Given all that, I reproach myself for missing a lovely paper in the Proceedings of the Royal Society 2000, by a group of Oxford colleagues, which is a perfect illustration of the principle.

The protozoan parasite Toxoplasma gondii, which lives inside the cells of its host, has cats as its definitive host, and rats as intermediate host. It leaves the cat in the faeces, from where it may then be ingested by a rat. It travels from rat to cat when an infected rat is eaten by a cat.

Natural selection would therefore favour Toxoplasma genes that change the behaviour of the rats they inhabit, making them more likely to be eaten by cats. This could be achieved as a "boring byproduct" of simply making the rats sick, so that they are less adept or speedy in running away from cats. However, Berdoy, Webster & Macdonald found something much more interesting. They showed experimentally that

"although rats have evolved anti-predator avoidance of areas with signs of cat presence, T. gondii’s manipulation appears to alter the rat’s perception of cat predation risk, in some cases turning their innate aversion into an imprudent attraction."

It's a beautiful example, which I wish had been available when I wrote The Extended Phenotype. The altered rat behaviour is an adaptation for the benefit of T.gondii genes, specifically those T.gondii genes that express themselves in the rat brains that they inhabit.

I repeat, it is especially pleasing that this effect is achieved, not in a boring way simply by making the rats sick and therefore more sluggish in escaping from cats. Indeed, they seem to have no obvious effect on the general health of rats. Their effect is a specific manipulation of rat behaviour vis-a-vis cats. It's as though they are pulling puppet strings in the rat's brain. A neurophysiologist would not be surprised to discover a way of doing exactly that, either with micro-electrodes in rat brain cells or with drugs. Or geneticists could do it by genetic manipulation of rat genes. It seems that natural selection, working on protozoan genes, has achieved exactly the same thing.