At first glance, organic brain chemistry seems as if it should be more flexible, faster and cheaper than the inorganic equivalent. A silicon-based electronic assembly with as many circuits as the brain would melt instantly if it was not surrounded by expensive cooling apparatus - and it would be many times bigger than the brain with current technology.

Of course it's fair to reflect that nature has had a few billion years to develop the brain, while scientists have been playing with silicon only for 50 years. Still, nature's choice of chemistry needs to be respected.

The question at issue is: how should the brains of human-scale robots be constructed? It may be that the problem of size will be bypassed with wireless or magnetic connectivity: if mental processing, memory, consciousness and external communication take place on a much more developed form of today's Internet, then the actual internal brain of a robot may need to be no more than a motor and sensory apparatus. But there will always be uses for autonomous robots (in warfare, and inter-planetary or inter-stellar mining, for instance).

Organic robot brains could be grown using the techniques kindly developed for us by evolution, once we have mastered the chemistry and its instruction set.

Research reported by the Weizmann Institute of Science is poniting towards the development of electronic components made of single layers of organic (carbon-based) molecules. Such components might be inexpensive, biodegradable, versatile and easy to manipulate, say the researchers.

Indeed, but maybe it would be simpler just to adopt animal brain chemistry lock, stock and barrel, and adapt it to the purpose of robotic control? People working on organic molecular switches still perhaps have a 'silicon' mind-set. Maybe they are pursuing a blind alley?