Humanoid intelligence requires humanoid interactions with the world, according to Professor Rodney Brooks, director of the MIT Artificial Intelligence Lab. By building a robot with a human-like form that can experience the world in the same way that humans do, Brooks and his colleagues believe that eventually someone will be able to build a robot as capable as a human being.

Meet Cog, a set of sensors and actuators that try to approximate the sensory and motor dynamics of a human body. Since the robot has humanoid form, it is both easy and natural for people to interact with it in a human-like way. And interaction with people is just what Cog needs to learn. Cog is not preprogrammed to perform tasks, but rather learns through interaction. Cog's "brain" is actually many sets of computers spread out around the robot's body and down its back. Each set of computers represents a different part of Cog's "nervous system" and confers perceptual abilities such as vision or basic motor skills such as moving its limbs.

The idea of this architecture is to allow simple interactions that will lead to simple behaviors that, in turn, will build on each other to make more complicated behaviors easier.