.³

The book defines an anticipatory system as "a system containing a predictive model of itself and/or of its environment, which allows it to change state at an instant in accord with the model's predictions pertaining to a later instant".² In plain words, the system carries a working model of itself or its surroundings, the model runs ahead of events, and the system acts now on what the model says will happen later.²⁶ In a 1997 interview Rosen set this against cybernetic control, which corrects a deviation from a desired behaviour once it appears; anticipation instead uses "a predictive model of the system you're trying to control" to generate the control in advance.⁶ He added that something in such a system has to run "faster than real time", or it is not anticipating anything.⁶

Before returning to that definition, the book works through what a model is, in chapters on natural and formal systems, the modelling relation between them, the encodings of time, and open systems.⁹ Rosen calls the modelling relation "by definition the heart of an anticipatory system".² As his daughter Judith Rosen explains it, a natural system is encoded into a formal one, the formal system's inferences stand in for the natural system's causes, and its predictions are decoded and checked against what the natural system does.⁸ A model that was accurate when built drifts out of step as the system changes, which Rosen called "temporal spanning", and only that check against the real system shows when it has happened.⁸ The book also distinguishes a true modelling relation from a simulation that is not based on one.⁷

Springer published a second edition in 2012, revised and edited, with an introduction by Judith Rosen, prolegomena by Mihai Nadin surveying the field since 1985, an added chapter by John Kineman, and autobiographical reminiscences by Rosen.¹⁵ Its publisher credits the book with founding the area of anticipatory systems theory.¹ Rosen recalled that its ideas "raised hackles" among scientists who saw science as dealing only with simple systems.⁶

Rosen's definition bears directly on what a digital twin of a place becomes when its forecasts are used to act: a model of the place and its surroundings, running ahead of events, whose predictions change what is done now.² He made the link to public questions himself.⁶ In 1997 he said that problems such as whether the planet is warming "are only amenable to anticipatory control".⁶ What people fought about, he said, "were models", and what the models told them about the future and what they ought to do about it.⁶ The chapters on the modelling relation, and the warning about temporal spanning, speak to anyone who has to say whether a twin still matches its place.⁸⁹