According to the New Scientist article, the parasitoid wasp Glyptapanteles lays its eggs, about 80 at a time, in young geometrid caterpillars. The eggs hatch and the larvae feed on the caterpillar's body fluids. When they are fully developed, they eat through the caterpillar's skin, attach themselves to a nearby branch or leaf and wrap themselves up in a cocoon.
At this point, something remarkable and slightly eerie happens: the caterpillar, still alive, behaves as though controlled by the cocooned larvae. Instead of going about its usual daily business, it stands arched over the cocoons without moving away or feeding. The caterpillar stays alive until the adult wasps hatch.
Arne Janssen of the University of Amsterdam and Brazilian colleagues noticed that when they moved a paintbrush towards parasitised caterpillars, the insects would thrash about, apparently in an attempt to protect the cocoons. To test the manipulation hypothesis, Janssen's team allowed wasps to infect caterpillars in a laboratory setting. Once the larvae emerged and formed their cocoons, the researchers separated half the cocoons and the caterpillars. The separated cocoons were attached to a leaf next to an unparasitised caterpillar, which was prevented from moving away by a ring of insect glue around the stem.
When they added a stinkbug, a voracious predator of wasp cocoons, the team found that 17 of the 19 parasitised caterpillars thrashed their heads around in the direction of the bug. More than half the time, this knocked the bug off the branch or made it retreat. Unparasitised caterpillars barely noticed the bug, even when it climbed on top of them.
To see if the behaviour affected the survival of wasp cocoons in the wild, the researchers placed over 400 parasitised caterpillars in guava fruit trees one day before the larvae were due to break through their skin. Once the larvae had cocooned themselves on the nearby branches, the researchers removed half of their bodyguard caterpillars and watched what happened. The survival rate of "guarded" cocoons was twice as high as that of unguarded cocoons.
Although Janssen and his colleagues do not know how the parasites make the caterpillars change their behaviour, they think that a few larvae in each brood may sacrifice themselves to help their brothers and sisters. "If we dissect the caterpillars, we find one or two parasitoid larvae have stayed behind, even after the rest of the brood has emerged and formed cocoons," says Janssen. It could be that the larvae that remain in the host control its behaviour in order to make it protect the rest of the brood.
Well it may not be the basis of a good horror film but still quite fascinating.