Remote control of behavior
I found out about this paper last week - it didn't seem to be picked up by a lot of the popular press despite its wackiness. Susana Lima and Gero Miesenbock at Yale have designed a fruit fly that can be activated to perform certain behaviors by remote control, using a laser. You put a non-endogenous ATP receptor into defined neurons that control a certain behavior, inject the flies with caged ATP, uncage the ATP with a laser, and - voila - those neurons are activated and the fly does the correct behavior. They got flies - even headless flies or blind flies - to perform stereotypical escape behavior (jumping and wing-flapping) when "activated" with a UV laser.
Abstract with link to full article here, layman's write-up here.