Sunday, December 21, 2008

Nature versus naturoid

[This is my Materials Witness column for the January 2009 issue of Nature Materials.]

Are there metameric devices in the same way that there are metameric colours? The latter are colours that look identical to the eye but have different spectra. Might we make devices that, while made up of different components, perform identically?

Of course we can, you might say. A vacuum tube performs the same function as a semiconductor diode. Clocks can be driven by springs or batteries. But the answer may depend on how much similarity you want. Semiconductor diodes will survive a fall on a hard floor. Battery-operated clocks don’t need winding. And what about something considerably more ambitious, such as an artificial heart?

These thoughts are prompted by a recent article by sociologist Massimo Negrotti of the University of Urbino in Italy (Design Issues 24(4), 26-36; 2008). Negrotti has for several years pondered the question of what, in science and engineering, is commonly called biomimesis, trying to develop a general framework for what this entails and what its limitations might be. His vision is informed less by the usual engineering concern, evident in materials science, to learn from nature and imitate its clever solutions to design problems; rather, Negrotti wants to develop something akin to a philosophy of the artificial, analogous to (but different from) that expounded by Herbert Simon in his 1969 book The Sciences of the Artificial.

To this end, Negrotti has coined the term ‘naturoid’ to describe “all devices that are designed with natural objects in mind, by means of materials and building procedures that differ from those that nature adopts.” A naturoid could by a robot, but also a synthetic-polymer-based enzyme, an artificial-intelligence program, even a simulant of a natural odour. This concept was explored in Negrotti’s 2002 book Naturoids: On the Nature of the Artificial (World Scientific, New Jersey).

Can one say anything useful about a category so broad? That might remain a matter of taste. But Negrotti’s systematic analysis of the issues has the virtue of stripping away some of the illusions and myths that attach to attempts to ‘copy nature’.

It won’t surprise anyone that these attempts will always fall short of perfect mimicry; indeed that is often explicitly not intended. Biomimetic materials are generally imitating just one function of a biological material or structure, such as adhesion or toughness. Negrotti calls this the ‘essential performance’, which itself implies also a selected ‘observation level’ – we might make the comparison solely at the level of bulk mechanical behaviour, irrespective of, say, microstructure or chemical composition.

This inevitably means that the mimicry breaks down at some other observation level, just as colour metamerism can fail depending on the observing conditions (daylight or artificial illumination, say, or different viewing angles).

This reasoning leads Negrotti to conclude that there is no reason to suppose the capacities of naturoids can ever converge on those of the natural models. In particular, the idea that robots and computers will become ever more humanoid in features and function, forecast by some prophets of AI, has no scientific foundation.

No comments: